The Oxford Handbook of Structural Transformation 0198793847, 9780198793847

Citation preview

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

   

STRUCTURAL TRANSFORMATION

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

   

.........................................................................................................................................

STRUCTURAL TRANSFORMATION ......................................................................................................................................... Edited by

CÉLESTIN MONGA and

JUSTIN YIFU LIN

1

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

3

Great Clarendon Street, Oxford, OX2 6DP, United Kingdom Oxford University Press is a department of the University of Oxford. It furthers the University’s objective of excellence in research, scholarship, and education by publishing worldwide. Oxford is a registered trade mark of Oxford University Press in the UK and in certain other countries © Oxford University Press 2019 The moral rights of the authors have been asserted First Edition published in 2019 Impression: 1 All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted, in any form or by any means, without the prior permission in writing of Oxford University Press, or as expressly permitted by law, by licence or under terms agreed with the appropriate reprographics rights organization. Enquiries concerning reproduction outside the scope of the above should be sent to the Rights Department, Oxford University Press, at the address above You must not circulate this work in any other form and you must impose this same condition on any acquirer Published in the United States of America by Oxford University Press 198 Madison Avenue, New York, NY 10016, United States of America British Library Cataloguing in Publication Data Data available Library of Congress Control Number: 2018949483 ISBN 978–0–19–879384–7 Printed and bound by CPI Group (UK) Ltd, Croydon, CR0 4YY Links to third party websites are provided by Oxford in good faith and for information only. Oxford disclaims any responsibility for the materials contained in any third party website referenced in this work.

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

C .............................................

List of Figures List of Tables List of Contributors

Introduction: Structural Transformation—Overcoming the Curse of Destiny

ix xiii xvii

1

C´  M  J Y L

PART I THEORIES AND FRAMEWORKS OF STRUCTURAL CHANGE 1. Structural Transformation, Deep Downturns, and Government Policy

35

J E. S

2. Structural Transformation and Growth: Theoretical Considerations

45

R R

3. Remodelling Structural Change

70

J Y L  Y W

4. Structural Transformation and Income Distribution: Kuznets and Beyond

96

R K

5. The Flying-Geese Theory: Reassessed and Reformulated in New Structuralist Perspective

109

T O

6. Changing Income Inequality During Structural Transformation: The Role of Agricultural Prices

127

C. P T

7. Structural Transformation: A Competitiveness-Based View C K

151

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

vi



PART II DRIVERS, CHANNELS, AND POLICY INSTRUMENTS 8. Trade and Structural Change over Two Centuries

175

G F  A T-J

9. Financial Reforms, Financial Development, and Structural Change

191

E K  L A. P  S

10. Location Fundamentals, Agglomeration Economies, and the Geography of Multinational Firms

218

L A  M X C

11. Sustainable Structural Change in the Context of Global Value Chains

253

P L

12. Participation in Global Value Chains: Challenges and Opportunities

273

X L  G G

13. Building Effective Clusters and Industrial Parks

291

X Z

14. Infrastructure Finance: Mobilizing Long-Term Liability Embedded Funds from International Institutional Investors to Emerging Markets

309

C M-P  K L

PART III EMPIRICS OF STRUCTURAL CHANGE 15. Measuring Structural Change

349

C´ M  S S

16. Transforming Traditional Agriculture Redux

363

J M. A  P G. P

17. Structural Transformation and Manufacturing Employment N H

381

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi



18. Global Megatrends and the Macroeconomics of Gender

vii

418

K K, S J-C,  M N

PART IV COUNTRY AND REGIONAL EXPERIENCES 19. Latin America’s Structural Transformation Patterns

439

J´ A O

20. India’s Path to Structural Transformation: An Exception and the Rule

465

D N

21. Structural Transformation in Egypt, 1965–2015

489

K I

22. Growth and Structural Transformation in Viet Nam: The Real Story Beneath

513

D T T H, F T,  D  S

23. Economic Reform and Structural Change: The Chinese Experience

531

B N

24. Financing Industrial Development in Korea and Implications for Africa

549

K L

25. How Taiwan Managed to Grow: Structural Transformation and Industrial Policy

571

W- C

26. Ethiopia: Lessons from an Experiment

591

A O

27. Economic Transformation in Africa from the Bottom Up: New Evidence from Tanzania

619

X D, J K,  M MM

28. Growth and Structural Transformation in the WAEMU Countries T´´ N’G

632

OUP CORRECTED PROOF – FINAL, 2/1/2019, SPi



viii

PART V CONCLUDING THOUGHTS 29. Truth is the Safest Lie: A Reassessment of Development Economics

659

C´ M

30. The Strength of American Federal Democracy: Lessons for Global Development

672

R B. M

31. Desirable Directions of Structural Transformation

684

E P

Index

693

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

L  F .......................................................................

3.1

How industrial structures are determined by endowment structures

3.2

How different industries evolve over time

5.1

The ‘double-helix’ development ladder à la Schumpeter: inter- and intra-industrial upgrading and diversification

6.1

Long-run patterns of structural transformation in Japan and Indonesia

6.2

Agricultural terms of trade for Asia and non-Asia separately (2000 = 100)

6.3

Agricultural terms of trade, global average, 1980–2010

7.1

What determines competitiveness?

7.2

Stages of development

7.3

Traded and local industries: mapping clusters

7.4

How does economic structure change? Hypotheses on the nature of development pathways

7.5

Comparison of development approaches

7.6

What drives prosperity? Relationship between competitiveness and structure

78 86 116 134 140 141 154 156 161 163 165

7.7

Cluster-based structural transformation

8.1

The growth of world trade, 1800–2014 (log scale)

8.2

Distribution of world exports at current prices, 1830–2010

8.3

World openness, 1830–2014

8.4

Openness, ‘1870 sample’, 1870–2014

8.5

Openness tradables, 1830–2014

9.1

Financial reforms and financial status quo, 1970–91

9.2

Cumulative distribution functions

9.3

GDP growth and employment growth

166 167 177 179 181 183 185 198 199 200

9.4

Productivity growth: advanced economies and emerging market economies

201

9.5 9.6

Productivity growth decomposition: advanced economies and emerging market economies Log of private credit to GDP

204 212

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

x

  

11.1

Export concentration trend by income, 1995–2014

11.2

Economic complexity and GDP per capita (PPP 2010 US$), 2014

11.3

Ordinary versus processing trade

14.1

Major economy 10-year bond yields, December 2009–July 2016

14.2

Global equity market indices

14.3

Current account as a percentage of world GDP, 1997–2015

14.4

Capital ratios: Basel II versus Basel III

14.5

Insurer and bank balance sheets in comparison

14.6

Credit losses for single A- and Baa-rated infrastructure debt and NFC

257 263 266 313 314 315 324 327

14.7

Correlations between select asset classes with unlisted infrastructure

334 334

14.8

Sharpe ratio comparison, 5-year results for the quarter to December 2014

334

14.9

Performance comparison, results for the quarter to December 2014 (annualized return)

14.10

Structure of Uruguay convertible bond with project funding

14.11

Process after project construction and ramp-up

14.12

Proposed structure of existing infrastructure ABS

14.13

Proposed structure of infrastructure CLO

14.14

Proposed structure of puttable project bond

14.15

Proposed structure of MIGA non-honouring of sovereign financial obligation bond

335 337 338 340 341 342

15.1

The four steps to constructing a composite index

15.2

Economic Complexity Index and GNI per capita

16.1

Agricultural GDP as share of GDP (by region), 1960–2014

16.2

Agriculture’s share of GDP versus GDP per capita

343 351 358 366 367

16.3

The latitudinal geography of rainfed versus irrigated crop production, 2005

369

16.4

Engel effects on the country-specific composition of agricultural output, 2013

16.5

Per capita agricultural production by region, 1961–2013

371 372

17.1

Value added as a function of the labour productivity and employment growth rate—early industries

389

Value added as a function of the labour productivity and employment growth rate—middle industries

390

Value added as a function of the labour productivity and employment growth rate—late industries

392

17.2 17.3

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

   17.4

Employment growth elasticity—early industries

17.5

Employment growth elasticity—middle industries

17.6

Employment level (EP ratio)—late industries

18.1

Dependency ratios, 2015–2100

18.2

Income inequality vs. gender inequality

18.3

Export diversification and gender inequality, 1990–2010

19.1

Manufacturing value added as a share of GDP, 1950–2015

19.2

GDP decennial growth rates (moving average annual growth rate over the decade that ends in the year indicated in the graph)

xi

393 396 398 420 426 427 445

19.3

Natural resource and technological contents of Latin American exports

448 455

20.1

Changes in the composition of output and employment in India, 1950–51 to 2013–14

471

Share of the primary, secondary, and tertiary sectors in GDP, dis-aggregated by sub-sectors in India, 1950–51 to 2013–14

477

20.2 20.3

Share of the primary, secondary, and tertiary sectors in employment, dis-aggregated by sub-sectors in India, 1950–51 to 2011–12

21.1

Structure of GDP, 1965 and 2015

21.2

Investment and savings, 1965–2015

21.3

Structure of employment, 1960–2015

478 493 494 497

22.1

Backward linkage multipliers for 2000 and the change between 2000 and 2012

519

Forward linkage multipliers for 2000 and the change between 2000 and 2012

520

Spill-over effects from activities on household income for 2000 and the change between 2000 and 2012

522

22.2 22.3 22.4

Feedback effects from households on activities for 2000 and the change between 2000 and 2012

23.1

Change in industrial composition, 1980–95

23.2

Composition of GDP: current prices

23.3

Composition of GDP: production side

23.4

Implicit deflators: secondary and tertiary sectors

23.5

Sector intensity of capital and human capital

26.1

Export shares of manufacturing sector by export value

26.2

Exports of floriculture and leather/leather products

523 533 542 543 543 544 597 598

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

xii

  

28.1

Evolution of the demand structure in several emerging countries

28.2

Evolution of the growth of the real GDP of sub-Saharan Africa

28.3

Evolution of the growth of the real GDP of UEMOA

28.4

Evolution of the demand structure in the countries of the WAEMU

28.5

Sectoral composition of the WAEMU countries’ GDP

634 639 640 643 644

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

L  T ...................................................................

3.1

Static equilibrium

8.1

Rates of growth of world trade, 1800–2014

8.2

Decomposition of change in world openness, long-term trends

8.3

Decomposition of change in world openness, 1970–2007

9.1

Financial reform and change in average economic performance

9.2

Change in labour productivity growth after financial reform

79 177 186 187 201 205

9.3

Change in labour productivity growth common component and financial reform

207

Change in labour productivity growth allocation component and financial reform

209

9.4 9.5

Financial reform, private credit, and labour productivity growth components

9A.1

Financial reform and financial status quo across countries

10.1

Descriptive statistics for MNC and domestic agglomeration indices

211 214 238

10.2

Location fundamentals, agglomeration economies, and MNC subsidiary agglomeration I

239

Location fundamentals, agglomeration economies, and MNC subsidiary agglomeration II

240

Location fundamentals, agglomeration economies, and MNC subsidiary worker agglomeration I

241

Location fundamentals, agglomeration economies, and MNC subsidiary worker agglomeration II

242

10.3 10.4 10.5 10.6

Location fundamentals, agglomeration economies, and MNC headquarters agglomeration

10.7

Comparing MNC subsidiaries with domestic plants I

10.8

Comparing MNC subsidiaries with domestic plants II

10.9

The endogeneity of agglomeration economy measures: the agglomeration patterns of MNCs in Europe

10A.1

Descriptive statistics for agglomeration economies

10A.2

Correlation of agglomeration economies

11.1

The world’s most and least complex products, 2014

11.2

The ten most and the ten least complex countries

243 244 245 247 248 249 262 263

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

xiv

  

14.1

Two-track approach (current and additional approach)

14.2

Major countries’ central bank key rates (as of 18 November 2016)

14.3

Major bond markets’ yield and yield curve (as of 4 August 2016)

14.4

Major equity market rates of return

14.5

Top 10 foreign currency reserve assets (as of April 2016)

14.6

Global interest rates (as of 18 November 2016)

14.7

Fixed income instruments’ capital charges and return on equity (ROE)

14.8

High quality liquid assets (HQLA)

14.9

Summary characteristics of relevant financial institutions

14.10

Largest percentage point change of the population aged 60 years or over

14.11

Size of sovereign pension fund

14.12

Size of sovereign wealth fund

14.13

Size of life insurance industry

14.14

Yield spread, compared to domestic 10-year bond rates

16.1

Regional value of agricultural production, 1961 and 2013

16.2

Conventional and unconventional inputs used in agriculture, 1961 and 2014

310 313 314 315 316 316 323 325 328 329 330 330 331 331 368

17A.1

Value added per capita

17A.2

Labour productivity

17A.3

Employment–population ratio

373 385 388 406 407 407

17B.1

Changes in the growth rate of value added per capita, employment–population ratio, and labour productivity

408

17.1

Manufacturing data classification used in this study

17.2

Value added shares in total MVA for large countries

17B.2

Changes in the growth rate of value added per capita, employment–population ratio, and labour productivity

17C.1

Changes in the level of employment (EP ratio)

17C.2

Changes in the level of employment (EP ratio)

17D.1

Changes in labour intensity (employment per value added)

17D.2

Changes in labour intensity (employment per value added)

19.1

Share in industrial value added

19.2

GDP growth rates

19.3

Manufacturing value added as a share of GDP, 1990–2015

410 412 413 414 415 447 451 453

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

  

xv

19.4

Composition of exports of goods and services, 1990 and 2013–14

19.5

Research and development expenditure

457 459

20.1

Structural changes in the composition of output in India: 1950–51 to 2013–14

473

Structural changes in the composition of employment in India: 1950–51 to 2011–12

474

20.2 20.3

Levels of, and growth in, GDP per worker in India, disaggregated by sub-sectors: 1977–78 to 2011–12

22.1

Economy-wide economic ratios, comparing 2000 and 2012

480 515

22.2

Activity real growth and nominal shares in GDP (at factor costs), comparing 2000 and 2012

517

Decomposition of change in value added between 2000 and 2012 due to changes in value added intensity, changes in technology, and changes in final demand

525

22.3

22.4

Decomposition of change in employment between 2000 and 2012 due to changes in employment intensity, changes in technology, and changes in final demand

23.1

Composition of industrial output

23.2

Non-agricultural workforce

24.1

Financing for the Pohang project (US$ million)

526 537 545 558

24.2

Industrial Base Technology Development Projects (IBTDPs), 1987–95

561

Outcomes of the survey to identify the ‘needed’ industrial technologies

562

24.3 24.4

Implementation plan of the technology development projects identified by the demand surveys

25.1

Major economic indicators I, 1951–2014

25.2

Major economic indicators II, 1952–2014

25.3

Distribution of manufacturing value-added, 1971–2014

25.4

Globalization, 1952–2015

25.5

Social indicators, 1952–2014

25.6

Changes in income distribution, 1964–2014

27.1

How high performing firms compare to the rest: individual and business characteristics

27.2

Obstacles to business

28.1

Basic indicators of the economies of the WAEMU, 2015

562 572 576 579 582 586 586 625 629 638

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

xvi

  

28.2

Average annual growth rates of the real GDP

28.3

Evolution of the agricultural workforce

28.4

Indices of export diversification of WAEMU countries

641 645 646

28.5

SWOT analysis of the structural transformation of the economies of the WAEMU

648

OUP CORRECTED PROOF – FINAL, 26/12/2018, SPi

L  C ......................................................................................................

Laura Alfaro, Harvard Business School, USA Julian M. Alston, University of California-Davis, USA Wan-wen Chu, Academia Sinica and National Taiwan University, Taiwan Xinshen Diao, IFPRI and Tuft University, USA Giovanni Federico, University of Pisa, Italy Gary Gereffi, Duke University, USA Nobuya Haraguchi, United Nations Industrial Development Organization, India Khalid Ikram, Harvard University, USA Sonali Jain-Chandra, International Monetary Fund, USA Ravi Kanbur, Cornell University, USA Christian Ketels, BCG, USA Enisse Kharroubi, Bank for International Settlements, Switzerland Kalpana Kochhar, International Monetary Fund, USA Josaphat Kweka, IFPRI and Tuft University, USA Pauline Lectard, University of Montpellier, France Keun Lee, Seoul National University, South Korea Justin Yifu Lin, Peking University, China Kevin Lu, Partners Group, UK Xubei Luo, World Bank, UK Cledan Mandri-Perrott, World Bank, UK Margaret McMillan, IFPRI and Tuft University, USA Célestin Monga, African Development Bank, Ivory Coast Roger B. Myerson, University of Chicago, USA Barry Naughton, University of California San Diego, USA

xviii

  

Deepak Nayyar, Jawaharlal Nehru University, India Monique Newiak, International Monetary Fund, USA Tchétché N’Guessan, University of Côte d’Ivoire, Ivory Coast José Antonio Ocampo, Columbia University and United Nations, USA Arkebe Oqubay, Government of Ethiopia, Ethiopia Terutomo Ozawa, Colorado State University, USA Philip G. Pardey, University of Minnesota, USA Luiz A. Pereira da Silva, Bank for International Settlements, Switzerland Edmund Phelps, Columbia University, USA Richard Rogerson, Princeton University, USA Samuel Standaert, Ghent University, Belgium Joseph E. Stiglitz, Columbia University, USA Finn Tarp, UNU-WIDER and University of Copenhagen, Copenhagen Antonio Tena-Junguito, University Carlos III de Madrid, Spain Dang Thi Thu Hoai, Central Institute of Economic Management (CIEM), Viet Nam C. Peter Timmer, Harvard University and Center for Global Development, USA Dirk van Seventer, UNU-WIDER, Helsinki, Finland Yong Wang, Peking University, China Maggie Xiaoyang Chen, George Washington University, USA Xiaobo Zhang, IFPRI and Peking University, China

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

......................................................................................................................

 Structural Transformation—Overcoming the Curse of Destiny ......................................................................................................................

ˊ      

S words, concepts, and phrases are unlucky throughout their history. They struggle more than others to establish their true meanings, and to claim their fate. ‘Structural transformation’ or ‘structural change’, used interchangeably in this volume, is one of these hapless phrases that have suffered decades of benign neglect, discontent, and outright abuse, before regaining glory and yet falling again into incomprehension and suspicion. Such a trajectory is puzzling, for structural transformation, which refers to the movement of a country’s productive resources (natural resources, land, capital, labour, and know-how) from low-productivity to high-productivity economic activities, is arguably the single most significant concept and social goal in the global quest for prosperity and world peace. As noted by some researchers decades ago, ‘[I]t is impossible to attain high rates of growth of per capita or per worker product without commensurate substantial shifts in the shares of various sectors’ (Kuznets 1979: 130). Structural transformation is the mysterious process through which societies push (or incentivize their productive resources) into higher-performing and dynamicallygrowing sectors, industries, and branches.¹ Once ignited and sustained, this process can generate both static and dynamic economic benefits to human societies. Static gains are usually defined as the rise in economy-wide labour productivity, as workers in a given country or region are increasingly employed in more productive sectors. Dynamic

¹ UNCTAD’s International Standard Industrial Classification (ISIC), Revision 4 uses the word ‘sector’ to refer to agriculture, industry, and services. The economic literature also refers to them, respectively, as the primary, secondary, and tertiary sectors. Sectors are further disaggregated into ‘industries’. For example, the industrial sector includes manufacturing, mining, utilities, etc. Within most of these industries, the disaggregation goes one step further into branches. For example, within manufacturing, one can distinguish branches such as food processing, garments, textiles, chemicals, metals, machinery, and so on.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

gains, which follow over time, accrue from skill upgrading and the positive externalities resulting from workers having access to better technologies and accumulating capabilities (McMillan and Rodrik 2011). As noted by an UNCTAD report, ‘Structural transformation can be particularly beneficial for developing countries because their structural heterogeneity—that is, the combination of significant inter-sectoral productivity gaps in which high-productivity activities are few and isolated from the rest of the economy—slows down their development’ (UNCTAD 2017). Before the First Industrial Revolution, there was little growth in the world economy and the income gap between countries was extremely small. For example, even in 1820, the between-country income differences represented less than 15 per cent of income equality across people in the world, whereas the between-country share rose to well over half of global inequality by 1950, and the richest countries’ per capita income was only just less than four times higher than the poorest (Lin and Rosenblatt 2012). The most recent period of world economic history (roughly since the mid-1990s) has offered a sharp reversal from a pattern of divergence to convergence—particularly for a few developing countries such as China, Vietnam, or Indonesia. The latter phenomenon is being fostered by increasing economic ties among developing countries, and on the intellectual scale, increased learning and knowledge-sharing opportunities among the developing countries. The patterns and dynamics of global economic progress have been transformed in recent years across many dimensions. While economic divergence is still the main reality, the experience of the handful of success stories, and the emergence of the multi-polar growth world justify the rethinking of development economics and policy. This historical record provides a challenge for economists attempting to fully understand the success of the rising economic powers and re-think the traditional views on economic development. Three major questions emerge: (i) Why was there so much divergence during the twentieth century? (ii) Why has the pattern changed recently and can it be sustained? (iii) What is the role of development institutions in facilitating sustained convergence? This introductory chapter briefly chronicles and assesses the idea of structural economic transformation. It starts with a brief discussion of some of the pervasive narratives of despair, which marked human societies everywhere for centuries. It then highlights the shift to new economic possibilities which occurred after the First Industrial Revolution and the Enlightenment period—and were reflected in the debate on the very idea of economic convergence among nations. The chapter then analyses the determinants and mechanics of structural change and gets into the ‘black box’ to examine the role of industrialization, defined more broadly and across the three traditional sectors (agriculture, industry, and services). The topics of deindustrialization, automation, and the future of work are then discussed, as they relate to structural transformation in times of the Fourth Industrial Revolution. Because Africa is the region of the world where structural change through economic diversification into the most productive sectors and industries has been occurring at the slowest pace, the chapter includes a section on meta-economic issues there. A summary of the contents of this volume concludes.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



T: N  D

.................................................................................................................................. Economists, philosophers, artists, and writers have always enjoyed predicting the future. Their conjectures have tended to be rather pessimistic, perhaps because doom and gloom tend to make for better stories. Homer famously suggested God may have afflicted humans with misfortune so that future generations would have tales to sing about. In the period known in the Western world as the Middle Ages, a time of pervasive poverty and violence, many popular works of art and fiction offered dark chronicles of what life was about, and the almost teleological dim fate that lay ahead for humans. Dante’s Divine Comedy, the most famous fourteenth-century poem on sin and redemption, was still mainly about the travails of human life. Its sober vision of mankind’s eternal fate offered no perspective or prescription for generating economic prosperity on earth. Even during the Renaissance, known in Western history as a period of rejuvenation, enthusiasm, and experimentation, branded as a period marked by extraordinary levels of optimism, a lot of powerful literature and works of art were also often chronicles of economic misery, emotional suffering, deep despair, collective self-doubt, and habituation to pain and evil. Leonardo Da Vinci’s views illustrate this paradox: he could paint both the Mona Lisa, an unsettling portrait of a Florentine woman with an enigmatic expression, elegant and aloof, and The Last Supper, his visual interpretation of Jesus Christ’s last night on earth, foretelling his unavoidable betrayal by one of his disciples——the painting specifically depicts the moment after Christ dropped the bombshell that one of his devotees around the table would betray him before sunrise, and all twelve of them react to the stunning news with shock, horror, and anger. Yes, even in times of optimism, influential artistic works tended to recount stories which epitomize the sad fate of men. Throughout the Enlightenment, economists did not stand out as voicing the democratization of economic well-being—what may be termed social positivism. They were not overly preoccupied with generating wealth and prosperity for the whole of society—not just the ruling classes favoured by mercantilist theorists. Fighting poverty, which was prevalent for millennia, and finding innovative ways and policies for improving the standard of living and the quality of life of the bottom deciles were not the focus of their work. Collective well-being, redistribution, and equity as the main public policy goal did not seem to be important topics of research. In that context, Adam Smith’s (1776) vigorous criticism of mercantilism and his belief in the need for and possibility of inquiring into the nature and causes of the wealth of nations, and for the blueprint for a just society that concerns itself with the well-being its least well-off members, sounds messianic, if not naïve. But Smith was not an economist: he was a moral philosopher. True, Smith’s masterpiece was preceded by the inspirational ethics of David Hume and paved the way to reflections by John Stuart Mill and David Ricardo, who seriously

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

considered the need for societies to set human well-being as an important goal. But despite its richness and brilliance, Hume’s work was dominated by his taste for empiricism and his scepticism. Mill’s and Ricardo’s brilliance did not lead them to focus on the common economic good or the plight of the poor. Great economists who carried the torch after them (from F. Y. Edgeworth to David Riccardo, Alfred Marshall or notable members of the Austrian School such as Carl Menger, Eugen Böhm Ritter von Bawerk, Friedrich Hayek, or Ludwig von Mises), developed new intellectual frameworks for increasing production and generating wealth but focused little on the issues of distribution. This seems even more surprising that during their time, the Western world was experiencing major sociopolitical transformations often fueled by mass poverty, exclusion, and sociopolitical conflicts. On purely moral grounds, famous literary works such as Victor Hugo’s Les misérables or Emile Zola’s Germinal were well ahead of the sophisticated theories produced by economists from various political and philosophical backgrounds. The veil of naturalist indifference about mass poverty and social change, which had covered much of the intellectual landscape for centuries, started to be lifted with the rise of Marxism, which brought the desire for economic change centre stage. Still, in the first half of the twentieth century, a period of major political transformations, social upheavals, and international tensions—culminating in two ignominious World Wars—the general intellectual mood was dominated by narratives of despair. No wonder that the post-World War II period, which witnessed the emergence and rise of macroeconomics as a discipline, still largely left the issues of mass poverty and income distribution on the fringe. Even the thinkers who conceptualized social transformations tended to do so in the darkest and most pessimistic terms. Orwell’s iconic 1984 is an illustration of such negative and pessimistic progressivism. Why did economic pessimism remain so well entrenched for so long? And why did economists ignore issues of collective well-being even as some of them studied the creation of national wealth through trade and finance? Looking at human history from an economic perspective, one can understand this long benign neglect: the potential for radical, transformational economic changes just seemed too remote and too improbable. For thousands of years indeed, there was little growth in the world economy and the income gap between countries was extremely small. Data compiled by Maddison (1982) shows that for some 1,400 years, the entire world was poor. For several centuries, income per capita in the city known today as Abidjan was identical to income per capita in the cities known today as Douala, London, Paris, Washington, Mexico, Tokyo, or Beijing. While there were a few relatively rich people in each of these places (the so-called aristocratic classes), most people everywhere on earth were poor. The quality of life then, even for the rich people, was bad: they died early of diseases that can now be prevented by a simple pill or a vaccine, and their life expectancy was very low. During the period known as Agrarianism (500–1500), estimates of what later became gross domestic product (GDP) grew by only 0.1 per cent on average—the same rate as the world’s population. Things improved marginally in the next couple of

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



centuries (1500–1700) when the world recorded an annual GDP growth rate of 0.4 per cent. Economies were largely based on agriculture and scientific progress was mainly divorced from technological innovation in production (Lin 1995). Agricultural productivity was also similar across nations and, as a result, the largest poles of economic production were in fact the largest population centres. China and India together contributed about half of world GDP during the seventeenth and eighteenth centuries. Then, something quite dramatic happened in the eighteenth century, which brought prosperity to some places and abruptly changed the course of human history: the Industrial Revolution. Scientific progress began to be applied to the means of production as machines were developed that both increased productivity, and also dramatically reduced transportation costs. This created the possibility for the countries that developed those technologies, or those that adapted the technologies first, to grow much faster than less technologically advanced countries. The Industrial Revolution marked a dramatic turning point in the economic progress of nations. Technological innovation introduced new tools that created the potential for a dramatic increase in productivity and living standards. Whereas per capita GDP growth worldwide was negligible in 1500–1820, it was about 1.5 per cent per annum following 1820. During the nineteenth century, several technological leaders and early adapters leapt ahead of the rest of the world, while others lagged behind. The result of this process was that (at least prior to the year 2000) the global economy was dominated by the few industrialized economies that existed in the world, and most of these few economies had become industrialized either as leaders or earlier followers of the nineteenth-century Industrial Revolution. There was a major caveat: Historical data dramatically reveal the divergent pattern of growth across country groupings: in the late nineteenth century, the Western European countries and their colonial ‘offshoots’ began to experience an historic take-off in incomes per capita. This was later matched by Japan in the middle of the twentieth century. The world economy was driven by several large Western European countries (Germany, France, Italy, the United Kingdom) and Anglophone-offshoots (Australia, New Zealand, the United States, and Canada), plus Japan. Many other countries, including the former Soviet Union, could rise to middle-income status and experience levels of average economic welfare that far surpassed prior centuries; however, their standard of living still lagged substantially behind the leading countries. In sum, the global economy was dominated by the so-called G7. During the twentieth century, the G7 maintained a large and stable share of global GDP. The twentieth century could have been a period in which technology spread across the entire world—allowing lagging economies to catch up with advanced economies. The predominant neo-classical paradigm in economic thinking suggested that this would be the case. Economic convergence might have been achieved through trade and capital flows based upon continued progress in transportation and communication technology. In principle, one would expect, and certainly hope, that the poorer countries in the world could catch up with the richer countries in the world. Instead, the twentieth century was an unfortunate period of continued and accelerated

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

divergence in living standards. Few countries have experienced convergence on a sustained basis. One approach to measuring relative progress is to look at per capita GDP relative to the United States, which has been the symbol of advanced industrialized countries since World War II. Pritchett (1997) analysed global economic performance during the twentieth century and concluded that it was marked by ‘Divergence, Big Time’. Lin and Rosenblatt’s (2012) historical overview of both the evolution of the economic performance of the developing world and the evolution of economic thought on development policy is by and large a sobering story. They confirm that the twentieth century actually accentuated divergence between high-income countries and the developing world, with only a limited number (less than 10 per cent of the economies in the world) managing to progress out of lower- or middle-income status to high-income status. Persistently, over 80 per cent of the countries in the world have GDP per capita levels that are half or less than half of the level in the United States. A few countries reached high-income status before falling back for a prolonged period into middle-income status. Argentina is a well-known example. In the fortythree years leading up to 1914, its GDP grew at an annual rate of 6 per cent, the fastest recorded in the world. It ranked among the ten richest in the world, ahead of Germany, France, or Italy. Its income per capita was 92 per cent of the average of the world’s sixteen richest economies. A century later, the income figure was 43 per cent of those same sixteen rich economies, trailing Chile or Uruguay. Russia is also an intriguing case: it was considered a middle-income country for some 200 years. Many African countries went from lower middle-income status at independence around 1960 to low-income in the 1980s. Since then, some have climbed back up to middle-income status. With a few exceptions (typically oil rich nations), the small number of developing economies that have recorded a substantial increase in income per capita are generally located in East Asia. They achieved sustained growth through rapid industrialization strategies underpinned by a comparative advantage in export dynamism and they also benefited enormously from the facilitating role of the state. The great divergence seen during the twentieth century may have been due, in part, to an interruption in trade and capital flows during the World Wars and the inter-war Great Depression that marked the first half of the twentieth century. Protectionism also persisted in many countries following World War II. It was only with the Uruguay Round of negotiations in the 1980s, leading to the eventual establishment of the World Trade Organization in 1995, that a clear institutionalized path towards opening up trade was established. Meanwhile, technological progress in communications and transport—but especially communications—facilitated the acceleration of global trade and capital flows in the last quarter of the twentieth century. Fortunately, the twenty-first century global economic landscape provides new possibilities for countries to catch up. While a general economic divergence is still the dominant story, there has been strong growth in the developing world, especially in several large developing countries, such as Brazil, China, India, Indonesia, and the Russian Federation. In some cases—most notably China and India—the high growth

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



period extends back some twenty or thirty years. In addition, there are numerous other countries that are taking advantage of growing trade and financial links—both with developed and developing countries—to accelerate economic growth. The global economy now exhibits a multi-polar system, with large developing countries leading the way as the new and most dynamic growth poles. In sum, there are opportunities out there for social and economic transformations, even in the most unlikely places (Lin and Monga 2017). Not too long ago, most countries in the world were poor. Yet some have managed to break out of the poverty trap, begin a sustained dynamic growth and become middle-income or even highincome economies, sometimes in a matter of just one or two generations. The governments in the small group of successful countries where positive change took place over the past century must have drawn valuable policy lessons or at least inspiration from the successful countries before them in the formulation of policies that unleashed their growth potentials. For economists engaged in research on poverty reduction, perhaps the biggest questions should consider why and how some countries succeeded whilst others failed to make it out of poverty. The countries that remain trapped in poverty might then be able to draw policy insights from those successful experiences and so avoid making mistakes, successfully initiating sustained, dynamic growth in their own countries. What did this elite group of successful countries do? How did they transform their economies and what lessons can the still lagging economies of the world draw from such experiences? These are perhaps the most important question in economics today.

T S M—W D S C

.................................................................................................................................. India’s modern mythology is rich with delicious parables which help make important points. A popular one is that of a poor nomadic merchant who travelled from one village to the other across the country, selling hats. On the road one evening, he was so exhausted that he fell asleep in an open field. On waking he was realized that his load of hats had disappeared. A group of monkeys sitting on neighbouring trees had picked up the hats and were wearing them and visibly mocking him from above. The merchant could not climb the trees to chase the monkeys. Desperate, he angrily threw his own hat on the ground. The monkeys observing him were so amused by his nervous gesture that they did the same: they all imitated him, took off the hats they were wearing and threw them on the ground. The hat seller couldn’t be happier: pleasantly surprised by the unexpected turn of events, he quickly collected his goods and continued his journey. Happy ending. The story does not end there. Some forty years later, his grandson who has become a hat seller too finds himself on the same route, going from one small town to the other to sell his goods. A similar

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

problem occurs: one day, as he is exhausted travelling, he drops his pile of hats and falls asleep. When he wakes up all the hats are gone. He looks up and, lo and behold, sees a group of monkeys wearing his hats, chanting and mocking him, having a good time. The hat seller does not panic. Remembering the lessons from his grandfather’s story, he purposely throws his own hat on the ground with the expectation that the monkeys will imitate him. To his surprise, nothing happens. The monkeys just stare at him. Then, one monkey comes down the tree, picks up the hat thrown on the ground, slaps the man on the cheek, and asks him the following question: ‘You think only you have a grandfather?’ The story of the Indian ‘signifying monkeys’,² told by Basu (2010), provides a gametheoretic metaphor for the need to take intergenerational learning and knowledge seriously. It also stresses the importance of understanding the many ways in which people learn and use knowledge—and warn about the risks of underestimating the ability to process information by other actors in the game. It is therefore a useful tale as we reflect on the reasons why societies evolved from being poor for centuries to generating prosperity for large sections of their populations. Studies about the Industrial Revolution—the event that changed the economic history of the world—have sparked some challenging questions, which eventually pointed to the importance of learning and knowledge: How did it occur (or manifested itself so obviously) at a particular time and place? Why did it take place in the Western world even though many other regions of the planet had recorded the rise and fall of great civilizations before then? Were some countries in the Western world blessed with enormous luck while those in Asia, Africa, or Latin America were cursed with poverty and sentenced to eternal poverty? Was it an unplanned, magical and random process? What major factors stimulated it? What was the role of past knowledge, perhaps transmitted silently across generations as in the case of the monkeys’ story? At first glance, there seem to be four interconnected sources of the post-nineteenth century growth acceleration (Fardoust 2006), although the sequencing and relative importance for each of them has long been the subject of debate: (i) improvements in transportation and communication technologies; (ii), trade liberalization and the intellectual doctrine of trade, as articulated by the likes of Adam Smith and David Ricardo; (iii) the peacefulness and stability (relative to earlier times) in Europe, where the likes of Metternich, Bismarck, and Castlereagh maintained the Concert of Europe and Britain assured a balance of power in its Pax Britannica; and (iv) an increase in ‘equity’ in European institutions beginning in the late seventeenth century and continuing through the early twentieth century. Economic historians and development theorists have struggled to explain the mystery of the Industrial Revolution and decipher its drivers and dynamics. The first set of possible explanations has centred on materialistic causes. Early economists such as Adam Smith and Thomas Malthus thought, erroneously, that the determinants of the

² We borrow this phrase from Gates (1988).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



prosperity of nations were to be found in physical objects and capital. Their analyses of the process of wealth creation was therefore dominated by notions of scarcity. For a long while, growth theories focused on how an economy transforms scarce resources like iron ore into machinery and so forth. It was all about capital and labour, with their combination leading to diminishing returns and to the steady state. Little attention was devoted to the underlying process of technological change (or to the possibility of increasing returns). A country or region or firm had a plot of land or an amount of capital that could be used only by them and there was a finite number of those basic sources of wealth. Some researchers have suggested that a combination of cheap energy costs at the time and high wages incentivized business people to devote more resources into technological innovation. Economists eventually recognized that other factors besides physical objects were important (things like institutions, rules, or recipes for how to rearrange physical objects and make them more valuable). A second set of explanations of the causes of the Industrial Revolution shed light on the benefits of colonial resource extraction, or on the social and political institutions that encouraged entrepreneurship. But researchers simply did not develop the rigorous analytical tools to integrate them into mainstream growth theory. The easy solution was to lump together all the possible elements other than capital and labour as ‘technological change’, and to treat them as exogenous—coming from outside the economic system. Then came John Maynard Keynes, who may not have heard the Indian monkeys’ story but had the right intuition when he stressed the importance of ideas in the very last pages of his General Theory: the ideas of economists and political philosophers, both when they are right and when they are wrong, are more powerful than is commonly understood. Indeed, the world is ruled by little else. Practical men, who believe themselves to be quite exempt from any intellectual influences, are usually the slaves of some defunct economist. Madmen in authority, who hear voices in the air, are distilling their frenzy from some academic scribbler of a few years back. . . . It is ideas, not vested interests, which are dangerous for good or evil. (Keynes 1936: 383)

The traditional arguments about capital, labour, and institutions, sound convincing but they are, at best, insufficient. Material and political conditions alone could not have done it. The Industrial Revolution was primarily the result of ideas. People and business leaders found innovative ways of adopting technology and making it commercially viable so that it could boost productivity. Some great inventions had been sitting on shelves for many decades. It took some wise and very practical people to design the institutions that would create the appropriate incentives and conditions for their broader use by firms and households, to bring benefits and rewards to all stakeholders, and to stimulate economic growth. As Azariadis and Stachurski (2005) point out, ‘While the scientific achievements of the ancient Mediterranean civilizations and China were remarkable, in general there was little attempt to apply science to the economic problems of the peasants. Scientists

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

and practical people had only limited interaction.’ Lin (1995) argues that the transition from innovation based on the experiences of artisan/farmers in the pre-Industrial Revolution period to innovation based on controlled experiments guided by science after the Industrial Revolution was the key factor. Societal incentives in pre-modern China did not favour the move toward the human capital accumulation needed for the new system of innovation. In fact, the significant waves of invention that occurred in China and the Islamic world prior to the Industrial Revolution did not snowball into a world-changing industrial revolution. Great inventions are rare and often potentially transformative for nations. But making them widely accepted and used by economic agents throughout society is really just the conditions for them to positively impact productivity and growth and to generate prosperity. McCloskey, who built on the analytical insights from Keynes, devoted a trilogy to refute these materialistic explanations (2006, 2010, and 2016). Slavery, imperialism, investment, coal, foreign trade, property rights, climate, genetics, or education, were not the root causes of the Industrial Revolution. Instead, he stresses the importance of the immaterial causes, including what he calls ‘bourgeois equality’—the idea that ordinary economic agents from all walks of life were emboldened to try innovative ways of doing things. It follows that the important questions are not just how innovations come about, but how they are adopted and translated into everyday improvements in living standards. Examples abound of great inventions that never turned out to become real innovations—or required time to do so. A well-known example is that of electricity, a discovery that dates to at least the late nineteenth century. Yet it took many decades before it could be used widely by firms and households and before researchers could estimate resulting improvements in the national productivity figures (David 1990). Besides designing and implementing the appropriate economic, financial, and marketing strategies and incentive systems to ensure the diffusion of innovations, nations also must undergo some social changes that allow for their large-scale adoption. Human history shows that such processes are not obvious or may not occur spontaneously. In his search for the specific conditions that turned the inventions of the late eighteenth and early nineteenth centuries into sustained, modern economic growth in Western Europe, Mokyr (2016) reexamines how the Enlightenment also created the right conditions for the emergence of a ‘Republic of Letters’, and sustained new forms of public debate and innovation that resembled what is now called ‘open science’. In Mokyr’s words, a culture of growth emerged that made everything possible during the Industrial Revolution. Thanks to that environment, knowledge could be converted from abstract scientific insights into practical technological know-how and readily useable common tools and processes. This process of a ‘democratization of knowledge’ often occurred simply because leading scientists and thinkers corresponded with their counterparts across Europe—a continent whose political fragmentation led ambitious rulers to compete in attracting the most prominent intellectual stars to their own territories. As Coyle (2014) puts it, ‘Competition among states to attract the best craftsmen, engineers and scientists spread knowledge, as did publishers vying for

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



new markets. This culture of public science, which had great prestige, paved the way for entrepreneurs to begin turning ideas into industrialisation and growth.’ Modern growth theorists such as Robert Solow and Joseph Schumpeter have indeed demonstrated quite convincingly that countries move from being low- to middle- and high-income not by accumulating more capital but through technological progress—that is, by learning how to do things better. In a world where labour and capital are quite mobile, the main explanation of the economic differences between rich and poor countries is not money: it is the difference in their ability to generate or borrow and use the best ideas available. ‘While some of the productivity increase reflects the impact of dramatic discoveries, much of it has been due to small, incremental changes. And if that is the case, it makes sense to focus attention on how societies learn, and what can be done to promote learning—including learning how to learn’ (Stiglitz and Greenwald 2014). In fact, formerly poor countries—especially those in East Asia—that have been able to converge toward the incomes of advanced economies have generally done so through learning. Economic development is therefore the process of technological diffusion and industrial upgrading. It involves making knowledge available to the largest number possible of economic agents and fostering constant learning. Then, the question becomes how good ideas emerge in any society, how they spread, are sorted out and validated, used, and shared widely enough to create a demand for them. Knowledge can be accumulated by some economic agents while others miss out. So, the production or underproduction of knowledge compounds over time, creating either positive or negative externalities for economic development. Kenneth Arrow’s work on ‘learning by doing’ showed, once learning starts, that it can be built dynamically and at scale. The dynamic nature and effects of learning can then far outweigh any short-term static losses in efficiency. Some economic sectors—manufacturing, above all—have greater learning spillovers than others. That justifies shaping policies to promote them over others, which would not occur randomly in any country environment. Yet knowledge is different from conventional goods. It is, in a sense, a public good. It is not excludable: people cannot be excluded from using it once it comes into being. Nor is it rivalrous: one person’s consumption of knowledge does not preclude another person’s consuming it. So, the marginal cost for another person or firm to enjoy the benefits of knowledge—beyond the cost of transmission—is zero. But markets— anywhere, whether in developed or developing countries—are not efficient at producing and distributing public goods. Producing, acquiring, sharing, and diffusing knowledge—through complex processes of social learning—make it the public good that is subject to the greatest market failures. In all societies, especially in low-income countries, knowledge brings many externalities, as gaps in production and acquisition by economic agents produce gaps between social and private returns. Producing knowledge is costly. Because individuals and firms do not always capture the full returns from their investments in learning (when they can afford such investments), knowledge is generally underproduced. The world has changed many times since the (first) Industrial Revolution. There is much to learn from the experiences of the few developing economies that have

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

managed to break out of the vicious cycle of poverty. Their stories, development strategies, and policy frameworks offer good news for other developing countries. In an increasingly globalized world economy where capital, labour, technology, and ideas are very mobile, it is still possible for countries at very low-income levels to borrow relevant knowledge from more advanced economies and use it at minimum costs to enrich their analytical and policy frameworks. But this process cannot be spontaneous or left just to market forces. Institutions must be in place to create incentives and to stimulate the process of learning across firms (even as they compete against each other) and across industries. International development agencies whose primary role is to facilitate the production and sharing of global public goods are best placed to stimulate and facilitate the process of sharing clever ideas, to foster the production and distribution of ideas in African countries, and to highlight the importance of meta-ideas (‘an idea that helps us get better at discovering ideas,’ as Paul Romer puts it).

D S C P: H  O

.................................................................................................................................. In their long quest to convert their discipline from a conjectural field of the social sciences into a hard science, economists have generally shied away from topics and issues that seemed difficult to grasp with rigorous, logical, and formal analytical tools. With the emergence and gradual dominance of mathematical models in economics during the second half of the twentieth century, some topics perceived as hard to model suffered from benign neglect. Krugman (1995) observed that this methodological shift, which caught some researchers off-guard, might have been the main reason for the reluctance of mainstream economists, for several decades, to tackle the important but challenging issues of development. Krugman’s conjecture may also explain the ups and downs of research on structural transformation, and the general intellectual shyness that kept it away from much of economics roughly from the 1960s to the late 1980s. Knowledge has emerged as the main driver of sustained structural transformation. Oxford dictionaries define it broadly as ‘facts, information, and skills acquired through experience or education; the theoretical or practical understanding of a subject’. Knowledge is also defined tautologically as the ‘sum of what is known’. But what exactly is structural transformation, and how has the economic research devoted to it evolved in recent decades? The general idea underlying the process has remained straightforward: structural transformation has always been understood as the process whereby the resources (labour, capital, and technology) in any economy are shifted out of traditional agriculture and other low-productivity primary activities into ‘modern’, higher productivity sectors (including non-traditional agriculture). This shift of resources (transformation), and the expansion of modern sectors, have been at the core of the sustained

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



productivity gains that characterize economic development. Indeed, there is ample consensus that rising productivity accounts for the bulk of long-term growth. Sustained economic growth is essentially about structural and technological upgrading. The specific elements underpinning structural transformation processes are difficult to identify and measure consistently. In fact, the economic notion of ‘structures’³ has evolved over time to cover both macro and micro issues, and to hold different meanings (Lin and Monga 2013). In the 1940s, a first wave of researchers working on low-income countries conceived development to be an interrelated set of long-run processes. Early development economists⁴ borrowed the notion of ‘structures’ from other social scientists and used it to design a first set of theories for growth and prosperity. While numerous variants of early economic structuralism⁵ can be traced to a very diverse body of work that spans over a century (from Karl Marx, David Ricardo, and Keynes to Michal Kalecki, Joan Robinson, Richard Nelson, and Sidney Winter), the fundamental assumption of all its various schools of thought is that ‘an economy’s institutions and distributional relationships across its productive sectors and social groups play essential roles in determining macro behavior’ (Taylor 2004: 1). Early structuralists argued that due to structural rigidities and coordination problems in developing country markets, modern heavy industries were unable to develop spontaneously there. They suggested that the virtuous circle of development essentially depended on the interaction between economies of scale at the level of individual firms and the size of the market. Specifically, they assumed that modern methods of production could only be made more productive than traditional ones if the market were large enough for their productivity edge to compensate for the necessity of paying higher wages. Yet the size of the market itself depended on the extent to which these modern techniques were adopted. Therefore, if the modernization process could be started on a very large scale, then the process of economic development would be

³ The concept of ‘economic structure’ refers to ‘the composition of production activities, the associated patterns of specialization in international trade, the technological capabilities of the economy, including the educational level of the labour force, the structure of ownership of the factors of production, the nature and development of basic state institutions, and the degree of development and constraints under which certain markets operate (the absence of certain segments of the financial market or the presence of a large underemployed labour force, for example) (Ocampo et al. 2009: 7). ⁴ The long list of these early development economists includes Rosenstein-Rodan (1943); Singer (1950); Lewis (1954); Nurkse (1956); Myrdal (1957); Prebisch (1959); Chenery and Bruno (1962), and Furtado (1964). ⁵ Dutt and Ross (2003) provide a comprehensive review of the main and often overlapping currents of early economic structuralism. They suggest that the first phase, which occurred from 1945 to the mid1950s, was launched by Rosenstein-Rodan, Lewis, and Nurske. A second sub-group working in this area from roughly the mid-1950s to the late 1960s and was dominated by contributions from Myrdal, Hirschman, Chenery and Bruno, and Furtado. A third sub-group, called ‘neo-structuralism’ or ‘late structuralism’, emerged in the early 1980s to respond to criticism from neo-classical economists and to modify and enrich development economics with lessons drawn from economic analysis and the actual experience of poor countries. It is represented by contributions from Taylor (1983, 1991), Ocampo and Taylor (1998), and Ocampo et al. (2009).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

self-reinforcing and self-sustaining. If not, countries would be trapped into poverty indefinitely (Rosenstein-Rodan 1943). The focus of the first wave of economic structuralism was therefore not on knowledge and the transfer of ideas but rather on structural change in production structure and on economy-wide phenomena such as agricultural transformation, industrialization, urbanization, and ‘modernization’. Early structuralists identified economic activities of a very different nature that existed side by side in the ‘periphery’ (developing countries), with an export sector of relatively high productivity of labour, and a subsistence agricultural sector of very low productivity. They hypothesized that poor economies had to specialize in the production of a few commodities whose exploitation could not generate any forward or backward linkages. They conjectured that poor economies were trapped in an external disequilibrium and could only occupy a marginal space on the international scene (especially given the long-term trend of declining terms of trade). As a result, they argued, these peripheral economies would not undergo the kind of transformation process that leads to modernization and prosperity. Leading that group, Kuznets (1966) studied the genesis and patterns of evolution of modern economic growth in high-income countries and approached structural analysis mainly through the lens of sectoral changes—that is, the evolution overtime of the relative contributions of agriculture, industry, and services to gross domestic product (Syrquin 1988). A second wave of development thinking dominated policy making in low-income countries in the 1980s and 1990s and tackled structural analysis only indirectly. Economists in that group approached structural change almost inadvertently through a broad examination of the general functioning of economies, their markets, institutions, mechanisms for allocating resources, regulatory and incentives systems, etc. The proponents of the ‘stuctural’ adjustment programmes implemented in many developing countries viewed the restoration of external and domestic balances as an essential precondition for launching the process of economic transformation and change. Throughout these two major waves of research, there was no consensus among researchers on its key features into the increasingly formal models of mainstream economic theory. Moreover, the definition and scope of structural transformation have been gradually broadened, making it even more challenging for economists to account for it consistently with the rigorous tools of mathematics. The dominant framework used to study the growth in output associated with the development process has been the one-sector neo-classical model, also used for as the basis for development accounting exercises. Under some assumptions, the model predicts a balanced growth path, which is viewed as providing a good description of the long-run behaviour of advanced economies. Despite its elegance and simplicity, the one-sector model typically only describes static, repetitive features of a balanced growth path in which economies become richer by doing the same things, with output, consumption, investment, or the capital stock all growing at the same constant rate, and with the interest rate and factor shares held constant while the real wage also grows at the same rate as output (the so-called Kaldor Facts). Yet the balanced growth path is

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



not the story of most economies. Thus, the need to build models that simultaneously generate outcomes that mimic balanced growth at the aggregate level while also generating structural transformation. A third and more recent wave of the development literature has sought to refine structural analysis.⁶ At the theoretical level, economists have developed multi-sector extensions of the one-sector growth model and made them consistent with the new theories of structural transformation (Ngai and Pissarides 2007; Herrendorf at al. 2014). They have also focused on building models that can quantitatively account for the properties of structural transformation and at the same time assess the importance of various economic mechanisms. The quest to the standard three-sector focus is felt more acutely in the case of advanced economies, which are increasingly dominated by distinct types of services; this constantly changing structure makes it necessary to disaggregate services. It has been noted, for example, that education and health care are very different activities than construction, in that they both represent an investment and tend to use very different skill intensities for the labour that they employ. Authors such as Jorgenson and Timmer (2011) and Duarte and Restuccia (2016) have pioneered work in this direction. The third wave of development thinking aims at reexamining the importance of knowledge in the process of structural transformation. Its main theoretical foundations can be found in a large corpus of interrelated topics of economic knowledge that includes: (i) the economics of information; (ii) the economics of ideas and diffusion of knowledge; (iii) the problem of agglomeration; and, perhaps most important, (iv) the problems of coordination and externalities. Stiglitz, who pioneered the economics of information, explained in his Nobel lecture how his encounter with developing country issues forced him to reassess his own views: ‘My first visits to the developing world in 1967, and a more extensive stay in Kenya in 1969, made an indelible impression on me. Models of perfect markets, as badly flawed as they might seem for Europe or America, seemed truly inappropriate for these countries. But while many of the key assumptions that went into the competitive equilibrium model seemed not to fit these economies well, the ones that attracted my attention was the imperfection of information, the absence of markets, and the pervasiveness and persistence of seeming dysfunctional institutions’ (Stiglitz 2001). It was not just the discrepancies between the standard neo-classical competitive model and its predictions that were being questioned. The model was not robust—even slight departures from the underlying assumption of perfect information had major analytical and policy consequences. In many areas of public policy (such as education, wage determination), the notion that had underlain much of traditional competitive equilibrium analysis—that markets had to clear—was simply not true if information were imperfect.

⁶ This section draws on Lin and Monga (2013) ‘Evolving Paradigms of Structural Change’.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

Since the nineteenth century, the most dominant idea in mainstream economics, which provided both the rationale for the reliance on free markets, and the belief that issues of distribution can be separated from issues of efficiency, was that competitive economies lead, as if by an invisible hand, to a (Pareto) efficient allocation of resources, and that every Pareto efficient resource allocation can be achieved through a competitive mechanism—provided only that the appropriate lump sum redistributions are undertaken. That big idea, still the fundamental theorem of welfare economics, also allowed the economist the freedom to push for reforms which increase efficiency, regardless of their seeming impact on distribution. As Stiglitz (2001) noted, ‘the economics of information showed that neither of these results was, in general, true’. Moreover, asymmetries of information have been shown to be related to absent or imperfect markets. They help explain why markets for used cars—as famously shown by Akerlof (1970)—or for credit or labour tend to work imperfectly. Information imperfections are pervasive in the economy and neither sustained economic growth nor structural change is possible without a reliable mechanism to address them (Greenwald and Stiglitz 1986). The fact that when there are asymmetries of information, markets are not, in general, constrained Pareto efficient implies there is a potentially vital role for government (Stiglitz 1997). That insight also opens up an avenue to discuss the economics of ideas and diffusion of knowledge. But what exactly is knowledge? Knowledge is typically considered a peculiar form of information, and many of the issues that are central to the economics of information and to the process of structural transformation—such as the problems of appropriability, the fixed costs associated with investments in research which give rise to imperfections in competition, and the public good nature of information—also point to a crucial role for government. An important conclusion from economic analysis is that the market economies of highincome countries (in which research and innovation play a vital role) are not well described by the standard neo-classical competitive model, and that the market equilibrium, without government intervention, is often not efficient. That conclusion also holds for developing countries where there is even less public and private funding available to generate knowledge. But these countries can learn by borrowing ideas and expertise from other more advanced economies. ‘Nations are poor because their citizens do not have access to the ideas that are used in industrial nations to generate economic value,’ Romer observed (1993a: 543). Developing countries remain trapped in poverty because households and firms there have not been able to improve their productivity levels by either inventing innovative ways of making better goods and services or by copying and use new industrial and technological tools available elsewhere. ‘In a world with physical limits, it is discoveries of big ideas, together with the discovery of millions of little ideas, that make persistent economic growth possible. Ideas are the instructions that let us combine limited physical resources and arrangements that are ever more valuable’ (Romer 1993b: 64).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



Another theoretical justification for the role that governments must play to foster development—is the economics of agglomeration. Balassa (1966) made a puzzling observation on the rise of intra-industry trade in Europe in the 1950s. Balassa noted each country produced only part of the range of potential products within each industry, importing those goods it did not produce. This trade allowed specialization in narrower ranges of machinery and intermediate products, permitting the exploitation of economies of scale through the lengthening of production runs. New trade theorists have consequently highlighted the fact that unexhausted economies of scale at the firm level necessarily imply imperfect competition. They have shown that increasing returns have been a powerful force shaping the world economy, and developed general equilibrium models of imperfect competition that confirm Marshall’s trinity of reasons for industry localization: knowledge spillovers, labour market pooling, and specialized suppliers.⁷ For developing countries that must rely on trade as their main source of growth in an increasingly globalized world, the policy implications of this are clear: it is essential that their governments are capable of solving the coordination and externalities issues that prevent the agglomeration of firms and activities from taking place (Rodrik 2007; Harrison and Rodriguez-Clare 2009). A third wave of structural transformation research has also been geared towards the specific experience of developing economies, seeking to highlight the dynamics of change and the factors that differentiate their experiences from those of advanced economies (Lin and Monga 2011, 2017; Lin 2012a, 2012b). Policy questions are being addressed systematically, covering the distribution of roles between the government and the private sector; the strategic selection of competitive industries according to the comparative advantage of developing countries; the determinants of the dynamics of sectoral contributions to growth; the evolution of the capital intensity of sectors over time—within and across countries; the processes that allow economies to move up the value chain; the various ways of organizing and fostering the adaptation and adoption of new technologies in poor countries; the determinants of a country’s ability to create employment; and the institutional arrangements that are necessary to support structural transformation, especially in the context of low-income countries where infrastructure, skills, and long-term financing are scarce. At the policy level, however, the big question remains: what exactly needs to be done to ignite, stimulate, or support processes of structural economic change, both from the macro and micro perspectives? Which sectors are more likely to record the highest productivity growth rates, and therefore contribute more effectively to positive social transformation? Theoretical and empirical economic research has long reached consensus: industrialization is key (UNCTAD 2017). But what type of industrialization? And has the very idea of industrialization become a moving target in a world of constantly changing production structures?

⁷ For a quick intellectual history of the importance of increasing returns on economics and a review of progress on theoretical analysis, see Krugman (2008).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

G   B B: I, U,  E F  C

.................................................................................................................................. One big idea seems to have survived the various waves of development thinking on structural transformation: industrialization. At times, it has been celebrated, revered, feared, and even challenged, but it has never been absent from the successive intellectual and policy frameworks seen as the essential feature of economic change. In fact, there is wide consensus among economists that industrialization is the single most important driver of structural change. The two concepts are indeed closely linked: structural transformation is the phenomenon whereby a society’s resources are moved from the sectors where they yield little economic benefit to those where the payoffs are the highest—and this occurs through industrialization. Indeed, prosperity is achieved in any country only when a country’s resources (human, natural, and capital) are shifted from subsistence and informal activities into high-productivity activities. Industrialization dynamics is therefore an unavoidable feature of structural transformation. It has long been recognized as one of the main engines of sustained economic growth, especially in the early stages of development.⁸ Its essential characteristics include: (i) an increase in the proportion of the national income derived from manufacturing activities and from secondary industry in general, except perhaps for cyclical interruptions; (ii) a rising trend in the proportion of the working population engaged in manufacturing; and (iii) an associated increase in the income per head of the population (Bagchi 1990). Few countries have been economically successful without industrializing. Only in circumstances such as an extraordinary abundance of natural resources or land have countries been able to do so (Unido 2009). The economic development of today’s industrialized countries was almost universally accompanied by an increase in agricultural productivity in the initial stages of development. Sustained economic development typically requires that agriculture, through higher productivity, provides food, labour, and even savings to the process of urbanization and industrialization. A dynamic agricultural sector raises labour productivity in the rural economy, pulls up wages, and gradually eliminates the worst dimensions of absolute poverty. Agricultural growth also stimulates growth in non-farm sectors, thus driving structural transformation and industrialization processes. The development of a competitive industrial sector yields an even higher payoff. Economists have established at least since the early1960s that manufacturing has always played a larger role in total output in ⁸ Earlier analyses of the process, dating back to the 1950s and 1960s (Datta 1952; Kuznets 1966), found that manufacturing specifically tends to play a larger role in total output in richer countries— a pattern corroborated by the UNIDO report (2009)—and that higher incomes are associated with a substantially bigger role of transport and machinery sectors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



richer countries, and that countries with higher incomes are typically those with a substantially bigger economic contribution from the transport and machinery sectors. The countries that manage to pull out of poverty and get richer are those that can diversify away from agriculture and other traditional products. Industrialization is an ever more powerful engine for economic and social change in the context of globalization, as it provides an almost infinite potential for growth— especially for many low-income countries. It has always played a key role in growth acceleration processes that are sustained over time and eventually transforms economies from ‘poor’ to ‘rich’. Whereas economic growth based on the exploitation of natural resources or agricultural land eventually faces the constraint of shortages of quantity, a development strategy based on producing manufactured goods for the global market benefits from economies of scale due to increasingly lower unit costs of production. In the early phases of modern economic growth, which started with the Industrial Revolution, manufacturing played a larger role in the total output of successful countries and their higher incomes were associated with a substantially greater role of transport and machinery sectors. Throughout the nineteenth and twentieth centuries, countries in North America, Western Europe, and Asia could transform their economies from agrarian to industrial powers, which included a rapidly growing services sector fuelled in large part by the multiplier effect of manufacturing. As a result, they built prosperous middle classes and raised their standards of living. In addition to the generally much higher levels of productivity in industry (especially manufacturing) than in traditional agriculture, the main reason for the growth of industrialization is the fact that its potential is virtually unlimited, especially in an increasingly globalized world. As agricultural or purely extractive activities expand, they usually face shortages of land, water, or other resources. In contrast, manufacturing easily benefits from economies of scale: thanks to new inventions and technological development, and to changes in global trade rules, transport and the unit costs of production have declined substantially during the past decades, which also facilitates industrial development. Several decades ago, low-income countries faced the constraints of their limited market size, high transportation costs, and trade barriers, and could not take advantage of the opportunities offered by manufacturing. With globalization, virtually any country can identify products for which it has overt or latent comparative advantage, facilitate the entrance of its firms into global value chains, and scale up production almost without limit, thereby creating its own niche in world markets. Today, almost any small country can access the world market, find a niche, and establish itself as a site of global manufacturing. For example, Qiaotou and Yiwu, two once small Chinese villages, have become powerhouses, producing more than twothirds of the world’s buttons and zippers, respectively! Industrialization also promotes inclusive development by expanding the fiscal space for social investments. In such a context, fiscal revenues are likely to increase due to: exports of higher value added; the rising profits of companies; and better incomes earned by a more productive and innovative labour force. Within the industrial sector,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

manufacturing has evolved and changed the dynamics of the world economy. Profound changes in geopolitical relations among world nations, the widespread growth of digital information, the decline of transportation costs and the development of physical and financial infrastructure, computerized manufacturing technologies, and the proliferation of bilateral and multilateral trade agreements have all contributed to the globalization of manufacturing. These developments have permitted the decentralization of supply chains into independent but coherent global networks that allow transnational firms to locate various parts of their businesses in various places around the world. The creative design of products, the sourcing of materials and components, and the manufacturing of products can now be done more cheaply and more efficiently from virtually any region of the planet while final goods and services are customized and packaged to satisfy the needs of customers in faraway markets. The globalization of manufacturing has thus allowed developed economies to benefit from lower-cost products driven by the lower wages used for production in developing countries such as China, India, Bangladesh, Costa Rica, Mexico, or Brazil while creating employment and learning opportunities in these formally poor nations. The intensity of these exchanges has led to new forms of competition and co-dependency.⁹ Yet, despite its importance, mainstream development economics paid only limited attention to industrialization for long decades. Several factors explain this benign neglect or even reluctance by researchers to think seriously about industrialization. First, tackling industrialization posed serious analytical challenges for a long time. In 1995 Krugman wrote about the intellectual shyness of early development economists when faced with the need to formalize ideas and concepts with the unsuitable mathematical toolbox available in the 1960s and 1970s. Following his intuition, it can be noted that issues of market size and economies of scale, central to the analytics of structural transformation as studied by some of the leading voices in the first generation of development economists (mainly Paul Rosenstein Rodan and Albert Hirschman) were not presented in formal models. A second reason for the long neglect of structural change in mainstream economic analysis was the intellectual shift from the type of deep, transformative long-term questions that had preoccupied economists since the eighteenth century, to a narrower focus on short-term, business cycles topics that dominated headlines. The quest for the sources and mechanisms for prosperity and social change, which started with classical economists such as Adam Smith, Alfred Marshall, or Allyn Young, slowed down after the Great Depression, as researchers turned their interest to short-run issues. Indeed, with the notable exception of the pioneering work of Robert Solow, for much of the twentieth century and certainly through the 1960 and 1970s, macroeconomists tended to study business cycle issues that characterized the post-war period. As they tried to better understand stabilization policies—monetary and fiscal measures to avoid ⁹ In recent decades, innovation, technological developments and new sources of economic growth have led some economists to question whether manufacturing still matters. See Monga (2014) for a critical assessment of the arguments in that debate.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



disruptive and costly inflation—few resources were devoted to the analysis of the long-run determinants of growth and transformation. Several waves of growth research produced valuable insights (Monga 2011). On the theoretical front, the analysis of endogenous technical innovation and increasing returns to scale has provided economists with a rich general framework for capturing the broad picture and the mechanics of economic growth. From Solow’s work, we know the importance of the role of capital accumulation (both physical and human) and technical change in the growth process. From contributions by Becker, Heckman, Lucas,¹⁰ and many others, we have also learned about the importance of human capital through the diffusion of new knowledge or on-the-job learning, often stimulated by trade, and the so-called college wage premium. From work by North (1981), with supporting theoretical and empirical analyses exemplified by the works of Greif (1993) and Acemoglu et al. (2001), we have learned that growth is in large part driven by innovation and institutions that have evolved in countries where innovative activity is promoted and conditions are in place for change to take place. From Romer and endogenous growth theorists, we have understood the need to change the focus of growth theory from accumulation to knowledge creation and innovation. In sum, we know quite a lot about some of the basic ingredients of growth. On the empirical side, the availability of standardized data sets—especially the Penn World tables—have stimulated interest in cross-country work that highlight systematic differences between high-growth and low-growth countries with regard to initial conditions (such as productivity levels, human capital, demographic features, infrastructure, financial development, and inequality), institutional variables (such as the rule of law, protection of property rights, and governance indicators), and policy variables (such as macroeconomic stability, financial regulation, or trade openness). However, growth research still faces significant challenges in identifying actionable policy levers to sustain and accelerate growth in specific countries. In recent years, growth researchers have confronted three types of challenges: the explanation of a lack of convergence among countries; the identification of robust determinants of economic performance; and the design of the supporting institutions for innovation and technological change, which are widely acknowledged to be the foundations for structural change and prosperity. The disappointments of growth research—most notably from the perspective of policy makers seeking specific action plans to generate prosperity—have led to a reassessment of the validity and usefulness of existing knowledge, and to a return to structural transformation. An important study by the World Bank (2005), focusing on lessons of the 1990s, highlighted the complexity of economic growth and noted that the reforms carried out in many developing countries in the 1990s focused too narrowly on macroeconomic stabilization and the efficient use of resources, not on the expansion of capacity for growth. While they enabled better use of existing capacity, thereby establishing the basis for

¹⁰ See, in particular, Becker (1992); Lucas (2004); Heckman (2006).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

sustained long-run growth, they did not provide sufficient incentives for expanding that capacity.¹¹ The report concluded that there is no unique, universal set of rules to guide policy makers. It recommended less reliance on simple formulas and the elusive search for ‘best practices’, and greater reliance on deeper economic analysis to identify the one or two most binding constraints on growth in each country. The rethinking of economic development—beyond growth theories—has brought attention back to the deeper issues of structural transformation and its key feature, industrialization.¹² However, economists have continued to debate the role and significance of industrialization in a world economy that has changed considerably since the days of Adam Smith and even Simon Kuznets.

D  A: T  O?

.................................................................................................................................. To many researchers, industrialization’s role has become marginally important in the global quest for economic prosperity. They cite as the most obvious piece of evidence the dramatic decline in employment in manufacturing as a share of total employment in the world’s most advanced economies, a phenomenon widely referred to as ‘deindustrialization’. This trend was first observed in the United States and Europe. Some critics saw deindustrialization as resulting from the rapid growth of North–South trade (trade between the advanced economies and the developing world) and explained that it was caused by the fast growth of labour-intensive manufacturing industries in the low-wage developing world. Viewing it as a threat to workers in the advanced economies, they branded it as a negative consequence of the globalization of markets, which generated fierce political debates in the Western world. Political leaders across the ideological spectrum seized on deindustrialization as the main explanation to widening income inequality in the United States and high unemployment in Europe. These popular explanations were inaccurate. Empirical research showed that when measured in real terms, the share of domestic expenditure on manufactured goods had been comparatively stable for decades in advanced economies. Thus, that round of deindustrialization was essentially the result of higher productivity in some capitalintensive manufacturing sectors rather than in services. The pattern of trade specialization among the advanced economies explained why some countries deindustrialize faster than others (Rowthorn and Ramaswamy 1997). In sum, deindustrialization was primarily a reflection of successful economic development and effective industrial

¹¹ Pritchett (2006) suggests that economists abandon the quest for a single growth theory and focus instead on developing a collection of growth and transition theories tailored to countries’ circumstances. ¹² Influential works include Ngai and Pissarides (2007); Rogerson (2008); Timmer and Akkus (2008); Rodrik (2008); Lin (2012a); Herrendorf et al. (2014); and Timmer (2014).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



upgrading strategies, and North–South trade has very little to do with it. This became even more apparent in Japan and in the successful Four Tiger economies of East Asia (Hong Kong, China, Korea, Singapore, plus Taiwan Province of China), which also experienced deindustrialization. Most recently, deindustrialization has emerged again as a concern not just for advanced economies but also for low-income countries. Rodrik’s (2016) seminal work on this topic has highlighted the changes in the relationship between industrialization (measured by employment or output shares) and incomes not just in advanced, post-industrial economies, but also in developing countries. He concludes that countries are running out of industrialization opportunities sooner and at much lower levels of income compared to the experience of early industrializers. In other words, industry’s share of employment in some developing countries seems to be peaking at a lower level than it used to, and at an earlier point in their development. Advanced economies have indeed lost considerable employment in some industries (especially of the lowskilled type), but they have done surprisingly well in terms of manufacturing output shares at constant prices. Rodrik’s analysis confirms that advanced economies have lost considerable employment (especially of the low-skilled type), but they performed well in terms of manufacturing output shares at constant prices. Surprisingly, Asian countries and manufacturing exporters appear to have been largely insulated from deindustrialization trends, while Latin American countries have suffered the most. Other empirical research examining developing countries as whole has shed light on the mystery of deindustrialization. Haraguchi et al. (2017) have analysed several decades of employment data on over 100 developing countries, going back to 1970. They explore whether the low levels of industrialization in developing countries are attributable to long-term changes in opportunities available to the sector around the globe. They find that manufacturing employment became geographically more concentrated after 1990, but no less important. Their study’s findings show that the manufacturing sector’s value added and employment contribution to world GDP and employment, respectively, have not changed significantly since 1970. The declining manufacturing value added and manufacturing employment share in many developing countries has not been caused by changes in the sector’s development potential but has instead resulted from a shift of manufacturing activities to a relatively small number of populous countries, thus resulting in a concentration of manufacturing activities in specific developing countries. While the average of each country’s manufacturing–employment ratio has indeed declined since the early 1990s, as observed by Rodrik (2016), the aggregate of manufacturing employment in developing countries is actually higher than in earlier decades. This counter-intuitive finding can be explained by the fact that the workforce in some developing countries—such as China—is so large that a stagnation or even a decline in the percentage of manufacturing in the labour force does not translate into a decline in the absolute, aggregate number of workers in that sector. Still, worries about deindustrialization and about the importance of industrialization in modern growth and structural transformation processes have been compounded by

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

the lacklustre global trade climate, which has characterized the world economy in the wake of the 2008 financial crisis.¹³ This new trade scepticism has led many researchers and policy makers to wonder whether today’s low-income countries could benefit from the same export opportunities that allowed rapid industrialization in Asia in the 1970s–90s—even if they could adopt the right policy frameworks and develop their manufacturing production bases. While it is indeed true that global trade grew at a lower rate than global GDP in the decade following the 2018 financial crisis, over the long term, the trade–GDP relationship is usually not a static one. Despite the resurgence of the protectionist discourse in some advanced economies, and the persistence of non-trade measures, the general, long-term trend of global trade is still a very positive one for developing countries. Moreover, the declining general trend in average tariffs around the world since World War II is unlikely to be rolled back given the structural changes they have induced in the global production system and the enormous win–win opportunities they have created for advanced and developing economies. The best indicator of that evolution is that many goods are now manufactured in several countries at the same time. Global trade is therefore no longer a series of transactions between countries producing individual goods and services within their national boundaries and exchanging them in international markets. It is often about collaboration and partnerships, even in an intensively more competitive world. Manufacturing is increasingly a network of global supply chains in which the various production stages take place in the most cost-efficient locations—regardless of where they are in the world (Lin and Monga 2017: ch. 7). Some researchers have observed that manufacturing is not the only driver of growth. In the words of Ghani and O’Connell (2014) and Enache et al. (2016), there is an ongoing Third Industrial Revolution led by services, which may now contribute substantially to output growth, productivity growth, and job growth in low-income countries. Services are invalidating some long-held tenets of economic development: for centuries, the service trade was limited because it required proximity and face-toface interaction between the buyer and the seller. However, this is no longer the case, as technology and innovation allow services to be produced and traded just like manufactured goods. Moreover, the cost of trading services that can be digitized has fallen dramatically, as services do not have to confront customs and other logistical barriers. And service-led growth is also greener and more gender-friendly. These observations have led Ghani and O’Connell (2014) to suggest that the services sector, branded as a ‘growth escalator for low-income countries’, be given priority in the design of structural transformation strategies. They conclude: ‘Unlike the goods sector, where developing countries already have a large market share, making it difficult for new entrants to become large-scale exporters, services appear to be steadily expanding, with catch-up ¹³ In the words of Davies, ‘world trade has lost its mojo’, and global trends support his observation. From 1990 to 2008 global real GDP expanded at an annual rate of 3.2 per cent, while world trade volume grew at 6.0 per cent. Since 2008, however, world trade has grown slightly slower than GDP, so the share of exports in GDP fell after a 25-year uptrend (Davies 2013).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



opportunities continuing to rise and entry for all. . . . A service-led growth can be sustained because the current globalization of services is only the tip of the iceberg, and service is the largest sector in the world, accounting for more than 70% of global output’ (Ghani and O’Connell 2014: 20 and 21). Today’s global economy certainly offers infinite opportunities for growth and transformation in the services sector but not to countries at all levels of development. Therefore, one should be careful not to draw swiping policy recommendations from the fact that an increasingly large services sector is driving global growth. First, there is a semantic issue to be addressed: manufacturing no longer means the type of old, capital-intensive industries that spurred the First Industrial Revolution in the eighteenth and nineteenth centuries. With the advent of the Second Industrial Revolution, manufacturing has become a continuum of activities that are interlinked. As noted by Schwieters and Moritz (2017), ‘One key indicator is that conventional boundaries between industries are eroding. It’s getting harder to tell the difference between, say, a telecommunications company and an entertainment producer, or between a retail bank and a retail store. The relationships among suppliers, producers, and consumers are also blurring, more rapidly than many business decision makers are prepared for.’ The definitions of ‘agriculture’, ‘manufacturing’, and ‘services’, should therefore evolve to reflect the constantly changing boundaries of these sectors. In its current meaning, manufacturing should be understood in its broadest sense as all trade based on the fabrication, processing, or preparation of all kinds of products from raw materials and commodities to chemicals, textiles, machines, equipment, and even modern services and virtual goods. The second reason why policy recommendations cannot necessarily follow from the increasing services sector growth is that even in developing countries where there has been a boom in the services sector without industrialization, a lot of these services are low-productivity, subsistence level, and sometimes even informal activities that may help households escape poverty but are not sustainable sources of growth. The type of high-productivity services that offer long-term growth prospects to nations (in sectors such as informational technology or banking and finance) are skill-intensive. Yet by definition, low-income countries have a weak skills base. That is certainly the case in most African and South Asian countries where the demographic structure and limited fiscal base do not allow for the rapid build-up of the kind of human capital necessary to sustain economic transformations driven by high-productivity modern services. Even developing countries such as India, Sri Lanka, Kenya, Cameroon, or Egypt, where substantial amounts of public funding have been devoted to the creation of strong education systems, too often end up exporting much of their skilled labour. Consistent with the basic rationale for structural transformation, which is to constantly move labour and capital into higher-productivity sectors, it is logical that advances in the modern service sector, rather than in traditional manufacturing, will drive the growth of living standards in the advanced economies in the future and also in the middleincome countries that successfully manage their industrial upgrading process. However, for low-income countries, low-skilled labour-intensive employment will still offer

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

sizeable growth opportunities—especially with the upcoming ‘graduation’ of large middle-income countries like China or Indonesia, which is freeing up substantial quantities of industrial employment (Lin 2011). Finally, there is the perceived threat of automation on industrialization. Improvements in the design of robots, and their increasing use in many industries around the world, have made economists wonder whether the long-held prescriptions for structural transformation are becoming obsolete. If sophisticated and smart robots, not people, can fill the factories and therefore lower production costs, would that not invalidate the Simon Kuznets insight that modern economic growth requires moving resources out of agriculture into industry, then out of industry into services? The question of whether robots hamper industrialization’s central development role is indeed important. However, the potential adverse employment and income effects of robots are being overestimated—most commentators neglect to consider that what is technically feasible is not always also economically profitable. For instance, it would be technically possible (if not necessarily economically sensible) to automate about twothirds of manufacturing employment in countries like India, Indonesia, or Thailand (McKinsey 2017). But the economic and social returns for doing this in the decades ahead are unclear, at best. The countries currently most exposed to robot-based automation are those with a large and well-paying manufacturing sector (UNCTAD 2017). While routine tasks in well-paid manufacturing and service jobs are being replaced by robots, low-wage manufacturing jobs in areas such as clothing factories are left largely unaffected by automation (UNCTAD 2017). So far, robotization has had a small effect on most developing countries, where mechanization continues to be the predominant form of automation. Despite the hype surrounding the potential of robotbased automation, today the use of industrial robots globally remains quite small and amounts to less than 2 million units.¹⁴ Industrial robots are concentrated in the automotive, electrical, and electronics industries, and only then in a small number of countries.¹⁵ Deindustrialization and automation need not to be viewed as worrisome phenomena. The appropriate response to the perceived or real threats that they pose is for developing countries to implement digital industrial policies to ensure that robotics supports—rather than threatens—inclusive development. This will require that countries at various levels of development constantly design and implement proactive upgrading strategies that are suited to their evolving endowments and comparative advantages. * * * ¹⁴ Of the 1.63 million industrial robots in operation worldwide in 2015, only 1,580 were in textiles, apparel, and leather. Of all the industrial robots shipped that year, a third ended up in middle-income countries. Source: International Federation of Robotics (https://ifr.org). ¹⁵ Almost half of operational industrial robots are in Germany, Japan, and the United States of America, but China has quadrupled its robot stock since 2010, and the Republic of Korea has the highest number of robots per worker globally (UNCTAD 2017).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



The importance of structural transformation as a process for generating prosperity and as the mechanics for improving the quality of lives around the world cannot be underestimated: since World War II, only two economies out of more than 200 have moved from low-income to high-income status: South Korea and Taiwan, China. Few countries have experienced economic convergence on a sustained basis. One approach to measuring relative progress is to look at per capita GDP relative to the United States, which has been the symbol of advanced industrialized countries since World War II. Persistently, over 80 per cent of the countries in the world have GDP per capita levels that are half or less than half of the level in the United States. There has also been some ‘churning’, with countries not only converging up the ladder, but also diverging down the ladder. This is the case of some former colonies in Africa: many have gone from being lower middle-income (MIC) economies at independence to lowincome in the 1980s. Since then, some have climbed back up to MIC status. Even some natural resource rich countries failed to diversify their economic base and, as a result, have experienced large declines in their relative income per capita. There are also countries at the high-income end of the distribution that have fallen back to MIC status, by this measure—most notably Argentina. Historical data on longrun growth compiled by Angus Maddison also suggests that many countries have remained stuck in the so-called middle-income trap: Russia, for instance, remained there for some 200 years. And in the most dynamic current middle-income countries (Vietnam, Thailand, Indonesia, Brazil, Peru, Mexico, Mauritius, South Africa, Botswana, etc.), there is the constant fear that they may not necessarily climb to high-income status and remain there. In sum, only a handful of developing countries have succeeded in reaching high levels of prosperity, and many of them are in Western Europe. The few developing economy success stories—with the exception of a few small oil rich countries—are generally located in East Asia and achieved rapid industrialization by following comparative advantage export dynamism, helped by the facilitating role of the state. This small group of exceptional examples of catching-up has not been studied systematically. Yet, their unique experience and the recent rise of the multi-polar growth world deserve scrutiny, especially with the adoption of the Sustainable Development Goals by the international community. Failure to understand the mechanics and policies for structural transformation is costly, not just for individual economies but for the world, as poverty is often associated with instability, conflicts, and mass migrations. This brings us to this Oxford Handbook of Structural Transformation. The existing academic literature on structural change lacks an in-depth, comprehensive compendium in which researchers and policy makers can find both a critical assessment of the major theories of the determinants of structure and its change, and a discussion of strategies, policies, and country experiences (best and worst practices) from which to learn. This volume tries to fill that gap. The Handbook is organized as follows: Part I discusses theories and frameworks of structural change. Part II focuses on the drivers, channels, and policy instruments for structural transformation. Part III discusses the empirics of structural change while

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

Part IV sheds lights on a number of country and regional experiences. Part V provides some concluding thoughts. The contributors have tried to strike a balance between academic literature and policy issues, between economic theory and practice, and between past knowledge and the unexplored ideas that remain for future research.

R Acemoglu, D., S. Johnson, and J. A. Robinson, 2001. ‘The Colonial Origins of Comparative Development: An Empirical Investigation’, The American Economic Review, 91 (5), pp. 1369–1401. Akerlof, G. A., 1970. ‘The Market for ‘Lemons’: Quality Uncertainty and the Market Mechanism’, Quarterly Journal of Economics, 84, pp. 488–500. Azariadis, Costas and John Stachurski, 2005. ‘Poverty Traps’, in Phillipe Aghion and Steven Durlauf, eds, The Handbook of Economic Growth, Amsterdam: Elsevier. Bagchi, A.K., 1990. ‘Industrialization’, The New Palgrave: Economic Development, New York: W.W. Norton & Co., pp. 160–73. Balassa, Bela, 1966. ‘Tariff Reduction and Trade in Manufactures among the Industrial Countries’, American Economic Review, June 1966, pp. 466–73. Basu, K., 2010. Beyond the Invisible Hand: Groundwork for a New Economics, Princeton, NJ: Princeton University Press. Becker, Gary S., 1992. ‘The Economic Way of Looking at Life’ (Nobel Prize Lecture). Chenery, Hollis B. and Michael Bruno, 1962. ‘Development Alternatives in an Open Economy: The Case of Israel’, Economic Journal, 72, pp. 79–103. Coyle, D., 2014. GDP: A Brief but Affectionate History, Princeton, NJ: Princeton University Press. Datta, Bhabatosh, 1952. Economics of Industrialization, first edition, Calcutta: World Press. David, P. A., 1990. ‘The Dynamo and The Computer: An Historical Perspective on the Modern Productivity Paradox’, American Economic Review, 80 (2), AEA Papers and Proceedings, May, pp. 355–61. Davies, G., 2013. ‘Why World Trade Growth Has Lost Its Mojo’, Financial Times, 29 September. Duarte, M. and D. Restuccia, 2016. Relative Prices and Sectoral Productivity, Working Paper No. 555, University of Toronto Department of Economics, 5 February. Dutt, Amitava Krishna and Jaime Ross (eds), 2003. Development Economics and Structuralist Macroeconomics: Essays in Honor of Lance Taylor, Cheltenham: Edward Elgar Publishing. Economist, 2014. ‘The Tragedy of Argentina: A Century of Decline’, 17 February. Edgeworth, F.Y., 1881. Mathematical Psychics: An Essay on the Application of Mathematics to the Moral Sciences, London. Enache, M., E. Ghani, and S. D. O’Connell, 2016. ‘Structural Transformation in Africa: A Historical View’. Policy Research Working Paper No. 7743. Washington, DC: World Bank. Fardoust, Shahrokh, 2006. ‘Predicting the Future, Looking Backward’, Background paper for the 2007 Global Economic Prospects, mimeo, Washington, DC: The World Bank. Furtado, Celso, 1964. Development and Underdevelopment, Berkeley and Los Angeles: University of California Press. Gates, H. L., 1988. The Signifying Monkey: A Theory of African American Literary Criticism, New York: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



Ghani, E. and S. D. O’Connell, 2014. ‘Can Service Be a Growth Escalator in Low Income Countries?’, Policy Research Working Paper No. 6971. Washington, DC: World Bank. Greenwald, B. and J. E. Stiglitz, 1986. ‘Externalities in Economies with Imperfect Information and Incomplete Markets’, Quarterly Journal of Economics, 101 (2), May, pp. 229–64. Greenwald, B. and J. E. Stiglitz, 2013. ‘Industrial Policies, the Creation of a Learning Society, and Economic Development’, in J. E. Stiglitz and J. Y. Lin, eds, The Industrial Policy Revolution I: The Role of Government Beyond Ideology, New York: Palgrave Macmillan. Greif, Avner, 1993. ‘Contract Enforceability and Economic Institutions in Early Trade: The Maghribi Traders’ Coalition’, The American Economic Review, 83 (3), pp. 525–48. Haraguchi, N., C. F. Chin Cheng, and E. Smeets, 2017. ‘The Importance of Manufacturing in Economic Development: Has this Changed?’ World Development, 93, May, pp. 293–315. Harrison, A., and A. Rodriguez-Clare, 2009. Trade, Foreign Investment, and Industrial Policy, Cambridge: NBER Paper No. 15261. Heckman, James J., 2006. ‘Skill Formation and the Economics of Investing in Disadvantaged Children’, Science, 312 (5782), pp. 1900–2. Herrendorf, B., R. Rogerson, and A. Valentinyi, 2014. ‘Growth and Structural Transformation’, in P. Aghion and S. N. Durlauf, eds, Handbook of Economic Growth, vol. 2, Amsterdam: Elsevier, pp. 855–941. Jorgenson, D. W. and M. P. Timmer, 2011. ‘Structural Change in Advanced Nations: A New Set of Stylised Facts’, Scandinavian Journal of Economics, 113, pp. 1–29. Keynes, J. M., 1936. The General Theory of Employment, Interest, and Money, London: Macmillan. Krugman, P., 1995. Development, Geography, and Economic Theory, Cambridge, MA: MIT Press. Krugman, Paul R., 2008. ‘Trade and Wages, Reconsidered’, Brookings Papers on Economic Activity, Spring, pp. 103–37. Kuznets, S., 1966. Modern Economic Growth: Rate, Structure and Spread, New Haven, CT: Yale University Press. Kuznets, S., 1979. ‘Growth and Structural Shifts’, in W. Galenson, ed., Economic Growth and Structural Change in Taiwan: The Postwar Experience of the Republic of China, London: Cornell University Press. Lewis, W. A., 1954. ‘Economic Development with Unlimited Supplies of Labour’, The Manchester School, 22 (2), pp. 139–91. Lin, J. Y., 1995. ‘The Needham Puzzle: Why the Industrial Revolution Did Not Originate in China’, Economic Development and Cultural Change, 41 (January), pp. 269–92. Lin, J. Y., 2011. From Flying Geese to Leading Dragons: New Opportunities and Strategies for Structural Transformation in Developing Countries, WIDER Annual Lecture 15, Helsinki: UNU-WIDER. Lin, J. Y. 2012a. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: World Bank. Lin, J. Y. 2012b. The Quest for Prosperity: How Developing Economies Can Take Off, Princeton, NJ: Princeton University Press. Lin, J. Y., 2014. ‘Structural Change in Africa’, in E. Aryeetey, S. Devarajan, R. Kanbur, and L. Kasekende, eds, The Oxford Companion to the Economics of Africa, New York: Oxford University Press. Lin, J. Y., and C. Monga, 2011. ‘Growth Identification and Facilitation: The Role of the State in the Dynamics of Structural Change’, Development Policy Review, 29 (3), pp. 264–90.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

Lin, J. Y. and C. Monga, 2013. ‘The Evolving Paradigms of Structural Change’, in Bruce Currie-Adler, Ravi Kanbur, David Malone, and Rohinton Medhora, eds, International Development: Ideas, Experience, and Prospects, New York: Oxford University Press, 277–94. Lin, J. Y. and C. Monga, 2017. Beating the Odds: Jump-Starting Developing Countries, Princeton, NJ: Princeton University Press. Lin, J. Y. and D. Rosenblatt, 2012. Shifting Patterns of Economic Growth and Rethinking Development, Policy Research Working Paper No. 6040, Washington, DC: World Bank. Lucas, Robert E., 2004. ‘Life Earnings and Rural-Urban Migration’, Journal of Political Economy, 112 (1), pp. S29–S59. Maddison, A., 1982. Phases of Capitalist Development, Oxford: Oxford University Press. McCloskey, D. N., 2006. The Bourgeois Virtues: Ethics for an Age of Commerce, Chicago: University of Chicago Press. McCloskey, D. N., 2010. Bourgeois Dignity: Why Economics Can’t Explain the Modern World, Chicago: University of Chicago Press. McCloskey, D. N., 2016. Bourgeois Equality: How Ideas, Not Capital or Institutions, Enriched the World, Chicago: University of Chicago Press. McKinsey Global Institute, 2017. Jobs Lost, Jobs Gained: Workforce Transitions in a Time of Automation, New York: McKinsey Global Institute. McMillan, M. and D. Rodrik, 2011. ‘Globalization, Structural Change and Productivity Growth’, in M. Bacchetta and M. Jense, eds, Making Globalization Socially Sustainable, Geneva: International Labour Organization and World Trade Organization, pp. 49–84. Mokyr, J., 2016. A Culture of Growth: The Origins of the Modern Economy, Princeton, NJ: Princeton University Press. Monga, Célestin, 2011. ‘Post-macroeconomics: Reflections on the Crisis and Strategic Directions Ahead’, Journal of International Commerce, Economics and Policy, 2 (2), pp. 277–304. Monga, Célestin, 2014. ‘The False Economics of Pre-Conditions: Policymaking in the African Context’, Journal of African Development, 16 (2), pp. 121–40. Myrdal, G., 1957. Economic Theory and Under-developed Regions, London: G. Duckworth. Ngai, L. R. and C. A. Pissarides, 2007. ‘Structural Change in a Multisector Model of Growth’, The American Economic Review, 97 (1), pp. 429–43. North, D. C., 1981. Structure and Change in Economic History, New York: W. W. Norton & Co. Nurkse, R., 1956. ‘The Relation between Home Investment and External Balance in the Light of British Experience, 1945–1955’, The Review of Economics and Statistics, 38 (2), pp. 121–54. Ocampo, José António and Lance Taylor, 1998. ‘Trade Liberalisation in Developing Countries: Modest Benefits but Problems with Productivity Growth, Macro Prices and Income Distribution’, The Economics Journal, 108 (September), pp. 1523–46. Ocampo, J. A., Codrina Rada, and Lance Taylor, 2009. Growth and Policy in Developing Countries: A Structuralist Approach, New York: Columbia University Press. Prebisch, Raúl, 1959. ‘Commercial Policy in the Underdeveloped Countries’, American Economic Review, 49, pp. 251–73. Pritchett, L., 1997. ‘Divergence, Big Time’, The Journal of Economic Perspectives, 11 (3), pp. 3–17. Pritchett, L., 2006. ‘Does Learning to Add up Add up? The Returns to Schooling in Aggregate Data’, in Eric A., Hanushek and F. Welsch eds, Handbook of the Economics of Education, vol. 1, Amsterdam: North Holland, pp. 635–95.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:     



Rodrik, D., 2007. One Economics, Many Recipes: Globalization, Institutions, and Economic Growth, Princeton, NJ: Princeton University Press. Rodrik, D., 2008. ‘The Real Exchange Rate and Economic Growth’, Brookings Papers on Economic Activity, 39 (2), pp. 365–439. Rodrik, D., 2016. ‘Premature Deindustrialization’, Journal of Economic Growth, 21 (1), pp. 1–33. Rogerson, R., 2008. ‘Structural Transformation and the Deterioration of European Labor Markets’, Journal of Political Economy, 116 (2), pp. 235–59. Romer, P. M., 1993a. ‘Idea Gaps and Object Gaps in Economic Development’, Journal of Monetary Economics, 32 (December), pp. 543–73. Romer, P. M., 1993b. ‘Two Strategies for Economic Development: Using Ideas and Producing Ideas’, Proceedings of the World Bank Annual Conference on Development Economics 1992, Washington DC: World Bank, pp. 63–91. Rosenstein-Rodan, P. N., 1943. ‘Problems of Industrialization of Eastern and South-Eastern Europe’, Economic Journal, 53 (June–September), pp. 202–11. Rowthorn, R. and R. Ramaswamy, 1997. ‘Deindustrialization: Causes and Implications’, Working Paper No. 97/42, Washington, DC: IMF. Schwieters, N. and B. Moritz, 2017. ‘10 Principles for Leading the Next Industrial Revolution’, strategy+business, Autumn 2017 / Issue 88. Singer, H. W., 1950. ‘Distribution of Gains between Investing and Borrowing Countries’, American Economic Review, Papers and Proceedings, 40 (May), pp. 473–85. Smith, Adam, [1776]. An Inquiry into the Nature and Causes of the Wealth of Nations, 2 vols., Chicago: University of Chicago Press. Stiglitz, J. E., 1997. ‘The Role of Government in the Economies of Developing Countries’, in E. Malinvaud and A. K. Sen, eds, Development Strategy and the Management of the Market Economy, Oxford: Clarendon Press, pp. 61–109. Stiglitz, J. E., 2001. ‘Information and the Change in the Paradigm in Economics’, Nobel Lecture, 8 December. Stiglitz, J. E. and B. C. Greenwald, 2014. Creating a Learning Society: A New Approach to Growth, Development, and Social Progress. Kenneth J. Arrow Lecture Series. New York: Columbia University Press. Syrquin, M., 1988. ‘Patterns of Structural Change’, in H. Chenery and T. N. Srinivasan, eds, Handbook of Development Economics, vol. 1, Amsterdam: Elsevier Science Publishers, pp. 203–73. Taylor, Lance Jerome, 1983. Structuralist Macroeconomics: Applicable Models for the Third World, New York: Basic Books. Taylor, Lance Jerome, 1991. Income Distribution, Inflation, and Growth, Cambridge MA: MIT Press. Taylor, Lance Jerome, 2004. Reconstructing Macroeconomics: Structuralist Proposals and Critiques of the Mainstream, Cambridge, MA: Harvard University Press. Timmer, C. P., 2014. Managing Structural Transformation Post-2015, WIDER Lecture No. 18, Helsinki: UNU-WIDER. Timmer, C. P. and S. Akkus, 2008. ‘The Structural Transformation as a Pathway out of Poverty: Analytics, Empirics and Politics’, Working Paper No. 150, Washington, DC: Center for Global Development.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ      

UNCTAD, 2017. Trade and Development Report: Beyond Austerity: Towards a Global New Deal, Geneva: UNCTAD. UNIDO, 2009. Industrial Development Report 2009: Breaking In and Moving Up: New Industrial Challenges for the Bottom Billion and the Middle-Income Countries, New York: United Nations. World Bank, 2005. Economic Growth in the 1990s: Learning from a Decade of Reform, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

P A R T I .............................................................................................................

THEORIES AND FRAMEWORKS OF STRUCTURAL CHANGE .............................................................................................................

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

               ,          ,      ......................................................................................................................

 . 

E are always changing, and one of the virtues of the market economy is its ability to adapt to these changes. Primitive agricultural economies face weather variability. Manufacturing economies are marked by new products. Rivals have to constantly adapt to the changing competitive landscape. But beyond these changes, there are a few major structural changes, large changes that occur very infrequently. The movement from feudalism to the post-feudal era was such a change. The industrial revolution was another—but even after the onset of the industrial revolution, the economy remained largely agrarian. It was not until years later that the structural change occurred—the move from a rural agrarian economy to an urban manufacturing society. That was a traumatic event.¹ Markets do not handle such changes well, nor typically do the political processes governing markets. The purpose of this chapter is (a) to explain why it is that markets on their own manage these transitions so poorly; (b) to show that when the structural transformation is not managed well, there may be a prolonged economic downturn (recession and depression), arguing that this is at least part of the explanation for the Great Depression and the Great Recession; (c) to demonstrate what government can do to help manage these structural transformations, and in particular, the role of industrial policy in facilitating these transitions; and (d) to set these industrial policies as a critical part of Keynesian counter-cyclical policies. We set much of our discussion in the context of the last major structural transformation confronting the advanced countries, the transition from an agrarian economy to one based on manufacturing, ¹ Described so forcefully by Polanyi (1944).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

because we can see the principles better from the perspective of 80 years. But in the last two sections of this chapter, we discuss their implications for the 2008 recession and the broader management of cyclical fluctuations.

1.1 T F  M  P  M S T

.................................................................................................................................. The reason for the failure of markets and politics to manage structural transformations is simple. The economic and political structures are designed for stability, including the maintenance of existing power relationships. The system is good at handling small shocks, but does not adapt well when managing big changes. In the economic sphere, big changes lead to large (and typically unanticipated) changes in asset values. In the transition to manufacturing, as farmers migrated out of the rural sector, the assets owned by farmers (in particular, their homes) decreased in value. Their human capital was even more affected: farmers were well attuned to the nuances of weather, disease, etc. in their locale as it related to the production of the particular crops in which they specialized, but those skills were largely unrelated to the skills required for manufacturing. Manufacturing occurs largely in urban centres (and for good reasons). The move from agriculture to manufacturing thus also required a massive change in the structure of housing. In a decentralized market economy, the individual is typically responsible for obtaining the human capital that he requires to be productive, beyond his basic education. Individuals are also responsible for relocation costs, including those associated with the purchase of a new home.² In short, moving from the old sectors to the new sectors is difficult and requires resources. Structural transformation requires upfront capital expenditure. Large numbers of individuals who should be making the transition do not have the resources to finance this transformation; and given the imperfections of capital markets—which can be explained in terms of imperfections and asymmetries of information—they cannot obtain financing. Indeed, the structural transformation itself makes it even more difficult to obtain the financing. Banks that have invested substantial amounts in the rural sector (in the ‘old’ sector)—that is, have lent ² There are good reasons for owner-occupancy; nonetheless, some market economies rely on rental housing, so that the provision of housing depends on specialized private enterprises. Here, the problem is that the net worth of these enterprises may suffer significant adverse effects in the process of structural transformation, and thus may not be able to provide the new housing required in the urban area. In this case, however, it is more likely that new enterprises will be created to provide housing for new migrants to the urban sector.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ,  



substantial amounts to that sector—also experience significant losses. The fact that the value of housing in the rural sector has diminished implies that a fraction of the loans will go into default, perhaps a large fraction. So, too, for loans made to finance other investments in the rural (old) sector. Thus, the net worth of banks will experience a negative shock. And this will reduce their ability and willingness to lend.³ In addition, these local institutions have the detailed information that allows them to judge the creditworthiness of the borrower. But the structural transformation has attenuated even the value of that information, since knowing an individual’s competence in the rural sector might provide only limited information about his performance in the urban sector. Moreover, moving is risky: there are clear up-front costs, with large uncertainties about the returns. Will the individual find a good job, an adequate home, a community in which the family thrives? There are no markets to which individuals can turn to obtain insurance against these risks. Using a model with costless mobility of resources to analyse such large changes in technology (and preferences) would suggest that a new equilibrium would emerge as the overall economy changed and adjusted.⁴ But these changes do not, in reality, occur on their own. We will explain shortly how government intervention can facilitate the transition. But even government attempts at facilitating transition encounter difficulties. Political institutions tend to reflect existing power structures. And these existing power structures derive their power, at least to some extent, from their existing economic power. The structural transformations under discussion undermine those power relations. Thus, rather than facilitating the transition, too often government lends its weight in the other direction, trying to preserve the status quo and the power structures associated with it. Nowhere is that more evident than in the example that is the focus of the discussion in this chapter—the movement from agrarian economies to manufacturing. The political institutions created in the nineteenth century gave undue weight to the rural agrarian communities, and the mindset of these communities was often at odds with that of the dominant urban communities, and remains so today. While this disparity between politics and the underlying economic realities is stark, the disparity is even greater when it comes to the movement now underway in postindustrial societies, to the service- and knowledge-based economies of the twenty-first century. A political system that gives disproportionate weight to a country’s rural and dying manufacturing regions may try to preserve their bygone industries and sectors and, rather than facilitating that transition, put significant hurdles in the path of the efficient transition from the ‘old’ economy to a new post-industrial economy.⁵

³ See Greenwald and Stiglitz (2003). ⁴ There are, in addition, large social costs. Individuals have built up networks of relationships that are not only a direct source of ‘utility’ but also provide strong systems of social support. ⁵ An interesting aspect of manufacturing in the United States today is that much of it has moved out of the urban areas to more rural locations. Low wages, low costs of land, and a good transportation system reversed the earlier advantages of urban locations.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

1.2 I   G D

.................................................................................................................................. The Great Depression provides a good illustration of the principles just discussed, and the consequences of impediments to an easy transition. The underlying ‘shock’ to the economy that led to the downturn was an increase in agricultural productivity.⁶ In the absence of frictions, this would have moved the utilities possibilities curve outward; that is, assuming that lump-sum redistributions were possible, everyone could have been made better off. (Without government intervention, the effects are ambiguous, since the competitive equilibrium could entail some group being worse off. This is the case with Hicksian labour-saving technological changes.) In this hypothetical new equilibrium, workers would have migrated from the rural sector to the urban sector simply because fewer workers are needed to produce the food required, since the income elasticity and price elasticity of food is low. The technological change leads to lower prices of agricultural goods, and this results in slightly higher demand—an increase in demand that is smaller than the increase in productivity. Hence, incomes in the rural sector decline. Absent frictions, workers would migrate from the rural to the urban sector. With the real frictions described in the previous section, however, this may not be the case. Assume for the moment that mobility is zero. Then those in agriculture will see their incomes go down, and as a result they will work harder (assuming income effects offset substitution effects) and this will lead to further decreases in agricultural prices and incomes. (Each farmer believes that by working harder, his income will increase, but, because of the inelasticity of demand, when output increases incomes actually fall.) Those in the urban sector are better off—at first. But with farmers

⁶ We do not present the evidence for this claim here. Note, however, that there was a drop in farmers’ income of some 50 per cent to 75 per cent, and that the rural sector represented some 70 per cent of the economy at the time. With reasonable multipliers, it is easy to see how this could translate into a macroeconomic downturn of the magnitude observed. There is a long-standing debate about the relative importance of different factors in contributing to the Great Depression, with some economists (e.g. Eichengreen 1992) emphasizing the role of the gold standard and others (e.g. Friedman and Schwartz 1963) that of monetary policy. Both of these clearly played a role, especially in the propagation and persistence of the downturn. We emphasize here, however, the role of the ‘productivity shock’ in agriculture as the source of the perturbation to the economy. The gold standard did introduce rigidities, making adjustment to the shock more difficult. It is often noted that countries that went off the gold standard performed better. But this says nothing about what would have happened if all had gone off the gold standard. Countries going off the gold standard early clearly had a competitive advantage over those that waited. A discussion of the role of monetary policy as a cause of the crisis would take us beyond this short chapter. Here, we simply note that the financial crisis occurred years after the onset of the Depression. Any deep and prolonged downturn will give rise to a financial crisis.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ,  



demanding fewer tractors and cars and other manufactured goods, the demand for urban goods decreases. Assume, again for simplicity, that wages are fixed—say at the efficiency wage. Then employment in the urban sector falls, leading to a decrease in demand for food, further depressing the price of food. The equilibrium that emerges entails lower food prices and lower urban employment—in both sectors, workers are worse off.⁷ What should have been an innovation that made everyone better off—if the structural transformation could have been efficiently carried out—actually leads to immiseration, with welfare in both the rural and urban sectors decreased.⁸ It is interesting that President Franklin D. Roosevelt’s first response to the Great Depression (embodied in the Agricultural Adjustment Act of 1933) was to restrict agricultural production. This would have increased incomes in both the rural and urban sectors. (Rural welfare would have been increased because of the higher prices they receive for the goods they sell.⁹) The 1933 law was struck down by the Supreme Court, and widely criticized by economists as an intervention in the workings of the market economy. But it was, in fact, a clever application of the principle of the second best. Given the market distortion (the inability of labour to move costlessly across sectors, and the inability to engage in lump-sum redistributions) such interventions may in fact have been desirable. In the end, it was World War II that brought the US economy out of the Great Depression. The demand for munitions and armaments and troops required moving people out of the rural sector, and training individuals for a manufacturing society helped in the transition. After the war, the GI bill, which provided a university education to all of those who had fought in the war (which was almost all men and many women, though it discriminated against African-Americans), provided the human capital needed for the transformation from an agrarian economy to a manufacturing economy. The forced savings during the war and deferred consumption helped to provide the basis of strong aggregate demand, substituting for government military expenditures which diminished rapidly after the war, thus averting the widely expected post-war recession. In short, war expenditures were more than a Keynesian stimulus; they constituted (unknowingly) an industrial policy, critical in engineering a structural transformation.

⁷ The lower wages increase the income of the owners of capital. So long as the marginal propensity to consume of these capitalists is lower than that of workers, the results described here hold. The adverse welfare effects hold so long as the marginal social utility of a dollar to the (higher-income) capitalists is lower than that to workers. Of course, workers who do retain jobs at the efficiency wage are better off, because food prices are lower, provided that the wage (denominated in the price of manufactured goods) does not change. ⁸ Formal models showing what has been discussed in the previous section are provided in Delli Gati et al. (2012a and 2012b). ⁹ We can express the welfare of farmers through an indirect utility function depending just on the price of agricultural goods relative to urban goods.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

1.3 I   2008 C

.................................................................................................................................. The 2008 crisis is often thought of as a financial crisis—and clearly, as part of the crisis, many financial institutions were close to collapsing. But to understand the crisis itself and what could and should have been done to help the economy emerge from it, we must dig deeper. By analogy to the Great Depression, one can think of globalization and the increases in productivity in manufacturing as the underlying drivers of the Great Recession. The growth in productivity in manufacturing exceeded the growth in demand for manufactured goods, and that meant that globally there had to be a decrease in manufacturing employment. Globalization meant that the advanced countries seized a diminishing share of this diminishing amount. But that in turn meant that these workers had to find employment elsewhere. One way of thinking of the real estate bubble in the United States was that it was one way that the country temporarily solved the problem. It provided jobs for the men who had lost jobs in manufacturing. So, too, the Federal Reserve’s low interest rates, which fed the housing bubble, were a reflection of the weak aggregate demand resulting from the underlying weakness in manufacturing. There were, of course, other ways by which the economy/society could have responded. There could have been an increase in fiscal expenditure, but the country’s political economy precluded that—the party controlling the government at the time was committed to downsizing government. Monetary policy also led to a real estate bubble, in part because of the prevailing ideology against regulations that might have circumscribed the growth of the bubble. Of course, the collapse of the financial sector amplified the downturn. Given that the obvious symptom of the crisis was the collapse of Lehman Brothers, which posed a real threat of the collapse of the entire financial sector, it was natural to refer to the crisis as a financial crisis. That led to a focus on the financial sector, including its recapitalization. But years later, when the banks were largely recapitalized, the downturn continued, suggesting at least that the downturn was not just a financial crisis. Some have said it was a ‘balance-sheet recession’. Essentially all major downturns are balance-sheet downturns, because the downturn leads to a weakening of the balance sheets of firms as well as banks, and this leads to a contraction both of production (in effect, a shift of the aggregate supply curve to the left) and demand—a shift in the demand curve for investment. There is nothing distinctive in this matter for the 2008 crisis.¹⁰ And again, by a few years after 2008, balance sheets were largely restored: large corporations were sitting on a couple of trillion dollars of cash. It was not balance sheets that were constraining investment, but aggregate demand. The question was, what was constraining aggregate demand?

¹⁰ For an analysis of balance-sheet recessions, see Greenwald and Stiglitz (1993a, 1993b, 2003) and Koo (2003).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ,  



Our analysis suggests it was the failure to advance on the necessary structural transformation of the United States from a manufacturing economy to a service sector economy. Just as farmers were ‘trapped’ in the agricultural sector, unable to move to the manufacturing sector, manufacturing workers are now also trapped, lacking the skills that would enable them to be productive in the expanding sectors of the economy and unwilling and unable to make the investments that would give them those skills and make it possible to move to the locations where the jobs were being created.¹¹ Again, Keynesian policies could have filled the void in aggregate demand, but because of the prevailing ideology government not only didn’t expand government spending to fill it, in many places, contracting spending (austerity) actually exacerbated the problem. The growth of public-sector employment fell short of what would have been expected on the basis of the growth of the working-age population. What was needed, though, was more. As in World War II, government was needed to push forward the structural transformation, with industrial policies supporting the new sectors, and retraining policies (active labour market policies) helping to move people from the old sectors to the new. But there was a further need for government: many of the sectors into which the economy was shifting were service sectors in which government naturally played a pivotal role, such as education, health, and care for the aged. Without government support, these sectors were constrained, and so as the manufacturing sector declined, the new sectors where workers might naturally have found employment did not grow.

1.4 R  I P  C P

.................................................................................................................................. The two previous sections have emphasized the market failures that emerge in a structural transformation.¹²,¹³ There is a natural role for government in correcting ¹¹ There are a host of other impediments to mobility, including the reluctance to leave one’s extended family and support systems. The absence of a good rental market for housing impedes mobility, as does the lack of affordable daycare, in those instances where members of the extended family provide such services. ¹² Industrial policies include any policies that help direct resources to or from a sector or encourage the adoption of a particular technology within a sector. They are not limited to the promotion of ‘industry’, as that term is usually understood. As I have noted elsewhere, all countries have industrial policies, hidden in the tax code or various aspects of the legal code. Markets don’t exist in a vacuum; they have to be structured and, inevitably, how they are structured affects resource allocations. Of course, government interventions in resource allocations become more compelling when there is a market failure—as here, a failure in the free mobility of labour. See Rodrik (2004); Greenwald and Stiglitz (2013); and Stiglitz (2017). ¹³ In December, 2017, and January 2018, the US almost doubled its fiscal deficit as a percentage of GDP (from nearly 3% to nearly 6%), and this large fiscal stimulus enabled the unemployment rate to fall to 3.7% (though the employment rate remained significantly below pre-recession levels), showing that if such a fiscal stimulus had been provided earlier, it would have assisted the recovery. Still, the underlying

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

these market failures. In particular, Keynesian policies can stimulate the economy. In the case of the transformation from agriculture to industry, such policies increased incomes in both the agricultural and urban sectors. The increased incomes by themselves would have facilitated movement from the rural to the urban sector, thus partially addressing one of the key market failures. But the expenditures during and after World War II were even more effective in enabling this transformation; unintentionally, they were industrial policies, helping people move and giving them the training and education needed to equip them to be productive in the expanding sectors of the economy. There is a general principle here. When the underlying problem facing an economy is the necessity of a major economic transformation, a key component of Keynesian cyclical policies should be industrial policies to facilitate that transformation. Managing such policies is, however, not always easy. When a sector is facing competition from outside the country, those already in the sector will often claim either that there is unfair competition from abroad and/or that the problems are only temporary, and a little help will enable the industry to recover and thrive. It is better to provide short-term support to the industry, it is argued, than to relocate the workers and see a long-term loss in human and organizational capital, which would result from the closure of enterprises. Both workers and firms have a self-interest in taking such a stance. Even a successful relocation of workers may be associated with a significant lowering of wages. Older workers trained for one sector may, even with retraining, be less productive in the new sector. There are several issues that have to be addressed when evaluating the best responses to foreign competition. For one, it must be considered whether or not the problem is actually temporary. Often, it is not. Comparative advantages do change. The United States almost surely does not have a comparative advantage in the production of cars. Germany, Japan, Korea, and China seem to have comparative advantages in different parts of the product spectrum, with Germany having a comparative advantage in high-tech cars, and Korea and Japan in more mass-produced cars. The US comparative advantage in large gas-guzzlers is not the basis of a successful twenty-first century automobile industry. The US car companies have returned to profitability, but only by lowering the wages of their workers to levels that are close to what should be the minimum wage in the United States. As noted above, both the firms and the workers in the declining industry have an incentive to claim that the industry is just facing temporary difficulties, and a little help—if not outright subsidies, then a little protection—would do the trick to get things back on track. Still, while inevitably politically contentious, it’s often possible to make an informed judgment with some degree of confidence whether an industry is facing temporary difficulties or has lost its long-term comparative advantage. America’s coal

transformational problems have been largely unaddressed. It is anticipated that the impact of the fiscal stimulus will wane, and it is unlikely that there will be another fiscal stimulus of this magnitude any time soon.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ,  



industry, for instance, is and should be a dying industry; it should not be the recipient of either direct or indirect assistance. Second, domestic firms are always going to claim that competition from outside— when it is successful—is unfair. They have to believe that they are more productive than firms elsewhere, so that if competition were fair, they would prevail. But the reality is often otherwise. It is not ‘unfair’ for a country to be poor and have low wages; it is unfortunate. The principle of comparative advantage says that even low-wage countries have comparative advantages and disadvantages. Again, of course, every firm complains about hidden subsidies. Those abroad complain about US government bailouts of American auto companies. Indeed, any firm that borrows from an American bank is a beneficiary of the hundreds of billions of dollars that went to the financial sector in the bailout. Those outside the United States claim that that gives American firms an unfair advantage. US monetary policy, which keeps interest rates at near zero, is also seen as giving American firms an unfair source of cheap capital. But there are cyclical fluctuations, and these fluctuations affect some industries more than others—the cyclically sensitive sectors. As a matter of policy, it would be best if monetary and fiscal policy stabilized the economy. It would also be good if the government sold state-contingent insurance, that is, insurance that paid off in the event of a cyclical downturn. Industrial policies, however, can be an important complement to these monetary and fiscal policies; and take on even greater importance as second-best measures when the government fails to fully and effectively implement them. Industrial policies can simultaneously help the industry weather the storm and restructure itself (sometimes downsizing, sometimes increasing its scale). Because of capital market imperfections, an economic downturn, especially in a capital-intensive industry with economies of scale, can lead to large (cash flow) losses, which inhibit its ability to modernize and compete. Roosevelt’s Agriculture Adjustment Act, as we noted, is an example of such a policy. So, too, are state-contingent tariffs, which increase tariffs or reduce quotas in a recession, so that producer prices are increased, reducing the losses confronting cyclically sensitive industries. Persistent unemployment is, in a sense, a symptom of the market economy not working well. We have also described how government interventions can, in such circumstances, provide symptomatic relief. While there may be policies that go more directly to the root of the problem, if they are politically unattainable then it is better to intervene with second- or third-best measures than to let the economy suffer from prolonged unemployment.

1.5 C R

.................................................................................................................................. Markets typically do not work as well as the textbook models of perfect markets suggest. At times, market failures become very significant, and government intervention is required. This chapter has discussed one such instance—the structural

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

transformation of an economy. We have explained why market failures are likely to be particularly significant when the economy is going through a major structural transformation, and described some of the market interventions that might be desirable. In particular, we have argued that Keynesian industrial policies can play an important role in simultaneously stimulating the economy, reducing unemployment, and facilitating the required transition.

A I wish to acknowledge the financial support of INET, research assistance of Matthieu Teachout and editorial assistance of Debarati Ghosh.

R Delli Gati, D., M. Gallegati, B. Greenwald, A. Russo, and J. E. Stiglitz, 2012a. ‘Mobility Constraints, Productivity Trends and Extended Crisis’, Journal of Economic Behavior and Organization, 83 (3). Delli Gatti, D., M. Gallegati, B. Greenwald, A. Russo, and J. E. Stiglitz, 2012b. ‘Sectoral Imbalances and Long-run Crises’, in F. Allen, M. Aoki, J.-P. Fitoussi, N. Kiyotaki, R. Gordon, and J. E. Stiglitz, eds, The Global Macro Economy and Finance, Basingstoke: Palgrave Macmillan UK, pp. 61–97. Eichengreen, B., 1992. Golden Fetters: The Gold Standard and the Great Depression, 1919–1939, Oxford: Oxford University Press. Friedman, M. and A. J. Schwartz, 1963. A Monetary History of the United States, 1867–1960, Princeton, NJ: Princeton University Press. Greenwald, B. and J. E. Stiglitz, 1993a. ‘Financial Market Imperfections and Business Cycles’, Quarterly Journal of Economics, 108 (1), pp. 77–114. Greenwald, B. and J. E. Stiglitz, 1993b. ‘New and Old Keynesians’, Journal of Economic Perspectives, 7 (1), pp. 23–44. Greenwald, B. and J. E. Stiglitz, 2003. Towards a New Paradigm in Monetary Economics, Cambridge: Cambridge University Press. Greenwald, B. and J. E. Stiglitz, 2013. ‘Industrial Policies, the Creation of a Learning Society, and Economic Development’, in J. E. Stiglitz and J. Yifu Lin, eds, The Industrial Policy Revolution I: The Role of Government Beyond Ideology, Houndmills and New York: Palgrave Macmillan, pp. 43–71. Koo, R., 2003. Balance Sheet Recession: Japan’s Struggle with Uncharted Economics and its Global Implications, Oxford: John Wiley & Sons. Polanyi, K. 1944. The Great Transformation, New York: Farrar & Rinehart. Rodrik, D. 2004. ‘Industrial Policy for the Twenty-first Century’, CEPR Discussion Paper No. 4767. Stiglitz, J. E. 2017. ‘Industrial Policy, Learning and Development’, in John Page and Finn Tarp, eds, The Practice of Industrial Policy: Government-Business Coordination in Africa and East Asia, Oxford: Oxford University Press, pp. 23–39.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  ......................................................................................................................

    Theoretical Considerations ......................................................................................................................

 

2.1 I

.................................................................................................................................. T one-sector neo-classical growth model is the dominant framework used by economists to study the growth in output associated with the development process. It also serves as the basis for development accounting exercises—the standard method that economists use to document the proximate causes of cross-sectional differences in living standards across countries. Part of the appeal of this framework is that under reasonably general conditions, it gives rise to a balanced growth path, which is viewed as providing a good description of the long-run behaviour of advanced economies. Along a balanced growth path, economic aggregates such as output, consumption, investment, and the capital stock all grow at the same constant rate, the interest rate and factor shares are constant, and the real wage grows at the same rate as output. These regularities are commonly referred to as the Kaldor facts. In the context of the one-sector growth model, an economy on its balanced growth path experiences growth over time, but the nature of this growth is such that the economy just does more of the same thing as it becomes richer. This feature of a balanced growth path is in sharp contrast to the growth experiences that we observe in actual economies. In his Nobel Prize address, Kuznets included structural transformation—the reallocation of economic activity across broad sectors that accompanies growth—as one of the salient features of growth and development. The onesector growth model necessarily abstracts from this process. Much recent work has argued that abstracting from structural transformation is not merely an issue of descriptive reality; models that explicitly include structural transformation also have important substantive implications for understanding the development process.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

My goal in this chapter is to review efforts to build models that simultaneously generate outcomes that resemble balanced growth at the aggregate level while also generating structural transformation. I will not summarize the empirical evidence on structural transformation; for this I refer the reader to the surveys contained in Syrquin (1988) and Herrendorf et al. (2014). I will break my presentation into two main parts. The first part will focus on expositing the three key economic forces that the literature has identified as the key driving forces behind structural transformation. These forces can be captured most transparently in a simple static model in which growth is captured by a simple comparative statics exercise. The first mechanism emphasizes income effects. As households become richer, expenditure shares will shift if income effects differ across broad categories of consumption. Generating heterogeneous income effects across sectors requires that preferences be non-homothetic. The other two mechanisms both emphasize relative price effects, although they differ in the manner in which they generate relative price effects. One of them emphasizes non-uniform productivity growth across sectors; that is, although development is associated with increases in productivity, this productivity growth may differ across broad sectors. If this is the case, then relative prices will also change. The other mechanism stresses changes in relative factor supplies as a source of relative price changes. Specifically, if development is associated with an increase in the capital to labour ratio and broad sectors differ in their capital intensity, the changing relative price of factors will also lead to changes in the relative prices of sectoral outputs. In either case, the resulting relative price changes can in turn lead to changes in the relative size of sectors. This will happen as long as preferences exhibit a non-unitary elasticity of substitution across broad categories of consumption. The second part will then focus on embedding these forces into standard models of growth and examine the possibility of obtaining balanced growth and structural transformation simultaneously. Embedding these economic mechanisms into a version of the standard growth model with technical progress by extending it to include multiple consumption sectors will naturally lead to a model that generates growth and structural transformation. But obtaining something that looks like balanced growth and structural transformation can be somewhat more challenging, and can depend on the details of how one specifies the general mechanisms just described. I summarize some of the specifications that have been found to successfully achieve this outcome. My analysis in this chapter will focus entirely on closed economy models, as I feel this is the most transparent way to present the key economic forces. Additionally, given that the vast majority of trade has historically been within the manufacturing sector, I think it is reasonable to think that trade has not been the fundamental driving force behind the reallocation of resources across broad sectors. But this does not mean that trade does not produce effects that reinforce some of the forces described above. Indeed, there is a growing literature devoted to understanding how trade interacts with the process of structural transformation and I will briefly comment on it in the concluding section.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



2.2 Q A  D F

.................................................................................................................................. Existing theories of structural transformation rely on three simple and intuitive economic mechanisms. In this section I study these three mechanisms in a simple static setting to best illustrate the basic economic forces at work. In the next section I embed these economic forces into a standard version of the growth model.

2.2.1 A Benchmark Model As a starting point, consider the following simple economy. There is a representative household with preferences over two goods defined by the following utility function: uðc1 ; c2 Þ ¼ αlogðc1 Þ þ ð1  αÞlogðc2 Þ The household is endowed with one unit of time, and as indicated above, does not value leisure. The economy has an endowment K of capital. Each of the two goods can be produced using a constant returns to scale production function, which here I assume to be Cobb-Douglas: ci ¼ Ai kθi hi1θ where hi is the amount of labour devoted to production of good i, ki is the amount of capital devoted to the production of good i, and the Ai are sector specific total factor productivities (TFPs). Note that at this stage we are assuming that the capital shares are the same across the two sectors. Straightforward calculation reveals that the optimal allocation of factors takes a very special form, specifically: h1 ¼ α; h2 ¼ ð1  αÞ; k1 ¼ αK; k2 ¼ ð1  αÞK The above expressions imply that the capital to labour ratios are equalized across the two sectors, implying that they are necessarily equal to the aggregate capital to labour ratio. The key property of this allocation for our purposes is that the sectoral allocation of labour does not depend on either the sectoral TFPs or the level of the capital stock. That is, growth in this economy, due to any combination of technological progress and capital accumulation, will increase total output (and welfare) but leave the allocation of labour across the two sectors unchanged. Note that this result is independent of the nature of technological progress, in the sense that it holds independently of whether technological progress is uniform across sectors or occurs at different rates across the two sectors. Put differently, in this economy, development will not be associated with structural transformation. The absence of structural transformation in this simple example makes

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

it a useful benchmark for understanding the factors that can give rise to structural transformation and in what follows we consider three generalizations of this example that will overturn this result.

2.2.2 Income Effects One property of the preferences in the above example is that they are homothetic, that is, the income elasticity for both goods is equal to unity. A robust and salient pattern found in household expenditure data is that expenditure shares vary systematically with household income, indicating that this property is violated in the data. A simple and tractable way to introduce non-homotheticities into the above example is to generalize preferences to be of the form: uðc1 ; c2 Þ ¼ αlogðc1 
c 1 Þ þ ð1  αÞlogðc2 þ
c 2 Þ where
c 1 and
c 2 are both positive constants. To relate this specification to household income elasticities, consider the household maximization problem: max uðc1 ; c2 Þs:t:p1 c1 þ p2 c2 ¼ I c1 ;c2

Assuming interior solutions for both c₁ and c₂, straightforward calculation gives the following expressions for optimal consumption expenditures: p1 c1 ¼ p1
c 1 þ α½I þ p2
c 2  p1
c 1  p2 c2 ¼ p2
c 2 þ ð1  αÞ½I þ p2
c 2  p1
c 1  It follows that the expenditure share for good 1 is decreasing in income and that the expenditure share for good 2 is increasing in income. Note also that as income becomes large, these effects tend to zero, so that asymptotically, the above commodity demands are identical to those from the previous case. Similar calculations can be carried out to solve for the optimal capital and labour allocations when we adopt this specification of preferences in our earlier example. An optimal allocation will still have the property that capital to labour ratios are equalized across sectors, so that factor allocations can be completely characterized by focusing on the allocation of labour. The Social Planner’s problem can then be written as: max αlogðA1 K θ h1 
c 1 Þ þ ð1  αÞlogðA2 K θ ð1  h1 Þ þ
c 2 Þ h1

Simple calculation using the first order condition for this problem gives: h1 ¼ h2 ¼ 

1 1 1
c 1 þ α½1 þ
c 2 
c 1  θ θ A1 K A2 K A1 K θ

1 1 1
c 2 þ ð1  αÞ½1 þ
c 2 
c 1  θ θ A2 K A2 K A1 K θ

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



From this we obtain: h1  h2 ¼

1 1 2ð1  αÞ
c 1 þ 2α
c 2 þ ð2α  1Þ A1 K θ A2 K θ

The right hand side is decreasing in each of A₁, A₂, and K. It follows that increases in productivity, whether uniform or non-uniform across sectors, and/or increases in capital will lead to a reallocation of labour from sector 1 to sector 2. That is, development driven by increases in productivity and/or capital will now lead to structural transformation. The intuition is straightforward: with non-homothetic preferences, changes in income lead to changes in desired expenditure shares. Increases in productivity and capital generate increases in income, and given the changes in desired expenditure shares, the economy reallocates labour and capital to accommodate the desired changes in expenditure shares. It is important to understand that the particular specification studied in this subsection is but one way to model non-homothetic preferences. The qualitative impact of introducing non-homotheticities is intuitive and so not surprisingly is robust to considering alternative formulations of non-homothetic preferences. But the quantitative implications of non-homotheticities are likely to depend on the exact manner in which they are introduced, an issue we will return to later.

2.2.3 Relative Price Effects I: Uneven Technical Progress Returning to the original example, another feature of the utility function considered was that it assumed a unitary elasticity of substitution between the two goods. This assumption is also inconsistent with micro evidence and so motivates consideration of a more general CES specification for preferences: ρ1

ρ1

ρ

uðc1 ; c2 Þ ¼ ½αc1ρ þ ð1  αÞc2ρ ρ1 where ρ is now the elasticity of substitution between the two goods. From the household perspective, the key feature associated with this generalization is that expenditure shares now respond to relative prices. Note that this utility function is homothetic, implying that increases in income will have no effect on expenditure shares. Simple calculations in the context of the household utility maximization problem yield the expression: c1 1  α p1 ρ ¼½  α p2 c2 which in turn implies that relative expenditure shares satisfy: p1 c1 1  α ρ p1 1ρ  ½  ¼½ α p2 c2 p2

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

Of course, if ρ=1, we get the same result as earlier, in which expenditure shares are independent of relative prices. But if ρ is not equal to unity, changes in relative prices do induce changes in relative expenditure shares. The direction of this effect depends on whether ρ is greater or less than unity. If ρ exceeds unity, then the relative expenditure share increases for the good whose relative price has declined, whereas if ρ is less than unity then the relative expenditure share increases for the good whose relative price has increased. Two extreme cases help to illustrate the intuition for this result. At one extreme, if two goods are perfect substitutes, then starting from a situation in which we have an interior solution, a decrease in the price of one of the goods will lead the household to spend all of its income on the cheaper good. In contrast, in the opposite extreme case of Leontief preferences, the household needs to consume the two goods in fixed proportions, and so the household optimally allocates more expenditure to the good that has become more expensive in order to maintain a perfectly balanced consumption bundle. Turning to the determination of optimal capital and labour allocations assuming this form of the utility function, simple calculation yields: h1 1  α ρ A1 ρ1  ½  ¼½ α h2 A2 If ρ is not equal to unity, it follows that changes in relative productivity will induce changes in the allocation of labour across sectors. Note that in contrast to the previous generalization, a proportional increase in A₁ and A₂ does not lead to any reallocation of labour, nor does an increase in capital have any effect on the optimal labour allocation. It follows that if this economy experiences development through productivity growth that is not uniform across sectors, development will be associated with structural transformation. From an empirical perspective the relevant question becomes whether non-uniform productivity growth is an important feature in reality. If not, this channel for structural transformation would be of interest purely as a theoretical possibility but with little practical relevance. In fact, it turns out that in reality, growth in TFP has been quite asymmetric across broad sectors, with the service sector experiencing much slower productivity growth than either the manufacturing or the agricultural sector in the last fifty or so years in advanced economies. Combined with an elasticity of substitution that is less than unity for these broad categories of consumption, this suggests a potentially important role for this channel as a driver of labour reallocation away from agriculture and manufacturing and into services.¹

¹ Baumol (1967) was the first to formalize this mechanism and argue for its empirical relevance in accounting for the movement of labour from manufacturing to services.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



2.2.4 Relative Price Effects II: Heterogeneous Capital Intensities Subsections 2.2.2 and 2.2.3 explored the possibility for development to affect labour allocations by considering alternative forms of preferences. In this subsection we consider the possibility of alternative production structures. Specifically, we will allow for capital intensities to differ across the two sectors. We continue to assume that each sector has a Cobb–Douglas production function, but, differently than before, we now assume that sector i has a capital share given by θi. If we continue to assume log-log preferences as in our simple benchmark, the Social Planner’s problem can be written as: 1  þ ð1  αÞlog½A2 ðK  k1 Þθ2 ð1  h1 Þθ1z  max αlog½A1 kθ11 h1θ 1

h1 ;k1

Deriving first order conditions and performing some simple algebra gives: h1 α 1  θ1 ¼  1  h1 1  α 1  θ2 k1 α θ1 ¼ K  k1 1  α θ2 The first equation implies that the labour allocation is still independent of both the level of productivity and the level of capital. The second equation says that the share of total capital devoted to each sector is independent of the level of capital or productivity. It follows that an increase in K will increase capital in each sector by the same percentage and leave labour unchanged. Output will thus increase by more in the sector with the higher value of θ, but importantly from our perspective, there will be no reallocation of labour across sectors. Intuitively, an increase in capital in this economy is akin to an increase in the relative productivity of the sector with the higher level of θ, but with Cobb–Douglas preferences such a change in relative productivities will not produce any reallocation of factors across sectors. The preceding logic in combination with our previous analysis of uneven technological progress suggests that if we depart from Cobb–Douglas preferences, an increase in the capital stock can lead to a reallocation of labour. In such a setting, even uniform productivity growth across sectors could lead to structural transformation if this is accompanied by an increase in the level of capital and there is heterogeneity in capital intensities across sectors. We now develop this result. Specifically, we continue to assume that the sectoral production functions are Cobb–Douglas with potentially different capital shares, and now also assume preferences are of the CES variety as in the previous subsection. The Social Planner’s problem now becomes: 1  max ðα½A1 kθ11 h1θ 1

h1 ;k1

ρ1 ρ

ρ1

ρ

þ ð1  αÞ½A2 ðK  k1 Þθ2 ð1  h1 Þ1θ2  ρ Þρ1

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

The two first order conditions for k₁ and h₁ are respectively given by: 1 1 k1 k2 1 ρ  A1 ½ θ1 1 ¼ θ2 ð1  αÞ½A2 kθ22 ð1  h1 Þ1θ2 ρ A2 ½ θ2 1 θ1 α½A1 kθ11 h1θ 1 h1 h2 1 1 k1 k2 1 ρ ð1  θ1 Þα½A1 kθ11 h1θ  A1 ½ θ1 ¼ ð1  θ2 Þð1  αÞ½A2 kθ22 ð1  h1 Þ1θ2 ρ A2 ½ θ2 1 h1 h2

Dividing the two first order conditions establishes that the ratio of sectoral capital to labour ratios is independent of both the Ai and K, and furthermore that the capital to labour ratio in each sector is a constant times the aggregate capital to labour ratio (which is equal to K). Using this fact and rearranging either of the two first order conditions yields: h1 A1 ¼ B½ K ðθ1 θ2 Þ ρ1  1  h1 A2 where the constant B depends on the preference and production parameters. This expression summarizes the intuition that we alluded to earlier. In particular, suppose that θ1 > θ2 so that higher capital ‘ favours’ sector 1. As long as ρ is not equal to unity, this will lead to a reallocation of labour across sectors. But, as in the previous subsection, the direction of this reallocation depends on the elasticity of substitution between the two goods in preferences. If the elasticity is less than unity, this increase in capital will lead to a reallocation of labour toward the less capital intensive sector. The key point here is that in contrast to the result in the previous section, one can generate structural transformation with homothetic preferences even if technological progress is uniform across sectors, if the technological progress is associated with an increase in the capital stock and capital intensities vary across sectors. If technological progress is uneven across sectors then the effect of differential capital intensities may either dampen or amplify the reallocation depending upon the correlation of productivity growth and capital intensity.

2.2.5 Other Formulations Subsections 2.2.2 to 2.2.4 generalized preferences beyond Cobb–Douglas to illustrate some basic economic forces that can drive structural transformation in the face of development associated with increases in productivity and capital. The extensions considered were very tractable and hence served well to illustrate these basic economic forces. Combining the two extensions would lead one to consider preferences of the following form: ρ1 ρ

uðc1 ; c2 Þ ¼ ½αðc1 
c 1 Þ

ρ1

ρ

þ ð1  αÞðc2 þ
c 2 Þ ρ ρ1

Formulations of this type have dominated the recent literature on structural transformation. But as noted previously, although this way of introducing non-homotheticities is

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



perhaps a natural starting point, the choice of specification may matter. Here we discuss two recent alternative formulations: Boppart (2014) and Comin et al (2015).² We begin by describing the specification in Boppart (2014). He proposes that we consider utility functions from the Price Independent Generalized Linearity (PIGL) class of preferences as defined by Muellbauer (1975, 1976). These preferences are defined in terms of the associated indirect utility function and do not in general admit a closed form representation for the direct utility function. Boppart studies preferences that are defined by the following indirect utility function: 1 I η p1 1 η vðI; p1 ; p2 Þ ¼ ð Þε  ð Þγ  þ ε p2 γ p2 ε γ where I is total income (and hence expenditure on the two goods), and the parameters ε, γ, and η satisfy: 0  ε  γ  1 and η>0. The parameter ε controls the extent of nonhomotheticities, and if ε = 0 then preferences are homothetic. The parameter γ influences the degree of substitutability between the two goods. If both ε and γ are set to zero then preferences are just Cobb–Douglas and the parameter η is the expenditure share on good 1. Applying Roy’s Identity we can uncover the following demand functions for the two goods: c1 ¼ c2 ¼

ηI p2 ε p1 γ ð Þð Þ p1 I p2

I p2 p1 ½1  ηð Þε ð Þγ  p2 I p2

Note that the income elasticity for good 1 is equal to 1 – ε, and so is constant and less than unity as long as ε > 0. This is in contrast to the earlier specification that delivered non-homotheticities, which we noted had the implication that income elasticities converged to unity as income grew. To examine the implications of these preferences for the allocation of labour we consider the economy studied above with Cobb–Douglas technologies with identical capital intensities but assuming that preferences are defined by this indirect utility function. Previously we solved the Social Planner’s problem of maximizing the representative household’s utility subject to feasibility, but we cannot proceed this way now given that we do not have an analytic representation of the utility function. Instead, we solve directly for the competitive equilibrium for this economy, which we know from the First Welfare Theorem yields a optimal allocation. As noted earlier, capital to labour ratios will be equalized across the two sectors in equilibrium, so that one can effectively eliminate the capital allocation decision from the problem and write the production function for sector i as:

² To be sure, there are other formulations in the literature as well. See, for example the papers by Echevarria (1997) and Foellmi and Zweimuller (2008).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



  ˜ i hi : ci ¼ Ai K θ hi ¼ A

To solve for the competitive equilibrium we normalize the price of labour to equal unity. This implies that income for the household will also equal unity. With linear production functions, the competitive equilibrium prices can be inferred immediately, ˜ i for i = 1,2. Simple calculation yields: with pi ¼ 1=A 1 ε A2 γ h1 ¼ ηð Þð Þ A2 K θ A1 1 ε A2 γ h2 ¼ ½1  ηð Þð Þ A2 K θ A1 From this expression we see all of the effects that we previously documented for the alternative specifications of preferences: a uniform increase in productivity will lead to a reallocation of labour from sector 1 to sector 2, as will an increase in A₁ holding A₂ constant. An increase in K holding technology constant will also lead to a reallocation of labour from sector 1 to sector 2, as an increase in capital with technology constant generates an increase in income. In contrast to the previous specification of nonhomotheticities, uniform growth in both A₁ and A₂ will lead to constant growth in h1 , which will prove to be important when one looks to find a balanced growth path. One limitation of this formulation is that it cannot be extended beyond the two sector case in a very general way. A recent paper by Comin et al. (2015) considers an alternative formulation. Specifically, they consider non-homothetic constant elasticity of substitution preferences, as first introduced by Hanoch (1975). This class also does not permit a closed form representation for utility, but instead offers an implicit definition of utility. In the two good cases that we have been studying, the utility u derived from consuming quantities c₁ and c₂ is defined as the solution to: αu

ε1 ρ ρ1 ρ ρ

c1 þ ð1  αÞu

ε2 ρ ρ1 ρ ρ

c2 ¼ 1

Note that if we pick ε₁ = ε₂ =1 then we can directly solve for utility u and we have the standard Constant Elasticity of Substitution Function: ρ1

ρ1

ρ

uðc1 ; c2 Þ ¼ ½αc1ρ þ ð1  αÞc2ρ ρ1 When the εi are not the same then the weights attached to the two goods are effectively changing as consumption changes, and as a result the preferences are not homothetic. To solve the household utility maximization problem one can write it as follows: max u

u;c1 ;c2

s:t:p1 c1 þ p2 c2 ¼ I αu

ε1 ρ ρ1 ρ ρ

c1 þ ð1  αÞu

ε2 ρ ρ1 ρ ρ

c2 ¼ 1

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



From the first order conditions associated with this problem one obtains: c1 p1 ¼ ð Þρ uε1 ε2 c2 p2 Two properties that Comin and coauthors note are that the elasticity of c₁ / c₂ with respect to utility holding relative prices constant is equal to ε₁ – ε₂. In the context of our simple production economy this can be understand as revealing the impact of a uniform increase in the Ai, as this will keep relative prices constant but lead to an increase in utility. Importantly, the implication is that heterogeneity in the εi generates non-homotheticities and the associated income effects are independent of the level of income. Second, the above expression also indicates that holding utility constant, the elasticity of relative consumption with respect to relative prices is equal to ρ. An important point that Comin and coauthors emphasize is that their specification does not place any implicit restrictions on the relative size of income and substitution effects, which they note is not the case for the specification used by Boppart (2014).

2.3 S T  B G

.................................................................................................................................. Section 2.2 illustrated three basic mechanisms that can drive sectoral reallocation of labour in the face of ongoing technical progress and capital accumulation. In this section we ask whether these basic mechanisms can be embedded into a version of the standard growth model in a way that would allow us to simultaneously retain the attractive feature of a balanced growth path at the aggregate level while also accommodating the reality of structural change.

2.3.1 Balanced Growth in the One Sector Growth Model It is useful to start by presenting results about balanced growth in the standard growth model before we consider extensions to generate structural transformation. There is an infinitely lived representative household, with preferences given by: 1 X

βt Uðct Þ

t¼0

There is a constant returns to scale aggregate production function that uses capital (kt) and labour (ht) to produce output in the presence of labour augmenting technical progress (At): yt ¼ Fðkt ; At ht Þ:

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

Output can be used for either consumption or investment (xt), and capital depreciates at rate δ: yt ¼ ct þ xt ktþ1 ¼ ð1  δÞkt þ xt The household is endowed with one unit of time in each period. It is well known that if we assume constant growth in At at rate gA, so that Atþ1 ¼ ð1 þ gA ÞAt , this economy will have a balanced growth path as long as we assume that the utility function U takes the following isoelastic form: Uðct Þ ¼

c1σ 1 t 1σ

Along this balanced growth path, each of y, k, c, and x will grow at the same rate as A. In the competitive equilibrium that decentralizes the solution to the Social Planner’s problem, the rental price of capital (and hence the interest rate) will be constant, while the wage rate will grow at the same rate as A. Note that this result does not require any restrictions on the functional form of the production function F other than to require constant returns to scale. While this result does require that preferences belong to a relatively narrow class, this class is commonly used in empirical work and so has not generally been interpreted as evidence of the fragility of balanced growth. We will return to this issue later.

2.3.2 Early Results on Structural Transformation and Balanced Growth We now extend this framework to allow for three separate categories of consumption that reflect the three broad sectors commonly studied in the context of structural transformation: agriculture (cat), manufacturing (cmt) and services (cst). In this subsection we describe the results obtained by Kongsamut et al. (2001), Ngai and Pissarides (2007), and Acemoglu and Guerrieri (2008) regarding the possible coexistence of balanced growth and structural transformation. Each of these papers can be understood as adopting specifications that reflect the first three extensions studied in Section 2.2. Specifically, in this subsection we will assume that the three categories of consumption are aggregated into the single composite ct via: h iρ ρ1 ρ1 ρ1 ρ1 Ct ¼ αa ðcat 
c a Þ ρ þ αm ðcmt Þ ρ þ αs ðcst þ
c s Þ ρ where the αis are all positive and sum to unity. We generalize the specification of technology to assume a separate production function for each sector, with sector specific technical change:

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



yit ¼ F i ðkit ; Ait hit Þ An issue that one needs to confront in extending the one sector model to allow for multiple sectors is how to handle the production of capital. There is some tradition in the literature of assuming that capital goods are produced in the manufacturing sector. If one identifies investment with equipment and structures then this assumption is probably reasonable. But in today’s economy, investment also reflects important contributions from categories such as software, which are perhaps more reasonably associated with the service sector than the manufacturing sector. In fact, total investment in the US economy now exceeds the size of the manufacturing sector. For this reason, and following the formulation in Herrendorf et al. (2014), I think a preferable alternative is to include a separate sector that produces the investment good. For this reason I add a fourth production function:³ yxt ¼ F x ðkxt ; Axt hxt Þ We are now ready to pose the question of interest: assuming that each of the Ait grows at a constant (but possibly different) rate, under what conditions will this economy exhibit a balanced growth path that includes structural transformation? Note at the outset that if we assume all of the production functions are identical, all of the Ait grow at the same rate, that
c a ¼
c s ¼ 0, and consider the limit as ρ goes to unity, then this economy will indeed have a balanced growth path as long as U is isoelastic. However, with a homothetic period utility function and unitary elasticity of substitution among the three goods, consistent with the simple example in Section 2.2, there will not be any structural transformation along the balanced growth path. Hence, the question is to what extent we can deviate from this benchmark specification, introducing features that can generate structural transformation in response to changes in productivity, and still find a balanced growth path. It is important to be clear about the question that we are asking here. If we were to adopt a specification consistent with the three mechanisms that we highlighted in the previous subsection, that is, we assume that
c a and
c s are both positive, assume that ρ < 1, assume different capital intensities across sectors and allow for differential productivity growth among the three consumption sectors, and then calibrated the model in a reasonable way, we would in fact find that the economy would grow over time and exhibit structural transformation. The question that the literature has asked is not whether we can get growth and structural transformation simultaneously, but rather whether we can get balanced growth and structural transformation simultaneously. In fact, if we define a balanced growth path as one in which all quantities grow at constant (but potentially different) rates, then one can show that one cannot

³ An alternative, as pursued by Acemoglu and Guerrieri (2008) and discussed further in Subsection 2.3.3, is to assume that each of the three broad sectors produces an intermediate input that is aggregated to a single final good which can then be used as either consumption or investment as in the standard one sector model.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

accommodate balanced growth with structural transformation with the above choice for preferences. A somewhat weaker restriction but one that still captures the essence of a balanced growth path in the standard one sector model is to look for a solution in which the aggregate capital stock grows at a constant rate and the marginal product of capital is constant, implying that in the associated competitive equilibrium the interest rate is constant. In the literature this type of path has been referred to as a generalized balanced growth path. Both Kongsamut et al. (2001) and Ngai and Pissarides (2007) offered sufficient conditions under which the above economy will possess a generalized balanced growth path accompanied with structural transformation. Acemoglu and Guerrieri (2008) only generate generalized balanced growth asymptotically. In what follows we look at each of these in turn.⁴

2.3.2.1 Structural Change Via Uniform Technical Progress We begin by summarizing the result from Kongsamut et al. (2001). This paper focuses on the channel that operates through non-homotheticities. The authors focus on the case in which ρ is equal to unity, so that with some abuse of notation the composite consumption good is now written as: Ct ¼ ½αa logðcat 
c a Þ þ αm logðcmt Þ þ αs logðcst þ
c s Þ Technologies are of the form: yit ¼ kθit ðAit hit Þ1θ for i ¼ a; m; s; x: and the rate of technological change is assumed to be the same in all sectors, so letting g be the common rate of growth we have that gi = g. Having imposed all of these conditions, it turns out that we are still not able to guarantee the existence of a generalized balanced growth path. To see why, consider the Social Planner’s problem of maximizing the utility of the representative household subject to feasibility. Because all of the production functions are identical, the problem in which the Social Planner allocates inputs to the four different sectors is equivalent to one in which the Social Planner simply uses any of the four (identical) production functions to produce a generic output, which can then be transformed one for one into any of the four different uses of output. This implies that the Social Planner’s problem can be recast as: X 1 ½αa logðcat 
c a Þ þ αm logðcmt Þ þ αs logðcst þ
c s Þ1σ max βt 1σ

⁴ The papers that I describe in detail are not the only ones to study structural change in the context of balanced growth. Foellmi and Zweimuller (2008) is a prominent example. However, their somewhat abstract formulation of the commodity space does not easily allow one to connect with data at the sectoral level. See also Ju et al. (2015) which also does not readily map into the three sector framework studied here.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



subject to: cat þ cmt þ cst þ ktþ1 ¼ kθt At1θ þ ð1  δÞkt plus the usual non-negativity constraints. Note that we have implicitly imposed that all labour will be allocated to production. Letting λt be the multiplier on the constraint for period t and assuming an interior solution for all variables, we obtain the following first order conditions: αa ¼ λt cat : βt Ctσ cat 
c a αm cmt : βt Ctσ ¼ λt cmt αs cst : βt Ctσ ¼ λt cst þ
c s kt kt : λt ½θð Þθ1 þ ð1  δÞ ¼ λt1 At Along a generalized balanced growth path, the marginal product of capital is constant, implying that kt grows at the same rate as At. Feasibility then implies that the sum cat þ cmt þ cst must also grow at the same rate as kt. Constant marginal product of capital and the FOC for capital implies that λt1 =λt is constant. When combined, the three FOCs for cat ; cmt , and cst imply that the ratios ðcst þ
c s Þ=cmt and ðcat 
c a Þ=cmt are both constant, so that each of ðcst þ
c s Þ; cmt and ðcat 
c a Þ grows at the same rate. It follows that Ct will also grow at this same rate. The FOC for cmt then implies that it must grow at a constant rate, and feasibility dictates that this constant rate must be the same as the growth rate in kt. We then have that the sum cat + cst must also grow at the same rate as kt, at the same time that the ratio ðcst þ
c s Þ=ðcat 
c a Þ is constant. But one can show that this can only happen if
c a ¼
c s . To summarize, Kongsamut and coauthors show that if uniform productivity growth combined with non-homothetic preferences as modelled above is the channel that drives structural transformation, then a generalized balanced growth that features structural transformation is possible only in the very special case in which
c a ¼
c s . Taken at face value, this result appears to suggest that simultaneously achieving balanced growth and structural transformation is very tenuous. We will return to this issue later.

2.3.2.2 Structural Change Via Non-Uniform Technical Change Ngai and Pissarides (2007) investigate the possibility of generating structural transformation along a generalized balanced growth path when preferences are homothetic but feature a non-unitary elasticity of substitution among goods in the presence of nonuniform technical change.⁵ That is, we now consider a consumption aggregator of the form: ⁵ As noted earlier, their work builds heavily on the earlier work of Baumol (1967), with their contribution being to embed the mechanism highlighted by Baumol into a standard growth model.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



  

ρ1 ρ

ρ1 ρ

ρ1 ρ

Ct ¼ αa cat þ αm cmt þ αs cst

ρ ρ1

Ngai and Pissarides also require that the utility function U feature a unitary intertemporal elasticity of substitution, that is, the case in which σ tends to one, so that UðCt Þ ¼ logðCt Þ. They also assume that all technologies are of the form: yit ¼ kθit ðAit hit Þ1θ for i ¼ a; m; s; x but allow for differential growth rates in technology across sectors. Here I sketch the proof of how to construct the generalized balanced growth path. The first step is to show that even if the Ait differ across sectors, it is optimal for the Social Planner to choose the same capital to labour ratio in each of the sectors. To see this, write the Social Planner’s problem including four constraints, one for production in each sector, in addition to feasibility constraints on total capital and total labour allocated to production: X X kit ¼ kt hit ¼ 1; Letting λit be the multiplier on the feasibility constraint for sector i in period t, and letting λht and λkt be the multipliers for the feasibility constraints for labour and capital allocations respectively in period t, we obtain the following first order conditions: kit kit : λit ð Þθ1 Ait1θ ¼ λkt hit kit hit : λit ð Þθ A1θ ¼ λht it hit It follows that kit =hit ¼ λht =λkt for all i, so that at each point in time we have equalization of capital to labour ratios across the four sectors. It of course follows that each sector will have the aggregate capital to labour ratio. Given that total labour will always equal unity, the capital to labour ratio in each sector will simply be equal to the aggregate capital stock. Using this fact, output in sector i can be expressed as: kit ¼ hit ð Þθ A1θ ¼ hit kθt A1θ yit ¼ kθit hit1θ A1θ it it it hit It follows that it is sufficient to focus on how labour is allocated across the four sectors. Moreover, because these outputs are linear in labour, we can represent the aggregate production possibilities frontier via the following single linear equation: cat cmt cst þ þ þ ktþ1 ¼ kθt A1θ xt þ ð1  δÞkt 1θ 1θ 1θ ðAat =Axt Þ ðAmt =Axt Þ ðAst =Axt Þ We can now write the Social Planner’s problem as maximizing the utility of the representative household subject to this feasibility constraint. Letting λt be the

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



Lagrange multiplier on this constraint in period t, we now obtain the following first order conditions for an interior solution: cat : βt

1 1=ρ 1=ρ λt Ct αa cat ¼ Ct ðAat =Axt Þ1θ

cmt : βt

1 1=ρ λt 1=ρ Ct αm cmt ¼ Ct ðAmt =Axt Þ1θ

cst : βt

1 1=ρ 1=ρ λt C αs cst ¼ Ct t ðAst =Axt Þ1θ

kt kt : λt ½θð Þθ1 þ ð1  δÞ ¼ λt1 Axt As in the previous case, a generalized balanced growth path will require that kt grows at the same constant rate as Axt, implying that λt will also grow at a constant rate. If kt grows at the same rate as Axt, the share of labour being allocated to the investment sector is constant along a balanced growth path. Additionally, if kt grows at the same constant rate cat cmt as Axt, the feasibility constraint implies that ðA =A þ ðA =A þ ðA =Acst Þ1θ must also Þ1θ Þ1θ at

xt

mt

xt

st

xt

grow at the same constant rate. Combining the first order conditions for each of the three consumptions we have that the consumptions are linearly related via: cat αm Amt cst αm Amt ¼ ½ ð Þ1θ ρ and ¼ ½ ð Þ1θ ρ cmt αa Aat cmt αs Ast

It follows that if we know the growth rate of cmt then we necessarily know the growth rates of the other two consumptions as well. If ρ is less than unity, then the three growth rates can be monotonically ordered based on the values of the sectoral technology growth rates (i.e. the gi), with the sector with the highest value of gi exhibiting the highest growth in consumption. Since both cat and cst can be expressed as linear cat cmt þ ðA =A þ ðA =Acst Þ1θ grows at the functions of cmt and we know that ðA =A Þ1θ Þ1θ at

xt

mt

xt

st

xt

same constant rate as kt, at each date we can solve for the implied growth of cmt. We can then solve for the implied growth rates for each of the two other consumptions. The growth rates in relative labour allocations can be inferred from growth rates in relative consumptions, using the fact that cit ¼ hit kθt Ait1θ . Substitution into the above expression gives: hat αm Amt hst αm Amt ¼ ½ ρ ½ ð1θÞð1ρÞ and ¼ ½ ρ ½ ð1θÞð1ρÞ hmt αa Aat hmt αs Ast Note that the relative growth rates of consumption and hours are inversely related: the sector with the highest growth rate of consumption is the sector with the lowest growth rate in hours. As a special case, note also that if ρ=1, then the above expression implies that labour allocations are constant, consistent with the results obtained in Section 2.2.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

From the above derivation it is not apparent why we needed to restrict the value of σ. To see why this is, multiply each of the three first order conditions by cit respectively and aggregate, to obtain: cat cmt cst þ þ ¼ 1θ 1θ ðAat =Axt Þ ðAmt =Axt Þ ðAst =Axt Þ1θ  ρ1  ρ1 ρ1 1 1 ρ ρ ρ t 1 ρ1 β Ct αa cat þ αm cmt þ αs cst ¼ βt λt λt The last equality holds on account of the fact that we have set σ equal to one. If we did not impose this condition, the above expression would still include a term with Ct on the right hand side. If this were the case, the implication would be that because the left hand side grows at a constant rate and λt grows at a constant rate, it must be that Ct also grows at a constant rate. But the only way that we could have Ct growing at a constant rate would be to have each of the consumptions growing at the same constant rate. But this is inconsistent with constant growth in the left hand side of the above expression. To summarize, assuming that σ = 1,
c a ¼
c s ¼ 0, that ρ is not equal to one and that there is heterogeneity in the gi’s, one can find a generalized balanced growth path that features structural transformation.

2.3.3 Structural Change and Heterogeneous Capital Intensities In this subsection we briefly describe the analysis in Acemoglu and Guerrieri (2008). Their specific environment is somewhat more specialized than the framework introduced in Subsection 2.3.2. In the context of the framework that I have introduced, the natural representation of their analysis would correspond to the case in which the
c i ’s are equal to zero, ρ is not equal to unity, and production functions are Cobb–Douglas but with heterogeneous capital intensities across sectors. They allow for technical change to differ across the two sectors, although this is not essential to their analysis. The reason that this does not correspond to their particular setup is that they assume only two sectors, but moreover interpret the two sectors as producing intermediate goods that are combined via a CES aggregator to produce a single final good that can be used for either consumption or investment. Put somewhat differently, investment and (aggregate) consumption are both produced from the same aggregation of sector level output. I will not present the details of their analysis for the reason that one of the results of their analysis is that the economy will only approach a generalized balanced growth path asymptotically as one of the sectors becomes vanishingly small in terms of input shares. That is, consistent with our simple static analysis in the previous section, as capital grows without bound, holding relative productivities constant, the sectoral shares of labour and capital will tend to zero in one of the sectors. With one sector

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



vanishing, the economy is effectively approaching a one sector economy, and aggregates will reflect the properties of the dominant sector. To understand why their setup will not generate generalized balanced growth except in the limit as one of the sectors becomes vanishingly small, it is sufficient to note that in each period, labour and capital will be allocated across the two intermediate sectors so as to maximize total output of the final good. This is tantamount to solving the static utility maximization problem that we solved in Section 2.2, and allows one to implicitly derive an aggregate production function that maps the total amount of capital into the (maximum) amount of the final good that can be produced. In the two previous analyses in this section, we effectively wrote the feasibility constraint for the economy as if we produced aggregate output using a Cobb–Douglas production function. The key difference is that because of the differential capital intensities in the Acemoglu and Guerrieri analysis, the marginal product of capital in the implicit aggregate production function does not behave nicely even if capital grows at a constant rate except in the limit when one of the two sectors is vanishingly small.

2.4 D  R P

.................................................................................................................................. Taken at face value, the results in the last section seem mixed at best. The analysis of Ngai and Pissarides (2007) suggests a somewhat robust result about the possibility of generating structural transformation and balanced growth if one assumes homothetic preferences. But the results regarding balanced growth and structural transformation when preferences are non-homothetic indicate a significant amount of fragility. And Acemolgu and Guerrieri (2008) could only find a generalized balanced growth path in the limit. However, some caution should be taken when interpreting these results. The motivation for seeking specifications in which we simultaneously have structural transformation and balanced growth comes from the fact that aggregate economic behaviour seems well approximated by balanced growth even during periods in which structural transformation has been ongoing. But the key part of this statement is that aggregate economic behaviour is well approximated by the case of balanced growth. This is quite different from saying that aggregate behaviour is consistent with exact balanced growth. This raises the key question of whether the conditions required for exact balanced growth (or exact generalized balanced growth) are misleading in terms of the ability of our specifications to deliver what one would view as approximate balanced growth. While this issue has not been studied in a thorough and systematic manner, the work that exists suggests that the fragility suggested by these results is perhaps somewhat overstated. For example, Kongsamut et al. (2001) also show that when their knife-edge condition on preference parameters is not satisfied the resulting dynamics still very

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

much resemble balanced growth for a very long period. In related work, Gollin et al. (2007) consider a two-sector specification in which
c 1 > 0 and
c 2 ¼ 0, and assume uniform technical change, and also find that the resulting aggregate behaviour does not deviate dramatically from the conditions of balanced growth. And Acemolgu and Guerrieri (2008) provide a numerical example to show that their model generates outcomes that resemble balanced growth during the transition to the asymptotic generalized balanced growth path. These issues notwithstanding, recent work has shown that one can obtain balanced growth in a more robust sense in settings that allow for both non-homothetic preferences and uneven technical change if one considers a different class of utility functions. In particular, each of the two classes discussed at the end of Section 2.2 have been shown to be consistent with balanced growth when placed in settings similar to what we assumed in the previous section, that is, isoelastic utility defined over the composite consumption good, and Cobb–Douglas production functions in all sectors that have the same capital intensity. Boppart (2014) shows this for the class of PIGL preferences, and Comin et al. (2015) show it for the class of the non-homothetic CES utility functions. To illustrate the mechanics of these two approaches I sketch the analysis found in Boppart regarding the existence of a balanced growth path. His analysis is restricted to the case of two sectors, which I will refer to as goods and services. As in the earlier analysis, we work directly with the competitive equilibrium rather than the Social Planner’s problem, and assume that each of the two sectors has a Cobb–Douglas production function with potentially different rates of technical change. The dynamic optimization problem of the household assuming a constant rate of return on saving is given by: max et ;at

1 X t¼0

1 et η ρgt 1 η βt ½ ð Þε  ð Þγ  þ  ε ρst γ ρst ε γ

s:t:atþ1 ¼ ð1 þ rÞat þ 1  et where et is total expenditure in period t, at are assets carried into period t and we have normalized the wage rate in each period equal to unity. Letting λt be the multiplier on the period t budget equation, the two First Order Conditions for this problem are: 1 ¼ λt et : βt ð Þε eε1 pst t at : ð1 þ rÞλt ¼ λt1 The second equation implies that λt will grow at a constant rate. Taking the first equation at dates t and t + 1 and dividing them yields: 1 pstþ1 ε etþ1 1ε λt ð Þð Þ ¼ β pst et λtþ1

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



Given that the right hand side grows at a constant rate, it follows that if the price of services grows at a constant rate the level of expenditure will also grow at a constant rate. Next consider the demand functions for goods and services. Again, from Roy’s identity we have: ηet pst ε pgt γ ð Þð Þ gt ¼ pgt et pst st ¼

et pst pgt ½1  ηð Þε ð Þγ  pst et pst

From the first expression it follows that if prices and expenditure grow at constant rates then so will gt, pgtgt and pgtg/et. Constant growth in gt will imply constant growth in hgt. Finally, to close the argument note that constant growth in prices is guaranteed by the fact that technological progress is constant. From a technical perspective, the analyses of Boppart (2014) and Comin et al. (2015) are nice because they establish the more robust possibility of exact balanced growth in settings that feature non-homothetic preferences, whereas the results in Kongsamut et al. (2001) suggested that we would have to be satisfied with attaining approximate balanced growth. These new results also have substantive implications that are potentially important for the analysis of structural transformation. In fact, the technical and substantive implications are intimately related. As noted earlier, in the specification studied in Kongsamut et al. (2001), the effect of the non-homotheticity vanished as the economy became richer, implying that the effects of the non-homotheticity will vary over time if the economy is growing. In contrast, the specifications used by Boppart and Comin et al. imply income effects that do not die out as the economy grows. This difference is potentially important from a substantive perspective because the specification of Kongsamut et al. necessarily imposes that these effects die out over time, a property that one would best view as an empirical issue. But this substantive property of time varying effects from the non-homotheticities is also what makes it difficult to obtain an exact generalized balanced growth path, as the time varying nature creates a challenge from the perspective of looking for constant growth paths. Having emphasized two features of preferences that are viewed as important for shaping structural transformation, an important empirical question is to assess the importance of the two features from an empirical perspective. The work of Herrendorf et al. (2013) emphasizes that this question raises an important issue about how to connect these models with the data. In particular, the issue is whether the arguments of the utility function represent final consumption categories or value added categories. Consider the case in which a consumer purchases a shirt. According to the final consumption approach, this would be viewed as consumption of the manufactured good, as the shirt is a non-food good. But from the value added perspective we would break the shirt down into its sectoral value added shares—the cotton from agriculture, the processing from manufacturing and a retail component from the service sector. Herrendorf et al. (2013) show how to implement both of these approaches using US

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

data for the post World War II period and show that the properties of preferences that best fit the relevant data in each case possess very different properties: in the case of the final consumption approach they find preferences that resemble those in Kongsamut et al. (2001), but in the case of value added preferences they find preferences that resemble those in Ngai and Pissarides (2007), with an elasticity of substitution that is close to zero. The preferences assumed in this work nested those of Kongsamut et al. (2001) and Ngai and Pissarides (2007), but not those of Boppart (2014) or Comin et al. (2015). Both of these latter studies find that income effects are an important source of structural transformation. Boppart only considers the USA, using final consumption data, whereas Comin et al. study a broad set of countries. An important message from these exercises is that the income effects that one estimates might be influenced by the manner in which income effects appear in the underlying utility function.

2.5 C

.................................................................................................................................. This chapter has reviewed progress on building models of structural transformation that are simultaneously consistent with the so-called Kaldor facts. In these models, aggregate growth is driven by exogenous productivity growth, just as in the versions of the one sector growth model. But the nature of this productivity growth and its interaction with preferences and technology also give rise to systematic movements of resources across sectors. While there is broad agreement that non-homothetic preferences and a non-unitary elasticity of substitution between goods are robust features that interact with productivity growth to generate structural transformation, there is less agreement at present about the relative importance of these two features. One message that has emerged is the importance of allowing for flexible specifications of preferences that do not impose ad hoc restrictions on the role of these two features. As noted in the introduction, all of the models discussed in this chapter have been closed economy models. One can also study the properties of structural transformation in an open economy setting, and several recent papers have pursued this.⁶ In principle, trade has the potential to affect structural transformation by allowing consumption to be disconnected from production, so that a country could specialize in one of the broad sectors at the expense of the others. For example, a country could systematically increase exports of services and increase imports of manufactured goods as it developed. Or at an earlier stage it could systematically increase imports of goods and increase exports of manufactured goods. While conceptually possible, as a practical matter most trade has historically taken place within the manufacturing sector, so that this dynamic is not a key driving force of ⁶ For example, see Matsuyama (2009); Uy et al. (2013); Teignier (2014); and Swiecki (2017).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



the structural transformation that we have witnessed over the last 150 years. To the extent that trade in services is now becoming more relevant it is possible that this may become more significant in the future. Trade of manufacturing goods for agricultural goods certainly occurs, and one sees evidence of this if one looks within the European Union, for example. But to a first approximation the key impact here seems to be in terms of level shifts rather than in dynamic effects. That is, in all countries we see the steady decline in the share of employment in agriculture, although the level of this curve is shifted in some countries relative to others. But even if trade across broad sectors is not the cause of structural transformation, trade is likely to have effects that are reflected in the closed economy analysis carried out in this chapter. For example, if trade within manufacturing allows countries to specialize in those goods for which they have comparative advantage then trade effectively serves to increase productivity in manufacturing, and so trade can be a source of uneven technological progress across sectors. Trade will thus affect relative prices across broad sectors and through the mechanisms studied earlier in this chapter, and also affect the allocation of factors across broad sectors. More generally, openness may affect the incentives for firms to innovate, and may affect the level of markups by affecting the extent of competition that local producers face. Once again, these forces can affect relative prices and income and so will generate effects of the type studied in this chapter. Although my focus has been on the forces that drive the reallocation of activity across broad sectors, it is also of potential interest to study the reallocation of activity across more narrowly defined sectors. Of particular interest is the extent of reallocation of activity within manufacturing or within services. Ju et al. (2015) provide evidence that capital intensity varies significantly across sectors within manufacturing and that activity tends to be concentrated in those manufacturing activities which have a capital intensity that is similar to the overall capital intensity in the economy. That is, as an economy develops and the capital to labour ratio increases, activity within the manufacturing sector tends to move to those industries that have higher capital to labour ratios. Buera and Kaboski (2012) show that as countries develop, much of the increase in activity within services is accounted for by what they refer to as ‘high-skill’ services. This suggests that the accumulation of both physical and human capital may play an important role in shaping the reallocation of economic activity during the development process. Rodrik (2015) argues that the pattern of structural change taking place among today’s developing economies looks different than the patterns found among early developers, and that in particular the manufacturing sector is peaking at a lower share of employment and at an earlier stage in the development process. An interesting issue for future study is to assess the extent to which this requires us to isolate additional driving forces or if it is simply that the nature of the driving forces is different in today’s world. For example, is it the case that productivity dynamics are different for today’s developing economies? Or does the dominance of China as a global producer of manufactured goods affect the nature of structural change in other economies? Huneeus and Rogerson (2016) is an early attempt to shed light on this issue.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



 

A I thank Justin Lin and an anonymous reviewer for helpful comments.

R Acemoglu, D. and V. Guerrieri, 2008. ‘Capital Deepening and Nonbalanced Economic Growth’, Journal of Political Economy, 116, pp. 467–98. Baumol, W., 1967. ‘Macroeconomics of Unbalanced Growth: The Origins of the Urban Crisis’, American Economic Review, 57, pp. 415–26. Boppart, T., 2014. ‘Structural Change and the Kaldor Facts in a Model with Relative Price Effects and Non-Gorman Preferences’, Econometrica, 82, pp. 2167–96. Buera, F. and J. Kaboski, 2012. ‘The Rise of the Service Economy’, American Economic Review, 102, pp. 2540–69. Comin, D., D. Lashkari, and M. Mestieri, 2015. ‘Structural Change with Long Run Income and Price Effects’, NBER Working Paper No. 21595. Echevarria, C., 1997. ‘Changes in Sectoral Composition Associated with Economic Growth’, International Economic Review, 38, pp. 431–52. Foellmi, R. and J. Zweimuller, 2008. ‘Structural Change, Engel’s Consumption Cycles and Kaldor’s Facts of Economic Growth’, Journal of Monetary Economics, 55, pp. 1317–28. Gollin, D., S. Parente, and R. Rogerson, 2007. ‘The Food Problem and the Evolution of International Income Levels’, Journal of Monetary Economics, 54, pp. 1230–55. Hanoch, G., 1975. ‘Production and Demand Models with Direct or Implicit Indirect Additivity’, Econometrica, 43, pp. 395–419. Herrendorf, B., R. Rogerson, and A. Valentinyi, 2013. ‘Two Perspectives on Structural Transformation and Preferences’, American Economic Review, 103, pp. 2752–89. Herrendorf, B., R. Rogerson, and A. Valentinyi, 2014. ‘Structural Transformation and Growth’, in P. Aghion and S. Durlauf, eds, Handbook of Economic Growth, Volume 2, Amsterdam: Elsevier, pp. 855–941. Huneeus, F. and R. Rogerson, 2016. ‘Deindustrialization Among Early and Late Developers’, Working Paper. Princeton University. Ju, J., J. Lin, and Y. Wang, 2015. ‘Endowment Structures, Industrial Dynamics and Economic Growth’, Journal of Monetary Economics, 76, pp. 244–63. Kongsamut, P., S. Rebelo, and D. Xie, 2001. ‘Beyond Balanced Growth’, Review of Economic Studies, 68, pp. 869–82. Matsuyama, K., 2009. ‘Structural Change in an Interdependent World: A Global View of Manufacturing Decline’, Journal of the European Economic Association, 7, 478–86. Muellbauer. J., 1975. ‘Aggregation, Income Distribution and Consumer Demand’, Review of Economic Studies, 62, pp. 526–43. Muellbauer. J., 1976. ‘Community Preferences and the Representative Consumer’, Econometrica, 44, pp. 979–99. Ngai, R. and C. Pissarides, 2007. ‘Structural Change in a Multisector Model of Economic Growth’, American Economic Review, 97, pp. 429–43. Rodrik, D., 2015. ‘Premature Deindustrialization’, NBER Working Paper No. 20935.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

   



Swiecki, T., 2017. ‘Determinants of Structural Change’, Review of Economic Dynamics, 24, pp. 95–131. Syrquin, M., 1988. ‘Patterns of Structural Change’, in H. Chenery and T.N. Srinivasan, eds, Handbook of Development Economics, Volume 1, Amsterdam: Elsevier, pp. 203–73. Teignier, M., 2014. ‘The Role of Trade in Structural Transformation’, Mimeo, Universitat de Barcelona. Uy, T., K. Yi, and J. Zhang, 2013. ‘Structural Change in an Open Economy’, Journal of Monetary Economics, 60, pp. 667–82.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  ......................................................................................................................

   ......................................................................................................................

     

3.1 I

.................................................................................................................................. E development is a continuous process of economic growth accompanied by structural change, including technology, industry, hard infrastructure, and institution (or soft infrastructure) (Kuznets 1966). The existing growth literature focuses mainly on the process of resource reallocation across the three sectors (agriculture, industry, and service) in the process of structural transformation (see, for example, Herrendorf et al. 2014). Following Kuznets, throughout this chapter, structural change covers a much broader range of changes in economic structures including endowment structure, industrial structure, financial structure, and governance structure, etc. We believe that a deep understanding of the nature of economic development requires thorough analyses and explicit characterizations of the determinants, evolution, and various development implications of each of these structures, which is the research agenda of New Structural Economics (NSE) proposed by Justin Yifu Lin (Lin 2012a, 2013a). The primary goal of this chapter is to introduce the key ideas and hypotheses of NSE, and, more importantly, to demonstrate by concrete examples the way structural change can be formally modelled in NSE. We argue that structural changes should be remodelled to highlight the central roles of endowment structure and firm viability, which deserve much more attention in the pertinent literature. The rest of the chapter is organized as follows. In Section 3.2, we briefly review the current dominant framework for macro-development analysis, which is almost structureless. In Section 3.3, we explain why structures are important for an understanding of economic development. In Section 3.4, we introduce a benchmark model of new structural economics. In Section 3.5, we discuss several theoretical extensions to the benchmark model. Section 3.6 concludes.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



3.2 S F

.................................................................................................................................. Most existing macroeconomic theories (including economic growth) have largely ignored structural differences between countries at different stages of development. The benchmark model for modern macroeconomics is the one-sector growth model with the following exogenous aggregate production function: Y ¼ AK α H β L1αβ ;

ð1Þ

where Y denotes total output, A denotes total factor productivity (TFP), K denotes physical capital, H denotes human capital, L denotes raw labour hours (often normalized to the head count of workers), α and β are parameters that measure the shares of contribution of physical capital and human capital to total output, respectively. Equation (1) is used to organize thinking on what explains the aggregate output difference across countries. To understand different levels of living standard across countries, one can derive the following per capita production function from (1): y ¼ Akα hβ ;

ð2Þ

where y  YL , k  KL , h  HL denote output per worker, physical capital per worker and human capital per worker, respectively. Equation (3), derived from (2), explains differences in growth rates across countries: gy ¼ gA þ αgk þ βgh ;

ð3Þ

where gx  dlogx dt denotes the growth rate of x for x ∈ fA; k; hg. This one-sector growth model is popular not only because it provides a simple conceptual framework of economic growth but also because it can successfully generate the Kaldor facts observed in the data of advanced economies. Moreover, it proves to be a useful quantitative framework for growth accounting, which decomposes economic growth into the contributions from the accumulation of each of the tangible and intangible inputs and TFP using (1)–(3). In addition, the obtained Solow residual term A plays a critical role in all kinds of macro-development analyses including, for example, economic fluctuations. However, when this one-sector framework is applied to address the fundamental question why some countries are richer (higher y) or growing faster (higher gy) than others, investigators are easily induced to only focus on the quantitative difference in A,k,h or their growth rates without seriously considering structural differences across countries at different development stages. For example, endowment structure, defined as the composition of different factor endowment (including land, labour, human capital, physical capital, etc.), is different for a country at a different stage of development. The composition of agriculture, industry, and service is also different at different income levels. Even within the manufacturing sector, the composition of sub-industries with different capital intensities (ranging from labour-intensive apparel industry to

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

capital-intensive precision equipment), or industrial structure, is also different at different income levels. Moreover, financial structure, defined as the composition of different forms of financial intermediaries (including banks of different sizes, stock market, and venture capital, etc.), is also likely to be different at different income levels; the composition, stock, and quality of public goods (such as infrastructure) and public services (such as property rights protections, human security protections, supervision of public health and financial risks, etc.) are also generally different at different income levels. Clearly, none of these structural differences and entailed policy implications can be effectively explored in the one-sector growth model.

3.3 W S M  D

.................................................................................................................................. Why do these different economic structures matter? Because negligence of these structural differences and their determinants could easily result in misleading policy suggestions that hamper economic development. In retrospect, the first wave of dominant thinking in development economics is the structuralism prevailing in the 1950s and 1960s. The proponents of structuralism observed that industrial structures are different between rich and poor countries. Industries in rich countries are generally more capital-intensive than in poor countries and the terms of trade are also in favour of rich countries. They argue that it is imperative for developing countries to establish the same industries as those in developed countries as quickly as possible, and that market failure prevents those heavy (capital-intensive) industries from emerging quickly enough in developing countries. The policy implication is, therefore, that government should provide large enough subsidies to capital-intensive industries together with import-substitution protectionist trade policies to give a big push to those ‘modern industries’. Unfortunately, it turns out that such development strategies have failed in practice. The key reason is the failure of these structuralists to understand that the optimal industrial structure is endogenous and should be consistent with the endowment structure of the economy at a given stage of development (Lin 2012a, 2013a). Promoting capital-intensive industries prematurely violates the comparative advantage in factor endowment of poor countries. It would result in the need to protect non-viable firms, encouraging rent seeking, deteriorating resource misallocation, triggering price regulations or even large scale nationalization, all of which lead to slow economic growth. The failure of the structuralism policy in practice became increasingly clear and it gradually lost favour by the 1970s. Meanwhile, Keynesian macroeconomics, which advocates active government intervention, was also increasingly challenged by the neoclassical school of macro-economists who had been formally introducing rational expectation and criticizing the ineffectiveness of Keynesian interventionist policies, especially when the Keynesian theory failed to explain stagflation in the 1970s.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



Ever since the 1980s, a new wave of social thought in development economics, namely, neo-liberalism, had gradually became dominant. The most representative policy advice of neo-liberalism is the so-called Washington Consensus, which emphasizes the fundamental role of the market rather than the state and highlights the importance of privatization, liberalization, and stabilization. Whereas this approach helps to initiate market-oriented reforms and deregulation, which are important in helping to improve the micro-incentives of individual households and firms, ameliorate efficiency of resource allocation, and create market environment that is more conducive to economic growth, the limitations of this approach are also enormous. Neo-liberalism sets the market institution of developed countries as the uniform target of reforms for all developing countries and advocates spontaneous structural transformation without a role for the state other than protecting private property rights and maintaining social order. One manifestation of the failure of the neo-liberalist approach is the ‘shock therapy’, which proposes that all market reforms (especially privatization) should be completed as rapidly as possible because all institutions and policies are interrelated, presumably making partial and gradual reforms more distorting and harmful than a thorough and once-and-for-all grand reform (Murphy et al. 1992). The former Soviet Union is the stereotype of a country that adopts the shock therapy, but it turned out to be a disaster: the state became too weak and society became unstable, the unemployment rate skyrocketed, and GDP growth plunged immediately due to the collapse of non-viable firms in previously protected industries. In fact, even today the Russian economy has not yet fully recovered since the reform began more than 25 years ago. Other East European countries that adopted the shock therapy also suffered similar problems. A further symptom of neo-liberalism is the low feasibility of implementing all the prescribed comprehensive reforms. Governments in poor countries are usually tightly constrained in terms of the fiscal resources, manpower, and political support necessary to allow them to complete radical and comprehensive across-the-board reform unless sufficient foreign support is freely available. However, foreign backing is not always available, and even when it is available, there is often a long list of preconditions that require comprehensive reforms beyond the capacity of the administration, especially when leaders have fixed incumbency terms and face tight time constraints. As a consequence, a reform agenda of this type is often tabled or poorly implemented. To summarize, the old structuralism fails because it mistakenly takes optimal industrial structures as exogenous and independent of the stage of development, and neo-liberalism fails because it erroneously takes optimal institutions as exogenous and independent of the stage of development. As a result, for any developing economy, the policy prescription from the old structuralism is an immediate and thorough imitation of the same industrial structures observed in the developed countries regardless of its own current endowment structure and stage of development. In constrast, the policy prescription from the neo-liberalism is an immediate and thorough transplantation of the same economic, political, and legal institutions prevailing in developed countries regardless of its own current stage of development and institutional history. The common mistake of these two approaches is a failure to recognize that the optimal

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

economic structures and institutions are both endogenously different for countries at different stages of development. In other words, we should not take the structures, including industries and institutions, of developed countries as the unique and timeinvariant optimal choice for all countries regardless of their stage of development. These two waves of problematic social thoughts have significantly influenced policy makers. As a consequence, most countries that have faithfully adopted these approaches do not achieve what was expected according to the theories. In fact, in the past sixty years, among more than 200 developing economies, only two economies (South Korea and Taiwan Province of China) have successfully upgraded from a lowincome status to a high-income status (Lin and Rosenblatt 2012), and only thirteen out of 101 middle-income economies have moved up the ladder and become high-income economies by 2008 (Agenor et al. 2012). Strikingly, among all the economies that have successfully escaped the low-income trap or the middle-income trap in the past eighty years, none has strictly followed either the structuralism approach or the neo-liberalism approach. Instead, each has adopted a more pragmatic approach by developing industries that are consistent with the endowment structures and by continuously upgrading their industries as their endowment structures change. Meanwhile, fast-growing transitional economies such as China have adopted a gradualist approach to institutional reform instead of overnight privatization and radical institutional transplanting as prescribed by neo-liberalism (Lin 2009). Such a huge discrepancy between mainstream theoretical prescriptions and realworld performance cannot be resolved without a new theory in development economics. This is where New Structural Economics enters the field.

3.4 A B M  N S E

.................................................................................................................................. A hallmark technical feature of NSE is that the aggregate production function is no longer taken as exogenous and time-invariant, as in (1). Instead, it is derived from the compositions of underlying industries which are in turn determined by the endowment structure. Moreover, when the endowment structure evolves over time, the optimal composition of industries also changes accordingly, which further implies that the functional form of the aggregate production function may also change over time. The key economic idea of NSE behind this technical feature is that endowment structure determines optimal industrial structures and that capital accumulation (improvement of endowment structure) serves as a fundamental mechanism that drives changes in industrial structures. An important technical challenge for such formal models is to characterize the dynamic optimization problem in the presence of many industries, each of which

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



evolves nonlinearly. More precisely, first, the model predictions should not only be consistent with the stylized facts at the disaggregated industry level (to be discussed in more details below), but should also be consistent with the Kaldor facts at the aggregate level.¹ Second, to keep track of the life-cycle dynamics of each industry along the whole path of aggregate growth, we must fully characterize transitional dynamics, which is well recognized to be difficult even for a two-sector model, but now we have infinite industries with an infinite time horizon.² Third, it turns out that endogenous structural change in the underlying industries eventually forces us to characterize a Hamiltonian system with endogenously switching state equations instead of a time-invariant state equation as in the Ramsey model. To be more concrete, we will illustrate how we model this dynamic evolution in industrial structures on the growth path. This part is mainly taken from Ju et al. (2015). We show that, despite of all these technical challenges, the Ju, Lin, and Wang (JLW hereafter) model is still highly tractable: We obtain closed-form solutions to fully characterize the whole process of the hump-shaped industrial dynamics for each of the infinite industries along the aggregate growth path. The model predictions are qualitatively consistent with all the stylized facts about the industrial dynamics at the micro industry level and the Kaldor facts at the aggregate level. We first develop a static model with infinite industries (or goods, interchangeably) and two factors (labour and capital). With a general CES production function for the final commodity, we obtain a version of the Generalized Rybczynski Theorem: for any given endowment of capital and labour, there exists a cut-off industry such that, when the capital endowment increases, the output will increase in every industry that is more capital intensive than this cut-off industry, while the output in all the industries that are less capital intensive than this cut-off industry will decrease. Moreover, the cut-off industry moves toward the more capital-intensive direction as the capital endowment increases. As a special case, when the CES substitution elasticity is infinity, generically only two industries are active in equilibrium and the capital–labour ratios of the active industries are the closest to the capital–labour ratio of the economy. The model implies that the structures of underlying industries are endogenously different at different stages of economic development. Then the model is extended to a dynamic environment where capital accumulates endogenously. The dynamic decision is decomposed into two steps. First, the social planner optimizes the inter-temporal allocation of capital for the production of consumption goods, which determines the evolution of the endowment structure. Then, at each time point the resource allocation across different industries is determined by the capital and labour endowments in the same way as the static model. Endogenous changes in the industrial composition of an economy translate into different functional ¹ Kaldor facts refer to the relative constancy of the growth rate of total output, the capital–output ratio, the real interest rate, and the share of labour income in GDP. ² See King and Rebelo (1993); Mulligan and Sala-i-Martin (1993); Bond et al. (2003); and Mehlum (2005).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

forms of the endogenous aggregate production function and the capital accumulation function; therefore, we must solve a Hamiltonian system with endogenously switching state equations because of the endogenous structural change.

3.4.1 Model Environment Consider a closed economy with a unit mass of identical households and infinite industries. Each household is endowed with L units of labour and E units of physical capital, which can be easily extended to incorporate intangible capital as well. A representative household consumes a composite final commodity C, which is produced by combining all the intermediate goods cn, where n ∈ f0; 1; 2; . . .g. Each intermediate good should be interpreted as an industry, although we will use ‘good’ and ‘industry’ interchangeably throughout the chapter. For simplicity, assume that the production function of the final commodity is C ¼ Σ∞ n¼0 λn cn ;

ð4Þ

where λn represents the marginal productivity of good n in the final good production.³ We require cn  0 for any n. The final commodity serves as the numeraire. The utility function is CRRA: U¼

C 1σ  1 ; where σ ∈ ð0; 1: 1σ

ð5Þ

All the technologies exhibit constant returns to scale. In particular, good 0 is produced with labour only. One unit of labour produces one unit of good 0. To produce any good n  1, both labor and capital are required and the production functions are Leontief:⁴
 k ;l ; ð6Þ Fn ðk; lÞ ¼ min an where an measures the capital intensity of good n. All the markets are perfectly competitive. Let pn denote the price of good n. Let r denote the rental price of capital and w denote the wage rate. The zero profit condition for a firm implies that p₀ = w and pn ¼ w þ an r for n  1.

³ It is not unusual in the growth literature to assume perfect substitutability for the output across different production activities; see Hansen and Prescott (2002). This assumption is relaxed and the general CES function is discussed in section 5 of Ju et al. (2015). ⁴ Leontief functions are also used in Luttmer (2007) and Buera and Kaboski (2012a, 2012b). It can be easily shown that our key qualitative results will remain valid when the production function is Cobb– Douglas, but that will enormously increase the nonlinearity of the problem in the multiple-sector environment, making it much harder to obtain closed-form solutions, especially for the dynamic analysis.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



Without loss of generality, the industries are ordered such that an is increasing in n. Empirical evidences suggest that a more capital-intensive technology is generally more productive, we assume that λn is increasing in n. To obtain analytical solutions, we assume λn ¼ λn ; an ¼ an ;

ð7Þ

a  1 > λ > 1:

ð8Þ

a > λ must be imposed to rule out the trivial case that only the most capital-intensive good is produced in the static equilibrium, and we strengthen the assumption further to a  1 > λ to simplify the analysis as good 0 requires no capital.⁵ The household problem is to maximize (1) subject to the following budget constraint: C ¼ wL þ rE:

ð9Þ

3.4.2 Market Equilibrium It is shown that at most two goods are simultaneously produced in the equilibrium and that these two goods have to be adjacent in the capital intensities. Suppose goods n and n + 1 are produced for some n  1, then the marginal rate of transformation (MRT) between the two intermediate goods must be equal to their price ratio: nþ1 r ¼ wþa MRTnþ1;n ¼ λ ¼ ppnþ1 wþan r , which yields n r λ1 ¼ n : w a ða  λÞ

ð10Þ

In addition, condition (8) ensures that good 0 is not produced. The market clearing conditions for labour and capital are given respectively by: cn þ cnþ1 ¼ L;

ð11Þ

cn an þ cnþ1 anþ1 ¼ E:

ð12Þ

The market equilibrium can be illustrated in Figure 3.1, where the horizontal and vertical axes are labor and capital, respectively. Point O is the origin and Point W = (L,E) denotes the endowment of the economy. When an L < E < anþ1 L; as shown in the current case, only goods n and n + 1 are produced. The factor market clearing conditions, (11) and (12), determine the equilibrium allocation of labour and capital in industries n and n + 1, which are represented respectively by vector OA and vector OB   in the parallelogram OAWB. Oan ¼ ð1; an Þcn and Oanþ1 ¼ ð1; anþ1 Þcnþ1 are the vectors ⁵ If λ ¼ 1, the equilibrium would be trivial because only good 0 is produced in this linear case. In section 5 of Ju et al. (2015), λ ¼ 1 is allowed when (4) is replaced by a general CES function.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



      E

an+1

B'

W' an W

B A' 0

an–1 A L

 . How industrial structures are determined by endowment structures

of factors used in producing cn and cnþ1 in the equilibrium. If the capital increases so the endowment point moves from W to W0 , the new equilibrium becomes parallelogram OA0 W0 B0 so that cn decreases but cnþ1 increases. When E ¼ an L, only good n is produced. Similarly, if E ¼ anþ1 L; only good n+1 is produced.⁶ More precisely, the equilibrium output of each good cn, the relative factor prices wr , and the corresponding aggregate output C are summarized in Table 3.1. The static equilibrium is summarized verbally in the following proposition. Proposition 1 Generically, there exist only two industries whose capital intensities are the most adjacent to the aggregate capital–labour ratio, EL : As EL increases, each industry n (n  1) exhibits a hump shape: the output first remains zero, then increases and reaches its peak and then declines, and finally returns to zero and is fully replaced by the industry with the next higher capital intensity. The equilibrium outcome, as summarized in the above proposition and Table 3.1, shows that the aggregate production function (C as a function of L and E) has different forms when the endowment structures are different, reflecting the endogenous structural change in the underlying industries. Accordingly, the coefficient right before E in the endogenous aggregate production function is the rental price of capital, and the coefficient before L is the wage rate. So the relative factor price is wr ¼ anλ1 ðaλÞ when E ∈ an L; anþ1 LÞ, and it declines in a stair-shaped fashion as E increases. This discontinuity results from the Leontief assumption. Observe that the capital income share in the total output is given by ⁶ This graph may appear similar to the Lerner diagram in the H-O trade models with multiple diversification cones (see Leamer (1987). However, the mechanism in our autarky model is different from the international specialization mechanism in the trade literature.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



Table 3.1 Static equilibrium 0  E < aL

an L  E < anþ1 L for n  1

c0 ¼ L  Ea

cn ¼

c1 ¼ Eai

cnþ1 ¼

cj ¼ 0 for 8j 6¼ 0; 1

cj ¼ 0 for 8j 6¼ n; n þ 1

r w

r w

¼ λ1 a

C ¼ L þ ðλ  1Þ Ea a , E0;1 ¼ λ1 ðC  LÞ

¼

Lanþ1  E anþ1  an E i  a n Li anþ1  an

λ1 an ða  λÞ

λnþ1  λn λn ða  λÞ L E þ a1 anþ1  an   λn ða  λÞ anþ1  an L nþ1 , En;nþ1 ¼ C  : a1 λ  λn

C¼

λ1 rE a1 E ¼ n λ1 rE þ wL E þ a ðaλÞ L a1

ð13Þ

a1

when E ∈ an L; anþ1 LÞ for any n  1. So the capital income share monotonically increases with capital within each diversification cone and then suddenly drops to ðλ1Þ a1 as the economy enters a different diversification cone, but the capital income share λ1 ðλ1Þa ; ða1Þλ for any n  1. This is consistent with the always stays within the interval ½a1 Kaldor fact that the capital income share is fairly stable over time.⁷

3.4.3 Dynamics Now we extend the above static model to a dynamic setting to fully characterize the industrial dynamics along the growth path of the aggregate economy, where the capital changes endogenously over time.

3.4.3.1 Environment There are two sectors in the economy: a sector producing capital goods and a sector producing consumption goods. Capital goods and consumption goods are distinct in nature and not substitutable. Moreover, they are produced with different technologies. Capital goods are produced using an AK technology: One unit of capital good produces A units of new capital goods, where A captures the effect of learning by doing. It also ⁷ See Barro and Sala-i-Martin (2003) for more discussion on the robustness of the Kaldor facts.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

highlights the feature that the technology progress is investment-specific, so it occurs in the capital (investment) goods sector rather than in the consumption goods sector (see Greenwood et al. 1997).⁸ Let KðtÞ denote the capital stock available at the beginning of time t, so the output flow coming out of the capital good sector is AK(t), which is then split between two different usages: AKðtÞ ¼ XðtÞ þ EðtÞ;

ð14Þ

where X(t) denotes capital investment and E(t) denotes the flow of capital used to produce consumption goods at t. E(t) fully depreciates, so capital in the whole economy accumulates as follows: 

KðtÞ ¼ XðtÞ  δKðtÞ;

ð15Þ

where δ is the depreciation rate in the capital goods sector. Substituting (14) into the above equation and defining ξ ¼ A  δ, we obtain 

KðtÞ ¼ ξKðtÞ  EðtÞ: At time t, capital E(t) and labour L (assumed to be constant) produce all the intermediate goods fcn ðtÞg∞ n¼0 with technologies specified by (6), which are ultimately combined to produce the final consumption good C(t) according to (4). Based on Table 3.1, define 8 ðλ  1Þ > < EþL if 0  E < aL a ; FðE; LÞ  ð16Þ nþ1 n n λ λ ða  λÞ > n nþ1 :λ L if a L  E < a Lforn  1 Eþ a1 anþ1  an which is the endogenous aggregate production function derived in Table 3.1. Therefore, CðtÞ ¼ FðEðtÞ; LÞ ¼ rðtÞEðtÞ þ wðtÞL;

ð17Þ

where r(t) and w(t) are the rental price for capital and the wage rate at time t, respectively. With some abuse of notation, let E(C(t)) denote the total amount of capital goods needed to produce final consumption goods C(t), so FðEðCðtÞ; LÞ  CðtÞ. Final consumption goods C and all the intermediate goods fcn g∞ n¼0 are non-storable. By the second welfare theorem, we can characterize the competitive equilibrium by resorting to the following social planner problem: Z

∞

max cðtÞ

0

c ðtÞ1σ  1 ρt e dt 1σ

ð18Þ

⁸ Notice that this dynamic setting differs from the most standard setting where capital goods and consumption goods are identical goods. Detailed comparisons and justifications are provided in subsection 4.1.3 and section 5 of Ju et al. (2015).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



subject to 

K ðtÞ ¼ ξKðtÞ  EðCðtÞÞ;

ð19Þ

Kð0Þ ¼ K0 is given; where ρ is the time discount rate. We assume ξ  ρ > 0 to ensure positive consumption growth and, to exclude the explosive solution, we also assume ξρ σ ð1  σÞ < ρ. Putting them together, we impose 0 < ξ  ρ < σξ:

ð20Þ

The social planner decides the inter-termporal consumption flow C(t) and makes optimal investment decisions X(t), which in turn determine the evolution of the endowment structure KðtÞ L and the optimal amount of capital allocated for consumption goods production E(t). Note that, at any given time t, once E(t) is determined, the optimization problem for the whole consumption goods sector is exactly the same as the static problem as in the previous subsection. From the bottom row of Table 3.1, we know that E(C) is a strictly increasing, continuous, piece-wise linear function of C. It is not differentiable at C ¼ λi L, for any i ¼ 0; 1; . . . . Therefore, the above dynamic problem may involve changes in the functional form of the state equation: (19) can be explicitly rewritten as 8 when C < L < ξK;  when L  C < λL ; K ¼ ξK  E0;1 ðCÞ; : n nþ1 ξK  En;nþ1 ðCÞ; when λ L  C < λ L; forn  1 where En;nþ1 ðCÞ is defined in the bottom row of Table 3.1 for any n  0.

3.4.3.2 Equilibrium Characterization We can verify that the objective function is strictly increasing, differentiable, and strictly concave while the constraint set forms a continuous convex-valued correspondence, hence the equilibrium must exist and also be unique. Let t₀ denote the last time point when aggregate consumption equals L (that is, only good 0 is produced), and tn denote the first time point when C ¼ λn L (that is, only good n is produced) for n  1: As can be verified later, aggregate consumption C is monotonically increasing over time in equilibrium, hence the problem can also be written as Z t0 ∞ Z tnþ1 X C ðtÞ1σ  1 ρt C ðtÞ1σ  1 ρt e dt þ e dt max 1σ 1σ cðtÞ 0 n¼0 tn subject to

8 when 0  t  t0 < ξK when t0  t  t1 ; K ¼ ξK  E0;1 ðCÞ; : ξK  En;nþ1 ðCÞ; when tn  t  tnþ1 ; forn  1 

Kð0Þ ¼ K0 is given; where tn is to be endogenously determined for any n  0.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

Table 3.1 indicates that goods 0 and 1 are produced during the time period [t₀,t₁] and a ðC  LÞ. When tn  t  tnþ1 for n  1; goods n and n+1 are EðCÞ ¼ E0;1 ðCÞ  λ1 n nþ1 n ðaλÞ L aλnþ1 a . If K₀ is suffiproduced. Correspondingly, EðCÞ ¼ En;nþ1 ðCÞ  ½C  λ a1 λn ciently small (this is more precisely shown in Proposition 3), then there exists a time period [0,t₀] in which only good 0 is produced and all the working capital is saved for the future, so E = 0 when 0  t  t0 . If K₀ is large, on the other hand, the economy may start by producing goods h and h+1 for some h  1, so t0 ¼ t1 ¼ . . . ¼ th ¼ 0 in equilibrium. To solve the above dynamic problem, following Kamien and Schwartz (1991), we set the discounted-value Hamiltonian in the interval tn  t  tnþ1 , and use subscripts ‘n,n+1’ to denote all the variables during this time interval: Hn;nþ1 ¼

CðtÞ1σ  1 ρt e þ ηn;nþ1 ½ξKðtÞ  En;nþ1 ðCðtÞÞ 1σ

nþ1 L  CðtÞÞ þ ζnn;nþ1 ðCðtÞ  λn LÞ þζnþ1 n;nþ1 ðλ

ð21Þ

n where ηn;nþ1 is the co-state variable, ζnþ1 n;nþ1 and ζn;nþ1 are the Lagrangian multipliers for nþ1 the two constraints λ L  CðtÞ  0 and CðtÞ  λn L  0, respectively. The first-order condition and Kuhn-Tucker conditions are

@Hn;nþ1 anþ1  an n ¼ CðtÞσ eρt  ηn;nþ1 nþ1  ζnþ1 n;nþ1 þ ζn;nþ1 ¼ 0; @C λ  λn

ð22Þ

nþ1 nþ1 ζnþ1 L  CðtÞÞ ¼ 0; ζnþ1 L  CðtÞ > 0; n;nþ1 ðλ n;nþ1  0; λ

ζnn;nþ1 ðCðtÞ  λn LÞ ¼ 0; ζnn;nþ1  0; CðtÞ  λn L  0: We also have η0n;nþ1 ðtÞ ¼ 

@Hn;nþ1 ¼ ηn;nþ1 ξ: @K

ð23Þ

n In particular, when CðtÞ ∈ ðλn L; λnþ1 LÞ, ζnþ1 n;nþ1 ¼ ζn;nþ1 ¼ 0; and equation (22) becomes

CðtÞσ eρt ¼ ηn;nþ1

anþ1  an : λnþ1  λn

ð24Þ

The left hand side is the marginal utility gain from increasing one unit of aggregate consumption, while the right hand side is the marginal utility loss due to the decrease in capital because of that additional unit of consumption, which by the Chain’s Rule can be decomposed into two multiplicative terms: the marginal utility of capital ηn;nþ1 and the marginal capital requirement for each additional unit of aggregate consumpnþ1 n (see Table 3.1). Taking the logarithm on both sides of equation (24) and tion aλnþ1 a λn differentiating with respect to t, we obtain the consumption growth rate from the regular Euler equation:

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  





CðtÞ ξ  ρ ¼ ; gc  CðtÞ σ

ð25Þ

for tn  t  tnþ1 for any n  0. The strictly concave utility function implies that the optimal consumption flow C(t) must be continuous and sufficiently smooth (without kinks); hence from (25) we obtain: CðtÞ ¼ Cðt0 Þ e gc ðtt0 Þ for any t  t0 > 0:

ð26Þ

Following Kamien and Schwartz (1991), we have two additional necessary conditions at t ¼ tnþ1 : Hn; nþ1 ðtnþ1 Þ ¼ Hnþ1;nþ2 ðtnþ1 Þ;

ð27Þ

ηn;nþ1 ðtnþ1 Þ ¼ ηnþ1;nþ2 ðtnþ1 Þ:

ð28Þ

Substituting equations (27) and (28) into (21), we can verify K  ðtnþ1 Þ ¼ K þ ðtnþ1 Þ. In other words, K(t) is also continuous. Observe that Cðt0 Þe gc ðtn t0 Þ ¼ Cðtn Þ ¼ λn L when t0 > 0;

that ð29Þ

which implies λ L log Cðt þ ξρ σ t0 0Þ n

tn ¼

gc

; when t0 > 0:

ð30Þ

Define mn  tnþ1  tn , which measures the length of the time period during which both good n and good n+1 are produced (that is, the duration of the diversification cone for good n and good n+1). We must have mn ¼ m 

logλ : gc

ð31Þ

The comparative statics for equation (31) is summarized in the following proposition. Proposition 2 The full life span of each industry n  1 is equal to 2m. The speed of 1 ) decreases with the productivity parindustrial upgrading (measured by frequency 2m ameter λ but increases with the aggregate growth rate gc. More precisely,the industrial upgrading is faster when technological efficiency ξ increases, or the inter—temporal elasticity of substitution σ1 increases, or the time discount rate ρ decreases. The intuition for the proposition is the following. Suppose good n and good n+1 are produced. When the productivity parameter λ is larger, the marginal productivity of good n (λn ) becomes bigger, making it pay to stay at good n longer; but the marginal productivity of good n+1 (λnþ1 ) also becomes bigger, making it optimal to leave good n and move to good n+1 more quickly. It turns out that the first effect dominates the second effect because (8) implies that, by climbing up the industrial ladder, the productivity gain l is sufficiently small relative to the additional capital cost reflected by the cost parameter a. Thus the net effect is that industrial upgrading slows down. On

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

the other hand, industrial upgrading is faster when the consumption growth rate gc increases because larger consumption is supported by more capital-intensive industries, as implied by Table 3.1. When the household is more impatient (larger ρ), it will consume more and save less and hence capital accumulation becomes slower and thus the endowment-driven industrial upgrading also becomes slower. When the production of the capital good becomes more efficient (ξ), capital can be accumulated faster, so the upgrading speed is increased. When the aggregate consumption is more substitutable across time (larger σ1), the household is more willing to substitute current consumption for future consumption, which also boosts saving and then causes quicker industrial upgrading. We are now ready to derive the industrial dynamics for the entire time period. The industrial dynamics depend on the initial capital stock, K(0) We show in the Appendix of Ju et al. (2015) that there exists a series of increasing constants, ϑ0 ; ϑ1 ;   ; ϑn ; ϑnþ1 ;   ; such that if 0 < Kð0Þ  ϑ0 ; the economy will start by producing good 0 only until the capital stock reaches ϑ0 ; if ϑn < Kð0Þ  ϑnþ1 ; the economy will start by producing goods n and n+1 for any n  0: Furthermore, we can show that Kðtn Þ  ϑn for any Kð0Þ < ϑn : That is, irrespective of the level of initial capital stock, the economy always starts to produce good n+1 whenever its capital stock reaches ϑn . To be more concrete, consider the case when ϑ0 < Kð0Þ  ϑ1 , where the threshold values ϑ0 and ϑ1 can be explicitly solved. That is, the economy will start by producing goods 0 and 1. Equation (26) and Table 3.1 jointly implies that when t ∈ 0,t₁], ξρ a a ðCðtÞ  LÞ ¼ ðCð0Þe σ t  LÞ: EðtÞ ¼ λ1 λ1 Correspondingly, 

K ¼ ξKðtÞ 

ξρ a ðCð0Þe σ t  LÞ: λ1

Solving this first-order differential equation with the condition K(0) = K₀, we obtain " # aCð0Þ ξρ  aCð0Þ aL aL þ K0 þ ξρλ1 þ eξt ; KðtÞ ¼ ξρ λ1 e σ t þ ξðλ  1Þ ξðλ  1Þ  ξ  ξ σ σ which yields

" #

ξσ aCð0Þ aL aL λL ξρ λ1 þ K0 þ ξρ : þ þ ϑ1  Kðt1 Þ ¼ ξρ ξðλ  1Þ ξðλ  1Þ Cð0Þ σ ξ σ ξ aλL  λ1

When t ∈ tn ; tnþ1 , for any n  1; the transition equation of capital stock (19) becomes   ξρ  λn ða  λÞ anþ1  an t σ L nþ1 when t ∈ tn ; tnþ1 ; for any n  1: K ¼ ξKðtÞ  Cð0Þe  a1 λ  λn

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



Solving the above differential equation, we obtain: ξρ

KðtÞ ¼ αn þ βn e σ t þ γn eξt when t ∈ tn ; tnþ1 ; for any n  1 where

ð32Þ

nþ1

a  an λn ða  λÞL ; αn ¼  nþ1 λ  λn ξða  1Þ nþ1

a  an Cð0Þ  ; βn ¼  nþ1 n ξρ λ λ  ξ σ

8 2 39 ξσ λn L ξρ < ðanþ1  an ÞL 4 1 ða  λÞ 5=  þ : ϑn þ γn ¼ ξρ Cð0Þ : λ1 ξða  1Þ ; ξ 

σ

Again the endogenous change in the functional form of the capital accumulation path (32) reflects the structural changes that underlie the aggregate economic growth. Note that fϑn g∞ n¼2 are all constants, which can be sequentially pinned down: ϑn  Kðtn Þ can be computed from equation (32) with Kðtn1 Þ known. For each individual industry, using equation (26) and Table 3.1, we obtain 8 ξρ > Cð0Þe σ t L > > when t ∈ tn1 ; tn  > n n1  < λ  λ ξρ λ  1 ; for all n  2 cn ðtÞ ¼ Cð0Þe σ t λL > > ; when t ∈ t þ ; t   nþ1 n nþ1 > >  λn λ  1 : λ 0; otherwise 8 ξρ > Cð0Þe σ t  L > > ; > < λ  1ξρ c1 ðtÞ ¼ Cð0Þe σ t λL > > ;  2 þ > > λ 1 λ  λ : 0; c 0 ðtÞ

¼

8 < :

when

t ∈ 0; t1 

when

t ∈ t1 ; t2 

;

otherwise

ξρ

L 0;

Cð0Þe σ t  L ; λ1

when

t ∈ 0; t1  ; otherwise

where C(0) can be uniquely determined by the transversality condition and the endogenous time points tn are given by (30) for any n  1. Recall t₀ = 0 in this case. The above mathematical equations fully characterize the industrial dynamics for each industry over the whole life cycle while aggregate consumption growth is still given by (25). If the initial capital stock is sufficiently small such that K0 < ϑ0 , then the economy will first have a constant output level equal to L (Malthusian regime) until the capital stock KðtÞ ¼ ϑ0 , which occurs at t0 > 0, after which the aggregate consumption growth

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



      cn∗ L

0

t1

t2 c∗0 :

t3 c∗1 :

t4 c∗2 :

t

c∗3 :

 . How different industries evolve over time

rate permanently changes to ξρ σ (Solow regime). All these mathematical results can be read as follows: Proposition 3 There exists a unique and strictly increasing sequence of endogenous threshold values for capital stock, fϑi g∞ i¼0 ; which are independent of the initial capital stock K(0).The economy starts to produce good n when its capital stock K(t) reaches ϑn–1 for any n  1.K(t) evolves following equation (32),while total consumption C(t) remains constant at L until t₀, after which it grows exponentially at the constant rat ξρ σ . The output of each industry follows a hump-shaped pattern: When capital stock K(t) reaches ϑn1 ; industry n enters the market and booms until capital stock K(t) reaches ϑn ; its output then declines and finally exits from the market at the time when K(t) reaches ϑnþ1 : The industrial dynamics characterized in Proposition 3 are depicted in Figure 3.2.

3.4.4 Empirical Relevance Sustainable economic growth relies on the healthy development of underlying industries, yet many important aspects still remain imperfectly understood within the context of economic growth, especially at the high-digit disaggregated industry level. Consider, for example, the automobile industry and the apparel industry. How different are the evolution patterns of two industries along the growth path of the whole economy? Which industry should we expect to expand or decline earlier than the other and why? How long, if at all, does a leading industry maintain its predominant position? What fundamental forces drive these dynamics? What is the relationship between individual industrial dynamics and aggregate GDP growth? These questions are interesting to economists, policy makers, and private investors. The JLW model is designed to shed some light on these issues by studying the dynamics of all high-digit industries simultaneously within a growth framework.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



In Ju et al. (2015), we establish four stylized facts about industrial dynamics using the NBER-CES data set of the US manufacturing sector, which covers 473 industries at the 6-digit NAICS level from 1958 to 2005: Fact 1 (cross-industry heterogeneity): There exists tremendous cross-industry heterogeneity in capital–labour ratios, capital expenditure shares, and labour productivity. Fact 2 (hump-shaped dynamics): An industry typically exhibits a hump-shaped dynamic pattern: its value-added share first increases, reaches a peak, and then declines. Fact 3 (timing fact): The more capital intensive an industry is, the later its valueadded share reaches its peak. Fact 4 (congruence fact): The further an industry’s capital–labour ratio deviates from the economy’s aggregate capital–labour ratio, the smaller is the industry’s employment share. Similar patterns are also found in the UNIDO data set, which covers 166 countries from 1963 to 2009 at the two-digit level (23 sectors). In fact, documentation and analyses of a subset of the above-mentioned patterns of industrial dynamics can be dated at least back to the 1960s. For example, Chenery and Taylor (1968) show that the major products in the manufacturing sector gradually shift from the labour-intensive ones to more capital-intensive ones as an economy develops. In the JLW model, we take Fact 1 as exogenously given and the model is able to simultaneously explain Facts 2, 3, and 4. Meanwhile, the theoretical results are also consistent with the Kaldor facts that the growth rate of total consumption remains constant and the capital income share is relatively stable, as shown in equation (13).

3.4.5 Related Literature The JLW model is most closely related to the growth literature on structural change. This literature mainly tries to match the Kuznets facts, namely, that the agricultural share in GDP has a secular decline, the industry (manufacturing) share demonstrates a hump shape, and the service share increases. However, such sectors are too aggregated to address questions such as those raised earlier. Insufficient effort is devoted to reconciling Kaldor facts with the aforementioned stylized facts about industrial dynamics at the disaggregated levels. The JLW model has two theoretical contributions to the literature. First, we develop a highly tractable growth model with infinite industries to fully characterize the industrial dynamics, which are qualitatively consistent with the four motivating facts. Second, more importantly, we show how capital accumulation serves as a new independent mechanism that drives the structural change. The existing literature mainly discusses two mechanisms of structural change in autarky. One is the

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

preference-driven mechanism, in which the demand for different goods shift asymmetrically as income increases due to the non-homothetic preferences.⁹ A weakness of this approach is that the change of income is treated exogenously. However, one of the main purposes of development economics is to explain income change. The second mechanism is that unbalanced productivity growth rates across sectors drive resource reallocation.¹⁰ However, for developing countries the technologies are mostly exogenously given to them due to the latecomer’s advantage. Therefore, the unbalanced productivity growth rates across sectors cannot be the main drive of resource reallocation. Unlike these two mechanisms, we propose that improvement of endowment structure (capital accumulation) itself is a new and fundamental mechanism that can independently drive industrial dynamics, which we refer to as endowment-driven structural change. To highlight the theoretical sufficiency and distinction of this new mechanism, we assume a homothetic preference to shut down the preferencedriven mechanism. We also assume that productivity is constant over time in all the industries to shut down the productivity-driven mechanism. Instead, the model, as motivated by Fact 1, assumes that industries differ in their capital intensities, which deviates from the standard assumption that different sectors have equal capital intensity, including models without capital.¹¹ An important exception is Acemoglu and Guerrieri (2008), who study structural change in a two-sector growth model with different capital intensities, but their model does not explain or generate the repetitive hump-shaped industrial dynamics because the life cycle of each sector in their model is truncated. In fact, their analytical focus is on the asymptotic aggregate growth rate in the long run, by which time one industry dominates the economy in terms of employment share and structural change virtually ends. In contrast, we have infinite sectors so the structural change goes on endlessly and this setting allows us to analyse the complete life-cycle dynamics of every industry at the disaggregated levels during the whole growth process. Ngai and Pissarides (2007) study structural change in a growth model with an arbitrary but finite number of sectors, which potentially allows for the life-cycle analysis of disaggregated industries. However, they do not treat capital accumulation as a major driving force for structural change, even in their appendix where they introduce different capital intensities across different sectors. Nor do they attempt to keep track of the life cycles of each industry to explain their dynamics along the growth ⁹ For instance, the Stone-Geary function is used in Laitner (2000); Caselli and Colemen (2001); Kongsamut et al. (2001); Gollin et al. (2007). Hierarchic utility functions are adopted in Matsuyama (2002); Foellmi and Zweimuller (2008); Buera and Kaboski (2012a); among others. ¹⁰ See, e.g., Hansen and Prescott (2002); Ngai and Pissarides (2007); Duarte and Restuccia (2010); Uy et al. (2013). ¹¹ Ngai and Samaneigo (2011) study how R&D differs across industries and contributes to industryspecific TFP growth. Acemoglu (2007) argues that technology progress is endogenously biased toward utilizing the more abundant production factors, which indicates that endowment structure is also fundamentally important even in accounting for TFP growth itself.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



path. Moreover, structural change will disappear in the long run because there are finite sectors in their model. Ju et al. (2015) is also closely related to the strand of growth literature that studies the life cycle dynamics of industries, firms, establishments, or products. The key mechanisms that drive the life-cycle dynamics are different in different models. For example, some highlight the role of innovation and creative destruction (see Stokey 1988; Grossman and Helpman 1991; Aghion and Howitt 1992; Jovanovic and MacDonald 1994); some highlight the role of specific intangible capital such as organizational capital (see Atkeson and Kehoe 2005) or technology-specific or industry-specific human capital (see Chari and Hopenhayn 1991; Rossi-Hansberg and Wright 2007); some focus on productivity change and destruction shocks (see Hopenhayn 1992; Luttmer 2007; Samaniego 2010)), still others highlight demand shift due to consumers’ heterogeneous preferences together with product awareness (Perla 2013) or nonhomothetic preferences (Matsuyama 2002). Ju et al. (2015) differ from and complement these approaches by focusing on the role of endowment structure via the endogenous relative factor prices.

3.4.6 Policy Implications The JLW model shows that the optimal industrial structures are different when endowment structures are different and optimal growth is achieved only when the industrial development follows the comparative advantage of the endowment structures of the economy. If a country follows a comparative-advantage-defying development strategy by prematurely boosting industries whose capital intensity is too high for the endowment structures of that country, it would lower the GDP growth rates and hurt the social welfare. In other words, the industries and technologies that prevail in developed countries are not necessarily suitable for developing countries to support and imitate immediately, in sharp contrast with the prescriptions by the old structuralism in the 1950s. This model also shows that aggregate GDP growth is synchronized with development of underlying industries with the appropriate capital intensities, which suggests that formulation or evaluation of sensible development strategies and macroeconomic policies must take into account the time-varying endowment structures, induced industrial structures and industrial dynamics. In particular, the fourth fact (congruence fact), which is explained in the JLW model, may provide a useful policy guidance for what kind of industries are most likely to be the dominant ones at each different stage of development. The fact that the model has infinite sectors potentially allows policy makers to take advantage of the increasingly available ‘big data’ for supply and demand information on products at the high-digit levels and formulate growth policies that are better micro-founded at the product or industry levels at each different stage of development.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

3.5 E   JLW M

.................................................................................................................................. The JLW model introduced in the previous section characterizes the first best scenario under the perfect market environment, so the first welfare theorem applies: Pareto efficiency is achieved by the market without any necessity of government intervention. While it serves as a useful benchmark, it must be further extended to incorporate all sorts of more realistic market imperfection before we can discuss the role of government more fruitfully. In this section, we show with several concrete examples how the JLW model could serve as a workhorse for NSE in various extensions.

3.5.1 International Trade Ju et al. (2015) discuss a simple extension of their model to a small open economy and the key results remain unchanged. Wang (2014a) extends the benchmark autarky setting in Ju et al. (2015) to an environment with two large countries, so terms of trade are now endogenous. Closed-form characterizations are provided to show how international trade and dynamic trade policies affect industrial dynamics and economic growth. Two main results are obtained: (a) both industrial upgrading and the aggregate growth of an economy are facilitated by the investment-specific technology progress (ISTP) of the trade partner if and only if the inter-temporal elasticity of substitution exceeds unity; (s) accelerating trade liberalization has a non-monotonic impact on aggregate output growth and industrial dynamics, depending on the level of trade cost and the inter-temporal elasticity of substitution.

3.5.2 Non-Competitive Market Structure Wang (2014b) relaxes the assumption of perfect competitive market structures in Ju et al. (2015) to capture a more realistic situation because the first adopters of a new (and also more capital intensive) technology in developing countries sometimes enjoy certain temporary de facto market power, which disappears after this technology is implemented for some time. It is shown that the temporary non-competitive market structure in the goods market indirectly distorts factor market price signals, which in turn affects the dynamic implementation decisions of the new technology through the general equilibrium effect, even though factor markets per se are perfect. In particular, under certain circumstances, an increase in initial capital endowment may delay rather than facilitate the adoption of a more-capital intensive technology because the monopoly profits are higher in the later period when capital becomes cheaper. The policy implication is that foreign aid may in some cases inefficiently delay rather than

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



accelerate industrial upgrading in developing countries when the final goods market is imperfectly competitive.

3.5.3 Marshallian Externality and Industrial Policies There is no role for industrial policies in the world of Ju et al. (2015) because there is no market failure. Ju, et al. (2011) explore optimal industrial policies by introducing Marshallian externality into the Ju et al. (2015) model. The model deviates from the standard setting in the existing literature of industrial policies on two important dimensions. The first deviation is that more than one sector exhibits Marshallian externality, so how to identify the right industrial target is an endogenous decision rather than taken as given. The second deviation is that both capital and labour are needed, not just labour, in the production of industries with different capital intensities, so relative price signals in factor markets change as the economy develops, which has an asymmetric impact on different industries. The model highlights the importance of factor market price signals in guiding the government to target the correct industries to support at each different stage of development. We show that the if government adopts an industrial policy to facilitate the growth of industries that is consistent with comparative advantages determined by endowment structures, the appropriate government intervention could overcome the coordination failure and create Pareto improvement over the laissez-faire market equilibrium allocation. However, if government targets an industry that violates the factor endowment-determined comparative advantage, such industrial policies would result in an outcome worse than the intervention-free market outcome, despite the existence of Marshallian externality. So NSE proposes a market-led-and-government-facilitated approach of industrial policies, which is different from both the old structuralism approach (ignoring the positive role of market) and the neo-liberalism approach (downplaying the positive role of state).

3.5.4 Frictional Labour Market Li and Wang (2017) study how a frictional labour market affects industrial upgrading and how labour market dynamics are shaped by industrial upgrading in the context of structural change and economic growth. To do this, we relax the assumption of perfect labour market in the JLW model by introducing search and match processes both within and across industries. Labour reallocation across the capital-intensive sunrise and the labour-intensive sunset industries is plagued by a mismatch between heterogeneous workers and the jobs which are created and destructed asymmetrically and endogenously as the economy develops. Mismatch is shown to delay industrial upgrading and depress growth by preventing workers from smoothly moving into the sunrise industries. On the other hand, industrial upgrading amplifies the role of mismatch in

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

affecting the dynamics of unemployment, wage inequality, and output volatility. Quantitative investigations suggest that those effects are significant.

3.5.5 Further Discussions NSE highlights the endogeniety of various dimensions of economic structures for understanding economic development, which can be potentially applied to many research topics. For instance, NSE holds the view that the characteristics of the financial services needed by different industries can be different because of the difference in their capital sizes and risks. Therefore the optimal compositition of different forms of financial intermediaries (such as banks of different sizes, stock market, and venture capital) presumably differs when industrial structures change with the endowment structures, which in turns implies that optimal financial structures should be different at different stages of development (see Lin et al. 2013; Lin et al. 2015). Also, macroeconomic topics such as economic fluctuations, fiscal and monetary policies, infrastructure investment, and international capital flows can all be remodelled through the lens of NSE. See Lin (2012a, 2012b, 2013a, 2013b) for further elaborations. To formalize those ideas, we could potentially follow an approach similar to the JLW model by examining heterogeneity in certain relevant dimensions across different sectors and exploring how this heterogeneity may evolve over time and its implications for economic structures, policies, and development outcomes. A hallmark feature of this NSE methodology is to stop assuming a time-invariant, development-stage-free exogenous economic structure, instead, we should pay sufficient attention to the endogenous differences in all dimensions of economic structures between countries at different economic stages of development.

3.6 C

.................................................................................................................................. In this chapter, we have introduced the key ideas of New Structural Economics and also shown in detail how various dimensions of structural changes can be formalized in NSE. A common feature of most NSE models is to highlight the role of endowment structures and capital accumulation in determining optimal industrial and other economic structures at each different stage of development in a multiple-factor and multi-sector environment. On the technical side, a common feature of NSE models is that the aggregate production function is often endogenously derived and may change over time, so we may have to solve a dynamic system with endogenously switching state equations. Moreover, for our purpose of understanding developing countries, the analysis on transitional dynamics is often more important than the long-run steady

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



state. All of these features of modelling structural changes can be clearly seen in the benchmark JLW model. We believe that remodelling structural change in this way can be more promising and fruitful than many of the existing approaches, especially when exploring economic growth for developing countries.

R Acemoglu, Daron, 2007. ‘Equilibrium Bias of Technology’, Econometrica, 75, pp. 1371–410. Acemoglu, Daron and Veronica Guerrieri, 2008. ‘Capital Deepening and Nonbalanced Economic Growth’, Journal of Political Economy, 116 (June), pp. 467–98. Agenor, Pierre-Richard, Otaviano Canuto, and Michael Jelenic, 2012. ‘Avoiding Middleincome Growth Traps’, Economic Premise No. 98, Washington, DC: World Bank. Aghion, Philippe and Peter Howitt, 1992. ‘A Model of Growth through Creative Destruction’, Econometrica, 60 (2), pp. 323–51. Atkeson, Andrew and Patrick Kehoe. 2005. ‘ Modeling and Measuring Organization Capital.’ Journal of Political Economy, 113 (5): 1026–53. Barro, Robert and Xavier Sala-i-Martin, 2003. Economic Growth, 2nd edn, Cambridge, MA: MIT Press. Bond, Eric, Kathleen Trask, and Ping Wang, 2003. ‘Factor Accumulation and Trade: Dynamic Comparative Advantage with Endogenous Physical and Human Capital’, International Economic Review, 44 (3), pp. 1041–60. Buera, Francisco J. and Joseph P. Kaboski, 2012a. ‘The Rise of the Service Economy,’ American Economic Review, 102 (6), pp. 2540–69. Buera, Francisco J. and Joseph P. Kaboski, 2012b. ‘Scale and the Origins of Structural Change’, Journal of Economic Theory, 147 (2), pp. 684–712. Caselli, Francesco and John Coleman, 2001. ‘The U.S. Structural Transformation and Regional Convergence: A Reinterpretation’, Journal of Political Economy, 109 (June), pp. 584–617. Chari, V. V. and Hugo Hopenhayn, 1991. ‘Vintage Human Capital, Growth, and the Diffusion of New Technology’, Journal of Political Economy, 99 (6), pp. 1142–65. Chenery, Hollis B. and L. Taylor, 1968. ‘Development Patterns: Among Countries and Over Time’, Review of Economics and Statistics, 50 (4), pp. 391–416. Duarte, Margarida and Diego Restuccia, 2010. ‘The Role of Structural Transformation in Aggregate Productivity’, Quarterly Journal of Economics, 125 (1), pp. 129–73. Foellmi, Reto and Josef Zweimuller, 2008. ‘Structural Change, Engel’s Consumption Cycles and Kaldor’s Facts of Economic Growth’, Journal of Monetary Economics, 55, pp. 1317–28. Gollin, Douglas, Stephen Parente, and Richard Rogerson, 2007. ‘The Food Problem and the Evolution of International Income Levels’, Journal of Monetary Economics, 54, pp. 1230–55. Greenwood, Jeremy, Zvi Herkowitz, and Per Krusell, 1997. ‘Long Run Implications of Investment Specific Technological Progress’, American Economic Review, 87 (June), pp. 342–62. Grossman, Gene and Elhanan Helpman, 1991. ‘Quality Ladders and Product Cycles’, Quarterly Journal of Economics, 106 (May), pp. 557–86. Hansen, Gary and Edward Prescott, 2002. ‘Malthus to Solow’, American Economic Review, 92 (September), pp. 1205–17.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



     

Herrendorf, Berthold, Richard Rogerson, and Akos Valentinyi, 2014. ‘Growth and Structural Transformation’, in Philippe Aghion and Steven Durlauf, eds, Handbook of Economic Growth, Volume 2B, Amsterdam: Elsevier. Hopenhayn, Hugo, 1992. ‘Entry, Exit, and Firm Dynamics in Long Run Equilibrium’, Econometrica, 60 (2), pp. 1127–50. Jovanovic, Boyan and Glenn MacDonald, 1994. ‘The Life Cycle of a Competitive Industry’, Journal of Political Economy, 102 (2), pp. 322–47. Ju, Jiandong, Justin Yifu Lin, and Yong Wang, 2011. ‘Marshallian Externality, Industrial Upgrading, and Industrial Policies’, World Bank Policy Research Working Paper No. 5796. Ju, Jiandong, Justin Yifu Lin, and Yong Wang, 2015. ‘Endowment Structures, Industrial Dynamics, and Economic Growth’, Journal of Monetary Economics, 76, pp. 244–63. Kamien, Morton I. and Nancy L. Schwartz, 1991. Dynamic Optimization: The Calculus of Variations and Optimal Control in Economics and Management, New York: Elsevier Science Publishing Co. Inc. King, Robert G. and Sergio Rebelo, 1993. ‘Transitional Dynamics and Economic Growth in the Neoclassical Model’, American Economic Review, 83, pp. 908–31. Kongsamut, Piyabha, Sergio Rebelo, and Danyang Xie, 2001. ‘Beyond Balanced Growth’, Review of Economic Studies, 68, pp. 869–82. Kuznets, Simon, 1966. Modern Economic Growth: Rate, Structure, and Spread, New Haven, CT: Yale University Press. Laitner, John, 2000. ‘Structural Change and Economic Growth’, Review of Economic Studies, 67 (July), pp. 545–61. Leamer, Edward, 1987. ‘Path of Development in Three-Factor n-Good General Equilibrium Model’, Journal of Political Economy, 95 (October), pp. 961–99. Li, Zhe and Yong Wang, 2017. ‘Industrial Dynamics and Mistmatch’. Working Paper. Lin, Justin Yifu, 2009. Economic Development and Transition: Thought, Strategy, and Viability, Cambridge: Cambridge University Press. Lin, Justin Yifu, 2012a. New Structural Economics: A Framework for Rethinking Development Policy, Washington, DC: World Bank. Lin, Justin Yifu, 2012b. The Quest for Prosperity: How Developing Economies Can Take Off, Princeton, NJ: Princeton University Press. Lin, Justin Yifu, 2013a. ‘New Structural Economics: The Third Wave of Development Thinking’, Asia Pacific Economic Literature, 27 (2), pp. 1–13. Lin, Justin Yifu, 2013b. Against the Consensus: Reflections on the Great Recession, Cambridge: Cambridge University Press. Lin, Justin Yifu, Xifang Sun, and Ye Jiang, 2013. ‘Endowment, Industrial Structure and Appropriate Financial Structure: A New Structural Economics Perspective’, Journal of Economic Policy Reform, 16 (2), pp. 1–14. Lin, Justin Yifu, Xifang Sun, and Harry Wu, 2015. ‘Banking Structure and Industrial Growth: Evidence from China’, Journal of Banking and Finance, 58, pp. 131–43. Luttmer, Erzo G. J., 2007. ‘Selection, Growth, and the Size Distribution of Firms’, Quarterly Journal of Economics, 122 (3), pp. 1103–44. Matsuyama, Kiminori, 2002. ‘The Rise of Mass Consumption Societies’, Journal of Political Economy, 110 (October), pp. 1035–70. Mehlum, Halvor, 2005. ‘A Closed Form Ramsey Saddle Path’, The B.E. Journal of Macroeconomics, 5 (1), pp. 1–15.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  



Mulligan, Casey B. and X. Sala-i-Martin, 1993. ‘Transitional Dynamics in Two-Sector Models of Endogenous Growth’, Quarterly Journal of Economics, 108, pp. 739–73. Murphy, Kevin, Andrei Schleifer, and Robert Vishny, 1992. ‘The Tradition to a Market Economy: Pitfall of Partial Reform’, Quarterly Journal of Economics, 107, pp. 889–906. Ngai, L. Rachel and Christopher A. Pissarides, 2007. ‘Structural Change in a Multi-Sector Model of Growth’, American Economic Review, 97 (January), pp. 429–43. Ngai, L. Rachel and Roberto Samaniego, 2011. ‘Accounting for Research and Productivity Growth Across Industries’, Review of Economic Dynamics, 14 (3), pp. 475–95. Perla, Jesse, 2013. ‘Product Awareness and the Industry Life Cycle’, Working Paper, University of British Columbia. Rossi-Hansberg, Esteban and Mark Wright, 2007. ‘Establishment Size Dynamics in the Aggregate Economy’, American Economic Review, 97 (5), pp. 1639–66. Samaniego, Roberto, 2010. ‘Entry, Exit and Investment-Specific Technical Change’, American Economic Review, 100 (1), pp. 164–92. Stokey, Nancy, 1988. ‘Learning by Doing and the Introduction of New Goods’, Journal of Political Economy, 96 (August), pp. 701–17. Uy, Timothy, Kei-Mu Yi, and Jing Zhang, 2013. ‘Structural Change in an Open Economy’, Journal of Monetary Economics, 60 (6), pp. 667–82. Wang, Yong, 2014a. ‘Industrial Dynamics, International Trade, and Economic Growth’, Working Paper. Wang, Yong, 2014b. ‘Market Structure, Factor Endowment and Technology Adoption’, Working Paper.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

     Kuznets and Beyond ......................................................................................................................

 

4.1 I

.................................................................................................................................. S transformation can mean many things and, once it is specified, it can be used for many purposes. The classical sense of structural transformation goes back to W. Arthur Lewis and Simon Kuznets in the 1940s and 1950s as the movement of population and economic activity from agriculture to industry. Lewis, Kuznets, and their peers recognized that these classifications were themselves too narrow. Lewis, for example, included the urban informal sector as part of his famous ‘unlimited supplies of labor’: [T]he phenomenon is not, however, by any means confined to the countryside. Another large sector to which it applies is the whole range of casual jobs—the workers on the docks, the young men who rush forward asking to carry your bag as you appear, the jobbing gardener, and the like. These occupations usually have a multiple of the number they need, each of them earning very small sums from occasional employment; frequently their number could be halved without reducing output in this sector. (Lewis 1954)

Thus, in a general sense, structural transformation was a move of the population from low productivity to high productivity sectors. The evolution of productivity within these sectors, for example through changing commodity mix or through learning by exporting, has generally been the focus of more recent literature. But the classical sense of structural transformation starts with an imperfection in the economy—labour having very different productivity in different sectors—and proceeds with a shift of this labour across sectors to high productivity sectors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



The Lewis (1954) model is an elegant formulation of this process where accumulation by capitalists in the industrial sector expands production in this sector and draws labour in from the low productivity sector at a constant wage (the wage being constant because of the ‘unlimited’ nature of its supply), thereby increasing profits, accumulation, and further expansion in the industrial high productivity sector. The factor distribution of income moves in favour of capital. But the movement of population will also have implications for the personal distribution of income. The process in this form ends when so much labour has been pulled out of the agricultural sector that labour is no longer in unlimited supply, and agricultural wages begin to rise and eventually match those in the industrial sector. The ‘Lewis turning point’ has been reached. The specific implications of this process of structural transformation for income distribution was explored by Kuznets (1955) in a classic paper, and it is this paper that provides the launch pad for this chapter in Section 4.2, which sets out the Kuznetsian basics. Section 4.3 presents an analytical framework for the assessment of inequality and poverty during structural transformation in a Lewis–Kuznets setting. Section 4.4 takes on critiques of the Kuznets framework, old and new. While accepting much of this criticism, the section argues that some of it is misdirected, and the general framework still has a lot to teach us. Section 4.5 outlines the main conclusions of this chapter.

4.2 K B

.................................................................................................................................. The classic Kuznets (1955) paper is not much read in the original these days, and many critics limit their focus to the ‘Kuznets curve’, the inverse-U shaped relationship between inequality and per capita income, the relationship which launched a thousand empirical investigations. It is worthwhile making a brief excursion into the original to situate the vast body of literature around it.¹ Kuznets was above all an empirical economist, well known for his work on national income accounts. He brought the same sensibility to income distribution, setting out requirements for data: First, the units for which incomes are recorded and grouped should be familyexpenditure units, properly adjusted for the number of persons in each. . . . Second, the distribution should be complete, i.e., should cover all units in a country rather than a segment either at the upper or lower tail. Third, if possible we should segregate the units whose main income earners are either still in the learning or already in the retired stages of their life cycle. . . . Fourth, income should be defined as it is now for national income in this country, i.e., received by individuals, including income in kind, before and after direct taxes, excluding capital gains. Fifth, the units should be grouped by secular levels of income, free of cyclical and other transient disturbances. . . . Furthermore, if one may add a final touch to what ¹ This is done in greater detail in Kanbur (2012).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  is beginning to look like a statistical economist’s pipe dream, we should be able to trace secular income levels not only through a single generation but at least through two—connecting the incomes of a given generation with those of its immediate descendants. (Kuznets 1955: 1–3)

Income distribution analysts today would do well to test their data against these Kuznetsian criteria. Kuznets went on to assess the evolution of inequality in three countries for which adequate data were available: The data are for the United States, England, and Germany—a scant sample, but at least a starting point for some inferences concerning long-term changes in the presently developed countries. The general conclusion suggested is that the relative distribution of income, as measured by annual income incidence in rather broad classes, has been moving toward equality—with these trends particularly noticeable since the 1920’s but beginning perhaps in the period before the first world war. (Kuznets 1955: 4)

According to Piketty (2014) it was this reading of the data that set the literature off in the wrong direction, leading to expectations of continuous declining inequality in the post-war period—a trend known to have discontinued half a century later. However, what is less appreciated is that Kuznets himself raised questions about the declining trend in light of underlying economic processes: The present instalment of initial speculation may be introduced by saying that a long-term constancy, let alone reduction, of inequality in the secular income structure is a puzzle. For there are at least two groups of forces in the long-term operation of developed countries that make for increasing inequality in the distribution of income before taxes and excluding contributions by governments. The first group relates to the concentration of savings in the upper-income brackets. . . . The second source of the puzzle lies in the industrial structure of the income distribution. An invariable accompaniment of growth in developed countries is the shift away from agriculture, a process usually referred to as industrialization and urbanization. (Kuznets 1955: 6–7)

The first set of forces above relates, of course, to Piketty’s (2014) focus on capital accumulation as a force for rising inequality. Kuznets (1955) discusses a number of countervailing factors to this tendency of inequality to increase. But our focus here is on the second set of forces identified by Kuznets, relating directly to the link with structural transformation. Kuznets (1955) proceeds with a detailed examination of the implications of structural transformation, the shift from agriculture to industry, and sets out an informal model which Anand and Kanbur (1985) refer to as the ‘Kuznets process’: The income distribution of the total population, in the simplest model, may therefore be viewed as a combination of the income distributions of the rural and of the urban populations. What little we know of the structures of these two component income distributions reveals that: (a) the average per capita income of the rural population is usually lower than that of the urban; (b) inequality in the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



percentage shares within the distribution for the rural population is somewhat narrower than in that for the urban population. . . . Operating with this simple model, what conclusions do we reach? First, all other conditions being equal, the increasing weight of urban population means an increasing share for the more unequal of the two component distributions. Second, the relative difference in per capita income between the rural and urban populations . . . tends to widen. . . . If this is so, inequality in the total income distribution should increase . . . [W]hy [then] does income inequality decline? (Kuznets 1955: 7–8)

Kuznets answers the question through a numerical simulation which captures the essence of many subsequent formal developments which will be laid out in Section 4.3. With the assumptions concerning three sets of factors—intersector differences in per capita income, intrasector distributions, and sector weights—varying within the limitations just indicated, the following conclusions are suggested: First, . . . Second, . . . Third, . . . Fourth, . . . Fifth, even if the differential in per capita income between the two sectors remains constant and the intra-sector distributions are identical for the two sectors, the mere shift in the proportions of numbers produces slight but significant changes in the distribution for the country as a whole. In general, as the proportion of A drifts from 0.8 downwards, the range tends first to widen and then to diminish. (Kuznets 1955: 14–15)

The fifth conclusion is, in effect, the Kuznets inverse-U shape. There is a pure effect of structural transformation: the mere shift of population from one sector to the other. Of course, changes in sectoral means and sectoral inequalities will also affect national inequality—this is what conclusions one to four are about. So, for example, if the gap between average incomes in the two sectors first widens and then narrows, then national inequality will also first widen and then narrow for that reason. If the sectoral inequalities follow an inverted-U shape, so will national inequality. But these patterns are not necessarily related to structural transformation—at least, a separate theory would have to account for such patterns. In the post-Kuznets literature, the inverse-U refers to national inequality as a whole, not just the pure effect of structural transformation. Empirical studies have attempted to estimate the relationship. The literature on the subject was launched by the work of Ahluwalia (1976a, 1976b), although there were precursors (Adelman and Morris 1973) and existing discussions on policy circles (Chenery et al. 1974). Given the lack of time series data for developing countries, this literature used cross-section data and searched for an inverse-U relationship between measures of inequality and per capita income. Early successes by Ahluwalia and others were soon questioned on grounds that ranged from reliability of data to econometric issues and lack of firm theoretical footing. An example of such interrogation is found in the papers by Anand and Kanbur (1993a, 1993b). Also, scepticism about the Kuznets inverse-U became the norm, even using more comprehensive data compilations (although primarily still cross-section) such as that by Deininger and Squire (1998). This view was summarized in surveys of the time, such as Fields (2001). The availability of time series data by the mid-2000s in countries such as India and China allowed an expansion of the research horizons. However, because these

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

countries have experienced rises in inequality during the period, there was little scope for finding an inverse-U (Kanbur and Zhang 2005). In Latin America, the time series data is insufficiently frequent to allow country specific econometric analysis. Nevertheless, in the last two decades there has been a steady decline in inequality, so here again the prospects for discovering an inverse-U curve are limited (Lopez-Calva and Lustig 2010). But, of course, the greater availability of global data has continued the quest for an inverse-U through combining cross-section and time series data (Barro 2008). The vast empirical literature attempting to test for an inverse-U shape relationship between inequality and per capita income has contributed little to the discussion on structural transformation per se. This body of literature focuses on the overall macro reduced form relationship, paying little attention to the mechanisms that give rise to the relationship, in particular the transformation of an economy through a shift of population from low productivity (agricultural or traditional) sectors to high productivity (industrial or modern sectors). A more disaggregated perspective is truer to the original Kuznetsian exposition and formulation and may provide greater insight into the relative weights of different forces impacting on national inequality. Section 4.3 presents an analytical formulation—a disaggregated picture of inequality change in structural terms.

4.3 S S A

.................................................................................................................................. The central analytical frame of the Kuznets exposition is that the national income distribution can be broken down into a population-weighted sum of sectoral distributions. With this framework, evolution in the overall income distribution can be broken down into its components and the shift in the population weights of these components. Kuznets numerical simulations elaborate on this structure, and the results give us a preliminary understanding of ways the national income distribution responds to shifts in different sectoral parts. In particular, as quoted in Section 4.2, the numerical calculations establish the possibility of an inverse-U simply through the shift of population from one sector to another—holding constant the sectoral distributions. A more precise general structure can be achieved through the use of specific inequality measures that allow the decomposition of inequality into sectoral components. In particular, for inequality measures which are decomposable in a precise sense, national inequality can be written as a function of inequality in each sector, the mean of each sector, and the population share of each sector. Robinson’s (1976) early use of such decomposition to explore the evolution of inequality in a Kuznetsian sense, demonstrated the possibility of an inverse-U purely as the result of population shift when the inequality measure was the variance of log-income. Anand and Kanbur (1993a) extended this exercise to six inequality measures, and used it to derive specific functional forms for the inequality—a per capita income relationship for each index

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



of inequality. Anand and Kanbur (1993b) then implemented this econometrically, estimating the appropriate functional form corresponding to each inequality index. One of the six indices of inequality in Anand and Kanbur (1993a) is the mean log deviation (MLD), also known as Theil’s second index. This index has strong decomposition properties and has increasingly come to be used as a standard measure of inequality (for example, in recent literature on inequality of opportunity, which is discussed in Section 4.4). Following Kanbur and Zhuang’s (2013) model, this section uses the MLD measure to set out the basic analytics of income distribution and structural transformation in the spirt of Kuznets. Let the national economy be divided into two sectors indexed 1 and 2. If the frequency density of the national distribution of income y is f(y), then the sectoral densities are presented as f₁(y) and f₂(y). Let the population share of sector 1 be denoted x, our key indicator of structural transformation. The national frequency density is then: f (y) ¼ x f1 (y) þ (1  x) f2 (y)

ð1Þ

Denote the MLD measure by L and the mean by m, with subscripts 1 and 2 to indicate each sector. Let k ¼ m1 =m2 denote the ratio of the two means. It is well known that the MLD measure can be decomposed into sectoral components as follows: L ¼ x L1 þ (1  x) L2 þ log [x k þ (1  x)]  [x log (k)]

ð2Þ

The first two terms on the right hand side together constitute a weighted sum of the sectoral inequalities, and the sum is known as the within-group component of national inequality: LW ¼ x L1 þ (1  x) L2

ð3Þ

The last term on the right hand side of (2) has an interesting and important interpretation. If the income distribution in each sector is equalized around its mean, the only inequality left would be that due to difference in the means of the two sectors. This is known as the between-group component of inequality and is given exactly by the last term on the right hand side of (2): LB ¼ log [x k þ (1  x)]  [x log (k)]

ð4Þ

Kuznets’s (1955) interpretations of his numerical simulations, in effect, follow the paths of (3) and (4), and of their aggregate, as different variables evolve, under empirically plausible assumptions that the more advanced sector has higher mean income and higher inequality. Let the more advanced sector be Sector 1, and consider what happens when the population share of this sector increases from a low value to a high value where the vast majority of the population derives its income from the advanced sector. In this case it is seen that: dLW =dx ¼ (L1  L2 )

ð5Þ

dLB =dx ¼ [(k  1)=(x (k  1) þ 1)]  log (k)

ð6Þ

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Thus in this Kuznetsian framework of structural transformation always increases the within-group component of inequality, as also argued by Robinson (1976) and Anand and Kanbur (1993a), since it consists of moving population from the low inequality to the high inequality sector. Where then does the inverse-U shape come from? The answer is from the between-group term, the evolution of which is given in (6). The intuition is straightforward. Recall that the between-group inequality is what is left when everybody has the mean income of their sector. The income distribution in this case is a two-point distribution, with a fraction x of the population at the high income, and the rest at the low income. At the two extremes, when everybody is at one income level or the other, between-group inequality is zero. In between, inequality is zero. Thus, we must have an inverse-U shape as population moves from the low income to the high income sector. While between-group inequality always follows an inverse-U shape, it is counteracted in its downward sloping portion by the increasing inequality of the within-group component. Thus, to get an inverse-U shape in overall inequality in the process of structural transformation, we need further conditions that (5) does not dominate (6), which will happen when the two sectoral inequalities are not too disparate. The specific condition is developed in Anand and Kanbur (1993a) and Kanbur and Zhuang (2013) and shown to be: L1  L2 < 1=k  1 þ log (k)

ð7Þ

The role of factors other than structural transformation, as measured by the population share x, raises the question of how these other factors affect inequality and indeed move with structural transformation. Kanbur and Zhuang (2013) address this question by looking at the total differential of (2) and relating it to a series of elasticities. dL=L ¼

Ex (dx=x) þ Ek (dk=k) þ EL1 (dL1 =L1 ) þ EL2 (dL2 =L2 )

ð8Þ

where Ex, Ek, EL1, EL2 are elasticities of L with respect to x, k, L₁ and L₂, respectively, and are given by Ex ¼ (x=L)[(L1  L2 ) þ [(k  1)=(x (k  1) þ 1)]  log (k)] Ek ¼ (k=L) [x=(1  x þ xk)  x=k] EL1 ¼ (L1 =L) x

ð9Þ

EL2 ¼ (L2 =L) (1  x) Expressions (8) and (9) quantify the consequences of changes in the key Kuznetsian parameters. We have already, in effect, discussed Ex in the examination of the inverse-U shape of the relationship with the structural transformation variable, x. What about the other parameters? The expression for Ek shows that a widening of the gap between sectoral means will increase inequality, and that, similarly, a narrowing of the gap will reduce national inequality. Thus, to generate an inverse-U in national inequality, the sectoral mean gap will also have to follow an inverse-U shape. The advanced sector will have to race away from the lagging sector in the early stages of development and then be

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



caught by the lagging sector in the later stages. It is this type of structural transformation that would re-enforce the inverse-U pattern induced by the pure effect of the population shift. Similarly, the expressions for EL1 and EL2 show that a widening gap between the inequalities of the two sectors will increase national inequality. Thus, if inequality in the advanced sector races ahead and is then caught up by inequality in the lagging sector, the result will be an alignment with the inverse-U path induced by other forces. The above discussions lead us to a broader conceptualization of sectoral transformation, rather than a simple shift of population from the low-productivity to the highproductivity sector. Arthur Lewis famously argued that inequality within the leading sectors would tend to increase: Development must be inegalitarian because it does not start in every part of an economy at the same time. Somebody develops a mine, and employs a thousand people. Or farmers in one province start planting cocoa, which will grow only in 10% of the country. Or the Green Revolution arrives, to benefit those farmers who have plenty of rain or access to irrigation, while offering nothing to the other 50% in drier regions. (Lewis 1976: 26)

Thus, certainly in the initial stages of development, inequalities within the advanced sector could augment, the mean income in this sector could increase rapidly, and these together with the population shift from the low-productivity to the high-productivity sector could increase inequality. What the expressions also make clear is that the impact of a change in x on national inequality depends on other sectoral variables as well. Thus, the bigger the average sectoral gap k, the more pronounced the impact of a shift in population on inequality will be. The same holds true for the gap in sectoral inequalities. Clearly, the links between structural transformation and national inequality are further complicated beyond the simple but powerful basic Kuznetsian forces identified in his classic paper.

4.4 K B K

.................................................................................................................................. Two major criticisms have been recorded against the Kuznetsian framework in the literature since 1955. The first, in the literature right up to today, but primarily in the 1970s, 1980s, and 1990s, is that we appear not to find an inverse-U relationship between national inequality and national per capita income in cross-country regression analysis. This is true even when, in more recent years, over time observations have been added to the mix for a few countries. The second and more recent and prominent, criticism, by Piketty (2006, 2014), is that Kuznets built a theory to explain the facts of the time—declining inequality over the previous few decades. Piketty’s position is summarized as follows: During the past half-century, the Kuznets’ curve hypothesis has been one of the most debated issues in development economics. And rightly so. In a nutshell, the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

hypothesis simply says that income inequality should follow an inverse-U shape along the development process, first rising with industrialization and then declining, as more and more workers join the high-productivity sectors of the economy (Kuznets 1955). This theory has strong—and fairly optimistic—policy consequences: if LDCs [less-developed countries] are patient enough and do not worry too much about the short run social costs of development, then they should soon reach a world where growth and inequality reduction go hand in hand, and where poverty rates drop sharply. . . . I will argue that recent historical research is rather damaging for Kuznets’ interpretation: the reasons why inequality declined in rich countries seem to be due to very specific shocks and circumstances that do not have much to do with the migration process described by Kuznets and that are very unlikely to occur again in today’s poor countries (hopefully). . . . Inequality dynamics depend primarily on the policies and institutions adopted by governments and societies as a whole. (Piketty 2006)

This position, which is restated in and is something of a launch pad for Piketty’s (2014) blockbuster book, Capital in the Twenty first Century, may be correct when referring to the literature as whole and the interpretation e of the Kuznetsian framework, but seems a little unfair to Kuznets himself, at least to his exposition in Kuznets (1955). This exposition is replete with discussions of many mechanisms beyond the simple population shift from the high productivity to the low productivity sector. Indeed, as discussed in Section 4.2, he characterizes the observed decline as a ‘puzzle’ given what he views as the forces responsible for the increase in inequality in the initial stages of industrialization—savings and capital accumulation (which appear to mesh with Pikkety’s own theories), and population shifts. In fact, institutional factors are prominent in Kuznets: One group of factors counteracting the cumulative effect of concentration of savings upon upper-income shares is legislative interference and ‘political’ decisions. These may be aimed at limiting the accumulation of property directly through inheritance taxes and other explicit capital levies. They may produce similar effects indirectly. . . . All these interventions, even when not directly aimed at limiting the effects of accumulation of past savings in the hands of the few, do reflect the view of society on the long-term utility of wide income inequalities. This view is a vital force that would operate in democratic societies even if there were no other counteracting factors. (Kuznets 1955: 8–9)

Kuznets goes on to discuss other factors, such as demography, which may counteract the fundamental forces of accumulation. Such institutional and other factors can be viewed, in the formalizing of Kuznets presented in Section 4.3, as influencing the within-sector inequalities and, through these, national inequality. Here is how he skilfully uses his numerical simulations to animate a theoretical discussion, recalling that A denotes agriculture and B denotes non-agriculture: If we grant the assumption of wider inequality of distribution in sector B, the shares of the lower-income brackets should have shown a downward trend. Yet the earlier summary of empirical evidence indicates that during the last 50 to 75 years there

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



has been no widening in income inequality in the developed countries but, on the contrary, some narrowing within the last two to four decades. It follows that the intra-sector distribution—either for sector A or for sector B—must have shown sufficient narrowing of inequality to offset the increase called for by the factors discussed. . . . This narrowing in inequality, the offsetting rise in the shares of the lower brackets, most likely occurred in the income distribution for the urban groups, in sector B. . . . Much is to be said for the notion that once the early turbulent phases of industrialization and urbanization had passed, a variety of forces converged to bolster the economic position of the lower-income groups within the urban population. . . . . Furthermore, in democratic societies, the growing political power of the urban lower-income groups led to a variety of protective and supporting legislation . . . (Kuznets 1955: 16–17)

These discussions belie any simplistic attempt to characterize Kuznets (1955) as putting forward a law that inequality would eventually decline. Rather, what we see is a sophisticated and open-minded reflection on different forces of inequality change, some making for increasing inequality, some for decreasing, organized in a sectoral framework. In this framework, structural transformation is anchored in the shift of population from the low-productivity to the high-productivity sector. But it can also be seen as a complex set of factors affecting within-sector distributions as well. Expressions such as those in (8) and (9) provide an entry point into the rich Kuznetsian discourse. For example, Kanbur and Zhuang (2013) present a contrast between India and China. Given the current values of their country-specific parameters, it is shown that a further increase in the share of the urban population would, all else constant, increase inequality in India and decrease inequality in China. Of course, all else is not constant, but the Kuznetisan framework provides a way of understanding the sharp increases in inequality in China in the 1980s, 1990s, and early 2000s and, crucially, a possible start of an inequality decreasing phase from the mid-2000s onwards (Fan et al. 2011; Kanbur et al. 2017). There are possible explanations for the peak in China’s inequality. First, a series of policy interventions have sought to contain the rise of inequality within urban and rural areas, including a broadening of health and social security provisions, and minimum wages (Kanbur et al. 2016). Second, the huge migration from rural to urban areas is beginning to tighten the labour market in rural areas, raising the mean in that sector and narrowing the average gap between the sectors (Zhang et al. 2011). Third, as argued by Kanbur and Zhuang (2013), urbanization has now reached a point in China where, given the other parameters, the shift of population from rural to urban areas is, in fact, contributing to falling inequality. Of course, these factors have to be set against the forces of accumulation and technical change pulling towards rising inequality. But the Kuznetsian framework goes beyond the simple population shift process to take in a range of forces acting on inequality. Thus, there are a number of ways in which we can go ‘Beyond Kuznets’ within the Kuznetsian frame itself. The Kuznets (1955) exposition is far richer than the simple characterizations of it in the literature. There is, however, one recent development—Roemer’s

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

(1998) formulation of the normative concept of ‘equality of opportunity’, which takes us beyond the conceptualizations in Kuznets (1955). Instructively, the basis of Roemer’s (1998) new arguments is the sectoral frame presented by Kuznets (1955). Following a long philosophical tradition, Roemer distinguishes between two types of determinants of income (or any other outcome variable)—‘circumstances’ and ‘effort’. Circumstances are the factors outside the control of the individual, while effort denotes factors that an individual controls. Gender, ethnicity, parental occupation or wealth, for example, may be thought of as circumstances. The variation of income attributable to variations in these circumstances is then deemed to be (in)equality of opportunity. The specific method is to decompose an inequality index like the MLD into betweengroup and within-group components and use the fraction of the former in total inequality as a measure of inequality of opportunity (see, for example, Paes De Barros et al. 2009). Place of birth is a key factor in identifying circumstances. This is legitimately thought to be outside the control of the individual. In the global context, for example, country of birth is considered to be morally arbitrary, and thus the fraction of global inequality attributable to per capita income differences across countries to be ‘global inequality of opportunity’ (around 75 per cent, see Milanovic 2016). Extending the analogy to within countries, as is indeed done in the literature, whether an individual is born into the rural or urban sector should be morally arbitrary, and thus inequality attributable to mean difference between the two sectors, LB in the notation of Section 4.3 should be inequality of opportunity. Of course, what we have in the Kuznets framework is a dynamic setting where migration takes place between the two sectors. Individuals born into the low-productivity sector could end up in the highproductivity sector. Paradoxically, however, if we start with a low share of the population in the rural sector, permitting some migration could, under certain conditions, increase LB and thus the measured inequality of opportunity. Thus, structural transformation in the early stages could increase not only inequality, but also inequality of opportunity as measured by Roemer (1998). These seemingly paradoxical outcomes need far greater investigation than has been accorded to them either by the literature on structural transformation and inequality of opportunity.

4.5 C

.................................................................................................................................. This chapter has explored the question of structural transformation and income distribution through the eyes of the pioneer of such analysis, Simon Kuznets. His 1955 paper is notable for its use of a combination of close attention to data and rich theorizing as a platform for a discussion of different forces acting on the evolution of inequality during the development process. While Kuznets’ exposition bears the marks

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



of its time, with the core theoretical development supported by numerical calculations rather than formal modelling, it stands the test of time in that its insights are relevant to the understanding of current phenomena such as the evolution of Chinese inequality. The literature following after Kuznets focused on the supposed inverse-U shape prediction for the relationship between overall national inequality and national per capita income. The evidence for such a relationship was not found in cross-section data or, more recently, in combinations of time series and cross-section data. Piketty (2006, 2014) focuses on the declining portion of the inverse-U, attributing to Kuznets an overly optimistic trickle-down view of growth and development. While it is true that some of the post-Kuznets literature drew this unwarranted inference, this chapter argues that Kuznets (1955) has a much more detailed and nuanced positive analysis of the components of income distribution change in a structural frame. Analysts should get the most they can out of this framework, rather than be pulled into a singular view of Kuznets and the inverse-U. Furthermore, the chapter has shown how the Kuznetsian framework could be used, for example, to predict the differential relationship between urbanization and inequality in India and China, to assess the detail of the contribution of sectoral mean and inequality evolution to overall inequality change, and to link the recent inequality of opportunity literature to rural–urban migration.

R Adelman, I. and C. T. Morris, 1973. Economic Growth and Social Equity in Developing Countries, Stanford, CA: Stanford University Press. Ahluwalia, M. S., 1976a. ‘Income Distribution and Development: Some Stylized Facts’, American Economic Review, Papers and Proceedings, 66 (2), pp. 128–35. Ahluwalia, M. S. 1976b. ‘Inequality, Poverty and Development’, Journal of Development Economics, 3, pp. 307–42. Anand, S. and R. Kanbur, 1985. ‘Poverty Under the Kuznets Process’, Economic Journal, 95 (380a), pp. 42–9. Anand, S. and R. Kanbur, 1993a. ‘Inequality and Development: A Critique’, Journal of Development Economics, 41 (1), pp. 19–43. Anand, S. and R. Kanbur, 1993b. ‘The Kuznets Process and the Inequality–Development Relationship’, Journal of Development Economics, 40 (1), pp. 25–52. Barro, R. J., 2008. ‘Inequality and Growth Revisited’, ADB Working Paper Series on Regional Economic Integration, No. 11. Manila: ADB. Chenery, H., M. Ahluwalia, C. Bell, J. Duloy, and R. Jolly, 1974. Redistribution with Growth, Oxford: Oxford University Press. Deininger, K. and L. Squire, 1998. ‘New Ways of Looking at Old Issues: Inequality and Growth’, Journal of Development Economics, 57 (2), pp. 259–87. Fan, Shenggen, Ravi Kanbur, and Xiaobo Zhang, 2011. ‘China’s Regional Disparities: Experience and Policy’, Review of Development Finance, 1 (1), pp. 47–56. Fields, G., 2001. Distribution and Development, A New Look at the Developing World, New York: Russell Sage Foundation; Cambridge, MA: The MIT Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Kanbur, R., 2012. ‘Does Kuznets Still Matter?’ in S. Kochhar, ed., Policy-Making for Indian Planning: Essays on Contemporary Issues in Honor of Montek S. Ahluwalia, London: Academic Foundation Press, pp. 115–28. Kanbur, Ravi and Xiaobo Zhang, 2005. ‘Fifty Years of Regional Inequality in China: A Journey Through Revolution, Reform and Openness’, Review of Development Economics, 9 (1), pp. 87–106. Kanbur, Ravi and Juzhong Zhuang, 2013. ‘Urbanization and Inequality in Asia’, Asian Development Review, 30 (1), pp. 131–47. Kanbur, Ravi, Yanan Li, and Carl Lin, 2016. ‘Minimum Wage Competition Between Local Governments in China’. Dyson School, Cornell University, Processed. Kanbur, Ravi, Yue Wang and Xiaobo Zhang, 2017, ‘The Great Chinese Inequality Turnaround’, CEPR Discussion Paper No. 11892. Kuznets, S., 1955. ‘Economic Growth and Income Inequality’, American Economic Review, 45, (1), pp. 1–28. Lewis, W. A., 1954. ‘Economic Development with Unlimited Supplies of Labor’, Manchester School of Economic and Social Studies, 22, pp. 139–91. Lewis, W. A., 1976. ‘Development and Distribution’, in Alan Cairncross and Mohinder Puri, eds, Employment, Income Distribution and Development Strategy: Problems of the Developing Countries (Essays in honour of H.W. Singer), New York: Holmes & Meier Publishers, Inc., pp. 26–42. Lopez-Calva, Luis Felipe and Nora Lustig, eds, 2010. Declining Inequality in Latin America: A Decade of Progress? Washington, DC: Brookings Institution Press. Milanovic, Branko, 2016. Global Inequality: A New Approach for the Age of Globalization, Cambridge, MA: Harvard University Press. Paes de Barros R., F. H. G. Ferreira, J. R. M. Vega, J. C. Saavedra, M. De Carvalho, S. Franco, S. Freije-Rodriguez, and J. Gignoux, 2009. Measuring Inequality of Opportunities in Latin America and the Caribbean, Washington, DC: World Bank. Piketty, T., 2006. ‘The Kuznets’ Curve, Yesterday and Tomorrow’, in A. Banerjee, R. Benabou, and D. Mookherjee, eds, Understanding Poverty, Oxford: Oxford University Press. Piketty, T., 2014. Capital in the Twenty-First Century, Cambridge, MA: Harvard University Press. Robinson, S., 1976. ‘A Note on the U-hypothesis Relating Income Inequality and Economic Development’, American Economic Review, 66 (3), pp. 437–40. Roemer, J., 1998. Equality of Opportunity, Cambridge, MA: Harvard University Press. Zhang, Xiaobo, Jin Yang, and Shengling Wang, 2011. ‘China has Reached the Lewis Turning Point’, China Economic Review, 22 (4), pp. 542–54.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

        -       Reassessed and Reformulated in New Structuralist Perspective ......................................................................................................................

 

5.1 I: A R

..................................................................................................................................

5.1.1 The USA as the Leader T ‘flying-geese’ (FG) theory was originally put forward by Japanese economist Kaname Akamatsu (1897–1974) in the mid-1930s. Although the FG theory was well known at its home in Japan, it began to be accepted abroad particularly after Japan’s then-Foreign Minister Saburo Okita had referred to, and helped to popularize, it in his speech at the Fourth Pacific Economic Cooperation Conference in Seoul in 1985. Undoubtedly, the ‘East Asian miracle’ (World Bank 1993) that unfolded in the late 1960s through the 1990s provided an ideal backdrop for the theory to draw worldwide attention from scholars, policy makers, and the news media. The Economist, for example, then observed that ‘When geese migrate, they fly in a V-formation. . . . Japan leads. Behind it follow the [NIEs: Singapore, Hong Kong, Taiwan, and South Korea]. In the third rank are [ASEAN-4: Thailand, Malaysia, Indonesia, and the Philippines] and coastal China. As with flying geese, the arrangement is purposeful, well-ordered and coordinated.’¹ However, this sequence of Japan—NIEs—ASEAN-4—China in growth, although often mentioned, gives an erroneous impression that Japan led the gaggle. True, Japan

¹ ‘Together Under the Sun: A Survey of the Yen Bloc’, Economist, 15 July 1989.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

has played a key role in the rise of Asia. But the United States has been, and still is, arguably the global lead goose that triggered an FG formation of staggered growth across East Asia by imparting to the follower geese advanced technologies, offering export markets, and encouraging catch-up. In fact, both Akamatsu and Okita stressed US leadership as the global engine of development.

5.1.2 Three Types of FG-style Structural Transformation Akamatsu was a structuralist in his analytical approach. He was deeply influenced by German Historical School economists, especially Friedrich List and Bernhard Harms; the former advocated ‘infant-industry’ protection to nurture national enterprises and analysed catch-up in stage-theoretic terms, while the latter conceptualized the notion of ‘economic structure’, birthing structural economics (Ozawa 2016). Indeed, Akamatsu’s framework is comprised of three types of structural transformation (or interchangeably, structural change) that are functionally inter-related in propelling catch-up growth in a cohort of aspiring emerging economies led by a leading economy: changes in trade structure, changes in industrial structure, and changes in the hierarchical structure of national economies. And in his mind was the Pacific Rim where Asian economies were underdeveloped vis-à-vis the West. However, he expressed these three types in very broad strokes by sketching them in terms of ‘patterns’: (i) ‘the sequence of importdomestic production-export’ (here restated as import substitution-cum-export promotion or IS-EP), (ii) production ‘from consumer goods to capital goods and from crude and simple articles to complete and refined articles’, (here recapitulated as inter-industry and intra-industry structural changes, respectively), and (iii) ‘the alignment from advanced nations to backward nations according to their stages of growth’ (i.e. a stages-ranked hierarchy of national economies) (Akamatsu 1962: 208). He then identified the first pattern as the fundamental one (kihonkei in his own Japanese word), in which the IS segment is carried out through the Alexander Hamilton-Friedrich List (H-L) approach of infant-industry protection (List 1841/56), but the subsequent EP process follows up on the IS segment in an outward-focused manner. By contrast, the second and third patterns were considered derivative/auxiliary (fukujiteki) (Akamatsu 1974: 165–6). Why, then, did he regard the first pattern as fundamental, and the other two as derivative? Most importantly, the IS-EP catch-up strategy, after all, enabled Japan to develop modern manufacturing industries, one after another, thereby diversifying and upgrading its production structure both inter- and intra-industrially. Thus, the second pattern is an outcome of the IS-EP drive—therefore, derivative in nature. Besides, the IS-EP progression is exactly the common thread he empirically found in the histories (1870–1939) of Japan’s manufacturing industries (such as woolen goods, cotton yarn, cotton cloth, spinning and weaving machines, general machinery, bicycles, and machine tools) (inter alia, Akamatsu 1935, 1962). Imports of a certain new manufacture first arose, but would be gradually replaced by locally manufactured goods, which

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



in turn would soon develop into exports. In fact, this sequential pattern inspired him to adopt the poetic allegory of FG formation for the theory. The third pattern (i.e. aligned nations) also displays an FG-formation, with the leader at the top, followed by second-ranking countries, and then by third-ranking ones, and so forth. Such a hierarchy constitutes a favourable external environment, if the leader advocates trade liberalization by opening up its own markets, an environment within which follower economies are tolerated to develop, under protection, national industries that will become comparatively advantaged in trade. But newly created, export industries spur growth in the followers, thereby impacting on the existent hierarchy and causing a realignment of countries. In this sense, the third pattern, too, is derivative. Although not analysed by Akamatsu, however, a newly realigned hierarchy of economies, in turn, creates a refreshed pattern of comparative advantages in trade. For that matter, diversified and upgraded production structures (the second pattern) likewise do the same. The evolving leader–follower relations are the vital source of interactive and integrative growth, since an innovative leader introduces a succession of new industries, rendering some existent ones comparatively disadvantaged at home but enabling its followers to enter the latter. Thus, such stages-ranked, hierarchical relations bound by the doctrine of comparative advantage continually promote structural transformation in each constituent economy precisely because of structural transmigration from more advanced to less developed economies (especially, transmigration of low-end manufacturing; see Section 5.5.1.1 below). In general, the greater the stages gap (divergence) between the more advanced and the less developed, the higher the (potential) speed of structural transmigration from the former to the latter, resulting in higher rates of catch-up growth (i.e. higher rates of convergence). This is because structural transmigration is fuelled by the logic of comparative advantage that is driven by the gap itself. In fact, at the end of World War II the USA stood as the world’s most dominant economy that enjoyed absolute advantages in practically all industries, yet was willing to open up its markets for comparatively disadvantaged goods, providing demand for the followers’ comparatively advantaged goods. This gap-based trade relationship was an ideal one that enabled both the USA and its war-ravaged allies to make mutual gains from liberalized trade, bringing greater prosperity to the former and reconstructing the latter’s economies. In addition, the US-led FG formation of economies across the Pacific has engendered, and captured within itself, the ‘neighbourhood/spill-over’ effect brought about by expanding markets due to the prosperity of higher-ranking economies as well as the successful catching up by lower-ranking ones.² Thus, growth agglomeration occurred endogenously across the Asian Pacific. Viewed in this light, Akamatsu’s second and third patterns actually become ‘fundamental’. And they also should be the key co-drivers of FG-style development.

² This integrative and interactive relationship also allows freer cross-border transmission of negative externalities, as seen in the 1997 Asian financial crisis and the 2008 Great Recession.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

As far as the FG formation of staggered growth across East Asia in the post-World War II period is concerned, the USA’ geopolitical motive as the leader of the Free World has been even more important than its economic rationale. The rise and spread of Soviet-led communism in the region compelled the USA to build a belt of allies around the Asian rim of the Pacific. First of all, America’s occupation policy for Japan changed diametrically from emasculation of the latter’s economic power in order to crush its military capability to assistance for its quick reconstruction, defensive rearmament, and reintegration into the world. Soon afterwards, the NIEs— and subsequently the ASEAN-4, too—were directly or indirectly treated favuorably in their commercial and military relationships with the USA. The Nixon and Kissinger diplomacy for opening China was motivated to pull China away from the Soviet orbit and into the East Asian FG formation, without which China’s galactic rise, and the way it was achieved so quickly, would have been unthinkable. This explains, largely if not totally, why a string of vigorous catch-up growth has so far transpired only in East Asia and not elsewhere, although these Asian countries themselves had the ability to capitalize on the favourable catch-up milieu created by America. This successful regional FG formation may thus be a historical happenstance of geopolitics, yet the economic dynamics of US-led capitalism that underlies such a formation, nevertheless, still remains fundamental and universal in principle.

5.1.3 The Role of Multinational Corporations as an Instrument of Catch-up—and an Endogenizer of Growth in the World Economy In today’s world, the old-fashioned H-L protectionism is no longer relevant for the emerging world to kick-start an industrial takeoff. Nowadays, a new strategy to invite foreign multinational corporations’ (MNCs) outsourcing activities in low-end manufacturing for export is a more expedient and more viable first step. They bring to host economies much-needed modern technology, managerial knowledge, skills, capital, and access to overseas markets, the necessary inputs that undeveloped hosts initially lack. MNCs are the cross-border transplantors of industries via foreign direct investment (FDI) and other off-shoring operations—especially in those industries whose comparative advantages wane due to a new wave of innovation and structural transformation at home. Hence, FG-style takeoffs among the followers can be endogenously sparked by MNCs among the stages-aligned economies for development. In this respect, Japan is the very last latecomer that managed to forge ahead largely by means of its old-style H-L strategy both prior to and immediately after World War II. By contrast, the NIEs and China have effectively crafted their own new catch-up models by capitalizing on MNC-driven global capitalism.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



This point also reveals that Akamatsu considered only trade (of the arm’s-length type) in neglect of MNCs’ activities. Before his death in 1974, however, Akamatsu had seen only the nascent rise of MNCs—hence, hardly their role in spurring development— in East Asia. In today’s world, however, MNCs’ FDI can take over the IS-EP progression in one fell swoop and establish export-driven, competitive manufacturing—practically overnight. After all, market-driven, profit-motivated MNCs are good at exploiting their host economies’ existing and latent comparative advantages. The upshot is the birth of new intra-firm trade in final and capital goods, tangible and intangible alike. MNCs (foreign and national alike) are a far more puissant catalyst of structural transformation than conventional traders. The former’s developmental role needs to be brought into our framework to study the MNC-involved process of structural transmigration and transformation, the salience of contemporary development (Ozawa 2016).

5.1.4 Friedrich List’s Legacy in the FG Theory It is worth noting that Akamatsu’s IS-EP sequence is built on List’s stages model of structural transformation. The latter’s stages framework shows how a nation, initially agrarian and dependent on resource exports and manufactured imports, develops import-substituting industries at home under infant industry protection, and finally reaches an advanced stage to export its own manufactures, thereby joining the advanced world (List 1841/56). For this, Akamatsu explicitly referred to List (Akamatsu 1955). There are, however, some important differences between Akamatsu’s IS-EP progression and List’s. For starters, Akamatsu focuses on individual industries, while List considers the whole economy as a unit of analysis. In Akamatsu’s model, the IS-EP process is replicated in a range of manufacturing industries as each develops over time in a staggered fashion, repeatedly nudging the national industrial structure away from primary-sector dominance and towards secondary-sector dominance—with an evolving pattern of comparative advantages. Also, Akamatsu takes into consideration the technological progress of each manufacture (i.e. inter- and intra-industry upgrading and diversification), thereby yielding strategic implications for catch-up industrial policy. Furthermore, he emphasizes a succession of interactions between leaders and followers in terms of a perpetual series of alternating structural changes from ‘heterogenization’ (divergence) to ‘homogenization’ (convergence) and vice versa in the global economy—as a result of technological infusion and diffusion.³ He christened this model ‘a theory of unbalanced growth in the world economy’ (Akamatsu 1961). ³ For details, see Akamatsu (1961, 1962) and Ozawa (2009: ch. 2).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

5.1.5 Hegelian Dialectics in Akamatsu’s Analysis Having been immersed in Hegelian dialectics (Hegel 1812/1991), Akamatsu was delighted to see an empirical verification of dialectics for the IS-EP progression—that is, imports (thesis), domestic production (antithesis), and exports (synthesis). His other structural changes are similarly couched in dialectics. In this regard, any evolutionary process of structural transformation itself can be interpreted in dialectic terms as a sequence of changes in terms of contradictions (dynamic tensions), disequilibrium (discrepancies), and transformation, a process that is driven by opposing forces and ultimately resulting in a new outcome. In this regard, a case in point is Schumpeter’s ‘creative destruction’, the hallmark of his development theory.⁴

5.2 T ‘(D-H) L  D  `  S’;  A  S T

.................................................................................................................................. Modern economic development can be defined in Schumpeterian terms as a ratcheting-up evolutionary process of structural transformation that is driven by innovation (technological, institutional, and organizational) in leading economies and spreads to follower economies that are capable of emulating via imitation and learning (Schumpeter 1934, 1942). This view is also in accordance with structural economics (inter alia, Little 1982, for old structural economics; Lin 2012a, 2012b, for new structural economics). One framework that more clearly conceptualizes structural shifts is what may be called the ‘(double-helix) ladder of development à la Schumpeter’ (Ozawa 2009, 2016) that can capture and expand on Akamatsu’s ideas about the interand intra-structural changes and the evolutionary hierarchy of economies aligned at different growth stages.

5.2.1 A Historical Progression of Structural Transformation The ‘ladder of development’ is widely and casually used to describe the progression of long-term growth. Yet, this analogy is nebulous. We must clearly specify the ‘ladder’— and its ‘rungs’ in particular—to understand the process of development as a succession of structural transformation.

⁴ See Ozawa (2009).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



In this regard, the ‘ladder of development à la Schumpeter’ is a stages model that delineates the process of development. It is based on a history of industrialization that traces out a progression of structural changes that has occurred ever since the Industrial Revolution, a historical sequence under capitalism that is punctuated and ratcheted up by stages/tiers/rungs—and testable as ‘structural breaks’ in econometrics. And in each stage a certain technological thrust and its accompanying leading-sector serve as the engine of structural transformation. The global economy has so far witnessed five tiers of leading-sector emerge in wavelike progression, and another tier appears in the making. The five tiers are (i) endowment-driven industries (exemplified by textiles in labour-abundant countries and by extractive industries in resource-endowed countries), (ii) resource-processing heavy and chemical industries (represented by steel, non-ferrous metals, basic chemicals, and heavy machinery), (iii) assembly-based industries (epitomized by automobiles), (iv) R&D-driven industries (e.g. microchips, computers, and pharmaceuticals), and (v) information and communications technology (ICT)-enabled industries (e.g. digital telecoms, operating platforms, the Internet, search engines, and social media—and apps). And (vi) a new emerging sixth tier; green and health technology (GHT)-based industries (e.g. renewable energies, energy-saving devices, pollution control, and life sciences—all designed to create a healthier environment for sustainable growth of economies, human habitats, and the human species itself, reflecting the rise of post-material values). These inter-industry sequential advances are the outcomes of Schumpeterian innovations. The first two tiers were basically introduced under British hegemony before World War I, representing the age of industrialism, and are endowment-based and productionfocused. In contrast, the higher tiers ((iii) through (vi)) have been engendered largely under American leadership after World War II, epitomizing the age of mass consumerism, and are more intensively knowledge-driven and consumption-oriented. The tier-(i) and -(ii) phase relies mostly on ‘endowed’ comparative advantages, while the highertier phase relies on ‘created’ comparative advantages.⁵ And morphing from the former into the latter is the most pivotal structural shift in any growing economy’s industrial development, as is currently faced by China. Also, Britain once engaged in ‘kicking away the ladder’ (in List’s words⁶) by jealously protecting its manufacturing at home, whereas America has been ‘providing the ladder’ to the emerging world—essentially because the latter’s consumption-focused industries need larger consumer markets, especially in fast-growing emerging economies. And a variety of MNCs have emerged pari passu with the historical progression of structural transformation under those two leaderships, giving rise to MNC-focused policies and regulations. Furthermore, these differential characteristics of two hegemon-led patterns of structural transformation each represent a varied set of aggregate demands as stimulus for ⁵ For detailed explanations of each tier, hegemonic differences, and tier-specific MNCs, see Ozawa (2005, 2009, 2016). ⁶ Chang (2002).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

growth (more on this in Section 5.4 below). In addition, the inter-industry ladder described above helps us understand how the extent of mutual gains from ‘interindustry trade’ changes over time for both the leader and the follower economies, depending on the speed of innovation-driven structural transformation in the former and the pace of catch-up growth in the latter. For example, when the leader economy is centred on knowledge-intensive higher-tiers, while the follower economies are still in tier-(i) and -(ii) growth, the gains are large. However, as the gap narrows, the magnitude of comparative advantage becomes smaller, hence, so does the trade gain. Moreover, as the followers advance into knowledge-intensive higher-tiers (albeit initially through low-end activities such as assembly operations), the modality of trade will morph from inter-industry to intra-industry. And for analysis of ‘intra-industry trade,’ we need a new additional framework.

5.2.2 Intra-industry Side-ladders In addition to the inter-industry ladder, each tier also has its own intra-industry sideladder, thereby turning the whole Schumpeterian ladder into a double-helix construct. Each side-ladder is composed of a vertical concatenation of sub-sectors, the upper end of which is highly capital-intensive and technologically and organizationally sophisticated, while the lower end is labour-intensive and technologically standardized. This intra-industry construct has opened up opportunities for a new international division of labour between more and less developed economies and has come to be actively cultivated by MNCs for their cross-country supply chains (see Figure 5.1). Such supply Inter-industry ladder Tier V

Capital/labour intensity ratio∗

Tier IV Tier III

Tier II Tier I

Supply chains

Capital-intensive activities Intra-industry side-ladder: vertical multi-process fragmentation Labour-intensive activities

Intra-industry investments: specialization and coordination gains (in global supply chains) Time

 . The ‘double-helix’ development ladder à la Schumpeter: inter- and intraindustrial upgrading and diversification Source: Based on Ozawa (2009, 2016).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



chains involve intra-industry trade (more precisely, intra-firm or intra-product or intra-process trade). Also, the side-ladders are where incremental adaptive innovations of all sorts (like apps in tier-(v) growth) follow the major technological breakthroughs that drive an inter-industry progression of structural transformation.

5.2.3 A Caveat to Stages Analysis 5.2.3.1 Customized Catch-up Strategies FG formation reflects the adage that history repeats itself—but only in very broad terms and never in exactly the same way. The historical sequence of structural changes is necessarily linear as a trace of things that have occurred over a long stretch of time in the past. The present allows policy makers to cherry-pick industries or their segments from the past and rearrange them in any time framework, adjusting their historical sequence. Any aspiring emerging economy must craft its own catch-up strategy suitable to its time-and-space-specific circumstances and capabilities. This view jibes with what Gerschenkron (1962: 26) emphasizes: ‘In every instance of industrialization, imitation of the evolution in advanced countries appears in combination with different, indigenously determined elements’. In this respect, Gerschenkron’s rebuttal of Rostow’s stages theory is often erroneously construed and cited as if it were the final dismissive say on stages models. A stages-theoretic framework is highly instrumental in understanding the ratcheting-up evolutionary process of structural transformation in any organism. A prime example is individual human growth from birth to childhood, to adulthood, and finally to senility, an evolutionary pattern that is physiologically classified in terms of changes in the major common characteristics of the human body at different stages of growth. Economic development similarly displays some common, stage-specific structural characteristics. Stages models are thus useful as an overall analytics so long as they are appropriately interpreted.

5.2.3.2 Industrial Fusion and Reconfiguration From today’s point of view, the ladder certainly no longer looks linear; the tiers are comingled, combined with new activities, and telescoped. Some original higher tiers are increasingly fusing with lower tiers, if not at the core but additively. One latest example is the fusion of ICT goods/services (of the tier-(v) type) and automobiles (of the tier-(iii) type) in their efforts to produce driverless vehicles (into which artificial intelligence is ‘assembled’) so that an ever-cheapening ride-sharing (‘ubering’) may lead to a transport revolution. Also, under the forces of liberalized trade and investment, original tier-(iii) assembly-based industries (notably automobiles) are transplanting their lowend assembly jobs from the advanced world (where those industries were originated) onto low-wage emerging economies, thereby joining the labour-intensive manufacturing (of the tier-(i) type) dominated once by textiles. Moreover, many original tier-(v) goods (especially digital telecoms, the Internet, and handheld devices like GPS and smart

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

phones) have become indispensable infrastructure goods in the present emerging world. In addition, corporate R&D activity, the key feature of original tier-(iv) industries, is now ubiquitous across knowledge-driven industries. And MNCs from these industries are off-shoring R&D in many host countries, including some emerging economies, especially in Asia (Ernst 2006). Indeed, emerging economies are entering many tiers simultaneously, although mostly from the lower end of each side-ladder. The upshot is constant industrial fusion and reconfiguration. In addition, some emerging economies may forge ahead on a newly opened path of technological advance instead of faithfully tracking earlier industrializers’ footprints— as was, for instance, the case with South Korea’s electronics industry (Lee 2013). In fact, post-war Japan earlier strove to acquire those latest, yet to be commercialized, technologies from the advanced West through patent licensing (Ozawa 1974). In this sense, South Korea has followed Japan’s strategy. China has successfully entered the frontiers of solar-energy and life sciences (nascent tier-(vi) industries)—and undertaken robotization (i.e. an upper-end activity on tier-(iii)’s assembly-based side-ladder) at an earlier point of time than expected. And many studies show how idiosyncratically each catching-up economy has chosen its own path to structural transformation (see Part IV in this Handbook). They reveal the customized catch-up strategies that result in diverse patterns of structural transformation with ‘different, indigenously determined elements’.

5.3 T R  G   M  C- G

.................................................................................................................................. In neo-classical economics the market is the only coordinating mechanism for economic activity. Yet it is merely the mechanism that allocates resources in response to price signals. The market is basically neither goal-setting nor goalpursuing; it is goal-neutral at best, and sometimes even goal-hindering (Ozawa 2016). What is needed, then, is a goal-focused government that can make the best use of the market for development. After all, catch-up itself is a public good, with which the market’s dealing is inept. There thus ought to be a ‘facilitating state’ (Lin 2012a, 2012b). And the government has to ‘govern’—and ‘lead’—the market (Wade 1990). Indeed, even market-primacy neo-classical economics justifies state interventions albeit temporarily when market failures occur and systemic risks rise, especially in financial markets. Structural economics stresses that undeveloped economies suffer structural rigidities and dysfunctional markets. In fact, the efficacy of the market is the increasing function of development (Little 1982). The market itself as a coordinating mechanism is necessarily underdeveloped prior to, and in the early stages of, industrialization. There is no functional market ‘in the beginning’—efficacious enough to autonomously

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



spur an industrial takeoff. Given such circumstances, therefore, a proactive government with an adequate administrative calibre and a strong national determination for development are what emerging economies really need for takeoff (Akamatsu 1962; Bresser-Pereira 2010). As the market develops pari passu with catch-up progress, both government and the market become the main collaborative drivers of sustainable development in line with new structural economics (Lin, 2012a, 2012b).⁷ Consequently, state–industry partnership takes place and tends to be both most intensive and extensive in labour markets (via education and skills training) and financial markets (e.g. capital creation and exchange rate stability) in the early stages of catch-up. This is particularly the case with tier-(i) and -(ii) growth spurred by infrastructure investment (more on this in Section 5.5.2). And as the economy moves into sustainable growth, industry is likely left more to market forces, and the state turns towards its regulatory role and away from its developmental one. Interestingly, however, as the economy matures—and especially falls into a ‘high-income trap’ (in which structural rigidities again emerge and hardens, popularly described as ‘sclerosis’), the government’s role as a structural rejuvenator is called for, resurrecting a closer state involvement in the economy (e.g. market reform, infrastructure redevelopment, worker immigration, overseas market expansion, and industry-specific basic research as well as general science promotion). Thus, in general, state–industry relationships are structurally determined by growth stages, tracing out a flattened U-shape curve over time, although each economy’s experience varies depending on its own political—and geopolitical—economy factors. Indeed, each economy’s unique state– industry relationship brings about a variety of capitalism.

5.4 T S L   S- G T

.................................................................................................................................. Economic development proceeds along an S-shaped trajectory, as is typical of any growing organism. Growth accelerates up to an inflection point during the early stages (tier-(i) and -(ii) combined phase) and then starts to decelerate during the maturing stages (from tier-(iii) and onwards). Viewed in terms of the Keynesian composition of aggregate demands for GDP (gross domestic product), that is, C (consumption) + I (investment) + G (net government spending) + [X (export)—M (import)], the preinflection phase of contemporary (MNC-involved) development is heavily reliant on G (infrastructure development spending), I (enterprises’ investment in productive

⁷ For differences between new and old structural economics, see Lin (2012a, 2012b).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

capacity), and X—M (net export), while it is least dependent on C, which actually remains relatively small. This compositional aggregate demand characteristic mirrors the legacy of UK-led industrialism. In contrast, however, as any taken-off economy moves beyond the inflection point, entering higher-tier, more knowledge-based, and more consumption-oriented industries, C rapidly emerges as a major aggregate demand that along with I (mostly to facilitate consumption and develop human resources, especially knowledge via R&D, education, and training), determines the pace of growth. This trend reflects the rise of US-led consumerism. And the C-primacy also becomes the major determinant of a growth rate—even stagnation—when consumption slows down in the advanced economies that have recently encountered deflation (more on this in Section 5.5.3 below). Interestingly, an economy operating in the tier-(i) and -(ii) married phase has historically been able to achieve double-digit rates of growth (roughly 10 per cent or more a year, the highest rate realizable in its entire economic history). And the phenomenon of national income doubling easily occurs during that phase. For example, America’s ‘Gilded Age’ registered such a doubling record during the period of 1881–91 (US Bureau of the Census 2006). Indeed, it was ‘the highest decadal rate . . . from 1805 to 1950’ (Friedman and Schwartz 1963: 93). More recently, Japan’s post-war period of modernizing heavy and chemical industries and infrastructure in the 1960s accomplished such a feat within seven years, a phenomenon for which the then government cleverly took credit by announcing a GDP-doubling plan in a timely manner, although such high growth is structurally inherent to that particular growth phase. Also, both South Korea and Taiwan underwent heavy and (petro-) chemical industrialization with their GDPs quickly more than doubling from the mid-1960s to the mid-1970s. Currently, China is in the midst of the same phase with its GDP-doubling plan (2010–20). Vietnam, too, has just entered its tier-(i) and -(ii) combined growth—with a similar GDPmultiplying opportunity. And such high growth helps economies that have already taken-off catapult into middle-income status. In this regard, Britain’s Dickensian years of the mid-nineteenth century set the precedent of high growth (which, however, spread over a much longer time span, during which a series of innovations were unveiled). Yet, contemporary catch-up is time-compressed with a much higher rate of growth thanks to the latecomer advantage to reconfigure industrial structures and borrow cutting-edge technologies. However, it is one thing to attain middle-income status after escaping from the low-income trap, but it is another for the economy to continue to grow fast enough to join the ranks of high-income economies without getting caught in the middleincome trap, a phrase coined by Gill and Kharas (2007). In other words, beyond the inflection point is hidden this trap that any sustainable growth must avoid. Moreover, a high-income (sclerosis) trap is another hurdle to surmount farther down the road.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



5.5 T R  N E R  M S G F

.................................................................................................................................. The structural changes as envisaged in the ‘(double-helix) ladder of development à la Schumpeter’ are basically a meso-structural phenomenon. There is, however, another type, a macro-transformative type of shift in the existent national ecosystem that governs economic activities, so as to drive growth forward. In other words, development needs a national ecosystem that is conducive to technological advance and structural upgrading and diversification in different growth phases. But such an ecosystem does not evolve autonomously and congruously with the pace of structural transformation, becoming obsolete in the very wake of economic advance and requiring occasional overhauls. Hence, any sustainable catch-up must undergo at least two rounds of major ecosystem reform. The first round aims at converting a primary-sector-dominant (povertystricken) economy into a secondary (manufacturing and construction, i.e., basically, tier (i) and (ii))-sector dominant (middle-income) economy, whereas the second round intends to create a tertiary (consumer-focused)-sector-dominant (higher-income) economy. The former is critical for the economy to succeed in terms of industrial takeoff—thereby extricating itself from the low-income trap—and the latter is essential to escape from the middle-income trap. These two rounds of overhaul take place along an evolutionary development path, a path that is envisaged in Clark’s (1935) threesector model and Kuznets’ (1966) agriculture-industry-commerce model—and in the Schumpeterian ladder of development described in Section 5.2. above. Furthermore, there is a need for another round if an advanced economy falls into a high-income (sclerosis) trap and stagnates or even declines. The common cause of the three traps is the structural and institutional rigidities that ensnare the economies.

5.5.1 The First Round—from Low-income Stagnation towards Middle-income Status The first round is intended to mobilize rural labour for industrialization via migration from the primary sector to the secondary sector, the vital step that raises productivity and creates investable capital out of profits. This process is theorized in terms of the labour market model (Lewis 1954). Here, a new ecosystem for the secondary-sectordominant economy must be built and government needs to demonstrate a strong determined leadership for planning and initiating catch-up by generating adequate

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

funding for the transportation, communications, and urbanization infrastructure suitable for the secondary sector’s development. Accompanying infrastructure development lays the ground for tier-(ii) growth spurred by heavy and chemical industrialization. The first round of transition thus enables the economy to escape from the low-income (poverty) trap, leading to tier-(i) and -(ii) married growth during which the growth rate can accelerate and reach the inflection-point (i.e. the highest possible rate), as discussed in Section 5.3 above.

5.5.1.1 ‘Grabbing low-hanging fruits’: Joining in Comparative Advantage Relaying in Low-end Manufacturing—with Magnified Comparative Advantage The most effective approach to the first round of reform is to get a ride on MNCs’ offshoring activities. To this end, emerging economies should open up their economies and entice MNCs to transplant tier-(i) type, labour-intensive manufacturing, an activity in which emerging, labour-abundant economies have an ‘endowed’ comparative advantage—and that offers them an opportunity for ‘grabbing low-hanging fruits’ in the words of Lin and Wang (2014), a phrase that describes a task that is relatively easy to undertake. Indeed, this has been the open-door, takeoff strategy pursued by successful East Asian economies (the NIEs, the ASIAN-4, China—and, most recently, Vietnam). Their staggered takeoffs have resulted in impressive reductions in extreme poverty across the region and an initial rise in trade and GDP. This strategy and its resultant sequential pattern of success are also conceptualized as ‘comparative advantage recycling (or relaying)’ (Ozawa 2009). This relaying is carried out by MNCs which transplant comparatively disadvantaged low-end manufacturing from advanced or rapidly catching-up economies to less developed ones. Such a process, in each turn, expands the basis of trade (i.e. magnifies hosts’ comparative advantages, existing and potential alike, via technology transfer), thereby boosting gains from trade, a phenomenon that is conceptualized as the ‘pro-trade’ type of FDI (as opposed to the ‘anti-trade’ type that works against the logic of comparative advantage) (Kojima and Ozawa 1984). Pro-trade FDI is prevalent during the tier-(i) phase, assisting rapid growth in labour-abundant host countries. (For knowledgedriven higher-tier industries, however, host governments may consider them ‘strategic’ and induce foreign MNCs to produce locally their home countries’ comparatively advantaged goods, reducing production in, and exports from, the advanced world— i.e. anti-trade FDI (Ozawa, 2016).) Also, it should be noted that Radelet and Sachs (1997) consider the FG theory as a major development doctrine alongside the ‘big push’ theory and the ‘importsubstitution’ approach. For them, a strong export orientation with the support of export processing zones is its defining characteristic. But, this feature applies largely to tier-(i) labour-driven growth.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



5.5.2 The Second Round—Bypassing the Middle-income Trap towards High-income Status The second round of reform is supported by a rapidly risen per-capita income stemming from combined tier-(i) and -(ii) growth, which creates demand for consumer goods and services, thereby leading to a consumption-oriented, new economy. In this sense, the market collaborates with the state’s determined efforts to sustain the momentum of catch-up growth. Yet, the second round may not be as ‘easy’ as the first one targeted at ‘low-hanging fruits’. The main reason is the political issues related to the inevitable contraction of tier-(ii) heavy and chemical industries that causes the ‘rustbelt’ and ‘industrial urban decay’ phenomena. Some industries (notably, steel, copper, aluminum, coal-fired power generation, heavy machinery, shipbuilding, and the like) are considered ‘strategic’ and mostly infrastructure-goods producers involving public interests—hence protected and promoted. Besides, being scale-driven, many tier-(ii) factories operate on a large scale to minimize costs—with their workers numbering in thousands and usually being unionized from the heydays of tier-(ii) growth. Job cuts and factory closures are not an easy matter to manage—particularly during the transition to the new economy where entirely different and higher skills are in demand and when growth rates inevitably start to decelerate (as seen in Section 5.3 above), resulting in a decline in job creation and a mismatch in skills. The upshot is the rise of vested interest groups, resisting any reform that affects them in adverse ways. Also, the ‘sun-rise’ period of tier-(ii) heavy and chemical industries is spurred by close collaborative actions between the state and industry and accompanied by a tilt of bargaining power in favour of workers and communities where factories are located. Consequently, the ‘sun-set’ period faces ‘legacy costs’ and other politicized problems, restraining the state from carrying out the necessary overhauls. Other tier(ii)-specific problems stemming from heavy industrialization and infrastructure construction include air and water pollution, corruption, overcapacity, and surplus dumping overseas. For instance, the ‘Gilded Age’ of America and the ‘construction nation’ days of Japan in the 1960s through the mid-1970s encountered a similar, if not identical, array of these problems. China, too, happens to be presently in the throes of withdrawing from tier-(ii) growth driven by state-owned enterprises and transforming itself fully into a new consumption-based (tier-(iii) and onwards) economy.

5.5.3 The Third Round: Escaping the ‘High-income (Sclerosis) Trap’ Advanced mature economies tend to stagnate for a multitude of reasons, including a declining rate of consumption and demographic ageing as accompaniments to GDP

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

growth, causing an ever-worsening secular downshift in structural transformation. Japan is a prime example. Its nominal GDP has barely budged over the past two decades. ‘Abenomics’ has been struggling to lift the economy out of the trap by means of ‘three arrows’ (monetary, fiscal, and structural reform measures)—so far largely to no avail. Deflation persists. And C and its related I remain in the doldrums, a phenomenon mentioned in Section 5.4 above. Other mature economies, too, are dependent on expansionary monetary policies, resulting in near-zero or even negative interests. These efforts have not yet met any success in truly rejuvenating their ecosystems.

5.6 S 

.................................................................................................................................. The ‘FG theory’ is now often mentioned in academia and the media alike. But Akamatsu’s original framework is now outdated, calling for elucidation and restatement in order to keep up with the times. However, if one considers the role of multinational corporations as a major driver in the catching-up process, his cornerstone ideas turn even more powerful in shedding light on the evolutionary process of structural transformation across the world. It highlights and explains the interactive and integrative process that is at the heart of contemporary economic development. True, his old-style IS-EP strategy is no longer relevant to the way any aspiring emerging economy should kickoff its industrialization, although the pro-active role of government in igniting and sustaining catch-up growth via ecosystem reforms is still necessary. However, his two other cornerstone ideas—the inter- and intra-industry processes of structural transformation and a closely knit hierarchy of economies aligned at different growth stages—serve as a pair of useful analytics when new thinking is based on them. In fact, what he considered ‘derivative’ turns ‘fundamental’. Catch-up structural transformation derives, in its very essence, from structural transmigration down the hierarchy—largely at the hands of MNCs (foreign and national alike). Trade- and investment-liberated, hierarchical relations are thus the vital source of interactive and integrative growth, particularly when MNCs engage in ‘pro-trade’—instead of ‘anti-trade’—activities that can expand the basis of trade— namely, reinforcing the power of comparative advantage. And for illustration, this chapter introduces the ‘(double-helix) ladder of development à la Schumpeter’ as a reformulated framework and examines the contemporary process of development and its related vital characteristics, namely state–industry relations and the S-shaped growth curve—in new structuralist perspective. Clearly, the financial and demand-side dimensions of development need to be examined, but are beyond the scope of this Chapter.⁸

⁸ For these dimensions of the FG theory, see Ozawa (2009, 2016).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 - :   



R Akamatsu, Kaname, 1935. ‘Wagakuni Yomokogyohinno Boeki Suisei [The Trend of Japan’s Trade in Woolen Goods]’, Shyogyo Keizai Ronso, 13, pp. 129–212. Akamatsu, Kaname, 1955. Boeki no Riron [Trade Theories], Tokyo: Chuo Keizaisha. Akamatsu, Kaname, 1961. ‘A Theory of Unbalanced Growth in the World Economy’, Weltwirtschafliches Archiv, 86, pp. 165–215. Akamatsu, Kaname, 1962. ‘A Historical Pattern of Economic Growth in Developing Countries’, Developing Economies, preliminary issue 1 (March–August), pp. 1–27. Akamatsu, Kaname, 1974. Kinhaika to Kokusaikeizai [Gold Demonetization and the International Economy], Tokyo: Toyokeizai Shinposha. Bresser-Pereira, L.C., 2010. Globalization and Competition: Why Some Emerging Countries Succeed While Others Fall Behind, Cambridge: Cambridge University Press. Chang, Ha-Joon, 2002. Kicking Away the Ladder: Development Strategy in Historical Perspective, London: Wimbledon. Clark, Colin, 1935. The Conditions of Economic Progress, London: Macmillan. Ernst, Dieter, 2006. Innovation Offshoring: Asia’s Emerging Role in Global Innovation Networks, East-West Center Special Report No. 10, Honolulu, Hi: East-West Center. Friedman, Milton and Anna J. Schwartz, 1963. A Monetary History of the United States: 1867–1960, Princeton, NJ: Princeton University Press. Gerschenkron, Alexander, 1962. Economic Backwardness in Historical Perspective, Cambridge, MA: Harvard University Press. Gill, Indermit and Homi Kharas, 2007. An East Asian Renaissance: Ideas for Economic Growth, Washington, DC: World Bank. Hegel, Friedrich W., 1812/1991. The Science of Logic, translated by A. V. Miller, Indianapolis, IN: Hacker (1991). Kojima, Kiyoshi and Terutomo Ozawa, 1984. ‘Micro- and Macro-Economic Models of Direct Foreign Investment: Toward a Synthesis’, Hitotsubashi Journal of Economics, 25 (1), pp. 1–20. Kuznets, Simon, 1966. Modern Economic Growth: Rates, Structure, and Spread, New Haven, CT: Yale University Press. Lee, Keun, 2013. Schumpeterian Analysis of Economic Catch-up, Cambridge: Cambridge University Press. Lewis, Arthur W., 1954. ‘Economic Development with Unlimited Supplies of Labor’, Manchester School of Economics and Social Studies, 12, pp. 139–91. Lin, Justin Yifu, 2012a. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: World Bank. Lin, Justin Yifu, 2012b. The Quest for Prosperity: How Developing Economies Can Take Off, Princeton, NJ: Princeton University Press. Lin, Justin Yifu and Yan Wang, 2014. ‘Africa’s Low-Hanging Fruits: The Right Investment at the Right Place’, UN-WIDER Newsletter, 23 April. List, Friedrich, 1841/56. The National System of Political Economy, Philadelphia, PA: Lippincott. Little, Ian M. D., 1982. Economic Development: Theory, Policy, and International Relations, New York: Basic Books. Ozawa, Terutomo, 1974. Japan’s Technological Challenge to the West: Motivation and Accomplishment, Cambridge, MA: MIT Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Ozawa, Terutomo, 2005. Institutions, Industrial Upgrading, and Economic Performance in Japan: The ‘Flying-Geese’ Paradigm of Catch-up Growth, Cheltenham: Edward Elgar. Ozawa, Terutomo, 2009. The Rise of Asia: The ‘Flying-Geese’ Theory of Tandem Growth and Regional Agglomeration, Cheltenham: Edward Elgar. Ozawa, Terutomo, 2016. The Evolution of the World Economy: The ‘Flying-Geese’ Theory of Multinational Corporations and Structural Transformation, Cheltenham: Edward Elgar. Radelet, Steven and Jeffrey Sachs, 1997. ‘Asia’s Reemergence’, Foreign Affairs, 76 (6), pp. 44–59. Schumpeter, Joseph A., 1934. The Theory of Economic Development, New York: Oxford University Press. Schumpeter, Joseph A., 1942/50. Capitalism, Socialism and Democracy, New York: Harper & Row. US Bureau of the Census, 2006. Historical Statistics of the United States, millennial edition, Cambridge: Cambridge University Press. Wade, Robert H., 1990. Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization, Princeton, NJ: Princeton University Press. World Bank, 1993. The East Asian Miracle: Economic Growth and Public Policy, New York: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

      The Role of Agricultural Prices ......................................................................................................................

.  

A sustainable structural transformation depends, causally, on a society’s widespread confidence in its food security. In return, however, sustaining this confidence in food security at a societal level requires, causally, a successful structural transformation. This mutual, two-way dependence of structural transformation and food security has bedevilled the development profession for decades. The profession has often ignored the critical role of agricultural development and food price stability as the underlying foundations to both structural transformation and food security. Without a clear understanding of how these four topics are linked, food policy analysis, advising, design, and implementation have usually been piecemeal and incomplete. The purpose of this short chapter is to clarify the issues, even if they can only be resolved in highly specific, country settings. I start with a focus on the historical evolution of thinking about these topics, as the foundation for how to think about them going forward. A suggested research agenda is at the end.

6.1 A P  P A  1950

.................................................................................................................................. Ever since colonies of Western powers began to achieve independence after World War II, the development profession has tried to understand the process of structural transformation and how it affects (and is affected by) the role of agriculture in the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

economic growth of poor countries. The interest stems from two distinct features of agricultural sectors in the early stages of development: (1) the sector tends to be the largest employer, producer of economic output, and earner of foreign exchange, and so gains in productivity among agricultural households are critical to broader gains in welfare and reduced poverty; and (2) typically a country’s agricultural sector is the main food producer. Both rural and urban food security depend heavily on the ability of farmers to produce significant surpluses, and on the marketing system to get this food to non-farm consumers. These two dimensions are linked, of course, and understanding the nature and dynamics of linkages among agricultural performance and broader stimulus to economic growth, reduced poverty, and enhanced food security is a continuing challenge.

6.1.1 Structural Transformation, Agricultural Development and Food Security The economics profession has thought a lot about these issues for a long time (Ricardo and Malthus debated them at length). Many Nobel Laureates in Economic Science have made notable contributions. In historical order, the list would include Gunnar Myrdal, Simon Kuznets, Arthur Lewis, T. W. Schultz, Robert Fogel, Douglas North, Amartya Sen, A. Michael Spence, Joseph Stiglitz, Eleanor Ostrom, and Angus Deaton. The third edition of the influential volume edited by Carl Eicher and John Staatz, International Agricultural Development (1998), contains contributions from North, Stiglitz, Sen, and Schultz. An earlier volume by Eicher and Witt (1964) had contributions from North, Kuznets, Schultz, Lewis, Griliches, and Hirschman. Clearly, this is an important topic.¹ Classic volumes on the theme from the 1960s and 1970s include Eicher and Witt on Agriculture and Economic Development (1964); Schultz on Transforming Traditional Agriculture (1964); Mosher on Getting Agriculture Moving (1966); Southworth and Johnston on Agricultural Development and Economic Growth (1967); Wharton on Subsistence Agriculture and Economic Development (1969); Hayami and Ruttan on Agricultural Development: An International Perspective (1971 and 1985); and Reynolds on Agriculture in Development Theory (1975). The 1960s and 1970s saw a flurry of interest in the field, with mainstream economists keenly interested in the linkages among structural transformation, agricultural development, and food security. The food crisis of 1972–74 sharpened this interest, but it had roots deep in work from the 1950s of W. Arthur Lewis and others.

¹ In many ways, the topic is also a personal odyssey. I was an undergraduate studying economics at Harvard from 1959 to 1963, just as development economics was taking shape (and Harvard was very much in the forefront, with Professor Ed Mason founding the Harvard Development Advisory Service (DAS) in 1962). In graduate school at Harvard from 1966 to 1968, I attended a seminar on agricultural development led by Wally Falcon, one of the early leaders in the field (and Wally was my PhD thesis advisor, along with Zvi Griliches and Chris Sims—another Nobel Laureate).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



Drawing out the lessons from this massive body of work since the 1950s is a real challenge, even if the scope is ‘narrowed’ to Asia. The region is far too heterogeneous to find many lessons for agricultural development that are applicable everywhere. Agriculture is very site-specific. That said, just a dozen countries in Asia contain nearly half the world’s population, so lessons from a few countries are relevant to many people. Nearly two-thirds of the world’s poor depend on rice as a staple foodstuff and the world’s rice economy is centred in Asia. Global food security is impossible without reasonably stable rice prices that are affordable by the poor. An Asian focus, with rice at the centre of the discussion, makes very good sense for this topic. It also makes sense to organize the discussion temporally in order to develop an intellectual history of how thinking about the topic has changed during the 1950s, and how, in turn, that thinking influenced policy approaches and development outcomes. This is, of course, a rolling circle of events and outcomes shaping analytical insights and development thought, which then influence strategic approaches and budget priorities, which determine investment plans and incentives for the entire agricultural system. Inevitably, it is a complicated story. But, as my recent book argues, ‘markets have done the heavy lifting’ (Timmer 2015a). Understanding the behaviour of food markets over time provides a simplifying framework for the chapter. Thus a unifying theme is how signals of changing scarcity of food, reflected (at least partially) in market prices, have influenced the intellectual and policy agendas. The historical discussion is organized by decades.

6.1.2 A Quick Review of the Literature: 1950 to the Present A review of the history of thought about drivers of structural transformation, agricultural development, and food security, needs to include the interplay of market events, analytical perspectives as represented in the professional literature, and impact on development strategies and policies. The most thoughtful analysis of these issues is in two chapters in Hayami and Ruttan (1985) on ‘Agriculture in Economic Development Theories’ and ‘Theories of Agricultural Development’, and in the introductory chapter by Staatz and Eicher (1998) on ‘Agricultural Development Ideas in Historical Perspective’. In the 1950s, the focus was on frameworks for development, especially the ‘surplus labour’ model of W. Arthur Lewis (1954) and the analysis of technical change as a driver of agricultural productivity that was stressed by Zvi Griliches (1957) in his work on hybrid corn in the USA. The decade of the 1960s was highly productive for development economics and understanding the role of agriculture in economic growth. The key publications included Johnston and Mellor (1961) on ‘the role of agriculture in economic development’ in the American Economic Review; T. W. Schultz (1964) on Transforming Traditional Agriculture; Eicher and Witt (1964) on Agriculture and Economic Development; Mellor (1966) on The Economics of Agricultural Development; Mosher (1966)

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

on Getting Agriculture Moving; and Southworth and Johnston (1967) on Agricultural Development and Economic Growth. The 1970s saw a renewed concern for income distribution and how to generate an economic growth process that reached the poor, especially in response to the world food crisis from 1972 to 1974. The key publications include Hayami and Ruttan (first edition in 1971) on Agricultural Development: An International Perspective; Johnston and Kilby (1975) on Agriculture and Structural Transformation: Economic Strategies in Late-Developing Countries; Reynolds (1975) on Agriculture in Development Theory; Lipton (1977) on Why Poor People Stay Poor: Urban Bias in Developing Countries; and Binswanger and Ruttan (1978) on Induced Innovation. The 1980s saw a number of important publications that grew out of the tumultuous record of the 1970s. To start, two important contributions from future Nobel Prize winners in economics redefined the basic analytics of food security: Sen (1981) on Poverty and Famines and Newbery and Stiglitz (1981) on The Theory of Commodity Price Stabilization: A Study in the Economics of Risk. An integrated analytical approach to food security was presented in Timmer, Falcon, and Pearson (1983) in Food Policy Analysis, and the first edition of Eicher and Staatz (1984) on Agricultural Development in the Third World appeared shortly after. In 1988, Timmer published ‘The Agricultural Transformation’ in the first volume of The Handbook of Development Economics, an article that has remained relevant to academic debates for a quarter of a century. By the 1990s, the full impact of the collapse of commodity prices in the 1980s was being felt by the economic development profession, which largely lost interest in the agricultural sector. An important empirical study of biases in agricultural price policy was published by Kreuger, Schiff, and Valdez (1991), The Political Economy of Agricultural Pricing Policy, and this too remained influential for decades. By 1995, it was possible for Timmer to ask ‘Getting Agriculture Moving: Do Markets Provide the Right Signals?’ At the time, the answer was ‘no’. Market signals started to change during the 2000s, and the profession took notice. The Timmer (2002) chapter ‘Agriculture and Economic Growth’ in the Handbook of Agricultural Economics synthesized the many reasons agriculture played an important role in economic growth, poverty reduction, and food security, no matter what shortrun price signals said. Fogel (2004) in an extension of his Nobel Prize address, explained The Escape from Hunger and Premature Death, 1700–2100. The rapidly changing food marketing sector was also coming on the policy agenda, with an article by Reardon, Timmer, Barrett, and Berdegue (2003) on ‘The Supermarket Revolution in Developing Countries’ leading the profession into this field. The rural non-farm economy finally received the serious attention it deserved in Haggblade, Hazell, and Reardon (2007), Transforming the Rural Non-Farm Economy: Opportunities and Threats in the Developing World. As an ‘official’ signal that agriculture was back on the agenda, the World Bank (2007) published the World Development Report, 2008: Agriculture for Development. The report was actually drafted before the world food crisis in 2007–08 but it served to mobilize professional and policy thinking about how to proceed in the face of the crisis.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



The early part of the 2010s (and we are only two-thirds of the way through it) saw many efforts to understand why the food crisis in the late 2000s occurred and what to do about it going forward. By 2014 attention was focused on the post-2015 Sustainable Development Goals and what role agricultural development should play in achieving them. Early in the decade, a publication by Timmer (2010) on ‘Reflections on Food Crises Past’ in Food Policy raised the issue of regularly recurring food crises and the cyclical nature of policy responses to them. Stanford University’s Center for Food Security and Environment hosted a series of lectures on global food policy that resulted in Frontiers in Food Policy: Perspectives on sub-Saharan Africa (edited by Falcon and Naylor 2014), which also contains very useful annotated summaries of the most influential literature in the fields of the lectures.² Early in 2015, Timmer’s Food Security and Scarcity: Why Ending Hunger is so Hard (2015a) was published, pulling together more than four decades of analysis of structural transformation, agricultural development, and food security. Where to next? The way forward must deal with accumulated historical issues, analytical challenges and policy approaches. Coping with climate change will be high on analytical and policy agendas. The ‘double burden’ of hunger and obesity seriously complicates food policy approaches aimed primarily at the link between poverty and undernutrition. Volatility in food prices is already a major concern and may worsen, although experience in Asia since 2008 with quiet coordination of government import and export policies for rice suggests a way forward. Sustainability of modern agriculture, the role of genetically modified organisms (GMOs) in the food chain, the market power of large-scale supermarkets and their concentrated supply chains, and increasing demand for ‘local’ foods are all likely to be on the agenda.

6.2 S T   O F

.................................................................................................................................. Understanding structural transformation is mainly an exercise in economic history, but learning how to manage the process involves understanding the political economy of policy making. The food system is at the core of this process in both the long run and short run. In the long run, the food system is a key element of the structural transformation, which historically has been the only sustainable pathway out of poverty. In the short run, the food system is the arena in which many of the poor make their living, and also face the risks of volatile food prices. A structural transformation that is successful in ending hunger requires that each society finds the right mix of market forces and government interventions to drive a process of economic growth that reaches the poor and ensures that food supplies are ² See Appendix 1 for the ‘Core Literature on Food Price Instability’, from Timmer (2011).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

readily, and reliably, available and accessible to even the poorest households. Finding this ‘right mix’ has been a major challenge, and serious efforts to provide food security to a society seem to have stimulated important ‘learning’ by governments on how to manage not just the essentials of rural markets, but also the broader dimensions of economic growth. Solving the ‘food problem’ is thus a key step—and a powerful catalyst—to solving the problem of poverty and finding the path to higher incomes. No country has been able to sustain rapid economic growth until its citizens and investors were confident that food was reliably available in the main urban markets. Rural poverty has always been a later concern. However, rural productivity and economic growth provide the ingredients to broad-based food security. The two are intimately linked, and identifying the many factors that must come together to generate a successful structural transformation aids an understanding of why. Two other transformations occur simultaneously with the structural transformation— sorting out which are cause and which are effect is mostly a fool’s game; they happen together. The agricultural transformation takes place within the sector at the same time the sector itself is changing its relationship to the rest of the economy— structural transformation (Timmer 1988). And the dietary transformation follows surprisingly robust ‘laws’ as societies become richer and more urbanized. Engel’s law describes the declining share of food in the budget of all families as they become richer. Bennett’s law describes the reduced role of starchy staples (cereals and root crops) and the increased diversity of calorie and protein sources in the diets of richer families (Bennett 1954). The persistence across countries and time of these common patterns of dietary change, and the agricultural transformations that make them possible, suggest verydeep seated global forces at work, no doubt some of them wired by evolution into humans’ brains. At the same time, there is widespread variance at the local level around these common pathways, so there is ample scope for unique behaviour and patterns as well. Here is the role for policy analysis at the country level.

6.2.1 Structural Transformation in Historical Perspective All successful developing countries undergo a structural transformation, which involves four main features: ³ a falling share of agriculture in economic output and employment; a rising share of urban economic activity in industry and modern services;

³ The heading of this subsection is the sub-title of my American Enterprise Institute (AEI) monograph A World without Agriculture (Timmer 2009a), from which much of this section is drawn. Detailed analysis of the gap between labour productivity in the agricultural and non-agricultural sectors was first presented there.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



migration of rural workers to urban settings; and a demographic transition in birth and death rates that always leads to a spurt in population growth before a new equilibrium is reached. The structural transformation involves declining shares of agriculture in GDP and employment, almost always accompanied by serious problems closing the gap in labour productivity between agriculture and non-agriculture. The basic cause and effect of the structural transformation is rising productivity of agricultural labour. The four basic dimensions of structural transformation are seen by all developing economies experiencing rising living standards; diversity appears in the various approaches governments have tried in order to cope with the political pressures generated along that pathway. Finding efficient policy mechanisms that will keep the poor from falling off the pathway altogether—managing the structural transformation—has occupied the development profession for decades. There are three key lessons. First, the structural transformation has been the main pathway out of poverty for all societies, and it depends on rising productivity in both the agricultural and nonagricultural sectors (and the two are connected). The stress on productivity growth in both sectors is important, as agricultural labour can be pushed off farms into even lower productivity informal service sector jobs, a perverse form of structural transformation that has generated large pockets of urban poverty, especially in sub-Saharan Africa and India.⁴ It is no accident that these are the two regions of the world where food insecurity remains severe. Second, in the early stages, the process of structural transformation widens the gap between labour productivity in the agricultural and non-agricultural sectors. This widening gap puts enormous pressure on rural societies to adjust and modernize. These pressures are then translated into visible and significant policy responses that alter agricultural prices. The agricultural surpluses generated in rich countries because of artificially high prices then cause artificially low prices in world markets and a consequent undervaluation of agriculture in poor countries (Timmer 1995). This undervaluation of agriculture since the mid-1980s, and its attendant reduction in agricultural investments, is a significant factor explaining the world food crisis in 2007–08 and continuing high food prices into the mid-2010s. Third, despite the decline in relative importance of the agricultural sector, leading to the ‘world without agriculture’ in rich societies, the process of economic growth and structural transformation requires major investments in the agricultural sector itself (Timmer 2009a). This seeming paradox has complicated (and obfuscated) planning in developing countries as well as donor agencies seeking to speed economic growth and connect the poor to it. Because of active policy concerns about providing food security to their citizens, countries in East and Southeast Asia largely escaped much of this paradox, but sub-Saharan Africa has not. ⁴ Both of these cases have been documented in the Stanford Symposium Series on Global Food Policy and Food Security in the 21st Century (Badiane 2011; Binswanger-Mkhize 2012).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.   100 1880

1880 50

1880

1880

2010 2010 2010 2010

2010

0 1880

1880

2010

–50 6

7

8 9 Logarithm of GDP per capita

10

Employment in agriculture (% of total employment) Agriculture, value added (% of GDP) Agriculture share of GDP minus share of employment Dashed lines indicate Japan’s trajectory; solid lines show Indonesia’s.

 . Long-run patterns of structural transformation in Japan and Indonesia Source: Author’s research.

For poverty-reducing initiatives to be sustainable over long periods of time, the indispensable necessity is a growing economy that successfully integrates factor markets in the rural sector with those in urban sectors, and stimulates higher productivity in both. That is, the long-run success of poverty reduction, and with it, improvements in food security, hinge directly on a successful structural transformation. The historical record is very clear on this path. Figure 6.1 shows the historical path of structural transformation from 1880 to 2010 for Japan and Indonesia. The similarity in paths is quite striking. Managing the ingredients of rapid transformation and coping with its distributional consequences have turned out to be a major challenge for policy makers. ‘Getting agriculture moving’ in poor countries is a complicated, long-run process that requires close, but changing, relationships between the public and private sectors. Donor agencies are not good at this. More problematic, the process of agricultural development requires good economic governance in the countries themselves if it is to work rapidly and efficiently. Aid donors cannot hope to contribute good governance themselves—and may well impede it (Pritchett 2013). The strong historical tendency toward a widening of income differences between rural and urban economies during the initial stages of the structural transformation is now extending much further into the development process. Consequently, with little prospect of quickly reaching the turning point, where farm and non-farm productivity and incomes begin to converge, many poor countries are turning to agricultural protection and farm subsidies sooner rather than later in their development process. The tendency of these actions to hurt the poor is then compounded, because there are so many more food-deficit, rural poor in these early stages.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



6.2.2 Agricultural Productivity and Structural Transformation No country has been able to sustain a rapid transition out of poverty without raising productivity in its agricultural sector (if it had one to start—Singapore and Hong Kong are exceptions). The process involves a successful structural transformation where agriculture, through higher productivity, provides food, labour, and even savings to the process of urbanization and industrialization. A dynamic agriculture raises labour productivity in the rural economy, pulls up wages, and gradually eliminates the worst dimensions of absolute poverty. Somewhat paradoxically, the process also leads to a decline in the relative importance of agriculture to the overall economy, as the industrial and service sectors grow even more rapidly, partly through stimulus from a modernizing agriculture and migration of rural workers to urban jobs. All societies want to raise the productivity of their economies. That is the only way to achieve higher standards of living and sustain reductions in poverty. The mechanisms for doing this are well known in principle if difficult to implement in practice. They include the utilization of improved technologies, investment in higher educational and skill levels for the labour force, lower transactions costs to connect and integrate economic activities, and more efficient allocation of resources. The process of actually implementing these mechanisms over time is the process of economic development. When successful, and sustained for decades, it leads to the structural transformation of the economy. The structural transformation complicates the division of the economy into sectors— rural versus urban, agricultural versus industry and services—for the purpose of understanding how to raise productivity levels. In the long run, the way to raise rural productivity is to raise urban productivity, or as Chairman Mao famously but crudely put it, ‘the only way out for agriculture is industry’. Unless the non-agricultural economy is growing, there is little long-run hope for agriculture. At the same time, the historical record is very clear on the important role that agriculture itself plays in stimulating growth in the non-agricultural economy (Timmer 2002, 2005a, 2005b, 2008, 2016). In the early stages of the structural transformation in all countries there is a substantial gap between the share of the labour force employed in agriculture and the share of gross domestic product (GDP) generated by that work force. Figure 6.1 shows that this gap narrows with higher incomes. This convergence is also part of the structural transformation, reflecting better integrated labour and financial markets. However, in many countries this structural gap actually widens during periods of rapid growth, a tendency seen in even the earliest developers. When overall GDP is growing rapidly, the share of agriculture in GDP falls much faster than the share of agricultural labour in the overall labour force. The ‘turning point’ in the gap generated by these differential processes, after which labour productivity in the two sectors begins to converge, has also been moving to the right over time.⁵ ⁵ This is not a temporal statement, but one driven by movements in real incomes per capita. If incomes per capita fall over extended periods, as they have in Brazil or Nigeria, for example, the pathway

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

This lag inevitably presents political problems as farm incomes visibly fall behind incomes being earned in the rest of the economy. The long-run answer, of course, is faster integration of farm labour into the non-farm economy (including the rural, nonfarm economy), but the historical record shows that such integration takes a long time. It was not fully achieved in the USA until the 1980s (Gardner 2002), and evidence shows the productivity gap is increasingly difficult to bridge through economic growth alone (Timmer 2009a). This lag in real earnings from agriculture is the fundamental cause of the deep political tensions generated by the structural transformation, and it is getting worse. Historically, the completely uniform response to these political tensions has been to protect the agricultural sector from international competition and ultimately to provide direct income subsidies to farmers (Lindert 1991; Anderson et al. 2013). Neither policy response tends to help the poor, even those remaining in rural areas. We now understand that the political economy of this process is driven by the structural transformation itself.

6.3 T   T P

.................................................................................................................................. Policy challenges can arise almost anywhere along the path of even the most successful structural transformation and many of them will be quite specific to the time and place where a problem occurs. In this sense, each country must find its own path, and solve its own problems along the way. Still, comparative economic history is rich with examples of common problems during the structural transformation and one is presented here: the widening gap between labour productivity in rural and urban areas as rapid industrialization takes place. How to measure and manage the growing gap in labour productivity between sectors occupies much of A World without Agriculture (Timmer 2009a) and my WIDER Lecture (Timmer 2015b), but important new evidence has been developed since the empirical work in those monographs, with hopeful implications for successfully managing a structural transformation.

6.3.1 Mind the Gap In the early stages of the structural transformation in all countries there is a substantial gap between the share of the labour force employed in agriculture and the share of GDP generated by that workforce. As can be seen in Figure 6.1, this gap narrows with ‘back’ is not likely to track the pathway ‘forward’ because of substantial stickiness in structural patterns of labour allocation.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



higher incomes. This convergence is also part of the structural transformation, reflecting better integrated labour and financial markets. The role of better technology and higher productivity on farms as a way to raise incomes in agriculture is controversial. Most of the evidence suggests that gains in farm productivity have been quickly lost (to farmers) in lower prices and that income convergence between agriculture and non-agriculture is driven primarily by the labour market (Johnson 1997; Gardner 2002). Moreover, in many countries this structural gap actually widens during periods of rapid growth, as was evident in even the earliest developers, the now-rich OECD countries. When overall GDP is growing rapidly, the share of agriculture in GDP falls much faster than the share of agricultural labour in the overall labour force. The turning point in the gap generated by these differential processes, after which labour productivity in the two sectors begins to converge, has also been moving ‘to the right’ over time, requiring progressively higher per capita incomes before the convergence process begins. Most empirical analysis of the structural transformation has focused on two variables—agriculture’s share in employment and in GDP. The ‘gap’ between the two has often been recognized, yet it has received little systematic analysis.⁶ In fact, the gap is an important element in understanding the political economy of structural transformation. For our purposes here, the gap is defined as the difference between the share of agriculture in GDP and its share in employment. This definition consciously causes the GAP variable to be negative in sign for virtually all observations, a visual advantage in Figure 6.1, which shows the gap approaching zero from below. One advantage of using the difference in shares rather than their relative values is that the gap variable then translates easily into a ‘sectoral Gini coefficient’ that indicates the inequality of incomes (labour productivity) between the two sectors.⁷ The negative of the GAP variable is equal to the Gini coefficient for agricultural GDP per worker compared with non-agricultural GDP per worker. On average, this ‘sectoral Gini coefficient’ accounts for 20–30 per cent of the variation in the overall Gini coefficient for most countries. Mitigating the rural–urban divide has turned out to be the key to maintaining political stability and rapid economic growth. This paradoxical connection stems from a failure of market-driven economic models based on free trade to cope with the powerful consequences of growing income inequalities. As incomes become more skewed, so too do the expectations, cultural values, and political orientation of those left behind. In China and India the increase in this gap since the early 1990s has generated serious political pressures. Rural America elected Donald Trump as president of the USA. Most rural households in the country are highly alienated from what is happening in more dynamic, liberal areas. This has turned out to be a poisonous political brew.⁸

⁶ The work by van der Meer and Yamada (1990) is an exception. ⁷ See Annex Table A-6 in Timmer and Akkus (2008) for details and an algebraic proof of this relationship. ⁸ See an especially thoughtful op-ed article by Robert Leonard, ‘Why Rural America Voted for Trump,’ The New York Times, 5 January 2017, p. A23.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

6.3.2 Widening Rural–Urban Income Gaps A worrisome aspect of the rural–urban income gap is that it tends to get larger during the early stages of economic growth. The turning point in the relationship is calculated from a regression explaining the size of the GAP variable as a function of the logarithm of per capita income and per capita income squared. The fact that labour productivity in the nonagricultural sector actually increases more rapidly than in the agricultural sector until this turning point is reached, thus exacerbating rural–urban income differences, has much to do with the political difficulties poor countries face during a rapid structural transformation. The turning point comes at a lower per capita income level when the domestic agricultural terms of trade variable is included in the statistical analysis. Individual countries use agricultural price policy as one way to manage the structural transformation by influencing their domestic terms of trade. This policy instrument helps the growth process to integrate agricultural labour into the rest of the economy, at least in terms of relative productivity. On the other hand, political efforts to influence the domestic terms of trade often run into powerful counter pressures from global commodity markets, and thus require large subsidies or trade barriers to make them effective. One possible advantage of higher food prices in world markets since 2007 might be less pressure on policy makers to protect their agricultural sectors from the forces of rapid structural transformation, a point discussed below.

6.3.3 Changes over Time One overarching question about the structural transformation is whether it has been a uniform process over time, or whether the very nature of economic growth, and its capacity for integrating ‘surplus’ agricultural workers into the non-agricultural sector, has been changing over the course of history. There are two ways to address the issue. The first is to examine the short-run record of growth using a sample of countries, with data from 1965 to 2000. The second approach is to examine the long-run record of the early developers to see how their patterns of structural transformation might differ from the modern record.

6.3.3.1 The Short Run There are a number of ways to slice the modern record (the 1965–2000 period) of structural transformation into smaller segments. There are systematic patterns over time in turning points. The clearest pattern occurs for the turning points in the gap relationship when the regression includes the terms of trade variable. These turning points are as follows: 1965–74: 1975–84: 1985–94: 1995–2000:

US$1,109 US$6,379 US$7,880 US$15,484

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



Unmistakeably, the turning point for the gap in labour productivity between the agricultural and non-agricultural sectors has been steadily rising since the mid1960s. Such results are strongly suggestive of a failure of modern economic growth processes to integrate the agricultural sector of poor countries into the rest of their economy despite relatively successful aggregate growth records (Ravallion et al. 2007). A widening sectoral income gap—as differences in labour productivity between urban and rural areas become larger—spells political trouble. It is no wonder that policy makers feel compelled to address the problem, and the most visible way is to provide more income to agricultural producers. The long-run way to do this is to raise their labour productivity and encourage agricultural labour to migrate to urban jobs, but the short-run approach—inevitable in most political environments—is to use trade policy to affect domestic agricultural prices (Olson 1965; Lindert 1991). In low-income economies, agricultural protection is a child of growing income inequality between the sectors during the structural transformation.

6.3.3.2 Long-run Patterns from 1820 to 1985 Concerns about the distributional impact of globalization are not new. The world economy experienced an earlier round of globalization from 1870 to World War I, and there may be lessons from the currently developed countries that participated in that process. Their economies were experiencing rapid economic growth (by the standards of the time) and facing challenges from the growing integration of labour and capital markets across countries (Williamson 2002). Thanks to recent work by economic historians, it is possible to examine the nature of these challenges empirically. The results are striking. First, the patterns from the early developers are remarkably similar to those for the sample of countries from 1965 to 2000. Although the small sample size (nine countries with just four observations for all but the UK) means the coefficients are measured with considerable error, they are still significant by most standards, with the same pattern of signs and magnitudes as for the full sample of contemporary economies (for details, see Timmer and Akkus 2008). Second, the tendency for the gap share variable to widen in the early stages of development seems equally strong in the early developers. But the turning point had been reached early in their development. The UK passed its turning point before 1800, the continental European countries reached it by the mid-1900s, and Japan followed early in the twentieth century. These growth patterns suggest that the early experience for these advanced countries was much more similar to the international growth patterns of the 1960s and 1970s than to those of the past several decades. Indeed, virtually the entire growth experience of modern developed countries has been spent on the convergent path of sectoral labour productivity. This is in sharp contrast to currently developing countries, which are mostly at income levels per capita where sectoral labour productivity is diverging.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

6.4 A P  D

.................................................................................................................................. Asian countries have a very different pattern of agricultural employment change with respect to per capita incomes than non-Asian countries. This difference seems to be accounted for by policy measures designed to manage the tensions that arise during rapid structural transformation. Asian economies tend to employ disproportionately more farm workers in the early stages of development. Indeed, statistical analysis shows that Asian countries were able to use the agricultural terms of trade as a policy instrument for keeping labour employed in agriculture, a pattern not seen in other developing countries. Average economic growth in the Asian sample was faster than in other countries, and the rapid decline in the share of GDP from agriculture reflects this rapid growth. Asian countries relied heavily on the agricultural terms of trade as a policy tool to mitigate the consequences of rapid growth: a widening gap in labour productivity between the agricultural and non-agricultural sectors. Thus Asian countries provided more price incentives to their agricultural sectors over this time period as a way to prevent the movement of labour out of agriculture being ‘too fast’. Certainly the pattern of movements in the agricultural terms of trade for the two sets of countries is strikingly different, with Asian countries seeing a longrun decline at half the pace of the non-Asian countries (see Figure 6.2). The net effect of these forces on the gap between labour productivity in the two sectors is that the turning point in the GAP relationship (after which labour productivity in agriculture begins to converge with labour productivity in non-agriculture) is 180

160

140

120

100 1960

1970

1980 Year Asian mean

1990

2000

Non-Asian mean

 . Agricultural terms of trade for Asia and non-Asia separately (2000 = 100) Source: Timmer (2009a).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



sharply lower in the Asian sample. The turning point for the Asian countries is just US $1,600, whereas it is over US$11,000 for the non-Asian countries—over six times higher. This difference underscores two distinctive features of the Asian economies— their more rapid growth and the greater role of agricultural productivity in that growth (Timmer 2005b). The reasons for these differences have been the source of considerable debate. An explanation that resonates with the empirical results reported here is that Asian countries were more concerned about providing ‘macro’ food security in urban markets and ‘micro’ food security to rural households because of large and dense populations engaged in farming on very limited agricultural resources. Political stability, and with it the foundation for modern economic growth, grew out of an approach to the provision of food security that connected poor households to improved opportunities (Timmer 2004, 2005a). The key to this story has been management of the domestic agricultural terms of trade. Did the world food crisis change things? Many Asian countries used domestic price policy to keep the agricultural terms of trade more favourable for their farmers, and thus kept political tensions from rapid structural transformation under control. Domestic policy interventions were necessary because global food prices had been steadily declining since the early 1980s (see Figure 6.3). Openness to those declining food prices, although very beneficial to the poor, was a real challenge to domestic farmers. The lack of successful agricultural transformation in many countries might thus be linked to the low profitability of agriculture, at least in world markets. Have things changed? 1.8

1.6

1.4

1.2

1

0.8 1980

1990 (mean) TOT

Year

2000 Fitted values

 . Agricultural terms of trade, global average, 1980–2010 Source: Author’s research, with the assistance of Selvin Akkus Clemens.

2010

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

Figure 6.3 shows that there was a significant ‘regime change’ in global food markets in the mid-2000s. For the first time since the early 1970s, average agricultural terms of trade were rising instead of falling. Openness to these higher food prices could be harmful to poor consumers, especially in the short run. But much more widespread agricultural dynamism also seems to have resulted in those countries that had lagged for decades. It is far too early to tell if the tide really has turned, and faster productivity growth in agriculture—made possible by sharply higher incentives—gets transmitted into successful structural transformations in sub-Saharan Africa and South Asia. The drop in commodity prices in late 2014 and early 2015 might signal the end of this boom. But the evidence from the increase in global agricultural terms of trade is promising in one respect: the gap in labour productivity between agricultural and nonagricultural labour is no longer so difficult to close. Using the new data set that generated the results in Figure 6.3, it is possible to calculate the turning point when the gap begins to close. That turning point had been moving rapidly to higher incomes in the earlier data set, which ended in 2000. The new data set starts only in 1980 but now extends to 2010. Having the extra decade, with its reversal in the downward trend in agricultural terms of trade, reveals a startling result. The new turning points are as follows: 1980–89: 1990–99: 2000–10:

US$15,493 US$97,838 US$5,668

As before, the turning point was moving to sharply higher income levels in the 1980s and 1990s. But in the first decade of the twenty-first century, not only did that trend stop, the level of the turning point returned to income levels easily within reach of transition economies. The ‘villain’ in the story of increased difficulty in integrating agricultural and non-agricultural sectors in developing countries was not globalization or even bad domestic policies (although they may have been players as well). The real driver was the rapid fall in global food and agricultural prices and the difficulties created for domestic policy makers as they tried to manage a smooth structural transformation. Without incentives to raise agricultural productivity, the sector stagnated. In turn, without the stimulus from a dynamic agricultural sector and rising labour productivity, the rest of the economy stagnated as well. A new window of opportunity opened with the world food crisis in 2007–08. It remains to be seen whether the window remains open and whether the opportunity is broadly seized.

6.5 A M F P P

.................................................................................................................................. Folk wisdom holds that ‘an ounce of prevention is worth a pound of cure’—prevention is sixteen times better than coping. Preventing food crises requires two separate, but integrated, approaches—a market-oriented approach to economic growth and structural transformation, and a stabilization approach to policy initiatives that prevent

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



sharp price spikes for staple foods. Both approaches require a behavioural perspective, and neither can work without the other. The pathway of structural transformation is long and hard. It is easy to get sidetracked or to miss the path altogether. The endpoint—an agricultural sector that is a small share of a large economy—is easily confused with a development strategy that squeezes agriculture from the start. Such a strategy has always been a catastrophe. Because of the unreliability of market prices in the short run as signals for long-run investments, both governments and private firms, easily miss the importance of investing in higher agricultural productivity, better food safety standards or social responsibility (Timmer 1989, 1995, 2009b, 2012). Changing income distribution is an important part of the problem. Even if the structural transformation goes smoothly, most rural households find growth in their incomes lagging behind growth in urban incomes. Changing relative incomes in rural and urban areas drive political dynamics, and the nearly universal tendency to increase agricultural protection during a successful structural transformation is easily understandable from the viewpoint of behavioural economics, thus explaining much of the ‘empirical’ political economy of food prices.⁹ Successful structural transformations have always been primarily a market-driven process. Markets process billions of pieces of information on a daily basis to generate price signals to all participants—no other form of institutional organization has evolved that is capable of the necessary information processing required for individuals and firms to make efficient allocation and investment decisions, and thus to raise longrun productivity. Without reasonably efficient markets, we are all doomed to poverty. The dilemma, of course, is that markets sometimes (or often, depending on political perspective and analytical training) fail at tasks that society regards as important, such as poverty reduction, nutritional well-being, or food price stability, even employment generation. We now understand that these failures are not just for technical reasons— externalities, spillovers, monopoly power, or asymmetric information, for example— but also have deep behavioural roots, based in loss aversion, widespread norms of fairness, and the regularity of ‘other-regarding preferences’. Fixing them is not easy unless these root causes are incorporated into the policy analysis, design and interventions (an example is in Thaler and Benartzi 2004). That said, a number of behavioural regularities are well documented, and building them into policy design simply requires paying attention. Norms of ‘fairness’ for example, are easy to build into food subsidy schemes—even when they conflict with economists’ sense of efficiency. The Raskin program of rice distribution to the poor in Indonesia, for example, has struggled with the ‘losses’ to rice distributed by village leaders on the basis of a ‘fairness’ mechanism

⁹ See Lindert (1991) for a summary of the empirical regularities in agricultural policy that cannot be explained by standard neo-classical economics. These include a bias against both imports and exports, an urban bias in poor countries when farmers are a majority of the population, and a rural bias when urban consumers are a majority of the population.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

rather than a ‘poverty’ mechanism. Knowing that such an approach was inevitable from the start would have significantly improved the performance of this programme. Beyond market failures, there are several problems with the process of structural transformation in the short- and medium-term. A health and nutrition transition seems to accompany structural transformation, but with lags and significant sectoral differences. Not all of the transitional impact is positive: significant increases in obesity, and accompanying chronic diseases, are linked to both the higher incomes and larger urban populations that come with successful structural transformation, as evidence from China and India is making apparent (Webb and Block 2012). Technical change, which is stimulated by high food prices, has paradoxically been the long-run mechanism for generating low food prices and better nutrition for the poor. There is considerable debate over the impact of cheap food, a processing-oriented commercial food sector, and urban lifestyles, on the rising tide of obesity. But again, the temporal disconnect between the poor losing access to food in the short run because of high prices, and a positive long-run technological response, requires public understanding and intervention, in the nutrition arena as well as in preventing food crises. By necessity, the poor live in the short run, but must place their hope for an escape from poverty in long-run forces that are mediated by efficient markets. The time inconsistent behaviour of most individuals and policy makers means this dilemma is very difficult to resolve. The growing importance of targeted social safety nets, including direct food deliveries to the poor in both normal times and during economic and food crises, may be pointing the way to a political resolution.

 1

..................................................................................................................................

C      Theodore W. Schultz, 1945. Agriculture in an Unstable Economy, New York: McGraw-Hill. Although concern over unstable agricultural prices and incomes is centuries old—the English Corn Laws date to 1688 and were concerned with both—the first modern treatment of the causes and consequences of instability in agriculture dates to this volume by T. W. Schultz. He was emphatic in attributing much of the causation of unstable agricultural prices to macroeconomic instability rather than the peculiarities of individual crop supply and demand, a position that put Schultz at odds with much of the agricultural economics profession at the time. In his later volume, The Economic Organization of Agriculture, published in 1953, Schultz carried his perspective to its logical conclusion: ‘The instability of farm prices is an important economic problem. It is, however, exceedingly difficult to organize the economy so that farm prices will be on the one hand both flexible and free and on the other hand relatively stable.’ Schultz resisted efforts to stabilize individual commodity prices from then on. David M. G. Newbery and Joseph E. Stiglitz, 1981. The Theory of Commodity Price Stabilization: A Study in the Economics of Risk, Oxford: Clarendon Press. This volume had a sharp impact on the development community when it appeared three decades ago. One of the first major efforts to put development economics on a firm micro

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



foundation, it treated commodity price instability as a problem for households and firms, which needed to cope with the risk of price fluctuations. A dynamic optimization model that incorporated risk into household decision making was expanded to prove that international commodity agreements (ICAs) to stabilize prices on world markets could not work— eventually they would run out of funds to buy at low prices or commodities to inject into markets at high prices. The profession has taken to heart the key conclusion from this analysis: it is impossible in theory and in practice to stabilize commodity prices. Of course, this holds only globally, not for individual countries, and all the costs and benefits are microbased. No costs to the macro economy stemming from unstable commodity prices, or benefits from stabilizing them, are dealt with in the analysis. C. Peter Timmer, 1989. ‘Food Price Policy: The Rationale for Government Intervention’, Food Policy, 14(1), pp. 17–27. At one level this paper is an attempt to confront the conclusions from Newbery and Stiglitz (1981) with the reality of successful food price stabilization efforts in a number of countries in Asia. The rationale for these stabilization programmes is developed at length, with considerable attention to the macro dimensions of food price instability, which rely heavily on signal extraction problems for investors. Without food stability at the macro level in major urban markets—proxied in Asia by stable rice prices—countries have a very hard time lengthening investors’ time horizons to fit the needs of modern economic growth. Stable food prices speed up that growth. Jeffrey C. Williams and Brian D. Wright, 1991. Storage and Commodity Markets, Cambridge: Cambridge University Press. This volume builds on a half-century of work on the supply of storage as the basic analytical framework for understanding inter-temporal price formation. A unique feature of commodity storage—it cannot be negative—is used to build a dynamic model of commodity prices. The model is very successful in reproducing the common features of commodity prices, especially their tendency to be low and stable for long periods of time, and then subject to sharp upward shocks. This volume remains the basic reference on how storage affects price formation. C. Peter Timmer, 1995. ‘Getting Agriculture Moving: Do Markets Provide the Right Signals?’ Food Policy, 20(5): 455–72. This paper appeared in a special issue of Food Policy that honoured Art Mosher and his insights on how to ‘get agriculture moving’. One of the key questions in the agricultural development literature is the role of price incentives to stimulate adoption of new technology. The basic argument in this paper is that prices on world markets for the key food staples— rice, wheat, and maize—often do not reflect either their long-run scarcity value with respect to investments in agricultural development, or their potential to create added value in the form of rural incomes, and thus faster poverty reduction. Donors should not use short-run prices in world markets to judge the impact of their investments in agricultural research and infrastructure, but should look at long-run trends and the feedback from current investment decisions to future food abundance and scarcity. C. Peter Timmer, 2000. ‘The Macro Dimensions of Food Security: Economic Growth, Equitable Distribution, and Food Price Stability’, Food Policy, 25(4): 283–95. This paper demonstrates the interactions among the rate of economic growth, of who participates in that growth, and the level of food prices, as they affect the numbers of people counted as ‘food insecure’. The basic methodology follows from earlier work by Reutlinger

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

and Selowsky (1976), but introduces food price instability as an important causal factor changing the level of food security. An important conclusion is that stable food prices make the achievement of ‘macro’ food security much easier, and ‘pro-poor’ growth makes ‘micro’ food security feasible. In combination, a rapid escape from poverty and hunger is possible. World Bank, 2005. Managing Food Price Risks and Instability in an Environment of Market Liberalization, Agriculture and Rural Development Department Report No. 32727GLB, Washington, DC. Many of the papers in this volume also appeared in a special issue of Food Policy edited by Derek Byerlee, Thom S. Jayne, and Robert J. Myers that appeared in May 2006. The volume was the result of a free-ranging conference arranged by the World Bank, but this summary reflects a clear neo-classical approach that allows unrestricted price formation with follow-up activities to protect food consumption of the poor if prices suddenly spike. Producers are urged to use modern financial derivatives to hedge their risks from price volatility, whereas poor consumers will need to rely on government-sponsored safety nets when food prices spike. This ‘Washington Consensus’ view of how to deal with food price instability has been challenged by the food crises in 2008 and 2011. Shahidur Rashid, Ashok Gulati, and Ralph Cummings, Jr., eds, 2008. From Parastatals to Private Trade: Lessons from Asian Agriculture, Baltimore, MD: Johns Hopkins University Press for the International Food Policy Research Institute. This volume makes the case that food price stabilization implemented via parastatals was necessary and effective for Asian countries to introduce Green Revolution technologies to small holders in the context of poor marketing infrastructure. However, as infrastructure and private marketing capacity have developed rapidly, and food parastatals have been subject to gross mismanagement and corruption, the time has come to turn most of food marketing in Asia over to the private trade. The editors/authors are especially knowledgeable about India. Philip C. Abbott, Christopher Hurt, and Wallace E. Tyner, 2008. What’s Driving Food Prices? (also supplements in 2009 and 2011), Farm Foundation Issue Report (FFIR), Oak Brook, IL. This was among the first scholarly efforts to understand what was driving the food price crisis in 2008 and has been the standard since. The update for 2011 argues that the drivers are somewhat different than in 2008, when exchange rate movements received a great deal of attention. In 2011, the authors place most of the blame on US and EU bio-fuels policies and on the Chinese decision to build substantial stocks of soybeans even as the world price was rising. They are increasingly nervous that demand growth for food will outstrip growth in production, with continuing high and unstable prices. C. Peter Timmer, 2010. ‘Reflections on Food Crises Past’, Food Policy, 35(1): 1–11. Similarities and differences between the rice price crisis in 1972/73 and the one in 2007/08 are analysed, especially from the perspective that long-run cycles in funding for agricultural research and infrastructure are the basic cause of periodic food crises. The changes in political economy of responses to spikes in rice prices between the two episodes are dramatic, and are determined largely by how well insulated domestic consumers were from world markets. Case studies of Indonesia, India, and Thailand also show a significant difference in policy response in the face of democratic pressures, which were present only in India in 1972/73, but were a force in all three countries in 2007/08.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



David Dawe, ed., 2010. The Rice Crisis: Markets, Policies and Food Security, London and Washington, DC: Earthscan. This volume grew out of a FAO-sponsored conference early in 2009 to examine what went wrong with the world rice market. It pulls together a number of country studies as well as several analyses of how the world rice market functioned in 2007/08. The Dawe and Slayton chapter in particular analyses the role of Japan and its WTO stocks of rice in pricking the speculative bubble in world prices that had formed as a result of panicked buying by the Philippines and widespread hoarding at all levels of the rice system—hoarding that was caused by the expectation of higher prices themselves. The need for more open trade policies, and larger rice reserves as a way to build confidence in such trade, is stressed in the conclusion. Christopher L. Gilbert and C. Wyn Morgan, 2010. ‘Food Price Volatility’, Philosophical Transactions of the Royal Society, 365: 2023–34. This commissioned review of the literature on food price volatility provides a very careful and sober assessment of recent claims that price volatility is increasing (the evidence is not in, but volatility in the 1970s was as great as now). Gilbert has done much of the high-quality analysis of commodity price trends and variations over the past two decades, and this article summarizes his findings very effectively. Evidence is provided that financial speculation did increase volatility of food prices in 2001, but not as much as in energy and mineral markets. The paper makes a clear case for why the world rice market is quite different from the markets for wheat, maize, and soybeans. Rosamond L. Naylor and Walter P. Falcon, 2010. ‘Food Security in an Era of Economic Volatility’, Population and Development Review, 36(4): 693–723. This paper summarizes results from a major research programme at Stanford on food security and the environment. It clarifies the debate over how to measure food price volatility and how those measures have changed over time, for the key food staples (and petroleum). The impact of food price volatility on the rural poor is examined in depth, perhaps for the first time. Concerns are raised about the restrictions on trade, and especially the widening of FOBCIF price bands for important food importing countries, that seem to represent a structural shift after 2008.

R Anderson, Kym, Gordon Rausser, and Johan Swinnen, 2013. ‘Political Economy of Public Policies: Insights from Distortions in Agricultural and Food Markets’, Journal of Economic Literature, 51 (2), pp. 423–77. Badiane, Ousmane, 2011. ‘Agriculture and Structural Transformation in Africa’. Stanford Symposium Series on Global Food Policy and Food Security in the 21st Century, Stanford University, 7 April. Bennett, Merrill K., 1954. The World’s Food, New York: Harper. Binswanger, Hans P. and Vernon Ruttan, 1978. Induced Innovation: Technology, Institutions and Development, Baltimore, MD: Johns Hopkins University Press. Binswanger-Mkhize, Hans P., 2012. ‘India 1960–2010: Structural Change, the Rural Nonfarm Sector, and the Prospects for Agriculture’. Stanford Symposium Series on Global Food Policy and Food Security in the 21st Century, Stanford University, 10 May.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

Eicher, Carl K. and John M. Staatz (eds), 1984. Agricultural Development in the Third World, Baltimore, MD: Johns Hopkins University Press. Eicher, Carl K., and Lawrence W. Witt (eds), 1964. Agriculture and Economic Development, New York: McGraw Hill. Falcon, Walter P. and Rosamond L. Naylor (eds), 2014. Frontiers in Food Policy: Perspectives on sub-Saharan Africa, Center on Food Security and the Environment, Stanford, CA: Stanford University. Fogel, Robert W., 2004. The Escape from Hunger and Premature Death, 1700–2100: Europe, America and the Third World, Cambridge: Cambridge University Press. Gardner, Bruce L., 2002. American Agriculture in the Twentieth Century: How it Flourished and What it Cost, Cambridge, MA: Harvard University Press. Griliches, Zvi, 1957. ‘Hybrid Corn: An Exploration of the Economics of Technological Change’, Econometrica, 25, pp. 501–22. Haggblade, Steven, Peter B. R. Hazell, and Thomas Reardon (eds), 2007. Transforming the Rural Nonfarm Economy: Opportunities and Threats in the Developing World, Baltimore, MD: Johns Hopkins University Press. Hayami, Yujiro and Vernon W. Ruttan, 1971 and 1985. Agricultural Development: An International Perspective, revised and expanded edition, Baltimore, MD: Johns Hopkins University Press. Johnson, D. Gale, 1997. ‘Agriculture and the Wealth of Nations (Ely Lecture)’, American Economic Review, 87 (2), pp. 1–12. Johnston, Bruce F. and Peter Kilby, 1975. Agriculture and Structural Transformation: Economic Strategies for Late Developing Countries, Oxford: Oxford University Press. Johnston, Bruce F. and John W. Mellor, 1961. ‘The Role of Agriculture in Economic Development’, American Economic Review, 51 (4), pp. 566–93. Krueger, Anne O., Maurice Schiff, and Alberto Valdez, 1991. The Political Economy of Agricultural Pricing Policy, Baltimore, MD: Johns Hopkins University Press. Leonard, Robert, 2017. ‘Why Rural America Voted for Trump’, The New York Times, 5 January, p. A23. Lewis, W. Arthur, 1954. ‘Economic Development with Unlimited Supplies of Labor’, The Manchester School, 22, pp. 3–42. Lindert, Peter H., 1991. ‘Historical Patterns of Agricultural Policy’, in C. Peter Timmer, ed., Agriculture and the State: Growth, Employment, and Poverty in Developing Countries, Ithaca, NY: Cornell University Press. Lipton, Michael, 1977. Why Poor People Stay Poor: Urban Bias in World Development, Cambridge, MA: Harvard University Press. Mosher, Arthur T., 1966. Getting Agriculture Moving: Essentials for Development and Modernization, New York: Praeger for the Agricultural Development Council. Newbery, David M. G. and Joseph E. Stiglitz, 1981. The Theory of Commodity Price Stabilization: A Study in the Economics of Risk, Oxford: Clarendon Press. Olson, Mancur, 1965. The Logic of Collective Action, Cambridge, MA: Harvard University Press. Pritchett, Lant, 2013. ‘Folk and the Formula: Fact and Fiction in Development’. WIDER Annual Lecture 16, Helsinki: UNU-WIDER. Ravallion, Martin, Shaohua Chen, and Prem Sangraula, 2007. ‘New Evidence on the Urbanization of Global Poverty’. World Bank Policy Research Working Paper No. 4199, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



Reardon, Thomas, C. Peter Timmer, Christopher B. Barrett, and Julio A. Berdegué, 2003. ‘The Rise of Supermarkets in Africa, Asia, and Latin America’, American Journal of Agricultural Economics, 85 (5), pp. 1140–6. Reutlinger, Shlomo and Marcelo Selowsky, 1976. ‘Malnutrition and Poverty: Magnitude and Policy Options’, World Bank Occasional Paper No. 23, Baltimore, MD: Johns Hopkins University Press for the World Bank. Reynolds, Lloyd G. (ed.), 1975. Agriculture in Development Theory, New Haven, CT: Yale University Press. Schultz, Theodore W., 1964. Transforming Traditional Agriculture, New Haven, CT: Yale University Press. Sen, Amartya, 1981. Poverty and Famines, Oxford: Oxford University Press. Staatz, John M. and Carl K. Eicher. 1998. ‘Agricultural Development in Historical Perspective’, in Carl K. Eicher and John M. Staatz, eds, International Agricultural Development, 3rd edition, Baltimore, MD: The Johns Hopkins University Press, pp. 8–38. Thaler, Richard, and Shlomo Benartzi, 2004. ‘Save More Tomorrow: Using Behavioral Economics to Increase Employee Saving’, Journal of Political Economy, 112 (1), pp. S164–87. Timmer, C. Peter. 1988. ‘The Agricultural Transformation’, in Hollis Chenery and T. N. Srinivasan, eds, Handbook of Development Economics, Volume 1, Amsterdam: North Holland, pp. 275–331. Timmer, C. Peter, 1989. ‘Food Price Policy: The Rationale for Government Intervention’, Food Policy, 14 (1), pp. 17–2. Timmer, C. Peter, 1995. ‘Getting Agriculture Moving: Do Markets Provide the Right Prices?’ Food Policy, 20 (5), pp. 455–72. Timmer, C. Peter, 2002. ‘Agriculture and Economic Growth’, in Bruce Gardner and Gordon Rausser, eds, Handbook of Agricultural Economics, Volume 2, Amsterdam: North Holland, pp. 1487–546. Timmer, C. Peter, 2004. ‘The Road to Pro-Poor Growth: The Indonesian Experience in Regional Perspective’, Bulletin of Indonesian Economic Studies, 40 (2), pp. 177–207. Timmer, C. Peter, 2005a. ‘Food Security and Economic Growth: An Asian Perspective’, Heinz W. Arndt Memorial Lecture, Australian National University, Canberra (November 22), in Asian-Pacific Economic Literature, 19, pp. 1–17. Timmer, C. Peter, 2005b. ‘Agriculture and Pro-poor Growth: an Asian Perspective’. Working Paper No. 63, Washington, DC: Center for Global Development. Timmer, C. Peter, 2008. ‘Rural Changes Stimulate Rising Giants’, Science, 321 (5889), pp. 642, Review of The Dragon and the Elephant: Agricultural and Rural Reforms in China and India, by Ashok Gulati and Shenggen Fan, eds, Baltimore, MD: Johns Hopkins University Press. Timmer, C. Peter, 2009a. A World without Agriculture: The Structural Transformation in Historical Perspective, Wendt Distinguished Lecture, Washington, DC: American Enterprise Institute. Timmer, C. Peter, 2009b. ‘Do Supermarkets Change the Food Policy Agenda?’ World Development, 37 (11), pp. 1812–19. Special Issue on ‘Agrifood Industry Transformation and Small Farmers in Developing Countries’, guest edited by Thomas Reardon, Christopher B. Barrett, Julio A. Berdegué, and Johan F. M. Swinnen. Timmer, C. Peter, 2010. ‘Reflections on Food Crises Past’, Food Policy, 35(1), pp. 1–11.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



.  

Timmer, C. Peter, 2011. ‘Managing Food Price Volatility: Approaches at the Global, National and Household Levels’. Stanford Symposium Series on Global Food Security and Food Policy, 26 May, Stanford University. Timmer, C. Peter, 2012. ‘Behavioral Dimensions of Food Security’, Proceedings of the National Academy of Sciences (PNAS), Agricultural Development and Nutrition Security Special Feature, 31 (109), pp. 12315–320. Timmer, C. Peter, 2015a. Food Security and Scarcity: Why Ending Hunger Is So Hard, Philadelphia, PA: University of Pennsylvania Press and the Center for Global Development. Timmer, C. Peter, 2015b. ‘Managing Structural Transformation: A Political Economy Approach’, 18th Annual WIDER Lecture, Helsinki, Finland. Timmer, C. Peter, 2016. ‘The Role of Agriculture in ‘Catching Up: ’A Gerschenkronian Perspective’, in Martin Andersson and Tobias Axelsson, eds, Diverse Development Paths and Structural Transformation in the Escape from Poverty, Oxford: Oxford University Press, pp. 68–92. Timmer, C. Peter and Selvin Akkus, 2008. ‘The Structural Transformation as a Pathway out of Poverty: Analytics, Empirics and Politics’. Working Paper No. 150, Washington, DC: Center for Global Development. Timmer, C. Peter, Walter P. Falcon, and Scott R. Pearson, 1983. Food Policy Analysis, Baltimore, MD: Johns Hopkins University Press for the World Bank. van der Meer, Cornelis and Saburo Yamada, 1990. Japanese Agriculture: A Comparative Economic Analysis, London and New York: Routledge. Webb, Patrick and Steven A. Block, 2012. ‘Support for Agriculture during Economic Transformation: Impact on Poverty and Undernutrition’, Proceedings of the National Academy of Sciences (PNAS), Agricultural Development and Nutrition Security Special Feature, 31 (109), pp. 12309–14. Williamson, Jeffrey G., 2002. ‘Globalization and Inequality, Past and Present’, World Bank Research Observer, 12 (2), pp. 117–35. World Bank, 2007. World Development Report 2008, ‘Agriculture for Development’, Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

  A Competitiveness-Based View ......................................................................................................................

 

7.1 I

.................................................................................................................................. T research on competitiveness aims to enhance our understanding of the drivers of prosperity differences across locations, focusing especially on aspects that can inform policy to support higher levels of prosperity (Porter 1990, 2000; Delgado et al. 2012). This chapter outlines key elements of the competitiveness framework, and discusses how it relates to the idea of structural transformation. What emerges are significant similarities and complementarities that connect the competitiveness approach with the new work on structural transformation (Lin 2012, 2016) as well as other related work on new industrial policy (Rodrik 2004; Stiglitz and Lin 2013; Warwick 2013), economic complexity (Hausmann and Klinger 2007; Hausmann et al. 2013), evolutionary economic geography (Neffke et al. 2011; Boschma et al. 2017), and innovation systems (Nelson 1993; Asheim and Gertler 2004). All of these approaches share a granular and often sector-specific perspective on microeconomic structures and systems, moving beyond macroeconomic, economy-wide, or single-factor microeconomic explanations of prosperity and development. But, focusing specifically on the relationship between competitiveness and structural transformation, the discussion also reveals meaningful differences: the competitiveness literature views sectoral composition as a largely endogenous part of development, and focuses on how productive a location is in the industries it has. The structural transformation literature instead views sectoral composition as a fundamental driver of development, and focuses attention on what set of industries a location should attract to achieve prosperity growth.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

These different views on what explains productivity differences are reflected in the implications drawn for economic policy. The competitiveness literature argues for a focus on upgrading competitiveness fundamentals in a highly context-specific way, using existing clusters of related industries as important platforms for action (Ketels and Memedovic 2008; Ketels 2011). It is deeply concerned with the process of how policies can be selected and implemented, and of who needs to be involved. It sees the choice of sectors in which this happens as largely operational and determined by the given composition of an economy. The structural transformation literature suggests pushing the development of industries found in economies that are similar but already more advanced, using industry-specific improvements in competitiveness fundamentals as well as support to firms exploring opportunities in these sectors (Lin 2016). It is focused on identifying those sectors that would enable a location to achieve higher levels of prosperity and are reliant on assets and capabilities that the location is able to develop. It sees the choice and implementation of policy instruments as largely operational and within reach for most relevant countries. In many ways the conceptual differences between these two research streams seem to be related to the different contextual situations in which they have emerged: The competitiveness work has largely originated in advanced economies with well diversified economies for which better alignment of microeconomic policies to cluster-specific needs is critical and where future pathways in terms of industrial diversification are unknown. New structural economics is instead focusing on developing and emerging economies where accelerating the transition into modern sectors is a powerful driver of prosperity growth and likely directions of structural transformation seem more predictable. They are also related to different views on what factors are most critical in holding back existing efforts to support development (or have been decisive in allowing some countries to succeed). The competitiveness framework sees two types of failures: countries that follow the traditional advice of the structural reforms and/or Washington consensus literature and try to upgrade the many cross-cutting dimensions of the general business environment often overstretching their ability to implement change and failing to create distinctive advantages in individual fields of economic activity for their location. Countries that instead target specific firms or industries often fail to enhance the underlying competitiveness fundamentals of their economy, and become susceptible for interest group capture. The structural transformation literature shares the scepticism about cross-cutting policies being sufficient to enable prosperity growth. But it sees the poor track record of firm- and sector-specific policies as being more related to poor sectoral choices, not primarily to more systematic failures in the implementation of policies. Despite these differences, there has been a visible convergence in the views on policy across the two approaches: there is agreement that changes in sector composition need to be anchored in upgrading underlying competitiveness fundamentals and changes in comparative advantages given by factor endowments. What remains is a potentially productive tension between the two that can trigger new research as well as policy

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



experimentation. Competitiveness-based approaches need to deal with the question of how countries can accelerate structural change, not just get better at what they have been doing in the past. And structural transformation-based approaches have to find policy instruments to enable the emergence of more productive activities that avoid the dismal experience of past industrial policy interventions. This chapter hopes to inspire more work along these lines.

7.2 C  E D

..................................................................................................................................

7.2.1 Defining Competitiveness Competitiveness is the level of prosperity that a location can sustain for its citizens, given the conditions it offers for firms to compete successfully in local and global markets (Porter 1990, 2000; Porter et al. 2013). This productivity-based definition is anchored in the research on cross-country differences in prosperity and long-term growth rates (Hall and Jones, 1999; Lewis, 2004; IADB, 2010). The empirical literature has operationalized this definition through different quantitative measures of productivity and prosperity (Delgado et al. 2012; Aiginger 2015). The focus on productivity as the key driver of prosperity and prosperity growth is shared by the productivity-based view of competitiveness and the literature on structural transformation. Competitiveness has been a controversial concept ever since it entered the debate in the early 1990s (Porter 1990; Krugman 1994; Boltho 1995; Kitson et al. 2004; De Grauwe 2010). The controversy was mainly driven by alternative definitions of competitiveness motivated by different policy questions, not by inherent disagreements with the productivity-based view. The cost/market share-based view, its main contender, defines competitiveness as the ability to sell on international markets. The ability to export has important repercussions for macroeconomic aggregates, especially the sustainability of external balances, and as a result of this relationship it is highly relevant for international financial institutions. Both concepts are thus in their own right meaningful, and productivity and the ability to sell are clearly also empirically related. But while they are related, they capture distinct aspects of economic performance and importantly can lead to diametrically opposing policy recommendations (Ketels 2016): Policies that lead to higher productivity are also positive for growing exports. But there are policies like devaluation and lowering wages that support higher exports while not raising productivity and maybe even reducing prosperity. It is this difference in policy recommendations that has fuelled the controversy about the term competitiveness and its different definitions. For the remainder of this chapter we will focus on competitiveness as defined by the productivity-based view.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

7.2.2 Drivers of Competitiveness Competitiveness has both an output component, that is, what level of productivity and ultimately prosperity is reached, and an input component, that is, what is the set of factors that causally drives these outcomes. The latter is particularly important for policy since it defines where changes have to be made in order to achieve sustainable improvements in prosperity outcomes. The competitiveness framework looks predominantly at those input factors that are shaped by current policy choices. It argues that there is broad range of factors that can be relevant, and provides an inclusive organizing structure for diagnosing a location. This contrasts with much of the empirical growth literature that instead tries to identify a small number of factors that are, on average, the most powerful in explaining productivity differences across all locations. The competitiveness framework distinguishes in Figure 7.1 at the first level macroeconomic and microeconomic factors (Delgado et al. 2012). Macroeconomic factors are those that shape the general context in which companies operate without having a direct impact on their productivity. This includes the quality of macroeconomic policies as well as that of public institutions and their services. Microeconomic factors are those that directly drive firm level productivity either through the behaviour of firms or via the assets that they can draw on. The competitiveness literature has particularly focused on understanding the role of different aspects of microeconomic competitiveness (Porter et al. 2006). First, it is the business environment conditions that cover the assets, capabilities, and structural market Microeconomic competitiveness Quality of the business environment

Sophistication of company operations and strategy

State of cluster development

Macroeconomic competitiveness Sound monetary and fiscal policies

Human development and effective public institutions

Endowments: location, legacy, and natural resources

 . What determines competitiveness? Source: Porter et al. (2006).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



conditions which shape the level of productivity firms can achieve. Prior and subsequent research has proposed a wide range of business environment conditions as being relevant, ranging from the availability and quality of input factors (e.g. skills, knowledge, capital, and infrastructure) to the openness of markets, and the costs imposed by rules and regulations. The ‘diamond’-framework introduced by Michael Porter provided a conceptual framework to combine these different factors and emphasize the interplay between them (Porter 1990). It also significantly raised the awareness of the role of local demand conditions as a critical driver of productivity growth and innovation (Fagerberg 2011). Second, clusters capture the presence of related industries in a particular location that, through multiple linkages and externalities, influence both the productivity level firms can achieve and the strategic options they face for positioning in the markets where they operate. Long featured in the literature on regions and economic geography, the competitiveness framework emphasized their role in understanding productivity differences across locations (Porter 1990; Ketels 2011). Empirical research has revealed a systematic relationship between cluster presence and economic performance (e.g. Audretsch and Feldman 1996; Delgado et al. 2010, 2014). It also showed that while clusters differ in depth and functions, they do exist in economies at very different stages of economic development (Zeng 2008; Long and Zhang 2012). And third, firm sophistication directly addresses the way firms compete, organize, and operate. While business environment conditions and the presence of clusters set the context, firms make many further internal choices that ultimately set the level of productivity they achieve. This idea has recently found strong support in the empirical literature on management quality across locations (Bloom et al. 2016). It showed that large differences in management quality exist and matter for prosperity, even after controlling for other factors. The competitiveness literature stands in contrast to the work on ‘deep roots’ (Spolaore and Wacziarg 2012) that explains prosperity differences across locations with different types of long-seated legacies. The role of institutions, in particular, their geographic location, and the connection between the two have been debated intensely (Acemoglu and Robinson 2012; McCord and Sachs 2013). Natural resources have been analysed in their dual role as a source of wealth and a ‘curse’ undermining prosperity growth (Sachs and Warner 2001; Frankel 2010). The ‘deep roots’ literature views the microeconomic factors that are the focus of the competitiveness research as largely endogenous to locations’ legacy. The empirical tests of the competitiveness framework finds conversely that, even controlling for ‘deep roots’, macro- and especially microeconomic competitiveness matter independently (Delgado et al. 2012).

7.2.3 Competitiveness and Economic Development In the competitiveness literature economic development is characterized as ‘a process of successive upgrading, in which a nation’s business environment evolves to support

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Factor-driven economy

Low cost inputs • Monetary and fiscal, political, and legal stability • Improving basic human capital • Efficient basic infrastructure • Lowering the regulatory costs of doing business

Investmentdriven economy

Productivity

Innovationdriven economy

Unique value

• Increasing local rivalry

• Advanced skills

• Market opening

• Scientific and technological institutions

• Advanced infrastructure • Incentives and rules encouraging productivity

• Incentives and rules encouraging innovation

• Cluster formation and activation

• Cluster upgrading

 . Stages of development Source: Porter (1990).

and encourage increasingly sophisticated and productive ways of competing’ (Porter et al. 2006: 56). This process can be described by characteristic stages in Figure 7.2 (Porter 1990): at the factor-driven stage, economies compete on low cost by providing access to cheap factors of production, particularly labour. At the investment-driven stage, their advantage shifts to high productivity driven by access to human and physical capital and other conditions that drive efficiency. At the innovation-driven stage, the unique value from the new products, services, and business models dominates, driven by further enhancement in competitiveness factors that encourage innovation and entrepreneurship. As economies move through these stages, there are changes in the relative importance of different aspects of competitiveness and thus of the policies that affects them. Quantitative work on competitiveness has found microeconomic factors to matter more at higher stages of development, much as had been predicted (Delgado et al. 2012). This thinking has found its reflection in measurements of competitiveness such as the Global Competitiveness Report (Sala-i-Martin et al. 2015). The stages framework suggests that locations face particularly complex upgrading challenges when they move from one ‘stage’ to another. In such situations, economies require systemic and coordinated changes across a broad range of policies, in some cases including steps that undermine previously important strengths. The middleincome trap, discussed elsewhere in this Handbook, can be understood as describing situations where countries fail to achieve such a broad-based shift in policies. Vietnam is an example of a country approaching the first transition from a factor- to an investment-driven stage (Cung et al. 2010). So far, Vietnam could focus on opening up to and enabling global investment and trade, which has driven significant growth and structural change towards manufacturing. Now it will have to build a broad range of capabilities and business environment qualities to enhance productivity in the industries that have emerged. Singapore is an example of an economy with strong

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



aspirations to be innovation-driven (Kang and Phang 2005). It has successfully added significantly more science and research capacity. More of a challenge is that the country also has to move beyond the strict rules-driven approach in firms and government policy that drove growth in the investment-driven stage. In addition, it has to create more local entrepreneurship in an economy traditionally driven by large multinationals and government-linked companies. While the competitiveness framework provides a structure within which to describe and analyse economic development, it does not conceptualize a dynamic model of selfsustaining development. The competitiveness upgrading that underpins development is instead largely seen as the result of specific policy choices and actions.

7.2.4 Policies for Competitiveness Upgrading Competitiveness has from the start been conceived as a framework not only to allow an understanding of outcomes but also to inform policy action (Porter 1990). But it is arguably the case that the academic competitiveness literature has focused more on positive aspects than on developing a policy framework to guide practitioners (Gordon 2011). The debate on competitiveness policy reveals a strong focus on clarifying the goals and motivations for economic policy, and contributing to a more effective design and implementation of programmes.

7.2.4.1 Setting the Right Goal for Policy The competitiveness framework suggests identifying productivity upgrading as the overarching goal for competitiveness policy. This focus on productivity stands in contrasts to other objectives that dominate practical policy debates, like jobs, investment, or exports. In the competitiveness framework these categories are seen as important symptoms of and intermediate steps towards higher productivity, but not as useful targets for government action. The problem is that, in the same way as for exports in the market-share driven definition of competitiveness discussed in Section 7.2.1, there are policies that impact such intermediate outcomes by raising the private profitability of activities while potentially decreasing productivity and prosperity. While there might be other rationales for some of these policies, they cannot be motivated by the objective of competitiveness upgrading.

7.2.4.2 The Case for Government Policy The general motivation for government action is in the competitiveness framework viewed in traditional terms of market failure: government should take action where there are market failures that policy is able to effectively and efficiently address. There is no general presumption on the widespread existence of economies of scale or other externalities that could motivate ‘strategic’ policies (Brander 1995). And even where these might exist, there is generally a concern about the ability of government to successfully pursue such polices.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

There is, however, a strong acknowledgement of local externalities affecting economic geography and in particular the emergence of clusters (Porter 1990; Ketels 2011). In a significant number of cases policy practitioners have interpreted this view as a reason to launch policies trying to create clusters. Such a policy has drawn significant criticism in the academic literature (e.g. Duranton 2011). Conversely, the competitiveness framework sees clusters as key elements of a location’s economic structure that emerge naturally in market processes. While government policies have a significant influence on cluster evolution, attempts to create clusters are subject to the well-known pitfalls that affect traditional industrial policies. Governments should instead focus on providing information about cluster presence, convening cluster groups, and investing in cluster-specific public goods (Porter 2007; Mills et al. 2008; Ketels 2013); all activities well motivated by standard market failure arguments.

7.2.4.3 The Design of Systemic Policies: Strategic Selection and Integration of Policy Actions Traditional policy analysis tends to ask whether a specific intervention, like more funding for upgrading workforce skills, is in general welfare enhancing, and which specific instrument is able to achieve this goal in the most efficient way. The competitiveness framework sees policy makers facing an important prior question: what area(s) of competitiveness should our location focus on now? This question is relevant because there are many dimensions of competitiveness that ultimately need to be improved, but there is only limited capacity to do everything at once. This is an even more pressing concern in developing or emerging countries with less robust institutions. The competitiveness framework argues that the answer to this question needs to be location-specific (Barca et al. 2012): the benefits of improving one dimension of competitiveness, say the level of skills in the workforce, depend on the quality of many other aspects of competitiveness in this location, such as the available infrastructure and the nature of market competition (Goni and Maloney 2014). Location-specific diagnostics are needed to drive the prioritization of policies (Hausmann et al. 2005; Rodrik 2007). The competitiveness framework further suggests that the impact of microeconomic policies is not only location- but is also often cluster-specific: a decision to upgrade workforce skills does require choices about the type of skills to provide, and the nature and value of these skills is both cluster-specific and dependent on the strength of the relevant cluster in that location. Again, diagnostics are critical for identifying areas in which specific cluster-specific actions are a priority (Shriram et al. 2013). The effectiveness of many policies and distinct efforts to upgrade competitiveness can be enhanced if they are organized around an existing cluster (Rodriguez-Clare 2007; Ketels and Memedovic 2008). Improvements in competitiveness depend on how individual policies, both crosscutting and cluster-specific, complement each other; this is another implication of the linkages between the different dimensions of microeconomic competitiveness. Effective growth policies, the competitiveness framework argues, are both the result of selecting

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



appropriate policy actions and of integrating them in a mutually reinforcing strategy. What such strategies should look like is the subject of an emerging literature on locational strategy. One question is whether they should focus on reducing relative weaknesses or creating unique advantages (Hausmann et al. 2005; Ketels 2015), another is whether they should concentrate on identifying new key transformative actions or on providing a broader framework for all relevant government policies (Foray 2015).

7.2.4.4 Implementing Policies for Upgrading Competitiveness The policy-oriented work on competitiveness has put a significant emphasis on how to effectively implement policies. Two dimensions have been identified as particularly relevant: the role of different levels of geography, and the way the public and the private sector collaborate. While the competitiveness literature initially focused on nations as the unit of analysis, there has been increasing recognition of the important role of choices and actions at other levels of geography. Subnational regions in particular have come into focus: Microeconomic conditions, sectorial composition, and economic performance differ not only across countries but also within countries. This observation has fuelled research on understanding and measuring regional competitiveness (Kitson et al. 2004; Dijkstra et al. 2011). Regional governments have a critical and unique role to play in integrating policy instruments and aligning them with the specific needs of their economy (Ketels 2017). More broadly, the allocation and coordination of policy choices across different levels of government has a critical influence on many policies that drive competitiveness. While policy discussions naturally focus on the role of government, work on competitiveness has led to a strong focus on the role of public–private dialogue to achieve competitiveness upgrading (Fernandez-Arias et al. 2016). Platforms for public– private collaboration are needed to inform and enable collective action when both knowledge and the ability to influence competitiveness fundamentals are dispersed across (many) different public and private entities (Porter and Emmons 2003). And if a significant part of these dynamics is cluster-specific in nature, cluster initiatives become key backbone institutions for upgrading competitiveness (Sölvell et al. 2003).

7.3 C  S T

..................................................................................................................................

7.3.1 Key elements of Structural Transformation The structural transformation literature builds on the long-standing empirical observation that the composition of economies differs systematically by stage of development (Herrendorf et al. 2014). In its initial form, composition was mainly understood as the relative size of the broad sectors of agriculture, industry, and

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

services. The literature proposed specific models of development that could explain structural transformation as an endogenous process in response to factor accumulation, increasing wealth, and sector-specific properties of demand and production functions. Subsequent work has emphasized that development is associated not only with a shift into different activities but also with diversification into a broader set of activities (Imbs and Wacziarg 2003; Hausmann and Klinger 2007). It has also linked structural transformation to changes in economic geography, in particular urbanization (Michaels et al. 2012) and inequality (Timmer and Akkus 2008). The new structural economics (Lin 2012, 2016; see also the other chapters in this Handbook) develops a novel set of recommendations for how economies can speed up the process of structural transformation. It argues that the market is the best mechanism for factor allocation but faces systematic failures in the exploration of new sectors. As policy makers consider how to overcome these failures without falling prey to the traditional pitfalls of industrial policy, two key suggestions from this new approach stand out: • First, as countries consider the direction of structural change to pursue, they can learn from the experience of peers with similar initial factor endowments that have already achieved higher levels of prosperity. The industries that they have seen emerge over time are the prime candidates to emerge in the economies that follow in their path. • Second, as countries look at the tools they can deploy to achieve structural change, they need to focus on encouraging the exploration and then exploitation of market opportunities, and on unblocking fundamental barriers in sector-specific competitiveness. Location-specific policies with these features can help if economywide changes are too hard to achieve.

7.3.2 Contrasting the New Structural Economics with the Competitiveness Framework New structural economics and the competitiveness framework both argue for a granular perspective in understanding development that moves beyond an analysis of macroeconomic aggregates. This section aims to develop in more detail where the two approaches differ and where they agree in their conceptualization of industrial structure, views on the drivers of prosperity differences across locations, the process of economic development, and the implications for economic policy.

7.3.2.1 Analytical Categories for Measuring Economic Structure The new structural economics and the competitiveness framework both take a granular look at industrial structure. But the analytical categories they use differ: structural economics follows the traditional identification of sectors defined by broad features of their respective production functions. The competitiveness framework instead

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



Traded vs. Local Share of the US Economy EMPLOYMENT

36%

64%

INCOME

51%

49%

PATENTS

91%

9%

Traded Industries • ‘Spiky’ across space; 2/3s of all traded industry employment is concentrated in strong clusters • Serve national and global markets • Exposed to competition from other regions and nations • Critical for prosperity through higher wages, productivity, and innovation; growth potential set by the global market

Local Industries • Present everywhere, proportional to overall size • Serve exclusively the local market • Little exposure to crossregional competition • Important for jobs but have lower wages; growth potential limited by size of the local market

 . Traded and local industries: mapping clusters Source: Delgado et al. (2010, 2014, 2016).

differentiates industries by their geographic footprint (‘traded’ = concentrated in a few places and serving markets beyond their home base vs ‘local’ = dispersed and present in all places, serving local markets as shown in Figure 7.3) and then organizes groups of related traded industries into specific cluster categories (Delgado et al. 2016). These categories and the focus on cross-industry linkages has similarities with work on economic complexity (Hausmann et al. 2013). Despite the differences in conceptual underpinnings and research method, there are clear similarities between the categories used in the structural transformation and the competitiveness literature: traded clusters tend to be dominated by industry, and local sectors by services. However, this similarity is getting weaker as advanced services become an increasingly important part of the traded economy.

7.3.2.2 The Relationship between Economic Structure and Prosperity In their analysis of prosperity differences across locations, structural economics and the competitiveness framework observe the same regularities: locations at different stages of development differ significantly in their economic composition and breadth. But they draw different implications from this observation. For the structural transformation literature—both old and new—sectoral composition is a critical driver of prosperity: What you do, that is, what the sectoral composition of your economy looks like, drives how prosperous you are. In the competitiveness framework sector, composition plays a different role: it is largely endogenous to underlying competitiveness drivers and thus viewed as a symptom, not as a driver of competitiveness. Productivity itself is not only determined by the sectors present in a location but also by the relative performance that has been achieved in a location for a given sector. The evidence in the literature supporting this view has largely been drawn from advanced economies: In the USA and Europe there are

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

significant wage and productivity differences across locations within sectors, clusters, and industries. Across US regions one-third of prosperity differences across locations are found to be related to differences in sectoral composition while two-thirds are explained by location-specific factors that influence the performance within sectors (Porter 2003). European data reveal regional specialization in high wage clusters (measured by their European average wage) to be statistically explained by business environment quality (Ketels and Protsiv 2013). There are also some findings involving emerging economies, showing that there are significant performance differences (static and over time) within industries and across locations that are systematically related to qualities of the business environment (Valencia Caicedo and Maloney 2014). However, further research covering structurally more heterogeneous economies will be needed to test the nature of these relationships for a broader sample of countries.

7.3.2.3 Patterns of Economic Development In their description of how economic structure changes in the course of economic development, structural economics and the competitiveness framework have a largely overlapping perspective: Development is characterized by a process of related diversification into more advanced economic activities, much in line with related work on the evolution of regional economies and economic complexity (Neffke et al. 2011; Hausmann et al. 2013). At a more detailed level, however, there are some differences, largely based on diverging approaches to conceptualizing and identifying the notion of relatedness. Structural economics identifies relatedness based on the historical experiences of fundamentally comparable countries that have already achieved higher levels of development (Lin 2016). The competitiveness-related cluster literature has instead identified relatedness through looking at patterns of co-location, input–output linkages, and overlaps in skill use at a given point in time (Delgado et al. 2016). These two approaches can lead to diverging results, especially in cases like Korea where countries have successfully ‘jumped’ to entirely new industries not aligned with existing factor endowments (Studwell 2013). For new structural economics this presents a path that can and should inspire others, even though these countries arguably also went significantly beyond their ‘latent’ competitive advantages (Rodrik 2011). For the competitiveness framework such cases are more of an exception to the rule—risky industrial policy worked because it was based on an unusually coherent and well implemented approach towards upgrading industry- or cluster-specific business environment conditions in traded industries, providing effective incentives for firms to raise productivity. Another difference between the different approaches relates to the nature of the development path as shown in Figure 7.4. New structural economics views development as a continuum of structures that change and upgrade over time. Importantly, it views this continuum as generic and stable, that is, all countries will follow essentially the same path, especially if they share similar starting conditions. There are similarities to the economic complexity work that sees commonalities among prosperous economies but more variability among less advanced economies (Hausmann et al. 2013).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  -  Structural economics

• All economies are on a common development path • Direction of structural transformation is known • Local data puts each country somewhere on the common path



Economic complexity

Competitiveness

• All economies are different but prosperous economies are structurally similar • End point of successful structural transformation is known • Local data identifies starting point and thus pathway to prosperity

• All economies are different, despite some similarities by stage of development • Local data identifies opportunity set for their structural transformation • Many different pathways depending on context and starting point

 . How does economic structure change? Hypotheses on the nature of development pathways

The empirical similarities in the historical development paths across countries have been a key motivation for these views. Some new data are suggesting, however, that the pathways of the past might not characterize current trends or opportunities (te Velde 2011; Gollin et al. 2013; Rodrik 2015; Newman et al. 2016; Rodrik et al. 2017). And new conceptual work suggests that related diversification might be an important but not the only path to diversification (Boschma et al. 2017). The competitiveness framework proposes a typical pathway in its stages model as well but emphasizes much more that individual countries will significantly diverge from this ‘average’ path (Porter et al. 2006: 57). Each location is seen as facing different opportunities to define its unique value proposition and develop competitive advantages for the specific types of activities where it aims to compete (Ketels 2015). This perspective reduces the value of benchmarking and copying the path of other countries, and raises the need for a country to develop its own distinct choices.

7.3.2.4 Implications for Economic Policy Finally, the question policy makers will ask is how the different conceptual approaches matter in terms of policy recommendations. Here the work on new structural economics (Lin 2012, 2016; for a discussion of related tools, see Altenburg et al. 2016; McMillan et al. 2017 discuss a broader set of policies to drive structural transformation) has provided significantly more clarity on how structural transformation is to be achieved. The policy approach emerging is to a significant degree compatible with the competitiveness- and cluster-based approach but some differences can remain that have the potential to result in divergent choices in practice.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

First, both approaches see the need to conduct diagnostics and act at a contextspecific, microeconomic level. General patterns of factor endowments are important, but these categories are not granular enough to be of much use in analysing the specific economic situation of locations or identifying specific policy actions. Policy makers need to understand the composition of their economies, and they need to understand the industry- and cluster-specific barriers and enablers to growth. Macroeconomic conditions matter but ultimately microeconomic upgrading is needed to achieve development and structural transformation. Second, both approaches see markets and policy playing complementary roles. They see markets and rivalry as critical for achieving efficient factor allocation and company sophistication. But they also see a clear need for policy and collective action in shaping business environment conditions that can enable market competition to occur at higher levels of productivity. Third, there has been convergence on the type of instruments that are being proposed. Both approaches are sceptical about interventions into the market process, and instead argue for more traditional investment in public goods and support for activities that generate positive externalities, including the exploration of new market opportunities. Despite these important agreements, there are also meaningful differences in thinking that can easily lead to significant divergence in policy practice. In fact, the differences are now much less in the ‘what to do’ but remain in the ‘how to implement’. But the ‘how’ is much more than an operational detail; it often has fundamental repercussions for what the ‘what’ ends up being. First and most fundamentally, the different views about composition being a key driver versus a symptom of competitiveness can easily lead to opposing choices on policy instruments. The new structural economics focuses policy practitioners on the question of how to attract and nurture the next line of industries; the tools needed in terms of upgrading industry-specific competitiveness fundamentals come second. The competitiveness approach focuses instead on upgrading these fundamentals, but argues for cluster-specific steps in doing so. The difference between the two perspectives, depicted in Figure 7.5, is less dramatic than between the traditional industrial policy (create an industry; the competitiveness upgrading will follow automatically) and policies to upgrade framework conditions (enhance general business environment conditions; the upgrading across and within industries will follow automatically). But a difference remains, and it is one that can turn out to be quite significant in practice. In particular, policy practitioners are faced with the challenge of how to enable new industries that are ‘next’ on their development path. What if just providing business environment conditions aligned with the target industry is not enough? How much should firm-specific incentives be used, even if just temporarily? Here the two approaches will tend to lead to different answers in practice, even if there has been a convergence of views in principle (so the related discussion in Ledermann and Maloney 2012). Second, the two approaches take a different view on the balance between crosscutting and location- and cluster-specific upgrading of the economy. Structural

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  -  Many, high productivity sectors



Advanced economies Structural transformation policy Industrial policy

Economic structure

Emerging economies Cluster-based competitiveness Policy

Developing economies

Few, low productivity sectors Low quality

Framework conditions policy

Competitiveness fundamentals

High quality

 . Comparison of development approaches

transformation focuses strongly on what is relevant for specific industries, and suggests the use of location specific approaches, for example the use of industrial zones (Lin 2016). The competitiveness approach, too, advocates cluster-specific approaches but focuses more on their role as instruments to enable the upgrading of business environment conditions that also benefit firms in the wider economy. At earlier stages of economic development in particular, it sees indications that many of the key competitiveness challenges are cross-cutting, related to institutional factors and more general rules of the game affecting the functioning of markets (Delgado et al. 2012). This issue has significant practical relevance as, for example, industrial parks have a mixed track record in spearheading such broader changes in developing and emerging economies, even if they were internally successful (Zeng 2010; Farole and Akinci 2011). Third, there are differences in the way governments at different levels of geography are seen. The structural transformation literature is fundamentally focused on the nature of the national economy and the policies set at this level. Location-specific interventions such as special economic zones and industrial parks are being discussed but remain a tool of national policy makers. The competitiveness literature has instead focused strongly on the complementary roles of different levels of government, and emphasized in particular the role of subnational regions in both analysis and action (Ketels 2017). Fourth, the structural transformation approach often ends up giving government a central role, making top down decisions about the direction of structural change with all the political economy complexities this entails. The cluster approach instead tends to lead towards much more business-led activities, where groups of firms collaborate

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

and engage with government based on the market conditions they experience. While the specific context matters, cluster efforts are an effective tool for public–private dialogue on competitiveness upgrading (Herzberg and Wright 2006; Fernandez-Arias et al. 2016). Finally, the new structural economics literature argues strongly that the experience of similar countries should be examined and their path followed. The competitiveness framework is sceptical about this advice, seeing the danger that such an approach might fail to create competitive advantage and put many developing and emerging economies on a path to head-on competition. Instead, it argues for more thinking about strategic choice and a location’s unique value proposition.

7.4 C

.................................................................................................................................. The discussion of the competitiveness framework and its relationship to the new literature on structural transformation has revealed differences but also a significant degree of conceptual affinity. What is emerging can be described as an integrated view that captures both the role of competitiveness fundamentals and industrial composition in driving productivity and prosperity outcomes; see Figure 7.6. Endowments from natural resources to geographic location and institutional legacies provide a unique foundation for any economy. The type of competitiveness conditions created on this basis, however, is wide open to the actions of policy makers and many public and private entities. Together, these fundamentals then give rise to Competitiveness fundamentals

Economic structure

What you have created How well you do it What you do

How prosperous you are

What you have inherited

 . What drives prosperity? Relationship between competitiveness and structure

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



economic activities at a certain level of performance in specific industries. Prosperity is ultimately created through these activities. Activities themselves create through learning new capabilities that enhance the existing competitiveness conditions and drive the presence of new industries. The level and sustainability of prosperity growth depends on the upgrading happening in terms of competitiveness fundamentals as well as the structural changes this triggers in terms of the composition of sectors and the sophistication of activities within them (Rodrik 2014). Growth remains episodic if there are no sustained improvements in competitiveness fundamentals, with the possibility of one-off growth spurts related to structural change. Growth remains low if improvements in competitiveness fundamentals do not translate into structural transformation across and within sectors. The key challenge that remains is how to enable structural change in situations where the market process does not seem to be working fast enough. This is an issue for developing and emerging economies, but also for many regions within advanced economies. New structural economics proposes an approach to tackle this challenge; it suggests how to identify an appropriate direction for change and proposes some principles for how to get there. Whether it will work in practice remains to be seen. Structural change and the nurturing of new sectors through cluster-based approaches has been tried. But the evidence from Europe is sobering: cluster efforts have had an impact in upgrading existing clusters but have a much less impressive track record in triggering transitions into new fields (Ketels and Protsiv 2013). This observation has been a main concern underpinning the EU’s regional policy ‘smart specialization’ approach to systematically identify interventions that can drive transformation (Foray 2015). It suggests moving towards a mixed approach that combines upgrading in existing clusters with the systematic exploration of opportunities in related fields and efforts to encourage more generally entrepreneurship and innovation as shown in Figure 7.7. Again, the jury is still out on how this will work.

Upgrade existing clusters

Encourage exploration of related clusters with high productivity potential

Encourage entrepreneurship economy-wide

 . Cluster-based structural transformation

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

R Acemoglu, Daron and James A. Robinson, 2012. Why Nations Fail: The Origins of Power, Prosperity, and Poverty, New York: Random House. Aiginger, K., 2015. ‘Competitiveness: from a Misleading Concept to a Strategy Supporting Beyond GDP Goals’, Competitiveness Review, Special Issue ‘25 Years after the Competitiveness Advantage of Nations’, December. Altenburg, Tilman, Maria Kleinz, and Wilfried Lütkenhorst, 2016. ‘Directing Structural Change: from Tools to Policy’, Discussion Paper No. 24, Bonn: German Development Institute. Asheim, Bjorn and Mark Gertler, 2004. ‘Understanding Regional Innovation Systems’, in Jan Fagerberg, David Mowery, and Richard Nelson, eds, Handbook of Innovation, Oxford: Oxford University Press. Audretsch, David and Maryann Feldman, 1996. ‘R&D Spillovers and the Geography of Innovation and Production’, American Economic Review, 86 (3), pp. 630–40. Barca, Fabrizio, Philip McCann, and Andrés Rodríguez-Pose, 2012. ‘The Case For Regional Development Intervention: Place-based Versus Place-neutral Approaches’, Journal of Regional Science, 52, pp. 134–52. Bloom, Nicholas, Sadun, Raffaella, and John van Reenen, 2016. Management as a Technology?, Harvard Business School Strategy Unit Working Paper No. 16–133, Stanford, CA: Stanford University Graduate School of Business Research Paper No. 16–27. Boltho, A., 1995. ‘The Assessment: International Competitiveness’, Oxford Review of Economic Policy, 12 (3), pp. 1–16. Boschma, Ron, Lars Coenen, Koen Frenken, and Bernhard Truffer, 2017. ‘Towards a Theory of Regional Diversification: Combining Insights from Evolutionary Economic Geography and Transition Studies’, Regional Studies, 51 (1), pp. 31–45. Brander, James E., 1995. Strategic Trade Policy, NBER Working Paper No. 5020, NBER: Cambridge, MA. Cung, Nguyen Dinh, Christian Ketels, Nguyen Thi Tue Anh, and Do Hong Hanh, 2010. Vietnam Competitiveness Report 2010, Singapore/Hanoi: ACI/CIEM. De Grauwe, Paul, 2010. ‘Introduction’, in P. De Grauwe, ed., Dimensions of Competitiveness, CESifo Seminar Series, Cambridge, MA: MIT Press. Delgado, Mercedes, Michael E. Porter, and Scott Stern, 2010. ‘Clusters and Entrepreneurship’, Journal of Economic Geography, 10 (4), pp. 495–518. Delgado Mercedes, Christian Ketels, Michael E. Porter, and Scott Stern, 2012. The Determinants of National Competitiveness, NBER Working Paper No. 18249, National Bureau of Economic Research, Cambridge, MA. Delgado, Mercedes, Michael E. Porter, and Scott Stern, 2014. ‘Clusters, Convergence and Economic Performance’, Research Policy, 43, pp. 1785–99. Delgado, Mercedes, Michael E. Porter, and Scott Stern, 2016. ‘Defining Clusters of Related Industries’, Journal of Economic Geography, 16 (1), pp. 1–38. Dijkstra, Lewis, Paola Annoni, and Kornelia Kozovska, 2011. ‘A New Regional Competitiveness Index: Theory, Methods and Findings’, Working Paper No. 2/2011, Brussels: European Commission. Duranton, Gilles, 2011. ‘California Dreamin’: The Feeble Case for Cluster Policies’, Review of Economic Analysis, 3 (1), pp. 3–45. Fagerberg, Jan, 2011. ‘Domestic demand, learning, and the Competitive Advantage of Nations: A Review of the Empirical Evidence’, in R. Huggins and H. Izushi, eds, Competition, Competitive Advantage, and Clusters, Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



Farole, Thomas and Ghokan Akinci (eds), 2011. Special Economic Zones: Progress, Emerging Challenges, and Future Direction, Washington, DC: World Bank. Fernandez-Arias, Eduardo, Charles Sabel, Ernesto Stein, and Alberto Trejos, 2016. Two to Tango: Publi—Private Collaboration for Productive Development Policies, Washington, DC: Inter-American Development Bank. Foray, Dominique, 2015. Smart Specialization: Opportunities and Challenges for Regional Innovation Policy, Abingdon: Routledge Publishing. Frankel, Jeffrey, 2010. The Natural Resource Curse: A Survey, RWP10-05, Cambridge, MA: Harvard Kennedy School of Government. Gollin, Douglas, Rémi Jedwab, and Dietrich Vollrath, 2013. ‘Urbanization with and without Structural Transformation’, Mimeo. Goni, Edwin and William F. Maloney, 2014. ‘Why Don’t Poor Countries do R&D?’, World Bank Policy Research Paper No. 6811, Washington, DC: World Bank. Gordon, Robert M., 2011. ‘National Economic Development and The Competitive Advantage of Nations’, in R. Huggins and H. Izushi, eds, Competition, Competitive Advantage, and Clusters, Oxford: Oxford University Press. Hall, R. E. and C. I. Jones, 1999. ‘Why Do Some Countries Produce So Much More Output per Worker than Others?’, Quarterly Journal of Economics, 114 (1), pp. 83–116. Hausmann, Ricardo and B. Klinger, 2007. The Structure of the Product Space and the Evolution of Comparative Advantage, CID Working Paper No. 146, Cambridge, MA: Harvard Kennedy School of Government. Hausmann, Ricardo, Dani Rodrik, and Andres Velaso, 2005. Growth Diagnostics, Harvard Kennedy School Working Paper, Cambridge, MA: Harvard Kennedy School of Government. Hausmann, Ricardo, César A. Hidalgo, Sebastian Bustos, Michele Coscia, Alexander Simoes, and Muhammed A. Yildirim, 2013. The Atlas of Economic Complexity: Mapping Paths to Prosperity, Cambridge, MA: Harvard University/MIT. Herrendorf, Berthold, Richard Rogerson, and Ákos Valentinyi, 2014. ‘Growth and Structural Transformation’, in Philippe Aghion and Steven N. Durlauf, eds, Handbook of Economic Growth, Volume 2B, Amsterdam: Elsevier Publisher, pp. 855–941. Herzberg, Benjamin and Andrew Wright, 2006. The Public–Private Dialogue Handbook: A Toolkit for Business Environment Reformers, Washington, DC: World Bank. IADB, 2010. The Age of Productivity: Transforming Economies from the Bottom up, Washington, DC: Inter-American Development Bank. Imbs, Jean and Romain Wacziarg, 2003. ‘Stages of Diversification’, American Economic Review, 93 (1), pp. 63–86. Kang, Kim-Song and Sock-Yong Phang, 2005. ‘From Efficiency-Driven to Innovation-Driven Economic Growth: Perspectives from Singapore’, World Bank Policy Research Working Paper No. 3569, Washington, DC: World Bank. Ketels, Christian, 2011. ‘Competitiveness and Clusters: Porter’s contribution’, in R. Huggins and H. Izushi, eds, Competition, Competitive Advantage, and Clusters, Oxford: Oxford University Press. Ketels, Christian, 2013. ‘Cluster Policy: A Guide to the State of Debate’, in Peter Meusburger, Johannes Glückler, and Edgar Wunder, eds, Knowledge and Economy, Berlin: Springer Publishing, pp. 249–69. Ketels, Christian, 2015. ‘What is Regional Strategy? Lessons from Business Strategy’, in: Valdaliso and Wilson, eds, Strategies for Shaping Territorial Competitiveness, Abingdon: Routledge Publishers.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Ketels, Christian, 2016. Competitiveness Frameworks Review, Dublin: Irish National Competitiveness Council. Ketels, Christian, 2017. ‘Upgrading Competitiveness: What Role for Regional Government?’, in Robert Huggins and Piers Thompson, eds, Handbook of Regions and Competitiveness, Cheltenham: Edward Elgar Publishing. Ketels, Christian and Olga Memedovic, 2008. ‘From Clusters to Cluster-based Economic Development’, International Journal for Technological Learning, Innovation and Development, 1, (3), pp. 375–92. Ketels, Christian and Sergiy Protsiv, 2013. ‘Clusters and the New Growth Path for Europe’, WWWForEurope Working Paper No. 14, Vienna: WIFO. Kitson, Michael, Ron Martin, and Peter Tyler, 2004. ‘Regional Competitiveness: An Elusive yet Key Concept?’ Regional Studies, 38 (9), pp. 991–9. Krugman, Paul, 1994. ‘Competitiveness: A Dangerous Obsession’, Foreign Affairs, 73 (2), pp. 28–44. Lederman, Daniel and William F. Maloney, 2012. Does What you Export Matter? Washington, DC: World Bank Group. Lewis, William, 2004. The Power of Productivity, Chicago: Chicago University Press. Lin, Justin Yifu, 2012. New Structural Economics: A Framework for Rethinking Development, Washington, DC: World Bank. Lin, Justin Yifu, 2016. The Quest for Prosperity: How Developing Economies Can Take Off, Princeton, NJ: Princeton University Press. Long, Cheryl and Xiabo Zhang, 2012. ‘Patterns of China’s Industrialization: Concentration, Specialization, and Clustering’, China Economic Review, 23, pp. 593–612. McCord, Gordon C. and Jeffrey Sachs, 2013. ‘Development, Structure, and Transformation: Some Evidence on Comparative Economic Growth’, NBER Working Paper No. 19512, Cambridge, MA: NBER. McMillan, Margaret, John Page, David Booth, and Dirk Willem te Velde, 2017. ‘Supporting Economic Transformation: An Approach Paper’, London: Overseas Development Institute. Michaels, Guy, Ferdinand Rauch, and Stephen J. Redding, 2012. ‘Urbanization and Structural Transformation’, The Quarterly Journal of Economics, 127 (2), pp. 535–86. Mills, Karen G., Elisabeth B. Reynolds, and Andrew Reamer, 2008. Clusters and Competitiveness: A New Federal Role for Stimulating Regional Economies, Brookings Metropolitan Policy Program, Washington, DC: Brookings. Neffke, Frank, Martin Henning, and Ron Boschma, 2011. ‘How Do Regions Diversify over Time? Industry Relatedness and the Development of New Growth Paths in Regions’, Economic Geography, 87, pp. 237–65. Nelson, Richard R., 1993. National Innovation Systems: A Comparative Analysis, Oxford: Oxford University Press. Newman, Carol, John Page, John Rand, Abebe Shimeles, Måns Söderbom, and Finn Tarp (eds), 2016. Manufacturing Transformation, Oxford: Oxford University Press. Porter, Michael E., 1990. The Competitive Advantage of Nations, New York: The Free Press. Porter, Michael E., 2000. On Competition, Boston, MA: Harvard Business School Press. Porter, Michael E., 2003. ‘The Economic Performance of Regions’, Regional Studies, 37 (6/7), pp. 549–78. Porter, Michael E., 2007. ‘Clusters and Economic Policy: Aligning Public Policy with the New Economics of Competition’, ISC Working Paper, Boston, MA: Harvard Business School.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 :  - 



Porter, Michael E. and Willis Emmons, 2003. Institutions for Collaboration: Overview, Harvard Business School Note 703–436, Boston, MA: Harvard Business School. Porter, Michael E., Christian Ketels, and Mercedes Delgado-Garcia, 2006. ‘The Microeconomic Foundations of Prosperity: Findings from the Business Competitiveness Index’, in Augusto Lopez-Claros, Michael E. Porter, Xavier Sala-i-Martin, and Klaus Schwab, eds, Global Competitiveness Report 2006–2007, Basingstoke: Palgrave Macmillan. Porter, Michael E., Jan Rivkin, and Rosabeth Kanter, 2013. Competitiveness at a Crossroads, Harvard Business School Survey on U.S. Competitiveness, Boston, MA: Harvard Business School. Rodriguez-Clare, Andres, 2007. ‘Clusters and Comparative Advantage: Implications for Industrial Policy’, Journal of Development Economics, 82, pp. 43–57. Rodrik, Dani, 2004. ‘Industrial Policy for the 21st Century’, KSG Faculty Research Working Paper Series, RWP04-047, Cambridge, MA: Harvard Kennedy School of Government. Rodrik, Dani, 2007. One Economics, Many Recipes: Globalization, Institutions, and Economic Growth, Princeton, NJ:Princeton University Press. Rodrik, Dani, 2011. ‘Comments on “New Structural Economics” by Justin Yifu Lin’, The World Bank Research Observer, 26 (2), pp. 227–9. Rodrik, Dani, 2014. ‘The Past, Present, and Future of Economic Growth’, In F. Allen et al., eds, Towards a Better Global Economy: Policy Implications for Citizens Worldwide in the 21st Century, Oxford: Oxford University Press. Rodrik, Dani, 2015. ‘Premature Deindustrialization’, Journal of Economic Growth, 21, pp. 1–33. Rodrik, Dani, Diao Xinshen, and Margaret McMillan, 2017. ‘The Recent Growth in Developing Economies: A Structural-Change Perspective’, mimeo, Cambridge, MA: Harvard Kennedy School of Government. Sachs, J. and A. Warner, 2001. ‘The Curse of Natural Resources’, European Economic Review, 45, pp. 827–38. Sala-i-Martin, Xavier et al., 2015. Reaching Beyond the New Normal: Findings from the Global Competitiveness Index 2015–2016, Global Competitiveness Report 2015–16, Geneva: World Economic Forum. Shriram, Urvi, Dennis Snower, and Mike Orszag, 2013. Economic Performance Index (EPI): An Industry-Centric Measurement Approach, Global Economic Symposium, Kiel: Kiel Institute for the World Economy and Towers Watson. Sölvell, Orjan, Christian Ketels, and Goran Linqvist, 2003. The Cluster Initiative Greenbook, Stockholm: Ivory Tower. Spolaore, Enrico and Romain Wacziarg, 2012. ‘How Deep Are the Roots of Economic Development?’, NBER Working Paper No. 18130, Cambridge, MA: NBER. Stiglitz, Jospeh E. and Justin Y. Lin (eds.), 2013. The Industrial Policy Revolution I: The Role of Government Beyond Ideology, New York: Palgrave Macmillan. Studwell, Joe, 2013. How Asia Works: Success and Failure in the World’s Most Dynamic Region, London: Profile Books. te Velde, Dirk, 2011. ‘Introduction, DPR Debate: Growth Identification and Facilitation: The Role of the State in the Dynamics of Structural Change’, Development Policy Review, 29, (3), pp. 259–63. Timmer, C. Peter and Selvin Akkus, 2008. ‘The Structural Transformation as a Pathway out of Poverty: Analytics, Empirics and Politics’, CDG Working Paper No. 150, Washington, DC: Center for Global Development.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Valencia Caicedo, Felipe and William F. Maloney, 2014. ‘Engineers, Innovative Capacity and Development in the Americas’, World Bank Policy Research Working Paper No. 6814, Washington, DC: World Bank. Warwick, Ken, 2013. ‘Beyond Industrial Policy: Emerging Issues and Trends’, OECD Science, Technology and Industry Policy Papers No. 2, Paris: OECD. Zeng, Douglas Zhihua, 2008. Knowledge, Technology, and Cluster-Based Growth in Africa, Washington, DC: World Bank. Zeng, Douglas Zhihua (ed.), 2010. Building Engines for Growth and Competitiveness in China: Experience with Special Economic Zones and Industrial Clusters, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

P A R T I I .............................................................................................................

DRIVERS, CHANNELS, AND POLICY INSTRUMENTS .............................................................................................................

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

       ......................................................................................................................

    -

8.1 I

.................................................................................................................................. S change, as an essential feature of modern economic growth, is strictly related to trade, and understanding their causal relation has been a key issue in economics since the time of Adam Smith. The effects of economic growth on levels and composition of trade by country are a staple in early development literature. Classics such as Maizels (1963), Kuznets (1966), and Chenery and Syrquin (1975) and more recently Prados de la Escosura (2005) analysed how growth and structural changes in domestic economy affected the patterns of trade. The early (or naïve) versions of gravity models of trade considered the size of the economies as the main determinant of bilateral trade flows (Head and Mayer 2014). A parallel literature, since the pioneering work by Edwards (1998), have tried to measure the effect of trade and openness on economic growth, with mixed results. Contrary to expectations, the increasingly sophisticated tests have failed to discover a consistently positive effect of trade and openness on growth (Singh 2010; Ackah et al. 2015; Costantinescu et al. 2016). Actually, some historical works have found a positive relation between tariffs and economic growth in the nineteenth century (O’Rourke 2000; Jacks 2006), although this result is not robust either (Tena-Junguito 2010; Schularick and Solomou 2011). These approaches are no longer popular. Growth regressions are deeply affected by the twin problem of omitted variables and endogeneity (Durlauf et al. 2005). Development economists focus on microeconomic behaviour rather than on macroeconomic changes in the long-run (Ray 1998). The recent micro-founded gravity models of trade consider explicitly only (proxies for) trade costs

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

as determinant of trade, dealing with all features of countries and their change in time, including structural transformation, by adding country and time fixed effects, and focus on static gains from trade (Costinot and Rodriguez-Clare 2014; Donaldson 2015). This chapter tackles the issue from a distinctive point of view. First, we adopt a truly long-term view, which we hope can enrich our understanding of more recent trends. Structural transformation started in Great Britain with the Industrial Revolution, in the USA and other European advanced countries in the early nineteenth century. In all these countries, it was largely over by the 1950s. Second, we do not attempt any explicit econometric testing: the data are too scarce and we do not have a good solution to the endogeneity problem. Last but not least, we focus on the effects of structural transformation on trade rather than trade on transformation. More precisely, we estimate the contribution of structural transformation to changes in openness, as measured by the export/GDP ratios. We prefer this metric to the ratio of imports and exports to GDP, because arguably it offers a glimpse of the trade to growth nexus. The more exports grow relative to GDP in any given period, the more likely is that they played a dynamic, growth-fostering, role. On the other hand, we cannot consider the effect of trade on structural transformation because comparative historical data on the composition of trade and GDP are not (yet) available. The available data constrains our analysis as well. Before 1970, we must deal with a subset of (mostly advanced) countries and we have to define structural transformation as the change in the share of tradables on total GDP. This is to some extent a blessing in disguise as the division between tradables and non-tradables is much less likely to be affected by trade than the composition of tradables. We start the chapter with a brief outline of the growth of world trade from 1800 to the present. Section 8.3 deals with long-term changes in export/GDP ratios at current prices for different time-invariant samples of countries (Federico and Tena-Junguito 2017). In Section 8.4, we measure the effect of structural transformation on world openness with a constant market share analysis. Section 8.5 concludes.

8.2 T G  W T, 1800   P

.................................................................................................................................. Figure 8.1 plots our series of world trade. After 1938 it reproduces the data of the United Nations (UN Statistical Yearbook) and of the WTO (www.wto.org). From 1850 to 1938 we obtain the data by summing our new estimates of exports by polity (independent countries, colonies, or native territories) at constant prices (Federico and Tena-Junguito forthcoming). In those years, our database includes about 130 polities—that is, all existing ones in each year, with very few and quantitatively negligible exceptions. Unfortunately, we have been unable to estimate exports in the first decades of the nineteenth century for quite a few polities, including the whole of sub-Saharan Africa. Thus, we estimate trade

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



9 8 7 6 5 4 3 2 1 0 1800

1820

1840

1860

1880

1900

trade

1920

1940

1960

1980

2000

Forecast 1817–1913

 . The growth of world trade, 1800–2014 (log scale) Sources: see text.

Table 8.1 Rates of growth of world trade, 1800–2014 Rates (*100) 1800–17 1817–65 1866–1913 1919–29 1929–38 1950–73 1973–80 1980–2007 2007–14

0.49 3.97*** 3.07*** 5.37*** 0.83 8.08*** 3.96*** 5.86*** 1.29

Cumulated change (%) 8.7 598.6 310.1 71.0 7.2 541.3 32.0 386.9 9.8

Notes: * significant at 10 per cent; ** significant at 5 per cent; *** significant at 1 per cent Source: see text.

from 1800 to 1850 by splicing together indexes for three time-invariant samples of polities—for 1800–22 (featuring ten polities, which account for 55 per cent of world trade in 1850), 1823–29 (62, accounting for 80 per cent of world trade in 1850) and 1830–49 (eighty-nine polities, accounting for 95 per cent of trade). The series show how astonishing the growth of world trade has been: the 1800–2014 rate (3.84 per cent yearly, significant at 1 per cent) corresponds to a cumulated increase of 3,705 times magnitude. This growth appears at a first glance not to be steady: a visual inspection, supported by some simple time-series analysis, suggests a division into nine periods, including the slowdown after 2007 (Table 8.1). World trade remained broadly constant during the French wars and it grew very quickly from 1815 to 1913. According to the best (and admittedly very uncertain) guess, trade during the wars was about a third below the level of the 1780s, but the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

recovery accounted for less than 2 per cent of the cumulated increase of the next one hundred years. The growth was faster in the first half of the period: if trade had continued to grow as quickly after 1867 as in 1817–66, it would have been 55 per cent higher at the end of the period. During World War I world exports fell by about a quarter, returning to the 1913 level seven years after its end. By 1929 world trade at current borders exceeded the pre-war peak by a third, but it collapsed by 35 per cent in 1933. The recovery after World War II was very fast: in 1950, trade was already 10 per cent higher than in 1929. It continued to growth very quickly until the oil crisis, and, after a short slow-down in the 1970s, from the 1980s onwards. The long-run rate of growth was significantly higher in 1950–2007 (5.10 per cent yearly) than in 1817–1913 (3.62 per cent). The Great Recession has so far been decidedly milder than the Great Depression. Trade did fall in 2009 (‘only’ by 12 per cent) but it rebounded the next year and since then has been growing very slowly or not at all. This sketch raises two further questions: (i) to what extent was the post-1950 growth a recovery from the shocks of the two world wars and the Great Depression? (ii) to what extent was the growth of trade affected by boundary changes? We address the first issue by adding to Figure 8.1 a counterfactual trade series growing as fast after 1913 as in 1816–1913. The boom in trade during the Golden Age almost brought it back to the pre-1913 path in the mid 1970s and further growth during the second globalization has caused actual trade to exceed the counterfactual level since 1994. However, the gap has been shrinking since 2007 and, without a quick recovery, trade is going to slide below the path again. We address the second question by estimating series of world trade at constant (1913) boundaries.¹ The differences with the baseline series at current borders are minimal before 1913 but substantial after 1918. The dissolution of Austria-Hungary and of the Ottoman Empire, ceteris paribus, caused actual trade in 1924 to be 3.1 per cent higher (6.8 per cent for Europe only) than the counterfactual, 1913 border, estimate. The gap reduced to 1.5 per cent (2.7 per cent) in 1938—that is, in the interwar years without the boundary changes trade would have been smaller but it would have grown a bit faster. With a different method, Lavallée and Vicard (2013) estimate that post-war boundary changes, such as the partition of British India in the 1950s and the fragmentation of the Soviet Union and Yugoslavia in the 1990s, accounted for about a sixth of the overall rise in trade from 1950 to 2007—that is, about a third of a point of rate of growth. In the long run, all polities joined the trade boom, but their performance differed widely: the simple coefficient of correlation of shares by polity between 1850 and 2014 is 0.52, and some changes are really huge. Saudi Arabia accounted for 0.002 per cent of world exports in 1938 and for 1.8 per cent (16th largest exporter) in 2014. The share of China declined from 2.3 per cent in 1850 (11th largest exporter) to 1.6 per cent in 1913 (17th place), remained constant until the 1970s and then rose steadily to 15 per cent in ¹ The estimate considers only major changes—such as the breakdown of polities and the creation of new polities—but not the swaps of territories between polities. Cf. for a justification of this choice and details on the method of estimation (Federico and Tena-Junguito forthcoming).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



2014. The UK had the opposite trajectory: it had been the largest exporter in 1850 by far (19 per cent of world exports) and in 1913, although only by a whisker, but in 2014 it was only the 10th largest (2.7 per cent). It would be impossible to map changes for all polities—so in Figure 8.2 we aggregate them by continent by level of development.² (a) shares by continent 100 90 80 70 60 50 40 30 20 10

EUROPE

AMERICA

ASIA

AFRICA

2010

2000

1990

1980

1970

1960

1950

1940

1930

1920

1910

1900

1890

1880

1870

1860

1850

1840

1830

0

OCEANIA

(b) shares by level of development 100 90 80 70 60 50 40 30 20 10

Rich 1870

Other OECD

Rest of Asia

2010

2000

1990

1980

1970

1960

1950

1940

1930

1920

1910

1900

1890

1880

1870

1860

1850

1840

1830

0

Rest of World

 . Distribution of world exports at current prices, 1830–2010 Sources: Federico and Tena-Junguito (forthcoming) and UN Statistical Yearbook.

² The ‘old rich’ group includes the UK and all countries whose GDP per capita exceeded half the British level in 1870 (Australia, Belgium, Canada, Denmark, France Germany, Netherlands, New Zealand, Switzerland, and the USA). The ‘other OECD’ countries are Austria, Greece, Finland, Ireland, Iceland, Italy, Japan, Norway, Portugal, Spain, and Sweden. Europe includes the former Soviet Union.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

These aggregate shares are of course less erratic than individual county ones, but one can still detect some massive changes, especially since the 1970s. In the early 1830s, Europe accounted for 62 per cent of world exports and the ‘old rich’ for about a half. This latter share increased by ten points in the 1850s and then remained around 60 per cent until World War I, while the Europe’s share was drifting slightly downwards to 56 per cent. The two world wars and the Great Depression caused substantial changes in shares, which reversed during the Golden Age. In 1970 Europe still accounted for 54 per cent of world exports and the ‘old rich’ for 59 per cent. Even more strikingly, the correlation of country shares remained quite high—0.69 between 1850 and 1970 and 0.72 between 1870 and 1970. Since 1972, the distribution of world exports started to shift towards Asia: its cumulated share rising from about a sixth to almost two-fifths of the total. Until the early 1990s, the fall in the share of ‘old rich’, from 56 per cent to about 40 per cent of world exports, had been compensated for by the relative increase of exports from the ‘other OECD’ countries—most notably Japan and Italy. Since the mid-1990s, exports from the ‘old rich’ decreased further to slightly over a third, but the ‘other OECD’ countries joined the decline, down to 12 per cent, the level of the 1960s. The big winner of the second globalization has been the ‘Rest of Asia’, whose exports exceeded exports of the ‘old rich’ for the first time in 2014.

8.3 O   T G

.................................................................................................................................. After 1970, computing the export/GDP ratio is easy, as the United Nations database provides data on GDP at current prices for all countries of the world. Unfortunately, this is not true before 1969, as series of GDP at current prices are available only for some, albeit very important, countries (Federico and Tena-Junguito 2017). Three series, for the USA, France, and Sweden start in 1800, most of the others sometimes in the nineteenth century and quite a few in the twentieth. It is possible to compute openness for thirty-eight polities in 1913, which comes out at about a half the ratio for the same polities in 2007 in aggregate (12.5 per cent vs. 22.5 per cent) and also for the overwhelming majority of cases (twenty-nine out of thirty-eight). The list includes all major countries, so the world would have been more open in 2007 than in 1913 even if all omitted polities had exported nothing. We analyse changes in time by building series of export/GDP ratios for two different samples, featuring seventeen polities since 1830 (thereafter ‘1830 sample’) and twentyseven since 1870 (‘1870 sample’). Figure 8.3 plots them, alongside the series for all countries since 1970. The series for the ‘1830 sample’ shows a massive increase in export/GDP ratio from 1830 to 1870, almost as fast as the rise during the second globalization. The sample is small, but rather representative, as these countries accounted for three-fifths of world

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



0.3 0.25 0.2 0.15 0.1 0.05 0 1830

1850

1870

1890

1830 sample

1910

1930

1870 sample

1950

1970

1990

2010

UN (all countries)

 . World openness, 1830–2014 Sources: Federico and Tena-Junguito (2017).

exports in 1850. Furthermore, the growing integration is confirmed by independent evidence about the convergence of prices within Europe (Federico 2011) and worldwide (Sharp and Weisdorf 2013; Chilosi and Federico 2015) and about the extensive liberalization of trade (Federico 2012; Tena-Junguito et al. 2012). The data suggest that, contrary to the conventional wisdom (O’Rourke and Williamson 1999; Collier and Dollar 2002; Bordo et al. 2003), the upward march of globalization halted after 1870. The export/GDP ratio for the ‘1870 sample’ fluctuated below the 1870 level until the very eve of World War I. However, it is likely that openness increased in the rest of the world. Export per capita of the polities outside the sample (103 in 1913) tripled from 1870 to 1913, while their GDP per capita according to the (admittedly tentative) estimates by Maddison (2010) increased by 45 per cent. The combination of these different trends would imply a 40 per cent increase in total openness. Openness, as trade, fell during the war, but, unlike trade, it did not recover in the 1920s. At the trough of the Great Depression, in 1932, the export/GDP ratio of the ‘1870 sample’ (6.6 per cent) was about half the 1913 peak and the ratio for the ‘1830 sample’ (5.7 per cent) was roughly at the level of the early 1830s. In the 1950s and 1960s the fast growth of trade was matched by an almost as fast increase in GDP of the advanced countries—so that the export/GDP ratio for the ‘1870 sample’ remained below its 1920s level (and thus, a fortiori, below the 1913 peak) until the outbreak of the oil crisis. This latter event caused the prices of tradables to jump relative to the GDP deflator(s) and consequently induced openness at current prices to soar.³

³ The jump disappears in the series of export/GDP at constant prices for a somewhat larger sample. It increased steadily in the 1960s (from about 8% to 9.7%) and in the 1980s (from 9.7% to 11.3%). See Federico and Tena-Juinguto (2016) for details.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

The export/GDPratio for all countries declined somewhat in the early 1980s, grew steadily in the late 1980s and early 1990s and then very quickly from 1993 to its all-time peak in 2008. By then the export/GDP ratio for all countries was three times higher than in 1970 (25.4 per cent to 8.2 per cent) and 75 per cent higher than in 1993.⁴ The ratio fell sharply in 2009, recovered in the following year, and then remained broadly constant thereafter, slightly below the 2008 peak. As is well known, ceteris paribus, the level of openness of a country is negatively related to its size (Alesina et al. 2005). A simple log–log regression for the thirty-eight countries, with population as proxy for size, yields coefficients of 0.16 in 1913 and 0.21 in 2007, both significant at 5 per cent, while dummies for continents and landlocked countries are not significant. Thus, the level of openness of any group of polities depends on its composition. Asia and the Americas come out as much less open than Europe or Oceania in the long run (Figure 8.4) because India and the USA account for most of the GDP of the two continents in the ‘1870 sample’.⁵ Also movements in openness and not just levels, differ across continents. The coefficient of correlations with the world ratio are quite high for Europe (0.87) and the Americas (0.84) but low for Asia (0.39) and even negative ( 0.14) for Oceania. Trends in Asia diverged sharply from worldwide ones during the Golden Age, when India closed itself to international trade, while the ratio for Oceania features huge swings before 1950, following the boom-and-bust pattern of Australian growth (MacLean 2013). The cases of Asia and Oceania highlight a key difference between trade and openness. The movements in the export/GDP ratio diverged from the worldwide trends in a substantial number of cases: the coefficient of correlation is negative for seven of the twenty-seven polities of the ‘1870 sample’, including India ( 0.09) and Australia ( 0.17) and the average is 0.37. These differences are concentrated in the central years of the period, from 1870 onwards. (i) From 1830 to 1870, the export/GDP ratio declined from 40 per cent to 25 per cent in Brazil where a national economy was starting to develop (Absell and Tena-Junguito 2016) and remained constant in the USA, where the boom of exports (from about 6.5 per cent to 8 per cent of world total) was matched by parallel growth in GDP. But these were exceptions. The ratio increased rapidly in exporters of primary products, such as Argentina (from 7.5 per cent to over 11 per cent) or Cuba (from 18 per cent to 26 per cent), but also in industrializing countries such as France (from 4 per cent to 12 per cent) Belgium (from 8 per cent to 14 per cent) and above all the UK (from 10 to 18.5 per cent). ⁴ The ratio for the ‘1870 sample’ rose by 42% from 1983 to 2008 and by 101% from 1970 to 2008. ⁵ Over the whole period 1870–2014 the export/GDP ratio for the polities of the ‘1870 sample’ was 6.1% in Asia (India 5.9%), 7.8% in the Americas (USA 5.9%), 17.7% in Oceania and 18.3% in Europe. Unfortunately, there are no suitably long series of GDP at current prices for Africa. The export/GDP ratio at constant prices for six countries (including Egypt and South Africa) is very high but poorly correlated with the worldwide one.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



(a) by continent 35 30 25 20 15 10 5 0 1870

1890

1910 America

1930

1950

Asia

1970 Europe

1990

2010

Oceania

(b) by level of development 25 20 15 10 5 0 1870

1890

1910

1930 rich

1950

1970

1990

2010

other

 . Openness, ‘1870 sample’, 1870–2014 Sources: see text.

(ii) From 1870 to 1913, the export/GDP ratio continued to grow in the peripheral countries, including Argentina (a further increase to 18 per cent), India (from 7 per cent to 11.7 per cent) and the Ottoman Empire (from 5.7 per cent to 11.6 per cent). In contrast, openness increased very little or declined in all advanced countries, except Germany. The ratio increased by one percentage point in the USA, but in the UK it remained below the level of the early 1870s until a surge in the years immediately before the war. (iii) In the interwar years, trends by polity are not so uniformly bleak as the aggregate series implies. About a third of the polities, including Canada, Belgium, Denmark, and the UK, were more open in 1950 than in 1913. Actually, the decrease in world openness depended largely on a composition

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

effect. The growing share of the USA on GDP of the polities of the sample (from 35 per cent in 1913, to 45 per cent in 1929 and to 59 per cent in 1950) accounts for about two-thirds of the decrease in total openness from 1913 to 1929 and for more than half from 1913 to 1950. (iv) trends during the Golden Age were mixed. The export/GDP ratio jumped in Germany, from around 10 per cent, historically quite a low level, to 18 per cent, remained roughly constant in France, Japan, and the USA, while it fell in the UK (from 24 to 15 per cent) and in almost all the poor countries in the sample. It halved in India (from about 9 per cent to less than 4 per cent) and Turkey (from 7.5 per cent to 3 per cent) and collapsed in Argentina (from 28 per cent to 5 per cent). As a result, openness in 1972 was lower than in 1913 in 25 out of 34 polities, including all the major ones. (v) during the second globalization, as during the first one, openness increased in almost all countries. The exceptions were concentrated in the Caribbean (including Cuba) and in Subsaharian Africa. As expected, openness increased a lot in former Socialist countries, tripling in the Soviet Union (from 3 per cent in 1972 to 31 per cent in 2007) and also in China (from 2.5 per cent to over 32.6 per cent).⁶ (vi) the Great Recession hit all countries, with few and hardly relevant exceptions, causing the export/GDP ratios to fall in 2009 by 15 per cent on average. Since 2009, the ratio has increased in about two-thirds of countries, but the rise has been large enough to bring openness back to its 2009 level in only a third of countries. This group included the USA, Italy, and (barely) the UK, but no other G7 country, nor China. It was the only sizeable country to have experienced a steep decline in openness from 2007 to 2009 (from 32.6 per cent to 22.5 per cent) without any recovery. As a result, in 2014 the export/GDP ratio of China was a third lower than in 2007.

8.4 S T   G  O

.................................................................................................................................. Following a consolidated tradition, we proxy the share of tradable GDP with the sum of shares of agriculture and manufacturing VA, which we can estimate for Australia and six European countries, including France and the UK, since 1830 and for fourteen, adding the USA, India, and five European countries (including Italy and Germany)

⁶ For the sake of comparability with earlier data, after 1991 we estimate openness as the ratio of cumulated exports of all former Soviet republics to cumulated GDP of the same countries. The ratios might thus overestimate openness as they classify formerly domestic flows as international trade.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



since 1870. These fourteen polities accounted for 66 per cent of world exports in 1913 and for 40 per cent in 2007. All these countries, except India, experienced a massive process of industrialization before 1913, which affected very little the share of non-tradables on GDP. The aggregate share increased marginally from 32 per cent of GDP in 1830 of the seven countries to 35 per cent in 1870, and from 42 per cent to 48 per cent of the GDP of the fourteen countries from 1870 to 1913. The trend accelerated in the 1920s, but paused thereafter, even reversing during the Great Depression. Services already accounted for 59 per cent of GDP of the fourteen countries in 1929, but they exceeded this level only in 1967 and continued to grow up to about 75 per cent in the early 2000s. The figures are lower for the rest of the world, but the trend is similar: the share of tradables on world GDP has risen from 55 per cent in 1970 to 68 per cent in 2014. The decline of tradables on GDP implies that exported goods were produced in (and imported ones competed with) a relatively shrinking portion of the economy. Thus the aggregate ratios arguably underestimate the impact of trade on the affected activities. A more accurate measure of this impact would be their ratio to gross output of tradables, but data on this are only available since 1973 and only for a small number of advanced countries (Federico and Tena-Junguito 2017). Thus, we use the value added of agriculture and manufacturing as denominator (Feenstra 1998). The ratio, or openness tradables, exceeds total openness by construction and the gap has been growing, especially since the 1960s (compare Figures 8.3 and 8.5). Openness tradables for the seven polities was 44 per cent higher than total openness in 1830 and it was double by 1870, while for the fourteen countries, the ratio was 1.70 in 1870, 2 in 1913, 2.5 in 1972, and 4 in 2007. By definition, structural transformation, in our definition, is bound to reduce world openness, ceteris paribus. When and by how much? We answer by decomposing the 1.0 0.9 0.8 0.7 0.6 0.5 0.4 0.3 0.2 0.1 0.0 1830

1850

1870

1890

1830 sample

1910

1930

1870 sample

1950

1970

1990

All countries

 . Openness tradables, 1830–2014 Sources: see text.

2010

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

total change in openness into three components—the structural transformation, changes in the distribution of GDP by country (or location effect), and a residual. Ceteris paribus, a rise (decline) in the share of non-tradables decreases (increases) openness, while an increase (decrease) in the share of relatively closed polity, such as the USA, would reduce (increase) it. The residual captures the effect of a change in trade costs, as well as other dimensions of the structural transformation, such as changes in the share of manufacturing on tradables and in the composition of manufacturing, and of course any data error. From 1830 to 1870, the residual explains all of the increase in openness, about 4.5 points, while structural change had a very modest negative effect (0.3 points) and the location effect was nil. Actually, the location effect would not have been nil had it been possible to include the USA in the sample.⁷ We report the result of the decomposition for the ‘1870 tradables sample’ in Table 8.2, inverting the signs of percentage changes in periods of declining openness to make reading easier. We have selected periods according to the structural breaks in main movements in openness, omitting the last period as it is too short and does not show any clear trend. Throughout the whole period, openness grew by 6.7 points (second column), exclusively due to the increase in the residual. It explains all the growth in openness after 1973 and in 1870–1913, while it was heavily negative during the Great Depression.

Table 8.2 Decomposition of change in world openness, long-term trends Openness Initial

Change

1870–1913 %

11.20

2.16

1913–50 %

13.35

1950–73 %

Structural transformation

Location effect USA

other

Total

0.98 45.6

1.16 53.7

0.36 16.6

0.80 37.1

3.94 182.8

5.86

0.16 2.7

4.52 77.2

1.59 27.2

2.93 50.0

3.09 52.7

7.49

3.34

2.35 70.3

2.22 66.2

0.04 1.2

2.18 65.1

1.18 35.4

1973–2007 %

11.29

7.07

3.24 45.7

0.32 4.5

0.49 6.9

0.17 2.4

10.48 148.1

1870–2007 %

11.20

6.72

5.34 79.5

1.88 28.0

1.24 18.5

0.64 9.5

12.69 189.0

Residual

⁷ The impact of the rise of the USA before 1870 can be measured by decomposing changes in openness of the ‘1830 sample’ (inclusive of the USA) between location effect and a residual, which includes the structural change. It reduced openness, ceteris paribus, by 0.4 points from 1830 to 1870. All other changes in the distribution of GDP augmented it by 0.9 points, for a total positive location effect of 0.5 points (i.e. about 10% of the total increase for that sample).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



In the long-run, the location effect is negative, by 0.6 points. The column ‘location USA’ shows that the decline depends exclusively on the rise of the USA: all other changes in the distribution of GDP by country increased openness. The effect was particularly large in 1913–50, when the United States rose from 43 per cent of GDP of the sample in 1913 (up from 25 per cent in 1870) to 66 per cent. The structural change reduced openness during the second globalization and also in the long run: without any other effect, ‘world’ openness in 2007 would have been about half its 1870 level. We use the same method to decompose changes in total openness since 1970 in Table 8.3, without distinguishing a specific location effect for the USA as the change in their share of world GDP was fairly small. The table confirms the two key points of the analysis of the ‘1870 sample’, the dominance of the residual and the openness-reducing effect of structural transformation, while location effect was small and positive, rather than very small and negative. The structural change compensated for about a quarter of the effect of the residual for all countries. This figure increases dramatically if we consider only the rich countries, which underwent the most far-reaching structural change. Without any other effect, the rise of non-tradables would have caused their openness to collapse to little more than 1 per cent.

Table 8.3 Decomposition of change in world openness, 1970–2007 Openness

Africa % America % Asia % Europe % Oceania % Rich 1870 % Other OECD % ROW % World % Source: see text.

Initial

Change

17.5

14.9

5.7

6.1

9.2

24

12.6

18.8

9.4

18.8

8.8

9.6

10.9

10.1

5.3

21.1

9.5

15.1

Structural transformation

Location effect

Residual

0.6 3.7 2.8 45.8 1.7 7.0 8.5 45.3 78.4 416.8

3.9 26.4 0.1 2.4 4.3 18.1 0.6 3.2 6.4 34.0

18.3 122.7 8.7 143.4 21.3 88.9 27.9 148.5 90.8 482.8

8.4 87.2 2.6

0.8 8.8 1.3

18.8 196.0 11.4

26.2 1.6 7.5

13.0 2.4 11.6

113.2 20.2 95.9

3.8 24.9

1.3 8.3

17.6 116.6

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

These results are subject to two caveats: (i) the omission of other dimensions of structural transformation would bias the contribution of residual upwards if the relatively growing sector(s) were more open than the rest of tradables and vice versa. As already mentioned, it is impossible to explore this issue further with the available historical data. Even if the necessary data had been available, any interpretation of results as causation would have been difficult, as changes in composition of—say—manufacturing were much more likely to be determined by trade than the rise of services. (ii) we implicitly assume that services are not tradable. This assumption is never wholly true. Shipping and financial services have always been traded, but in absolute terms their trade was small before 1913. Exports of services from the UK, by far the largest supplier of services, accounted for 1.1 per cent of the GDP of the ‘1870 sample’ and only for 0.8 per cent on the eve of World War I (Mitchell 1988: 871). This decline might reflect a growing competition on markets for internationally traded services, but it is highly unlikely that adding exports from other countries would increase openness by much. According to UN data, trade in services accounted for 2.9 per cent of world GDP in 1970 and for 5.5 per cent in 2008. The rise (2.6 points) compensates for about three-quarters of the loss in openness from the structural transformation (cf. Table 8.3).

8.5 C

.................................................................................................................................. We can sum up our results in three main stylized facts: first, contrary to a widely, although not unanimously, held view, the current level of globalization is unprecedented. The growth in openness from 1993 to 2008 exceeds by far the growth before World War I for any comparable group of polities and, in all likelihood, also if one factors in the omitted polities. Furthermore, the difference between the two periods is substantially larger if one considers tradables only. Second, the rise in openness—and thus the potential for trade-fostered economic growth—concentrated in 1830–70 and in 1972–2007. The period 1870–1972 featured widely different trends by country, but worldwide openness grew very little. Third, the movements in openness were largely driven by the residual, which in our constant market share analysis reflects mostly, albeit not exclusively, changes in trade costs and, in more recent times, the development of international supply chains (Costantinescu et al. 2016). The long-run effect of changes in distribution GDP by countries is small, while structural change dampened the increase in openness since the 1970s. However, this last effect was largely compensated by the increased trade in services. The financial crisis suddenly stopped the onward march of openness, which has not yet resumed after seven years. This ‘Great trade slowdown’ has attracted a lot of scholarly interest (Baldwin 2009; Costantinescu et al. 2015; Hoekman 2015) but it is

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



too early to tell whether the slowdown is the harbinger of a long-term stagnation, or simply an interlude after which the march will resume.

R Absell, C. D. and A. Tena-Junguito, 2016. ‘Brazilian Export Growth and Divergence in the Tropics, in the Nineteenth Century’, Journal of Latin American Studies, 48 (4), pp. 677–706. Ackah, Charles, Vincent Leyaro, and Oliver Morrissey, 2015. ‘Trade, Tariffs, Growth and Poverty’, in Oliver Morrissey, Ricardo Lopez, and Kishor Sharma, eds, Handbook on Trade and Development, Cheltenham: Edward Elgar Publishing, pp. 19–37. Alesina, Alberto, Enrico Spolaore, and Romain Wacziarg, 2005. ‘Trade Growth and the Size of Countries’, in Aghion Philippe and Steven N. Durlauf, eds, Handbook of Economic Growth, Amsterdam: Elsevier, vol IB, pp. 1499–542. Baldwin, Richard, 2009. The Great Trade Collapse: Causes, Consequences and Prospects. A VoxEU.org Publication, Geneva: London Graduate Institute and CEPR. Bordo, Michael, Alan M. Taylor, and Jeffrey G. Williamson, 2003. Globalization in Historical Perspective, Chicago: University of Chicago Press. Chenery, Hollis and Moses Syrquin, 1975. Patterns of Development 1950–1970, Oxford: Oxford University Press. Chilosi, David and Giovanni Federico, 2015. ‘Asian Globalizations: Market Integration, Trade and Economic Growth, 1800–1938’, Explorations in Economic History, 57, pp. 1–18. Collier, Paul and David Dollar, 2002. Globalization, Growth and Poverty, Washington and Oxford: World Bank and Oxford University Press. Costantinescu, Cristina, Aaditya Mattoo, and Michele Ruta, 2015. ‘The Global Trade Slowdown Cyclical or Structural?’ World Policy Research Working Paper No. 7158, World Bank. Costantinescu, Cristina, Aaditya Mattoo, and Michele Ruta, 2016. ‘Does the Global Trade Slowdown Matter?’ Journal of Policy Modeling, 38 (4), pp. 711–22. Costinot, Armand and Andres Rodriguez-Clare, 2014. ‘Trade Theory with Numbers: Quantifying the Consequences of Globalization’ (NBER WP 18896), G. Gopinath, E. Helpman, and K. S. Rogoff, eds, Handbook of International Economics vol 4, Amsterdam: ElsevierNorth Holland. Donaldson, Dave, 2015. ‘The Gains from Market Integration’, Annual Review of Economics, 7, pp. 619–47. Durlauf, Steven, Paul Johnson, and Jonathan Temple, 2005. ‘Growth Econometrics’, in Aghion Philippe and Steven N. Durlauf, eds, Handbook of Economic Growth, Amsterdam: Elsevier, vol IA, pp. 555–677. Edwards, Sebastian, 1998. ‘Openness, Productivity and Growth: What Do We Really Know?’ Economic Journal, 108, pp. 383–98. Federico, Giovanni, 2011. ‘When Did the European Market Integrate?’ European Review of Economic History, 15, pp. 93–126. Federico, Giovanni, 2012. ‘The Corn Laws in Continental Perspective’, European Review of Economic History, 16 (2), 166–87. Federico, Giovanni and Antonio Tena-Junguito, 2017. ‘Tale of Two Globalizations: Gains from Trade and Openness 1800–2010’, Review of World Economics, 153, pp. 601–26.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    -

Federico, Giovanni and Antonio Tena-Junguito, forthcoming. ’World Trade 1800–1938: A New Synthesis’, Revista de Historia Economica/Journal of Iberian and Latin Economic History. Feenstra, Robert, 1998. ‘Integration of Trade and Disintegration of Production in the Global Economy’, Journal of Economic Perspectives, 12 (4), pp. 31–50. Head, K. and T. Mayer, 2014. ‘Gravity Equations: Workhorse, Toolkit and Cookbook’, in G. Gopinath, E. Helpman, and K. S. Rogoff, eds, Handbook of International Economics, vol. 4, Amsterdam: Elsevier, pp. 131–95. Hoeckman, Bernard, 2015. The Great Trade Slowdown: A New Normal? CEPR on-line book. Jacks, David, 2006. ‘New Results on the Tariff-Growth Paradox’, European Review of Economic History, 10, pp. 205–30. Kuznets, Simon, 1966. ‘Quantitative Aspects of the Economic Growth of Nations X Level and Structure of Foreign Trade: Long Term Trends’, Economic Development and Cultural Change, 15, pp. 1–140. Lavallée, Emmanuelle and Vincent Vicard, 2013. ‘National Borders Matter . . . Where One draws The Lines Too’, Canadian Journal of Economics, 46, pp. 135–63. MacLean, Ian, 2013. Why Australia Prospered. The Shifting Sources of Economic Growth Princeton, NJ: Princeton University Press. Maddison, Angus, 2010. The World Economy. A Millennial Perspective, Paris: OECD. Maizels, Alfred, 1963. Industrial Growth and World Trade, Cambridge: Cambridge University Press. Mitchell, B. R., 1988. British Historical Statistics, Cambridge: Cambridge University Press. O’Rourke, Kevin, 2000. ‘Tariffs and Growth in the Late 19th Century’, Economic Journal, 110, 456–83. O’Rourke, Kevin and Jeffrey G. Williamson, 1999. Globalization and History. The Evolution of the Nineteenth Atlantic Economy, Cambridge, MA: MIT Press. Prados de la Escosura Leandro, 2005. ‘Gerschenkron Revisited. European Patterns of Development in Historical Perspective’, Economic History and Institutions Carlos III Working paper 79–05. Ray, Debraj, 1998. Development Economics, Princeton: Princeton University Press. Schularick, Moritz and Solomos Solomou, 2011. ‘Tariffs and Economic Growth in the First Era of Globalization’, Journal of Economic Growth, 16, pp. 33–70. Sharp, Paul and Jacob Weisdorf, 2013. ‘Globalization Revisited: Market Integration and the Wheat Trade between North America and Britain from the Eighteenth Century’, Explorations in Economic History, 50, pp. 88–98. Singh, Tarlok, 2010. ‘Does International Trade Cause Economic Growth? A Survey’, World Economy, 33, pp. 1517–64. Tena-Junguito, Antonio, 2010. ‘Bairoch Revisited. Tariff Structure and Growth in Late 19th Century’, European Review of Economic History, 14, pp. 111–14. Tena-Junguito, Antonio, Markus Lampe, and Felipe Tamega-Fernandes, 2012. ‘How Much Trade Liberalization was there in the World Before and After Cobden-Chevalier?’, Journal of Economic History, 72, pp. 708–40. United Nations Yearbook, ad annum. Yearbook of International Trade Statistics, New York: United Nations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  ......................................................................................................................

 ,  ,    ......................................................................................................................

    .   

T relationship between financial development and economic growth as well as the role of reforms in financial development are relatively old issues in development macroeconomics. In this debate, the question of whether a more liberalized financial system—through its effect on structural change—is conducive of higher or lower productivity growth has been a focal point for policy makers and a fundamental element in assessing the desirability (beneficial effect) of financial reforms. Yet, defining and measuring structural change is by no means an easy task. One approach consists in considering changes over time in the fraction of income devoted to different types of goods and services. Under that view, changes can be measured by variations in components of aggregate demand that then affect the supply structure of the economy. Hence it would imply a reallocation not necessarily of only the labour force, which would move from one sector to another but also of the investment pattern (capital). Over a long time horizon, new capital may be invested in a new sector as the economy develops. Another way to look at structural change comes from the oldest and most wellknown focus on the distribution of the labour force across the economy, originating from the seminal Lewis (1954) two-sector model which posits that in the course of economic development, the labour force first tends to move from a lower productivity backward (agriculture) sector toward a higher productivity modern (manufacturing) sector. Given that we will be looking more closely at labour productivity, structural change is hereby understood as the process of labour re-allocation across economic

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

sectors, in such a way that it ignites, maintains or even accelerates economic growth. Subsequently, when the pool of labour reallocation is exhausted, wages and thus incomes start to rise in line with the higher productivity of the manufacturing sector.¹ Notwithstanding aggregate productivity improvements, economic growth therefore depends on the economy’s ability to engineer sectoral reallocation of the work force to take advantage of differentials in productivity gains across sectors.² This chapter therefore takes the latter view and intends to test the impact of financial reform on changes in labour productivity. Early intuitions (Bagehot, Schumpeter) regarding the role of the financial sector on economic growth have, despite disagreements, fed a large body of empirical literature relating to a vast array of financial sector indicators (e.g. financial depth, institutional stability, market-based incentives to take adequate risk, etc. (Levine 2005)) to economic growth. Despite the recognition of the costs of financial crises and hence the need for proper regulation, especially after the Global Financial Crisis (GFC) of 2008, diverse econometric evidence suggests that financial development precedes and contributes to improved economic performance by mobilizing savings, allocating funds more efficiently, transforming risk and therefore increasing productivity and not simply facilitating capital accumulation and investment. The process of reallocating production factors even if limited to labour, might face significant barriers; financial development also requires quite a large number of supporting conditions as diverse as the availability of regular domestic savings, the robustness of the rules of law to enforce contracts, etc. There could be initial rigidities to overcome before a surge in development can take place. This includes obstacles such as hampered labour mobility as the result of a negotiated political economy equilibria after a phase of prosperity, regulations in the labour market, limits to the access to finance, oligopolistic extraction of rents, etc. All such factors can hinder the reallocation process, either by slowing it significantly, making it more costly or even impeding it altogether. In particular, labour market regulations by modifying the risk–return trade-off to open or close a vacancy can have a direct effect on labour demand and hence on the extent to which high productivity gain sectors may be willing or able to absorb the ‘excess’ workforce located in other low productivity gain sectors. Labour market regulation also affects the supply of labour as the existence of rents in the low productivity sector may also deter reallocation by reducing incentives to move to the high productivity sector that is more competitive. Similar arguments can be made about finance. For example, the extent to which firms can easily access financing at a reasonable price for the level of perceived risk is

¹ This theory may be complemented by saying that at later stages of development, the labour force tends to move form manufacturing to services as higher relative productivity gains in manufacturing tends to ‘free’ labour for other activities. ² In this chapter, we focus on labour reallocation because of data limitations. Our argument should, however, apply equally to capital reallocation.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



key in determining how much they can expand and thereby attract new workers. If, during an upswing, the treatment of any demand for credit is exceedingly high, the virtuous cycle of investment with job creation will be slowed down. Conversely, we also know that the opposite excesses of excessive debt build-up and exuberance might accelerate the probability of a crisis. The fact is that finance has more than quantitative effects related to firms’ ability to expand. It also has qualitative implications for growth as credit allocation across firms and sectors determines in an important way, how much structural change can occur, and how much overall growth an economy can benefit from, both in the short and in the medium term. This chapter tests whether liberalizing the functioning of the financial sector and lifting a number of regulations through financial reforms affects structural change and growth. To do so, we proceed as follows. As explained above, we define structural change as the contribution of labour reallocation across sectors to aggregate productivity growth.³ To put it simply, fluctuations in aggregate productivity growth can be related either to changes in productivity growth that are common to all sectors in the economy or to factors of production shifting between sectors with heterogeneous productivity gains. We choose to identify structural change with the latter source of labour productivity growth. We make use of the difference-in-difference methodology to compare the evolution of productivity growth—and its different sources including structural change—before and after a financial reform, and contrast this evolution with the one observed where no financial reform takes place, that is, with the cases that we will call financial status quo. Of course, the assumption that financial reforms can be considered exogenous is pretty strong. Yet, when it comes to the relationship with productivity, the data—as will be discussed in the next sections—does not suggest that this assumption can be rejected out right.⁴ Moreover, it is important to bear in mind that we do not pass any judgement on the ‘nature’ of the financial sector reform per se, and its transmission mechanisms into the rest of the economy. We use the IMF database on financial liberalization as our metric for reforms (Abiad et. al. 2008). This dataset allows us to move away from the analytical work that is based on measuring financial outcomes, for example growth in financial sector employment, growth in credit, or any other similar indicator of size. Instead we look at changes in policies/regulations affecting the functioning of the sector (financial sector reforms) and try to capture how these changes affect productivity. Our objective is to explore the impact of financial sector reforms on productivity and refine the previous general finding on the generic contribution of finance to growth and

³ Although arguably simplistic, this definition has the advantage that it can easily be implemented empirically. ⁴ The main argument for the endogeneity of financial reforms would consist in arguing that financial reforms are undertaken as a response to a productivity slow-down. Yet the evidence produced later shows that if anything, the opposite holds: productivity growth is stronger prior to financial reforms than prior to the financial status quo.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

productivity (e.g. Hausmann and Rodrik 2003). Our assumption is that financial sector reforms can contribute to structural change and we test the hypothesis that finance might play a labour productivity growth enhancing role under specific conditions. Empirical results indeed confirm that labour productivity tends to accelerate following financial reforms, relative to what happens following periods of financial status quo. Moreover decomposing productivity growth suggests that financial reforms are associated with a stronger contribution of labour reallocation to productivity growth, but only in the case of advanced economies. Conversely, financial reforms are associated with stronger economy-wide productivity gains across sectors but only in advanced economies. Hence if financial reforms seem to affect productivity growth in a similar way in advanced and emerging market economies, the data suggest that the mechanisms and channels are very different. To account for these different patterns, we test whether differences in financial development could be an explanation.⁵ This is predicated on two arguments. On the one hand, advanced economies are close to the technological frontier. A well-developed financial system is therefore key to identifying growth opportunities and allow sectors with such opportunities to expand. On the other hand, emerging market economies tend to be further away from the technological frontier and as a result, economy-wide improvements in productivity tend to be more prevalent. Moreover, because the financial sector is not that sophisticated, sectors benefiting from a number of distortions (e.g. natural monopoly rents, large commodity producing firms, state-owned enterprises, etc.) may find it easier to expand their labour force than sectors with stronger growth opportunities. Labour reallocation may hence not necessarily contribute to aggregate productivity growth in a positive way. And indeed the evidence confirms that the pay-off to financial reforms in terms of economy-wide productivity improvements is negatively correlated with financial development, while the pay-off in terms of labour reallocation driven productivity improvements is positively correlated with financial development. The rest of the chapter is organized as follows: we begin with a short literature review to illustrate that the relationship between financial reforms and productivity growth has some gaps. We then expose the characteristics of our data set, explain our methodology, and provide a first overview of the results coming from a simple analysis of the data. We then contrast advanced and emerging economies (AEs and EMEs) to highlight differences in productivity growth patterns. We then use a variant of the Olley-Pakes decomposition and apply a proper difference-in-difference regression approach to identify changes in productivity growth comparing situations with and without financial sector reforms. We finally analyse our results in concluding remarks.

⁵ Although recent literature questions the use of credit to GDP as an indicator of financial development (see Sahay et al. 2015), we will stick to the former, essentially because of data limitations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



9.1 L R: F  G  M S C  R

.................................................................................................................................. The literature on the relationship between the development of the financial sector and growth performance grew significantly pari passu, with large data sets that were assembled and allowed a variety of econometric testing of pure cross-country growth regressions (see Valickova et al. 2015 for a meta-analysis of this literature), panel data and microeconomic studies at the industry and firm-level of detail. However, the specific role of financial reforms has not been extensively explored. The initial reviews (King and Levine 1993) typically related aggregate measures of growth over long periods and measures of financial development. The increase in this type of literature came essentially from the growth and sophistication of the databases to include more countries and more financial sector indicators. The major results were that, using the proper controls over various other determinant factors explaining economic growth, financial development variables are indeed significant and associated with productivity growth and capital accumulation. Further developments in this literature tested for simultaneity bias and used instrumental variables to explain financial development across countries that is uncorrelated to growth, in particular variables that capture characteristics of the ‘legal framework’, the efficiency of law enforcement, etc. (La Porta et al. 1998; and also Beck et al. 2000). In parallel to these cross-section and panel studies, others (Rajan and Zingales 1998), focus on industry-level data to test whether financial development positively affects growth while providing a rigorous treatment of the issue of causality. Their argument and findings are that financial improvement removes the friction that prevent firms from accessing financing and therefore the industries that use finance more intensively suffer less ‘friction’ than others and grow faster. A similar idea is tested using firm-level data (Demirguc-Kunt and Maksimovic 1998; Demirguc-Kunt and Levine 2001) by trying to get deeper into the financing needs of individual firms and confirm the positive relation between financial and economic growth. With different methodologies it seems that the positive relation between financial development and economic growth holds, with specific variables such as financial depth, the size of the banking sector, the liquidity of stock markets, the wellfunctioning of the legal system and access to external financing. This brand of literature is summarized in compendium studies (World Bank 2001) and surveys (Levine 2005) but should not be interpreted as a naïve advocacy piece in favour of unregulated, fully open, globalized, limitless financial sectors. There was already (even in the early 2000s) a clear understanding of the incentive problems, regulatory capture, distortions, moral hazard problems, and all the array of issues that can lead toward the accumulation of vulnerabilities that end in costly financial crises. As we know, almost all the policy papers and literature on financial regulation post-GFC has emphasized these

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

dimensions (for a good succinct summary, see Fischer (2014), for a more complete narrative see Blinder (2010); for the technical reports, naturally, one should refer to the various contributions by the Bank for International Settlements (BIS) and the Financial Stability Board (FSB)). Along the same lines, another line of work has explored (see Arcand et al. 2012; Cechetti and Kharroubi 2012; or Samargandi et al. 2015) the idea that the level of financial development produces positive results for growth up to a certain point, after which it could become a drag. That insight is important: to be prudent and carefully explore further trade-offs to avoid rushing into conclusions that an unrestricted rise of finance can only bring more growth. But perhaps a more fundamental critique of these approaches (Trew 2006) is that there has been little effort made to match the empirical work with a full-fledged analytical model of the finance and growth nexus on important critical dimensions such as the relationship with endogenous growth, the type of finance mechanism and the treatment of asymmetric information. Implicitly, financial intermediation is treated as a mechanical ‘neutral’ activity—albeit very useful—that transmits resources from households which save to entrepreneurs that invest. Therefore, in most cases the policy recommendations that emerge are related to facilating the functioning of the financial sector by liberalizing it, opening it up to more ‘foreign’ competition and strengthening the institutional framework upon which finance works. An example of such an approach is the case for increasing competition to allow industries that are more financially dependent to grow faster (Claessens and Laeven 2005). Shifting now to the role of the financial sector in fostering structural change, innovation, and reform, it is quite symptomatic that there are relatively few leads in this direction. There are naturally some important insights regarding the role of financing in supporting innovation (Rajan and Zingales 2003; Mohan 2008) but there are surprisingly few contributions that try to construct theoretical models to explain the link between finance and growth (Trew 2007) with the standard set of agents and their micro-founded behaviour. Such a modelling effort would endogeneize the provision of financing/credit to firms that is costly to monitor and whose return is uncertain while at the same time capable of triggering endogenous growth (e.g. a rise in total factor productivity). On another related front, the growth literature has moved away from variations around the canonical aggregate model of growth (Solow 1956) and even further from the role of accumulating physical and human capital, later with endogenous technological change (Grossman and Helpman 1991; Aghion and Howitt 1992) towards exploring the capacity to mobilize resources to innovate, create new products, new processes, and understand what is needed to trigger reform that engineers productivity and structural changes. With the progress made by the literature on the importance of institutions (Acemoglu and Robinson 2012) and reforms, a greater emphasis was put on how to overcome the ‘structural transformation’ challenge (Rodrik 2013), that is, how to ensure precisely that resources flow from low to high productivity sectors as well as on how to strengthen ‘fundamentals’—accumulate a critical mass of resources, to maintain incentives, and macroeconomic balances in such a way that it favours longterm investment decisions, etc.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



This type of reflection abandoned the traditional route of thinking growth/development as a process where a country needs to follow a relatively well-known recipe of principles and policies to think, instead, of studying what are the ‘binding constraints’ that might impede reforms, development, and eventually sustainable growth even if the candidate country has everything else right (Hausmann et al. 2006). This new approach to reform tries to see development as being much more contingent on idiosyncratic characteristics of countries, including features of political economy. Instead of throwing policy makers a ‘profession’s list of consensus measures/policies’, it is better to focus perhaps on one or two single problems that embodies the biggest obstacle to growth. In other words, the process of structural change is more multi-dimensional than just shifting production factors around. One illustration of that complexity can be found in the detailed study (McMillan and Rodrik 2011) of the significant gaps in labour productivity between traditional and modern parts of economies in several regions of the world where the initial intuition (i.e. the ‘natural’ movement from low- to high-productivity activities) is not present in Latin America and Africa. Why? In those areas, typically, countries are characterized by a large share of natural resources in exports that actually makes structural change growth reducing. Interestingly enough, this ‘resource curse’ makes these export sectors function with high productivity, but simultaneously prevents them from absorbing the surplus labour from lower productivity sectors such as agriculture. As reviewed, it is difficult to find a natural role to assign to finance in this recent literature. The emphasis on reform is more on specific ‘institutional’ or ‘market-based’ regulatory impediments rather than on the capacity for credit/financing to facilitate resource reallocation. Nevertheless, this chapter will explore this hypothesis and we now move to detailing our procedure to evaluate it.

9.2 D  E M

.................................................................................................................................. We start by using the IMF database on financial liberalization and, as stated earlier, we are consciously agnostic about the underlying microeconomic and macroeconomic channel of transmission of ‘liberalization’ into the real economy. This data set covers seven different areas for ninety-one countries (credit controls, interest rate controls, barriers to entry, banking supervision, privatization, international capital markets, and securities markets)⁶ of financial reform over the period 1971–2005. Each area receives a score from 0 to 3, higher readings being associated with a more liberalized financial sector. An increase in the index for a given area therefore captures liberalization while a decrease represents some rollback on previous liberalization. To simplify the analysis, ⁶ See Annex 1 for the detailed characteristics of the financial reform data set.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .    Financial status quo

5

5

4

4

3 2

3 2

1

1

0

0

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991

7 6

# of reform areas

average # of reform areas over 3 last years

1970 1971 1972 1973 1974 1975 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991

Financial reforms 7 6

# of status quo areas

average # of status quo areas over 3 last years

 . Financial reforms and financial status quo, 1970–91 Notes: Bars in the left hand panel (right hand panel) represent the number of areas in which a financial reform (no financial reform) took place for the year under consideration. The solid line represents the 3-year backward looking moving average. Source: Authors’ computations.

we focus on two ‘states of the world’ that we want to compare: (1) reform or liberalization and (2) status quo, the latter being represented either by a decreasing or a stable index. Based on this dichotomy of the two states, we count for each country– year pair the number of areas (out of the seven listed above) in which reform was undertaken, and the number of areas in which the status quo prevailed. Last to make sure we really pick up periods of significant financial reforms and conversely periods of significant ‘status quo’, we consider a three-year backward looking moving average and set the highest average obtained through this calculation as, respectively, a ‘reform year’ or a ‘status quo’ year. The example in Figure 9.1 provides a graphical representation of the results from our accounting methodology. The left hand panel provides for each year from 1970 to 1991 the number of areas in which financial reform took place. It shows that in 1976 the economy experienced reform in 5 out of the 7 areas listed above and in 6 out of 7 areas in 1987. Hence, considering the 3-year backward looking moving average (the grey line), financial reform activity reached a maximum during those two years, and as a result, we will pick up those two years— 1976 and 1987—as ‘financial reform’ years for this country. Using a similar methodology, the right hand panel shows that the financial status quo peaked in 1983 and, as a result, 1983 will be considered as a financial status quo year for this country. Applying this approach to our sample of 24 countries, we end up with 125 observations, 51 of which are financial reform years and 74 of which are status quo years. In more detail, financial reform years appear to be almost perfectly split between advanced and emerging market economies (25 vs. 26) while 40 status quo years are found in emerging market economies and 34 in advanced economies. To check that our methodology is rightly picking up periods of financial reform and periods of financial status quo, we look at the change in the financial liberalization index, this index just being the sum of the scores for the seven different areas of financial reform listed above, normalized to be between 0 and 1. A priori, we would expect periods of financial reform to be associated with a significantly larger increase in the financial liberalization index compared to periods of financial status quo. In Figure 9.2, we plot the

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



Cumulative distribution functions 1 0.8 0.6 0.4 0.2 0 –0.1

0

0.3 0.2 0.1 Change in financial liberalization index Financial reform

0.4

Financial status quo

 . Cumulative distribution functions Notes: The black (grey) line represents the cumulative distribution function of the change in the financial liberalization index for country–year pairs identified as financial reform (financial status quo) episodes. Sources: IMF financial reforms database and authors’ computations.

cumulative distribution for changes in the financial liberalization index separately for financial status quo and financial reform years. As is visible in the grey line representing the c.d.f. for changes in the financial liberalization index, conditional financial reform is always below or to the right of the red line representing the c.d.f. for changes in the financial liberalization index conditional on the financial status quo. Hence, the probability that the change in the financial liberalization index is above some given value is always greater under financial reform, whatever the given value is.

9.3 A F A   D R

.................................................................................................................................. We first run a simple eye browsing exercise by which we compare the change in the average performance of economies before and after ‘financial reform’ to the same change but before and after ‘financial status quo’, financial reform and financial status quo being defined as the dummy variable described in Section 9.2. Specifically, we consider the period ‘before’ financial reform (status quo) as the 3-year period ending on the financial reform year (status quo year), financial reform years (status quo years), being defined as n Section 9.2. Conversely, we consider the period ‘after’ financial reform (status quo) as the 5-year period starting on the financial reform year

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .    GDP growth

Employment growth

All countries

0.04

Status quo

All countries

Financial reform

Status quo

0.03

0.03

Financial reform

0.02

0.02 0.01

0.01

0

0 –0.01

–0.01 before

after

change

 . GDP growth and employment growth Notes: The black (grey) bar in the left hand panel represents the 3-year average growth in aggregate GDP preceding (following) financial reform or financial status quo episodes. The black (grey) bar in the left hand panel represents the 3-year average growth in aggregate employment preceding (following) financial reform or financial status quo episodes. The light grey bars in both panels represent the change in average growth rates. Sources: GGDC and IMF financial reform databases, authors’ calculations.

(status quo year). The 3-year choice for the ‘before’ period is directly linked to the methodology used to identify financial reform and status quo years which looks at the extent to which there were financial reforms over the 3-year preceding period. Concerning the ‘after’ period, we consider a longer time span, as financial reforms can take time to deliver real effects, in particular, when some time may elapse between changes in the laws, proper implementation, and practical changes in economic agents’ behaviour. To look at the real effects of financial reforms, we use data on GDP and employment from the GGDC 10-Sector Database (de Vries et al. 2015) which provides, on top of aggregate figures, a sectoral breakdown of GDP and employment, something which we will rely on to investigate the relationship between financial reforms and structural change. For now, turning to the evidence on aggregate GDP, the left hand panel in Figure 9.3 shows that, on average, GDP growth tends to drop after the financial status quo (black vs. grey bars in the status quo panel). By contrast, GDP growth is relatively stable after financial reform (black vs. grey bar in the financial reform panel), which suggests that financial reforms are associated with an increase in GDP growth relative to financial status quo. The right hand panel in Figure 9.3 uses the same approach to look at employment. As is the case for GDP, employment growth is on average weaker after the financial status quo, while it is slightly stronger after financial reform, suggesting that financial reforms coincide with an accelerating pattern in employment.⁷ If both GDP and employment display an accelerating pattern following financial reform relative to financial status quo, what about labour productivity? Table 9.1 suggests that financial reforms have, on average, a relatively muted effect on ⁷ Note that the difference between the dark grey and the light grey bars in Figure 9.3 and similar graphs is not directly interpretable because of the difference in the lengths of the time periods considered before financial reform/status quo (3 years) and after financial reform/status quo (5 years). Only the differencein-difference effect, i.e. the difference between the medium grey bars can be properly interpreted.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



Table 9.1 Financial reform and change in average economic performance Change in Growth variable

GDP

Financial status quo Financial reform Prob. negative diff-in-diff

Employment


0.55 0.08 7.60

Labour Productivity


0.27 0.21 3.45


0.26
0.13 35.5

Note: Numbers in the last row represent the probability that a positive difference-in-difference effect can be rejected. A positive difference-in-difference effect is the case where the change in the growth variable after reform is larger than the change in the growth variable after status quo. Source: GGDC and IMF financial reform databases, authors’ calculations.

Productivity growth

Productivity growth

Advanced economies

Status quo

Emerging market economies

Financial reform

0.02

0.03

0.015

0.02

0.01

Status quo

Financial reform

0.01

0.005 0

0

–0.01

–0.005 before

after

change

 . Productivity growth: advanced economies and emerging market economies Notes: The black (grey) bar in the left hand panel represents the 3-year average growth in aggregate labour productivity preceding (following) financial reform or financial status quo episodes in advanced economies. The black (grey) bar in the left hand panel represents the 3-year average growth in aggregate labour productivity preceding (following) financial reform or financial status quo episodes in emerging market economies. The light grey bars in both panels represent the change in average growth rates. Sources: GGDC and IMF financial reform databases, authors’ calculations.

productivity growth, which more formal statistical tests tend to confirm. While the hypothesis of a positive difference-in-difference effect cannot be rejected for the case of GDP and even more clearly so for employment, it is not the case for labour productivity. This is because the difference between labour productivity growth before and labour productivity growth after financial reform is not statistically significantly larger than the same difference between labour productivity growth before and labour productivity growth after financial status quo. Running this exercise for advanced and emerging market economies separately provides interesting and different insights. Labour productivity tends to decelerate more strongly following financial reforms in advanced economies⁸ (Figure 9.4, left ⁸ The advanced economies are Denmark, Spain, France, United Kingdom, Italy, Japan, Netherlands, Sweden, and the USA; the emerging market economies are Argentina, Brazil, China, Costa Rica, Ghana,

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

hand panel), with a loss in labour productivity growth around 0.4 percentage points. Conversely, an opposite result holds for emerging market economies where financial reforms coincide with an acceleration in labour productivity (Figure 9.4, right hand panel), about 0.3 percentage points.

9.4 D L P G

.................................................................................................................................. These preliminary results suggest that finance affects productivity in very different ways depending on how advanced countries are. To push the investigation further, we choose to decompose aggregate labour productivity growth using the sectoral breakdown of output and employment. To put it simply, aggregate labour productivity can essentially fluctuate either because sectoral labour productivity is fluctuating in a similar fashion across all sectors in the economy, or because production factors, the labour force in particular, is shifting across sectors with heterogeneous productivity gains. More formally, following Borio et al. (2016), let us denote respectively Y and L, aggregate output and aggregate employment, and their respective sector-level counterparts ys and ls. Then, using the following definitions: ð ð Y ¼ ys ds and L ¼ ls ds’ ð1Þ we can write aggregate labour productivity as the sum of two terms: one denoting average labour productivity across sectors, and another one denoting the covariance across sectors between sectoral labour productivity and sector employment share.   ð Y ys ys ls ¼ ds þ cov ; Ð ð2Þ L ls ls li di As is clear from this decomposition, when employment shares are randomly allocated across sectors, aggregate labour productivity is just equal to the simple average of sector-level labour productivities. Conversely, when sectors with higher productivities also tend to be larger, then the covariance term is positive and aggregate labour productivity is higher than the simple average of sector level labour productivities. Using a similar approach and denoting αs sector s share in total output, we can write the growth rate of labour productivity as Ð   ð Δls = li di ΔY=L Δys =ls Ð ¼ 1þ 1þ αs ds ð3Þ 1þ Y=L ys =ls ls = li di

Indonesia, India, Kenya, Mexico, Malaysia, Philippines, Senegal, Thailand, Taiwan province of China, and South Africa.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



Hence using the definition of a covariance, we can decompose aggregate labour productivity growth as the sum of two terms: the first is the product of average sector-level productivity growth and average growth in sector-level employment shares. Given that, in practice, average growth in sector-level employment shares across sectors is close to zero, this term essentially depends on average sector-level productivity growth, with the important qualification that sector-level productivity growth is weighted using output shares. Ð Ð      ð  ð  Δls = li di Δls = li di ΔY=L Δys =ls Δys =ls Ð Ð ¼ 1þ αs αs ð4Þ ds : 1þ ds þcov ; 1þ 1þ Y=L ys =ls ys =ls ls = li di ls = li di |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} common component

allocation componentðSCÞ

The second term measures the covariance across sectors between growth in sector-level employment shares and growth in sector-level output size-weighted labour productivity. As such it measures alternatively whether productivity tends to grow more quickly in sectors where employment is expanding faster or whether stronger employment growth takes place in sectors with higher productivity gains. For instance, if employment tends to grow in sectors where productivity is shrinking, then the covariance term would be negative and hence would act as a drag on aggregate productivity growth. Conversely if the employment share of sectors with high productivity gains is increasing, then the covariance term is positive and would act to lift aggregate productivity growth. In what follows we will call this covariance term Structural Change (SC), that is, the contribution of labour reallocation across sectors to aggregate productivity growth. Now having this decomposition, we can revisit the evidence produced so far and determine whether heterogeneity in the sources of labour productivity growth can actually account for differences in the productivity response to financial reforms.⁹ Figure 9.5 actually shows that in advanced economies, financial reforms tend to be associated with decelerating labour productivity, essentially because the common component tends to weaken.¹⁰ It is true that the contribution of the allocation component goes up, but this is not enough to compensate for the negative trend on the common component (Figure 9.5, left hand panel). In contrast, in emerging market economies, there is a clear increase in the allocation component following financial reform while the common component does not change significantly, comparing status quo to financial reform episodes (Figure 9.5, right hand panel). Financial reforms in emerging markets are therefore associated with structural change that ⁹ This exercise requires using annualized growth rates to compute meaningful changes in the productivity growth components. As a result, the allocation and common components used here are simply first order approximations of the ‘true’ allocation and common components. The sum of these ‘annualized’ common and allocation components therefore does not sum up exactly to the annualized labour productivity growth numbers, used in previous graphs and tables. ¹⁰ The change in the common component after financial reform is significantly lower than the change after financial status quo, the probability of wrongly accepting this hypothesis, being around 1.7%.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .    Productivity growth decomposition

Productivity growth decomposition

Advanced economies

Emerging market economies

0.04

0.02

0.03

0.015

0.02

0.01

0.01 0

0.005 0

–0.01 before

after

Status quo

before

after

Financial reform Common component

before

after

Status quo

before

after

Financial reform

Allocation component

 . Productivity growth decomposition: advanced economies and emerging market economies Notes: The black (light grey) bars in the left hand panel represent the common component (the allocation component) of the 3-year average growth in aggregate labour productivity preceding (following) financial reform or financial status quo episodes in advanced economies. The black (light grey) bars in the right hand panel represent the common component (the allocation component) of the 3-year average growth in aggregate labour productivity preceding (following) financial reform or financial status quo episodes in emerging market economies. Sources: GGDC and IMF financial reform databases, authors’ calculations.

fosters stronger labour productivity growth and hence raises the growth potential of these economies.

9.5 U  D-- R A

.................................................................................................................................. Now we can turn to a proper difference-in-difference analysis. For this purpose we propose to estimate regression (1), where the dependent variable—denoted ½ΔA lnðy=lÞit —is labour productivity growth in country i after year t financial reform or status quo. On the right hand side, we include labour productivity growth in country i prior to year t financial reform/status quo ½ΔB ln ðy=lÞit and our treatment variable Fit indicating whether year t in country i was a financial reform or a financial status quo year. To be comprehensive, we will be using alternatively, the qualitative variable which takes a value  for financial reform years and  for financial status quo, or the quantitative variable indicating the change in the financial reform index during the -year period of reform or status quo. Finally we saturate this regression with country and time fixed effects, to focus on the within-country effect of country-specific financial reforms. ½ΔA lnðy=lÞit ¼ αi þ αt þ β½ΔB ln ðy=lÞit þ γ Fit þ εit

ð5Þ

Column (i) in Table 9.2 shows that the treatment variable has a positive and significant coefficient, meaning that financial reforms are indeed associated with an acceleration of

Table 9.2 Change in labour productivity growth after financial reform Dependent variable: Subsequent labour productivity growth

Past productivity growth Common component

(i)

(ii)

0.277** (0.125)

0.278** (0.129)

Allocation component Financial reform dummy Financial reform dummy in AE

4.156*** (1.292)

Financial reform dummy in EME

0.339*** (0.109) 0.920*** (0.245) 5.035*** (1.195)

(iv)

0.357*** (0.112) 0.980*** (0.267) 6.129*** (1.731) 3.965** (1.773)

Financial reform index

(v) 0.263* (0.140)

0.246*** (0.0729)

Financial reform index Observations R-squared

116 0.214

115 0.219

116 0.292

115 0.303

110 0.251

(vi) 0.253* (0.143)

0.189 (0.132) 0.278*** (0.0966) 110 0.254

(vii)

(viii)

0.352*** (0.131) 0.884*** (0.298)

0.353** (0.133) 0.888*** (0.313)

0.259*** (0.044)

110 0.316

0.264* (0.134) 0.257*** (0.0795) 110 0.316

Note: Each column represents the estimation of regressions where the dependent variable is the 5-year growth in labour productivity (Subsequent labour productivity growth) following a financial reform or status quo year. Independent variables are the 3-year growth in labour productivity prior to the financial reform or status quo year (Past productivity growth), the common and allocation component of productivity growth (Common component and Allocation component ), a dummy variable indicating whether the observation corresponds to financial reform or to a financial status quo (Financial reform dummy ), the financial reform dummy interacted with the dummy for advanced economies (financial reform dummy in AE ), or the dummy for emerging market economies (financial reform dummy in EME ), the number of areas of financial reform during the 3-year period prior to the financial reform or financial status quo year (financial reform index), the financial reform index variable interacted with the dummy for advanced economies (financial reform index in AE ), or the dummy for emerging market economies (financial reform index in EME ). All estimations include country and time fixed effects. Robust standard errors in parentheses. Statistical significance at 1%/5%/10% is indicated with ***/**/*. Source: GGDC and IMF financial reform databases, authors’ calculations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Financial reform index

5.293*** (1.764) 3.110 (1.877)

(iii)

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

labour productivity. Interestingly, column (ii) suggests that financial reforms have a more significant effect in advanced economies than they do in emerging market economies, a result we will investigate more deeply in Section 9.6. Quantitatively, using estimates from columns (i) and (ii), labour productivity typically accelerates by 0.65–0.9 pp per year over five years after a financial reform compared with a situation of financial status quo. Regressions (i) and (ii) also suggest that there is some time persistence in labour productivity growth, meaning that the full benefit of financial reforms that countries eventually earn, in terms of increased productivity is actually larger that the numbers given above. In columns (iii) and (iv), we allow different sources of past labour productivity growth to affect subsequent productivity growth differently. Namely, we use the decomposition derived in equation (4) which we re-write in this case as ½ΔB ln ðy=lÞit ¼ comBit þ allocBit and estimate now the following regression: ½ΔA lnðy=lÞit ¼ αi þ αt þ βc comBit þ βc allocBit þ γFit þ εit

ð6Þ

Two main conclusions can be drawn comparing (i) and (ii) to (iii) and (iv). First, financial reforms are followed by significantly stronger labour productivity growth, this holds in both advanced and emerging market economies. Second, the estimated coefficient for the allocation component is significantly larger than the one estimated for the common component (roughly 3 times larger). Hence the source of past productivity growth also matters: productivity improvements stemming from labour reallocation tend to be followed by much stronger productivity growth than productivity improvements that are common across sectors. Columns (v)–(viii) run the same exercise to that run in columns (i)–(iv), except that we now use the change in the financial liberalization index (denoted financial reform index in Table 9.2), instead of the dummy for financial reform. Except for one qualification, the results tend to be very similar. A larger increase in the financial liberalization index is followed by significantly stronger productivity growth (columns (v) and (vii)). Second, sources of labour productivity growth still matter: an increase in labour productivity growth coming from the allocation component is followed by significantly stronger productivity growth than an increase in labour productivity growth coming from the common component (columns (vii) and (viii)). Lastly, a difference arises as to the specific effect of financial reform in advanced vs. emerging market economies. While the results are more significant for advanced economies when using the qualitative variable indicating financial reform or financial status quo, the opposite holds when using the quantitative variable measuring the change in the financial liberalization index. This is because the results tend to be more precisely estimated for emerging market economies. Next, we want to understand whether the positive effect of financial reforms comes from an improvement in the common component or from the allocation component of labour productivity growth. To do so, we estimate specifications (5) and (6), replacing the left hand side variable alternatively with the common and the allocation component of subsequent labour productivity growth following the decomposition (4).

Table 9.3 Change in labour productivity growth common component and financial reform Dependent variable: Subsequent common component of labour productivity growth

Past productivity growth Common component

(i)

(ii)

0.376** (0.145)

0.388** (0.146)

Allocation component Financial reform dummy Financial reform dummy in AE

5.063*** (1.447)

Financial reform dummy in EME

0.394*** (0.146) 0.565* (0.318) 5.323*** (1.427)

(iv)

0.418*** (0.147) 0.655* (0.341) 4.917** (1.965) 5.622** (2.457)

Financial reform index in AE

(v)

(vi)

0.371** (0.148)

0.334** (0.151)

0.316*** (0.0815)

Financial reform index in EME Observations R-squared

116 0.260

115 0.263

116 0.266

115 0.273

110 0.335

0.124 (0.128) 0.426*** (0.0871) 110 0.363

(vii)

(viii)

0.397** (0.150) 0.556 (0.346)

0.346** (0.156) 0.413 (0.380)

0.320*** (0.0797)

110 0.339

0.133 (0.135) 0.423*** (0.0907) 110 0.364

Note: Each column represents the estimation of regressions where the dependent variable is the common component of the 5-year growth in labour productivity (Subsequent common component of labour productivity growth) following a financial reform or status quo year. Independent variables are the 3-year growth in labour productivity prior to the financial reform or status quo year (Past productivity growth), the common and allocation component of productivity growth (Common component and Allocation component), a dummy variable indicating whether the observation corresponds to a financial reform or to a financial status quo (Financial reform dummy), the financial reform dummy interacted with the dummy for advanced economies (financial reform dummy in AE) or the dummy for emerging market economies (financial reform dummy in EME ), the number of areas of financial reform during the 3-year period prior to the financial reform or financial status quo year (financial reform index), the financial reform index variable interacted with the dummy for advanced economies (financial reform index in AE), or the dummy for emerging market economies (financial reform index in EME). All estimations include country and time fixed effects. Robust standard errors in parentheses. Statistical significance at 1%/5%/10% is indicated with ***/**/*. Source: GGDC and IMF financial reform databases, authors’ calculations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Financial reform index

4.599** (2.049) 5.298** (2.436)

(iii)

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

Table 9.3, which focuses on the determinants of the common component of labour productivity growth, suggests that irrespective of how they are precisely measured, financial reforms are followed by significantly stronger labour productivity growth (columns (i), (iii), (v), and (vii)). Moreover, as was the case in Table 9.2, separating the effect of reforms in advanced and emerging market economies provides mixed results: using the qualitative variable for reforms suggests that advanced and emerging market economies are very similar. However, using the change in the financial liberalization index as a measure of financial reforms suggests a more significant effect in emerging market economies, implying that productivity improvements that are common to all sectors of the economy tend be stronger following financial reforms in emerging market economies (columns (vi) and (viii)). To some extent, this result is not altogether surprising if we consider that in advanced economies, sectors are more heterogeneous in their distance to frontier, with some sectors typically being at the frontier (usually in manufacturing) and some others lying pretty far from their respective frontiers (usually non-tradable sectors). In this case, engineering cross-sectoral productivity improvements can be more difficult than in the case of emerging market economies where all sectors tend to have some distance to their frontier. Let us now turn to the effect of financial reforms on the allocation component of productivity growth. Using a similar methodology to the one used up to now, two striking conclusions can be drawn from Table 9.4. First, financial reforms captured with a dummy variable do not have any significant effect on the subsequent allocation component (columns (i)–(iv)). This nonsignificance result holds irrespective of whether we estimate a unique or a different effect for advanced and emerging market economies. Interestingly, however, the financial reform dummy takes a negative coefficient for emerging market economies, thereby possibly implying that financial reforms are actually followed by labour reallocation in favour of sectors with relatively low productivity gains and to the detriment of sectors with relatively high productivity gains. Second, when measuring financial reforms with the change in the financial liberalization index, there is strong evidence that in emerging market economies, financial reforms are detrimental to the allocation component of labour productivity growth (columns (vi) and (viii)), while in advanced economies, there is some evidence of a positive effect, meaning that financial reforms are followed by labour reallocation in favor of sectors with relatively high productivity gains. This therefore suggests that financial reforms can have a positive effect on structural change understood as a higher contribution of labour reallocation to productivity growth. However, this benefit seems to accrue only in advanced economies. In emerging market economies, the evidence points to a lower contribution if anything. This effect has been somehow alluded to earlier (McMillan and Rodrik 2011). Hence, summarizing the empirical results from Tables 9.2, 9.3, and 9.4, it appears that in emerging market economies, the bulk of the improvement in labour productivity growth which follows financial reforms comes from the common component, not

Table 9.4 Change in labour productivity growth allocation component and financial reform Dependent variable: Subsequent allocation component of labour productivity growth

Past productivity growth Common component

(i)

(ii)


0.098 (0.092)


0.110 (0.088)

Allocation component Financial reform dummy Financial reform dummy in AE


0.907 (0.991)

Financial reform dummy in EME


0.055 (0.079) 0.354** (0.160)
0.287 (0.874)

(iv)


0.061 (0.077) 0.325* (0.166) 1.212 (1.248)
1.658 (1.510)

Financial reform index

(v)

(vi)


0.107 (0.095)


0.081 (0.095)


0.070 (0.059)

Financial reform index Observations R-squared

116 0.063

115 0.106

116 0.217

115 0.237

110 0.103

0.0650 (0.0665)
0.148** (0.070) 110 0.175

(vii)

(viii)


0.046 (0.082) 0.327* (0.175)

0.006 (0.070) 0.474*** (0.161)


0.060 (0.054)

110 0.229

0.131* (0.066)
0.167*** (0.059) 110 0.361

Note: Each column represents the estimation of regressions where the dependent variable is the allocation component of the 5-year growth in labour productivity (Subsequent allocation component of labour productivity growth) following a financial reform or status quo year. Independent variables are the 3-year growth in labour productivity prior to the financial reform or status quo year (Past productivity growth), the common and allocation component of productivity growth (Common component and Allocation component), a dummy variable indicating whether the observation corresponds to a financial reform or to a financial status quo (Financial reform dummy), the financial reform dummy interacted with the dummy for advanced economies (financial reform dummy in AE), or the dummy for emerging market economies (financial reform dummy in EME ), the number of areas of financial reform during the 3-year period prior to the financial reform or financial status quo year (financial reform index), the financial reform index variable interacted with the dummy for advanced economies (financial reform index in AE ), or the dummy for emerging market economies (financial reform index in EME). All estimations include country and time fixed effects. Robust standard errors in parentheses. Statistical significance at 1%/5%/10% is indicated with ***/**/*. Source: GGDC and IMF financial reform databases, authors’ calculations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Financial reform index

0.693 (1.092)
2.188 (1.641)

(iii)

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

from the allocation component. By contrast, the evidence for advanced economies points to the fact that productivity acceleration after financial reforms is due to both productivity growth components.

9.6 C   P  F D

.................................................................................................................................. In emerging economies why would financial reforms come instead from an improvement in the common component of labour productivity growth and much less from the allocation component? Why do financial reforms favour economy-wide productivity improvements as opposed to productivity enhancing labour reallocations? There could be several reasons for this. In many emerging market economies, the economy’s industrial (and hence employment) structure can be biased to favour low productivity gain sectors (e.g. large agroindustrial sectors, large commodity exporters in natural resources such as oil or mining, etc.). In those cases, financial reform might have a low, insignificant, or even negative effect on productivity. Another reason may reside in the fact that domestic capital markets might not be sufficiently mature. Oligopolistic behaviour, a lack of competition might result in the difficulty to identify, finance, and/or implement new investment projects with positive higher returns. Along the lines of the ‘binding constraint’ argument, in many emerging markets, there are different areas within financial reforms that might have different pay-offs. The examples that are captured in our data set do not necessarily produce the proper dichotomy between the ‘reform’ and ‘status quo’ episodes. Finally, similarly to the previous argument, there are also a number of distortions that might supersede the results of financial reform. For example, many emerging markets do have significant portions of their credit being directly or administratively allocated, there are also a number of restrictions related to capital account regulations. Last but not least, emerging economies are the ones that display the higher volatility in their macroeconomic and policy environment (e.g. higher fiscal deficits, higher inflation, higher debt levels, etc.). Therefore, it might be more difficult to capture the effects of financial reforms using our difference-in-difference evaluation technique. While it is beyond the scope of this chapter to carry out a proper investigation of all these possibilities, we can still draw some evidence relating to the domestic capital market hypothesis, namely that financial market depth may help to account for how labour productivity growth responds to financial reforms differently in advanced and in emerging market economies. For this purpose, we consider an extended specification which allows the estimated coefficient for financial reforms to depend on the amount of credit extended to the private sector, arguably a measure of financial market depth, which we draw from the World Bank financial structure and development database

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



(Beck et al. 2001). Denoting Cit the log of private credit to GDP in country i in year t, and Fit.Cit the interaction between the financial reform indicator and the log of private credit to GDP, we now estimate the following specification:   comAit ¼ αi þ αt þ βc comBit þ βc allocBit þ γFit þ δCit þ θFit :Cit þ εit ð7Þ allocAit In these regressions, the effect of financial reforms on the different productivity components is γ þ θ:Fit . The results provided in Table . yield two main conclusions. First, financial reforms have a positive and significant effect on the common component

Table 9.5 Financial reform, private credit, and labour productivity growth components Common component Dependent variable: Common component Allocation component Financial reform index in AE Financial reform index in EME Financial reform index

(i)

(ii)

(iii)

(iv)

0.346** (0.156) 0.413 (0.380) 0.133 (0.135) 0.423*** (0.0907)

0.307** (0.153) 0.483 (0.358)

0.00649 (0.0704) 0.474*** (0.161) 0.131* (0.0661)
0.167*** (0.0592)

0.0754 (0.0621) 0.212 (0.168)

Financial reform index and private credit to GDP Private credit to GDP Observations R-squared

Allocation component

110 0.217

1.234*** (0.415)
0.236** (0.0960)
0.00697 (0.0418) 105 0.237

110 0.103


0.766*** (0.250) 0.171*** (0.0627)
0.00445 (0.0125) 105 0.175

Note: Each column represents the estimation of regressions where the dependent variable is either the common component of the 5-year growth in labour productivity (Common component, in cols. (i) & (ii)) or the allocation component of the 5-year growth in labour productivity (Allocation component, in cols. (iii) & (iv)) following a financial reform or status quo year. Independent variables are the 3-year common and allocation component of productivity growth prior to financial reform or status quo year (Common component and Allocation component), the number of areas of financial reform during the 3-year period prior to the financial reform or financial status quo year (financial reform index), the financial reform index variable interacted with the dummy for advanced economies (financial reform index in AE ) or the dummy for emerging market economies (financial reform index in EME ), the log level of private credit to GDP on the financial reform or status quo year (private credit to GDP ), and the product of the financial reform index with the log level of private credit to GDP (Financial reform index and private credit to GDP ). All estimations include country and time fixed effects. Robust standard errors in parentheses. Statistical significance at 1%/5%/10% is indicated with ***/**/*. Source: GGDC, World Bank financial structure and development and IMF financial reform databases, authors’ calculations.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

of labour productivity growth (column (i)). However, this positive effect tends to dampen in countries with higher private credit to GDP (column (ii)) as the interaction term carries a negative and significant coefficient. Hence, it can be argued that in advanced economies, where financial markets are deeper, financial reform carries less of a benefit to the economy. Conversely, in emerging market economies, financial reform can have a larger pay-off, which is consistent with our earlier finding that the effect of financial reforms on the common component of labour productivity growth is larger and more precisely estimated for emerging market economies. Second, turning to the allocation component, things start to reverse, as the financial reform variable coefficient is negative while the interaction term carries a positive and highly significant coefficient (column (iv)). Financial reforms therefore tend to be followed by a higher contribution of the allocation component only to the extent that private credit to GDP is sufficiently large. In emerging market economies lacking deep capital market, the effect would be either non-significantly different from zero, or even negative. It can therefore be argued that in emerging market economies, financial reforms are followed by labour reallocation acting as a drag on aggregate productivity growth, precisely because of the lack of deep financial markets. Financial depth can therefore be interpreted as an (inverse) proxy for the existence of economy-wide productivity gains. With deep financial markets, these ‘low-hanging’ fruits are already exhausted, hence the negative correlation between private credit to GDP and the benefit of financial reforms on the common component of productivity growth (black line in Figure 9.6). Conversely, financial depth can also be considered as 2 1.5 1 0.5 0

0

1

2

3

4

5

6

–0.5 –1 –1.5 Log of private credit to GDP Change in the common component

Change in the allocation component

 . Log of private credit to GDP Notes: The solid black (solid grey) line represents the estimated change after a financial reform in the common component (the allocation component) of aggregate labour productivity growth as a function of the log of private credit to GDP at the time of the reform. Dashed lines represents the confidence bands around estimated changes at the 95 per cent level.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



a proxy for the financial sector’s ability to identify sectors with significant growth potential. As a result, the allocation component can benefit from financial reform only to the extent that financial markets are sufficiently developed (grey line in Figure 9.6). And here comes a chicken and egg problem for emerging market economies. While financial reforms are considered as a possible blue print to deepen financial markets, it is the case that deep financial markets are actually a necessary condition for financial reforms to deliver benefits in terms of productivity enhancing structural change. Still, the good news is that overall, the pay-off to financial reforms, in terms of enhanced labour productivity growth, is somewhat larger with less developed capital markets (summing the black and the grey lines in Figure 9.6), suggesting that for emerging market economies, financial reforms are not bad after all.

9.7 C

.................................................................................................................................. This chapter attempts to shed some light on the question ‘Do financial reforms contribute to structural change?’ We show some evidence of the positive impact of financial reforms on the contribution of structural reform to productivity growth. Our analysis is based on a cross country, cross time difference-in-difference approach. It compares before and after reforms against a ‘status quo’ situation. Our preliminary results show that financial reforms do have a significant effect on productivity growth and that part of this effect goes through the contribution of structural change. As stated earlier, we are not passing judgement on the nature of the financial sector reforms that are collected in our data set and hence not drawing conclusions or policy recommendations regarding the relationship between reforms pertaining to the area of ‘financial liberalization’ and enhancements to productivity. The next steps are precisely about achieving a better understanding of the economic transmission channels through which financial reforms affect the contribution of structural change. That could include refining the measurements of ‘financial reforms’, their ‘intensity’, and using other evaluation techniques (e.g. country studies, macroeconomic models, etc.)

 1

..................................................................................................................................

T M C   D Countries in the sample: Argentina, Brazil, China, Costa Rica, Denmark, Spain, France, UK, Ghana, Indonesia, India, Italy, Japan, Kenya, Mexico, Malaysia, Netherlands, Philippines, Senegal, Sweden, Thailand, Taiwan, province of China, United States, and South Africa.

Country Argentina Brazil China Costa Rica Denmark Spain France UK Ghana Indonesia India Italy Japan Kenya Mexico Malaysia Netherlands Philippines Senegal Sweden Thailand Taiwan USA South Africa

1973 1976 1977 1978 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 0 1

1

1

1

0

0

0

0 0 0

0

0

0

0

0

0

0

0

0

1

1 0

1 0

1

1

0

0

0

0

0

0 1 0 0

0 1 1 0

1 1

0 1

1 1

1

1 1

1

0

0

1

1

1 1

0 0

0

1

0

1 0

0

0 1

0 0

0

Notes: Financial reform years are coded as ones, financial status quo years are coded as zero. Source: IMF financial reform databases, authors’ calculations.

1

0

1

0

1 1

1

0

0

0

0 0

1 0 1

1 0

1

0

0

0

0

1 0 0 0

0

0

1 1

1 1

1 1 0 0

0

0

1

1

1 0

0

0

0

1 0 1 0 1 1 0

0 0 0

0

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Table 9A.1 Financial reform and financial status quo across countries

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



Table 9A.1 denotes financial reform years as defined in Section 9.2 in each country of the sample with a ‘1’ for this country/year pair. It denotes the financial status quo years as defined in Section 9.2 in each country of the sample with a ‘0’ for this country/year pair. The financial reform database covers seven different areas of financial regulation. More details on each of the seven different variables are given below: 1. Credit controls. Is there regulation imposing that minimum amounts of credit should be channeled to specific ‘priority’ sectors (e.g. agricultural firms, selected manufacturing sectors, or small-scale enterprises), or to the government to ease budget deficits financing? Are directed credits required to carry subsidized rates? Is there a ceiling on overall credit extended by banks, or on credit to specific sectors? How high are reserve requirements? 2. Interest rate controls. Are there administrative restrictions on lending and/or borrowing interest rates, like ceilings, floors, or interest rate bands? An intermediate regime allows interest rates to fluctuate within a band. Interest rates are considered fully liberalized when all ceilings, floors, or bands are eliminated. 3. Entry barriers. Are there government restrictions to entry into the financial system of new domestic banks or of other potential competitors, for example foreign banks or non-bank financial intermediaries? These may be restricting foreign bank participation; restricting the scope of new banks’ activities or the geographic area where new banks can operate. 4. State ownership in the banking sector. This variable is coded using the share of banking sector assets controlled by state-owned banks. Thresholds of 50 per cent, 25 per cent, and 10 per cent are used to delineate the grades between full repression and full liberalization. 5. Capital account restrictions. These restrictions include multiple exchange rates for various transactions, as well as transaction taxes or outright restrictions on inflows and/ or outflows specifically regarding financial credits. 6. Prudential regulations and supervision of the banking sector. Here, a greater degree of government intervention is coded as a reform. This variable is coded according to the answers to the following questions: Does a country adopt risk-based capital adequacy ratios based on the Basle I capital accord? Is the banking supervisory agency independent from the executive’s influence and does it have sufficient legal power? Are certain financial institutions exempt from supervisory oversight? How effective are on-site and off-site examinations of banks? 7. Securities market policy. This variable captures government policies to either restrict or encourage development of securities markets, e.g. auctioning of government securities, establishment of debt and equity markets, tax incentives to encourage development of these markets, development of depository and settlement systems. Also included here are policies on the openness of securities markets to foreign investors.

R Abiad, Abdul, Enrica Detragiache, and Thierry Tressel, 2008. ‘A New Database of Financial Reforms’, IMF Working Paper No. 266. Acemoglu, Daron and James Robinson, 2012. Why Nations Fail: The Origins of Power, Prosperity, and Poverty, London: Crown Publishing Group.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



    .   

Aghion, Philippe, and Peter Howitt, 1992. ‘A Model of Growth Through Creative Destruction’, Econometrica, 60, pp. 323–51. Arcand Jean-Louis, Berkes Enrico, and Ugo Panizza, 2012. ‘Too Much Finance?’ IMF Working Paper No. 161. Beck, Thorsten, Ross Levine, and Norman Loayza, 2000. ‘Financial Intermediation and Growth: Causality and Causes’, Journal of Monetary Economics, 46, pp. 31–77. Beck, Thorsten, Asli Demirgüç-Kunt, and Ross Levine, 2001. ‘The Financial Structure Database’, in A. Demirgüç-Kunt and R. Levine, eds, Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development, Cambridge, MA: MIT Press. Blinder, Alan, 2010. ‘It’s Broke, Let’s Fix It: Rethinking Financial Regulation’, International Journal of Central Banking, pp. 277–330. Borio, Claudio, Kharroubi Enisse, Upper Christian, and Zampolli Fabrizio, 2016. ‘Labor Reallocation and Productivity Dynamics: Financial Causes, Real Consequences’, BIS Working Paper No. 534. Cecchetti, Stephen and Kharroubi Enisse, 2012. ‘Reassessing the Impact of Finance on Growth’, BIS Working Papers No. 381. Claessens S. and L. Laeven, 2005. ‘Financial Dependence, Banking Sector Competition, and Economic Growth’, World Bank Policy Research Working Paper No. 3481. de Vries Gaaitzen, Klaas de Vries, Reitze Gouma, Stefan Pahl, and Marcel Timmer, 2015. ‘GGDC 10-Sector Database: Contents, Sources and Methods’, Groningen Growth and Development Centre. Demirgüç-Kunt, A. and Ross Levine, 2001. Financial Structure and Economic Growth: A Cross-Country Comparison of Banks, Markets, and Development. Cambridge, MA: MIT Press. Demirgüç-Kunt, A. and Vojislav Maksimovic, 1998. ‘Law, Finance, and Firm Growth’, Journal of Finance, 53 (6), pp. 2107–37. Fischer, Stanley, 2014. ‘Financial Sector Reform: How Far Are We? Speech at the Martin Feldstein Lecture’, National Bureau of Economic Research, Cambridge, MA. Available at: https://www.federalreserve.gov/newsevents/speech/fischer20140710a.htm Grossman, Gene, and Elhanan Helpman, 1991. Innovation and Growth in the Global Economy, Cambridge, MA: MIT Press. Hausmann, Ricardo and Dani Rodrik, 2003. ‘Economic Development as Self-Discovery’, Journal of Development Economics, 72, pp. 603–33. Hausmann, R. D. Rodrik, and A. Velasco, 2006. ‘Getting the Diagnosis Right’, Finance and Development, 43 (1). King, Robert G. and Ross Levine, 1993. ‘Financial Intermediation and Economic Development’, in Colin Mayer and Xavier Vives, eds, Capital Markets and Financial Intermediation, London: Centre for Economic Policy Research, pp. 156–89. La Porta, Rafael, Florencio Lopez-de-Silanes, Andrei Shleifer, and Robert W. Vishny, 1998. ‘Law and Finance’, Journal of Political Economy, 106 (6), pp. 1113–55. La Porta, Rafael, Florencio Lopez-de-Silanes, and Andrei Shleifer, 2002. ‘Government Ownership of Banks’, Journal of Finance, 57 (1), pp. 265–301. Levine, Ross, 2005. ‘Finance and Growth: Theory, Mechanisms and Evidence’, in P. Aghion and S. N. Durlauf, (eds), Handbook of Economic Growth, Amsterdam: Elsevier. Lewis, W. Arthur, 1954. ‘Economic Development with Unlimited Supplies of Labor’, Manchester School of Economic and Social Studies, 22, pp. 139–91.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

 ,  



McMillan, Margaret and Dani Rodrik, 2011. ‘Globalization, Structural Change, and Economic Growth’, in M. Bachetta and M. Jansen, eds, Making Globalization Socially Sustainable, New York: International Labor Organization and World Trade Organization. Mohan, Rakesh, 2008. ‘Innovation and Growth-Role of the Financial Sector’, Lecture at Entrepreneurship Development Institute of India. Rajan, Raghuram and Luigi Zingales, 1998. ‘Financial Dependence and Growth’, American Economic Review, 88 (3), pp. 559–86. Rajan, Raghuram and Luigi Zingales, 2003. Saving Capitalism from the Capitalists, Princeton, NJ: Princeton University Press. Rodrik, Dani, 2013. ‘The Past, Present, and Future of Economic Growth Growth’, Challenge, 57 (3), pp. 5–39. Sahay, Ratna, Martin Čihák, Papa N’Diaye, Adolfo Barajas, Ran Bi, Diana Ayala, Yuan Gao, Annette Kyobe, Lam Nguyen, Christian Saborowski, Katsiaryna Svirydzenka, and Seyed Reza Yousefi, 2015. ‘Rethinking Financial Deepening: Stability and Growth in Emerging Markets’, IMF Staff Discussion Notes No. 15. Samargandi, Nahla, Jan Fidrmuc, and Sugata Ghosh, 2015. ‘Is the Relationship Between Financial Development and Economic Growth Monotonic? Evidence from a Sample of Middle-Income Countries’, World Development, 68, pp. 66–81. Solow, Robert M., 1956. ‘A Contribution to the Theory Economic Growth’, Quarterly Journal of Economics, 70, pp. 65–94. Trew A. W., 2006. ‘Finance and Growth: A Critical Survey’, The Economic Record, 82, pp. 481–90. Trew, A, W., 2007. ‘Endogenous Financial Development and Industrial Takeoff ’, CDMA Working Paper Series 200702, Centre for Dynamic Macroeconomic Analysis. Valickova, Petra, Tomas Havranek, and Roman Horvath, 2015. ‘Financial Development and Economic Growth: A meta-analysis’ Journal of Economic Surveys, 29 (3), pp. 506–26. World Bank, 2001. ‘Finance for Growth’, Policy Research Report.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

 ,           ,       ......................................................................................................................

     

10.1 I

.................................................................................................................................. A exponential increase in flows of goods, capital, and ideas is one of the most prominent economic trends in recent decades. A key driver of this phenomenon is cross-border production, investment, and innovation led by multinational corporations (MNCs). Multinational affiliate sales as a share of world GDP have more than doubled in the past two decades, increasing from close to 25 per cent in 1990 to more than 50 per cent in 2014.¹ This explosion of MNC activities is rapidly transforming the global landscape of industrial production, precipitating the emergence of new industrial clusters around the world. Firms that agglomerated in, for example, Silicon Valley and Detroit now have subsidiary plants clustering in Bangalore and Slovakia (termed, respectively, the Silicon Valley of India and Detroit of the East). Understanding the agglomeration, and more broadly the economic geography of MNCs and their role in

¹ UNCTAD, World Investment Report (2015).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



shaping the global industrial landscape as a result has become increasingly important for shaping policies and promoting the benefits of globalization. Recent evidence shows that MNCs exhibit distinct agglomeration patterns (Alfaro and Chen 2014). In contrast to domestic production which emphasizes domestic geography and natural advantage, multinational production stresses foreign market access and international comparative advantage. Moreover, as highlighted in a growing literature (e.g. Helpman et al. 2004; Antràs and Helpman 2004, 2008; Yeaple 2009; Chen and Moore 2010), the economic attributes and organization of multinationals are, by selection, distinctively different from average domestic firms. Their vertically integrated organization, greater productivity, and higher capital- and knowledge-intensities relative to domestic firms all suggest that MNCs are likely to exhibit different agglomeration motives. This chapter seeks to grasp the complexity in the various drivers of MNC location decisions. In particular, we run a horse-race between two distinct economic forces: location fundamentals (also referred to as ‘first nature’) of multinational production (MP) and agglomeration economies (also known as ‘second nature’). The location fundamentals of MP, as stressed in the international trade literature, consist of primarily foreign market access (multinationals choose to produce in large foreign markets to avoid trade costs) and comparative advantage (multinationals locate production in countries with desired factor abundance and low factor prices).² In contrast, agglomeration economies, the study of which dates back to Marshall (1890), stress the benefits of geographic proximity between firms, including lower transport costs between input suppliers and final good producers, labour and capital-good market externalities, and technology diffusion. While existing studies offer separate evidence on the roles of the above two categories of factors in multinationals’ location decisions, how these factors jointly influence multinationals’ global economic geography given their organizational structure and capital- and knowledge-intensive production requires further exploration.³ An evaluation of the patterns and causes of MNC economic geography faces, however, several key challenges. First, the measurement of agglomeration has been a central issue in the economic geography literature. Traditional indices that define agglomeration as the amount of activity located in a particular geographic unit omit agglomerative activities separated by administrative and geographic borders and can be affected by the extent of industrial concentration. Second, distinguishing between the effects of MP’s location fundamentals and agglomeration economies is complicated by the difficulty of measuring the factors quantitatively. Further, their common propensity to lead MNCs to locate next to each other makes it difficult to separate their relative ² While comparative advantage is defined here in the context of neo-classical trade theory, other country factors such as institutional characteristics and physical locations can also play a role in firms’ location decisions and are sometimes considered as part of comparative advantage (see, e.g., Nunn 2007). As described in Subsection 10.4.1, our empirical specification controls for all host-country specific factors when constructing the location fundamentals of multinational production measures. ³ See Alfaro and Chen (2018b) for a review of the literature assessing the role of transportation costs in multinational production.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

effects. Third, identifying the causal effects of location fundamentals and agglomeration economies is a key challenge in empirical analyses of economic geography. Both types of factors can endogenously reflect the patterns of MNC agglomeration. Finally, quantifying the global patterns of MNC economic geography requires cross-country data that document multinational production at the plant, instead of firm, level. To overcome the above challenges, our empirical analysis proceeds in the following steps. First, we quantify the global agglomeration of MNCs using WorldBase, a worldwide plant-level data set that provides detailed location, ownership, and activity information for establishments in more than 100 countries. Its broad cross-country coverage enables us to depict worldwide patterns of MNC economic geography. Moreover, the data set’s detailed location and operation information for over 43 million plants, including multinational and domestic, offshore and headquarters establishments, makes it possible to compare the geographic patterns of different types of establishments. The physical location of each establishment allows us to construct indices of agglomeration using precise latitude and longitude codes for each plant and the distance and trade cost between each pair of establishments. Second, we construct a spatially continuous index of agglomeration for pairwise industries (also referred to as coagglomeration).⁴ We obtain latitude and longitude codes for each establishment in the data based on plant-level physical location information and compute not only the distance but also the trade cost that accounts for other forms of trade barriers between each pair of establishments. Following an empirical methodology introduced by Duranton and Overman (2005) and extended in Alfaro and Chen (2014), we then employ a Monte-Carlo approach that compares the actual geographic density of plants in each industry pair with counterfactual densities. This procedure separates agglomeration from the general geographic concentration of multinationals and deals with the effect of industrial concentration. Industry pairs that exhibit greater geographic density than the counterfactuals are considered to exhibit significant evidence of agglomeration. As in Alfaro and Chen (2014), we construct the agglomeration index for each pairwise industry to help disentangle the effects of location fundamentals and various agglomeration forces. As noted by Ellison et al. (2010), while location fundamentals and agglomeration economies tend to predict agglomeration among firms in the same industry, their predictions of which industry pairs should agglomerate vary significantly. Compared to firms in the same industries, firms from different industry pairs often exhibit greater variation in their relatedness in production, factor markets, and technology space, thereby displaying different agglomeration incentives.⁵ Exploring the

⁴ We use the term ‘agglomeration’ broadly to refer to both within- and between-industry agglomeration (the latter sometimes referred to as ‘coagglomeration’). The broad usage of the term ‘agglomeration’ is fairly common in the literature. ⁵ For example, firms in the automobile industry may agglomerate because of both location fundamentals and any of the agglomeration economies whereas firms in the automobile and steel industries are likely to agglomerate mainly because of their production linkages.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



pairwise-industry agglomeration of MNCs thus makes it possible to separate the effects of location fundamentals and the various agglomeration economies. Third, after computing the actual agglomeration index of MNCs, we construct an expected index of MNC agglomeration to capture the effect of location fundamentals. This index reflects the geographic distribution of MNC plants predicted exclusively by the location fundamentals of multinational production including, among others, foreign market size, trade costs, and comparative advantage. Specifically, we invoke a twostep procedure. In the first step, we estimate a conventional gravity-type MP equation and examine the effects of market access and comparative advantage as well as other location factors in multinationals’ location decisions. Based on the estimates, we obtain the location patterns of MNC plants predicted by the location fundamentals and, in the second step, construct an index of agglomeration using the predicted, instead of actual, locations. This index represents the expected degree of pairwise-industry agglomeration based on industry pairs’ common location fundamentals. Fourth, controlling for the agglomeration predicted by location fundamentals and all industry specific factors, we examine the degree to which proxies of agglomeration forces, including between-industry input–output linkages, labour demand similarity, technology spillover, and a new measure of capital-good market externality, explain the variations in the agglomeration index of multinational firms. We construct the proxies of agglomeration forces using lagged, disaggregated US industry account data to mitigate the potential reverse causality concern, as it is not very likely that US industries’ production, factor, and technology linkages are a result of worldwide MNCs’ agglomeration patterns. To further alleviate concerns of endogenous agglomeration economy measures, we examine regional agglomeration patterns from which the United States is excluded. If US domestic industry-pair relationships could be affected by the agglomeration of MNCs in the United States, then one would expect that the former would not be affected by the agglomeration of MNCs located in other regions, such as Europe.⁶ Our analysis presents a rich array of new findings that shed light on the global agglomeration of MNCs. First, the location fundamentals of multinational production, although playing a significant and vital role, are not the only driving forces in the patterns of MNC offshore agglomeration. As shown in Alfaro and Chen (2014), agglomeration economies, especially capital-good market externality and technology diffusion, are crucial determinants of MNCs’ overseas location decisions. When comparing the relative importance of location fundamentals and agglomeration economies, however, we find the effect of location fundamentals to exceed the cumulative impact of agglomeration forces. A one-standard-deviation increase in the former is associated with a 0.31 standard-deviation increase in the extent of MNCs’ offshore agglomeration ⁶ In Alfaro and Chen (2014), we investigate the process of agglomeration. Exploring the dynamics in MNCs’ offshore agglomeration sheds light on the formation of MNC clusters and mitigates the possibility of reverse causation between our measures of location fundamentals, agglomeration economies, and MNCs’ agglomeration patterns.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

at the 200 km level, whereas the cumulative effect of agglomeration economies is around 0.17. Second, as suggested by the agglomeration patterns, the relative importance of location fundamentals and agglomeration economies varies significantly between MNC offshore and domestic plant agglomeration and between MNC offshore and headquarters agglomeration. Comparing the agglomeration of MNC offshore and domestically owned plants, we find MNC plants, reflecting their high capital- and innovationintensities, to be significantly more influenced by capital-good market and technological agglomeration factors. The under-provision of capital goods in many host countries increases MNCs’ incentives to locate in proximity to one another overseas and take advantage of agglomeration economies. Moreover, location fundamentals and capital-good market externality exert a stronger effect on the offshore agglomeration of MNC subsidiary establishments, while technology diffusion and labour market externalities are the leading forces behind the agglomeration of headquarters. Vertical production linkages, in contrast, matter for offshore clustering only. These results are consistent with the increasing segmentation of activities within the boundary of multinational firms, in particular, the market-seeking and the inputsourcing focuses of offshore production and the emphasis of headquarters on knowledge intensive activities such as R&D, management, and services. The findings also remain largely robust when examining regional agglomeration patterns, and restricting the analysis to Europe which yield additional insights. The rest of the chapter is organized as follows. Section 10.2 reviews the related literature. Section 10.3 discusses the methodology used in this chapter to construct pairwise-industry agglomeration indices. Section 10.4 describes the methodology used to measure location fundamentals and agglomeration economies. Section 10.5 describes the cross-country establishment data. Section 10.6 reports the stylized facts and econometric evidence on the determinants of MNC economic geography. Section 10.7 presents additional analysis that examines agglomeration in Europe. The last section concludes.

10.2 R L

.................................................................................................................................. Our work builds on three broad strands of literature. First, an extensive literature in international trade has shed important theoretical and empirical light on the role of location fundamentals in MNCs’ decisions to invest abroad. Two main motives of foreign investment have been stressed by studies in this literature. First, firms may choose to produce overseas to avoid trade costs. This strategy, referred to as the market access (or tariff jumping) motive, leads firms to duplicate production processes in countries (see, e.g., Markusen 1984 and Markusen and Venables 2000). Second, firms may choose to locate different stages of production in countries where the factor used intensively is abundant. This strategy is referred to as the comparative advantage

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



motive (see, e.g., Helpman 1984). These two motives, leading to horizontal and vertical FDI respectively, have been synchronized in the knowledge-capital model developed by Markusen and Venables (1998) and Markusen (2002) and examined in a number of empirical studies.⁷ This strand of literature provides a theoretical and empirical foundation for the location fundamental portion of our research. A second related literature consists of the extensive body of research in regional and urban economics that has been devoted to evaluating the importance of Marshallian agglomeration forces in domestic economic geography.⁸ Marshall (1890) first introduced the idea that concentrations of economic factors, such as knowledge, labour, and inputs, can generate positive externalities. Data restrictions have impeded the progress of studying economic geography at a global scale; most related research has focused on a geographic area such as the United States (Rosenthal and Strange 2001) or the United Kingdom (Overman and Puga 2009). As in Alfaro and Chen (2014), the Worldbase database allows us to explore the global economic geography of MNCs in this chapter. As noted in Section 10.1, a central issue in studies of agglomeration concerns the measurement of agglomeration. In an influential paper, Ellison and Glaeser (1997) introduce a ‘dartboard’ approach to construct an index of spatial concentration. The index compares the observed distribution of economic activity in an industry to a null hypothesis of random location and controls for the effect of industrial concentration, an issue noted to affect the accuracy of previous indices. Using Ellison and Glaeser’s (1997) index to evaluate the importance of agglomeration forces in explaining the localization of US industries, Rosenthal and Strange (2001) find both labour market pooling and input–output linkages to have a positive impact on agglomeration. Also employing Ellison and Glaeser’s (1997) index, Overman and Puga (2009) examine the role of labour market pooling and input sharing in determining the spatial concentration of UK manufacturing establishments. They find sectors whose establishments experience more idiosyncratic employment volatility and that use localized intermediate inputs are more spatially concentrated. Duranton and Overman’s (2005) study advances the literature by developing a spatially continuous concentration index that is independent of the level of geographic disaggregation. Applying this index, Ellison et al. (2010) employ an innovative empirical approach that exploits the coagglomeration of US industries to disentangle the effects of Marshallian agglomeration economies. They find, as in Rosenthal and Strange (2001), a particularly important role for input–output relationships. Exploring the role of agglomeration economies in MNCs’ location patterns also relates the present chapter to a third strand of literature in international trade that emphasizes the advantage of proximity between customers and suppliers. Several studies (see, e.g., Head et al. 1995; Bobonis and Shatz 2007) show that MNCs with vertical production

⁷ The analyses by Carr et al. (2001), Yeaple (2003a), and Alfaro and Charlton (2009), for example, offer empirical support for both types of motives. ⁸ See Ottaviano and Puga (1998), Duranton and Puga (2004), Head and Mayer (2004), Ottaviano and Thisse (2004), Rosenthal and Strange (2004), Puga (2010), and Redding (2010, 2011) for excellent reviews of these literatures.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

linkages tend to agglomerate regionally within a country. Alfaro and Chen (2014), also using the Dun & Bradstreet WorldBase data, construct a spatially continuous index of agglomeration extending the Duranton and Overman (2005) methodology and analysing the different patterns underlying the global economic geography of multinational and non-multinational firms. We uncover new stylized facts that suggest the offshore clusters of multinationals are not a simple reflection of domestic industrial clusters. We find that agglomeration economies including technology diffusion and capital-good market externality play a more important role in the offshore agglomeration of multinationals than the agglomeration of domestic firms. These findings remain robust when exploring the process of agglomeration. Our current analysis contributes to the above literature and extends it in several important ways. Instead of focusing on domestic agglomeration patterns in industrialized countries like the US, our analysis offers a perspective on the structure of industrial agglomeration at the world and regional level. In particular, we investigate how the most mobile and distinctive group of firms—multinationals—agglomerate domestically and overseas. Importantly, we incorporate the location fundamentals of MNCs into the analysis of agglomeration and develop a new quantitative measure to quantify the role of location fundamentals in MNCs’ spatial concentrations. Further, we evaluate how agglomeration economies, particularly the value of external scale economy in capital goods and knowledge, affect MNCs relative to domestic firms, given MNCs’ vertically integrated organizational form and high demand for capital goods and technologies. Our results show that location fundamentals matter and that capital-good externality and technology diffusion, factors that have not been emphasized in this literature, exert an important effect on the agglomeration of MNCs.

10.3 Q A: M

.................................................................................................................................. In this section, we describe the empirical methodology used to quantify the global agglomeration of multinational firms.

10.3.1 The Agglomeration Index As noted in Head and Mayer (2004), measurement of agglomeration is a central challenge in the economic geography literature.⁹ Continuous effort has been devoted to designing an index that accurately reflects the agglomeration of economic activities. ⁹ More recently, Duranton and Kerr (2015) have also noted the difficulty of obtaining appropriate data in order to measure agglomeration. They emphasize the importance and potential contribution that

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Duranton and Overman (2005) construct this index to measure the significance of same-industry agglomeration in the UK. The index has then been adapted by Ellison et al. (2010) to investigate the coagglomeration of US industries. There are two requirements for the construction of the Duranton and Overman (2005) index. First, availability of physical location information for each establishment at the most detailed level. The WorldBase data set, described in Section 10.5, supplemented by a geocoding software, satisfies this requirement. Second, the empirical procedure adopted to construct the index uses a simulation approach that is computationally intensive, especially for cross-country studies and large data sets. We extend this index to a global context to measure the degree of coagglomeration of multinational firms worldwide. Because it accounts for continuity in space, the index is well suited for cross-country studies as shown in Alfaro and Chen (2014). In this chapter, we also expand the original index’s focus on distance as the main form of trade cost to a measure that accounts for various forms of trade costs (distance, tariffs, etc.).

10.3.2 Empirical Procedure The empirical procedure to construct the index involves three steps. To compare global location patterns of MNC subsidiaries, headquarters, and domestic firms, we repeat the procedure for each type of establishment.

10.3.2.1 Step 1: Kernel Estimator We first estimate an actual geographic density function for each pair of industries. Note that even when the locations of nearly all establishments are known with a high degree of precision (such as in the data we use, as described in Section 10.5), distance (as well as estimated trade cost) is an approximation of the true trade cost between establishments. One source of systematic error, for example, is that travel time for any given distance might differ between low- and high-density areas. Given the potential noise in the measurement of trade costs, we follow Duranton and Overman (2005) in adopting kernel smoothing when estimating the distribution function. Let τij denote the distance between establishment i and j. For each industry pair k and ~k, we obtain a kernel estimator at any point τ (e.g., Kk~k ðτÞ): fk~k ðτÞ ¼

1 Xnk Xnk~ τ  τij  K i¼1 j¼1 nk n~k h h

ð1Þ

where nk and n~k are the number of plants in industries k and ~k, respectively, h is the bandwidth, and K is the kernel function. We use Gaussian kernels with the data

new data will allow for research into agglomeration-related issues. In line with our research, they suggest the need for further research into the role of multinationals and their subsidiaries in agglomeration economies given new availability of data sources.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

reflected around zero and the bandwidth set to minimize the mean integrated squared error.¹⁰ This step generates a kernel estimator for each of the 7,875 (=126125/2) manufacturing industry pairs in our data.¹¹ In addition to estimating the geographic distribution of establishment pairs, we can also treat each worker as the unit of observation and measure the level of agglomeration among workers. To proceed, we obtain a weighted kernel estimator by weighing each establishment by employment size, given by τ  τ  Xnk Xn~k 1 ij ð2Þ r r K fkw~k ðτÞ ¼ Pnk Pn~k i¼1 j¼1 i j h h i¼1 j¼1 ðri rj Þ where ri and rj represent, respectively, the number of employees in establishments i and j. We do this for each of the 7,875 industry pairs.

10.3.2.2 Step 2: Counterfactuals and Global Confidence Bands To obtain counterfactual estimators, we estimate the geographic distribution of the manufacturing multinationals as a whole in order to control for factors that affect all manufacturing multinational plants. We proceed by drawing, for each of the 7,875 industry pairs, 1,000 random samples, each of which includes two counterfactual industries. Given our goal of obtaining, in this step, the overall agglomeration patterns of MNCs, the random samples are drawn from the entire set of MNC establishment locations.¹² Note that to control for the potential effect of industry concentration, it is important that the counterfactual industry in each sample has the same number of observations as the actual data. We then calculate the bilateral distance between each pair of establishments and obtain a kernel estimator, unweighted or weighted by employment, for each of the 7,875,000 samples. This gives 1,000 kernel estimators for each of the 7,875 industry pairs. ¹⁰ Although we follow Duranton and Overman (2005) and Ellison et al. (2010) in obtaining kernel estimators, a less computationally intensive approach that yields similar properties would be to look at cumulative distances. ¹¹ Identical industry pairs (126 observations) are dropped from the analysis because, as explained earlier, we rely on industry-pair variations in relatedness in production, factor demand, and technology to disentangle the effects of location fundamentals and various agglomeration economies. Identical industry pairs exhibit all dimensions of relatedness and lack the needed variation. Moreover, as we explain in Section 10.4, the measures of location fundamentals and agglomeration economies used in this chapter, by design, capture only between-industry relationships. The main empirical analysis is performed at the SIC 3-digit level. This level of industry disaggregation is dominated by the availability of control variables, as described in Section 10.4. ¹² An alternative approach would be to use all existing, including domestic and MNC, establishment locations as the counterfactuals. This would help to control for the effect of general location factors instead of those that affect primarily MNCs’ location decisions. In Subsection 10.6.3, we perform an analysis in that direction by employing domestic establishments as the benchmark and comparing the agglomeration patterns of MNC and domestic plants.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



We compare the actual and counterfactual kernel estimators at various distance thresholds, including 200, 400, 800, and 1,600 kilometers (the maximum threshold being roughly the distance between Detroit and Dallas and between London and Lisbon). We compute the 95 per cent global confidence band for each threshold distance. Following Duranton and Overman (2005), we choose identical local confidence intervals at all levels of distance such that the global f k~k ðτÞ confidence level is 5 per cent. We use to denote the upper global confidence band of industry pair k and ~k. When fkk~ðτÞ> f k~k ðτÞ for at least one τ ∈ 0; T, the industry pair is considered to agglomerate at T and exhibit greater agglomeration than counterfactuals. Graphically, it is detected when the kernel estimates of the industry pair lie above its upper global confidence band.

10.3.2.3 Step 3: Agglomeration Index We now construct the agglomeration index. For each industry pair k and ~k, we obtain   XT f ~ ðτÞ; 0 agglomerationk~k ðTÞ  max f ðτÞ  ð3Þ ~ kk kk τ¼0 or employment-weighted agglomerationwk~k ðTÞ 

  w f w~ ðτÞ; 0 : max f ðτÞ  ~ kk kk τ¼0

XT

ð4Þ

The index measures the extent to which establishments in industries k and ~k agglomerate at threshold distance T and the statistical significance thereof. When the index is positive, the level of agglomeration between industries k and ~k is significantly greater than that of the counterfactuals.

10.4 M L F  A E

.................................................................................................................................. We now turn to economic factors that could systematically account for the observed agglomeration patterns of MNCs. Incorporating the multinational firm theory with the literature of economic geography, the location decisions of multinational firms can be viewed as a function of two categories of factors. One consists of location fundamentals of MP that motivate MNCs to invest in a given country including market access and comparative advantage. The other is agglomeration economies, which includes vertical production linkages, externality in labour markets and capital-good markets, and technology diffusion.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

10.4.1 MP Location Fundamentals To quantify MP location fundamentals, we construct a measure that incorporates an empirical approach from the multinational firm literature with the agglomeration index methodology and invoking a two-step procedure.

10.4.1.1 Step 1 In the first step, we seek to obtain estimates of multinational actitivity predicted by location fundamentals including market size, trade cost, comparative advantage, and natural advantage, among other related characteristics. To obtain such estimates, we consider two alternative specifications. In the first specification, we estimate a conventional empirical equation following Carr et al. (2001), Yeaple (2003a), and Alfaro and Charlton (2009). Using a conventional empirical specification enables us to assess how MP location fundamentals commonly stressed by previous studies affect MNCs’ agglomeration patterns. Specifically, we consider the following specification: yc~c k ¼ γ0 þ γ1 marketsize sizec~c þ γ2 distancec~c þ γ3 skill diffc~c þγ4 skill diffc~c  skillintensityk þ γ5 tariffc~c k þ γ6 tariffc~ck þ μck þ μ0c~k þ εc~c k

ð5Þ

where yc~c k denotes either the number or the total employment of subsidiaries in country c~ and industry k owned by MNCs in country c, marketsize avec~c is average market size proxied by the GDP of home and host countries,¹³ distancec~c is the distance, skill diffc~c represents the difference in skill endowment, measured by average years of schooling, between the home and the host countries (i.e., skillc~  skillc ), skillintensityk is the skilled labour intensity proxied by share of non-production workers for each industry, tariffc~c k and tariffc~ck are the levels of tariffs set by the host country c~ on the home country c and vice versa in industry k, and εc~c k are the residuals. In addition to the above variables, host-country characteristics such as institutional and physical infrastructure could also affect multinationals’ location decisions.¹⁴ We thus include vectors of country-industry dummies, μck and μ0 c~k , to control for all countryindustry specific factors such as institutional quality, physical infrastructure, domestic industry size, and economic policies.¹⁵

¹³ We consider, in addition to GDP, market potential which is the sum of domestic and distanceweighted export market size of the home and host countries. ¹⁴ As noted by Helpman (2006), firms’ sorting patterns and organization choices are dependent on the characteristics of the firms and the contractual environment (see, e.g., Grossman and Helpman 2002; Antràs 2003). Existing empirical evidence also suggests that institutional development (such as the rule of law and intellectual property rights) exerts a positive effect on the receipt of foreign investment (Alfaro et al. 2008). ¹⁵ Note that the effect of agglomeration forces such as the size of upstream and downstream industries is controlled for in equation (5) by country-industry dummies. Ideally we would like to estimate equation (5) at more disaggregated geographic levels such as cities and provinces, but the explanatory variables in equation (5) are mostly available only at the country level.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



We estimate equation (5) using Poisson quasi-MLE (QMLE).¹⁶ If market access is a significant motive in MNCs’ investment decisions, we expect the effects of market size and trade cost (measured by distance and tariffs) to be positive, that is, γ1 >0, γ2 > 0, and γ5 > 0. If comparative advantage is a significant motive, we expect the effect of trade cost to be negative and the effect of difference in skilled labour endowment to be negative for unskilled-labour intensive industries, that is, γ2 < 0, γ4 >0, γ5 < 0, and γ6 < 0. Our estimates are largely consistent with the literature (see, e.g., Yeaple 2003a; Alfaro and Charlton 2009). Consistent with the market access motive, MNCs are found more likely to invest in countries with a larger market size (γ1 > 0). Consistent with the comparative advantage motive, MNCs have a greater probability of investing in unskilled labour abundant countries (γ3 0), and trade cost exerts a negative effect on MNCs’ investment decisions (γ2 < 0 and γ5 < 0).¹⁷ Based on the estimates of equation (5), we obtain and sum, for each host country ~c and industry k, the values of yc~c k predicted by market access and comparative advantage factors. To construct predicted MNC activities at a more disaggregated location level, we use the actual share of multinationals in each city to capture cross-city variations in attractiveness (e.g. port access and favourable industrial policies). Multiplying the actual share by ^y c~k gives ^y sk for each city s and industry k. In the alternative specification, we directly estimate MNC activity at a disaggregated regional level. To proceed, we re-consider equation (5) to examine MNC activity at the regional, instead of country, level and include a series of regional characteristics as additional regressors to capture the effect of regional location fundamentals. The main advantage of this specification is that it enables us to examine the role of regional characteristics, such as market size and natural and comparative advantages, in MNCs’ location decisions, instead of relying on the role of country characteristics alone and then using a region’s share of MNCs as a proxy for regional attractiveness. However, the disadvantage of this specification is the difficulty in obtaining disaggregated regional data for a wide sample of countries. In the end, we compiled a detailed database of regional characteristics from a number of national sources. For most countries, we were limited by the information available to primarily state or province level data. Specifically, for Europe, data were compiled from the Eurostat Regional Database at the NUTS 2 level disaggregation, both to compare with other data and for reasons of availability. For other countries, such as the USA, Australia, Brazil, Canada, China, Japan, Mexico, and South Korea, we used state- or province-level data. Because of constraints on data availability, the regional characteristics systematically available across countries and included in the final sample are income, schooling (educational attainment, percentage of labour with ¹⁶ See discussion in Santos Silva and Tenreyro (2006) and Head and Ries (2008). We also considered a two-step Heckman selection procedure following Helpman et al. (2008) in which we estimated the decision to trade and volume of trade; the results were similar. ¹⁷ Results are suppressed because of space considerations and are available upon request.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

tertiary education), infrastructure (roadways, ports, and airports), and taxes, measured in 2004 or the closest year available (to mitigate potential causality concerns).¹⁸ Based on this database, we estimate the following equation: yc~c sk ¼ γ0 þ γ1 marketsizes izec~c þ γ2 distancec~c þ γ3 skilld iffc~c s þγ4 skilld iffc~c s  skillintensityk þ γ5 tariffc~c k þ γ6 tariffc~ck þγ7 taxc~s þ γ8 roadwayc~s þ γ9 portc~s þ γ10 airportc~s þ μck þ μ0c~k þ εc~c k :

ð6Þ

where yc~c rs now denotes either the number or the total employment of subsidiaries in country c~’s region s and industry k owned by MNCs in country c, skilld iffc~c s represents the difference in skill endowment, measured by percentage of labour with tertiary education, between the home country and the host region (i.e., skillc~s  skillc ), taxc~s is the region’s corporate tax level, roadway is the length of roadway in each region s, and portc~s and airportc~s are indicators of ports and airports in the region. Again, we estimate the equation using Poisson quasi-MLE (QMLE) and find estimated parameters to be largely similar to the results from the first specification. In addition, we find regional skill level and infrastructure characteristics to matter significantly in multinationals’ location decisions. Similar to the first specification, we then obtain and sum, for each host country c, region s, and industry k, values of ^y c~c sk predicted by the market access, comparative advantage, and infrastructure variables.

10.4.1.2 Step 2 In the second stage, we repeat step 1 of Duranton and Overman’s (2005) procedure to obtain a geographic distribution function for each pair of industries k and ~k. We use the predicted levels of MNC activity (either predicted number or total employment of MNCs) in each region and industry (i.e. ^y sk and ^y~s ~k ) as the weight when estimating the kernel function. This generates, for each pair of industries, an expected geographic density function based exclusively on the estimated effects of location characteristics including market size, comparative advantage, and trade costs. We compare in Section 10.6 the role of these characteristics relative to that of agglomeration forces in determining the spatial patterns of multinational firms.

¹⁸ The US data were collected at the state level. Population and education attainment data were collected from the US Census; GDP and income/compensation statistics were collected from the Bureau of Economic Analysis; roadway statistics were from the Federal Highway Administration; employment data were collected from the Bureau of Labor Statistics. Australian data were compiled from the Australian Bureau of Statistics at the state level (ABS). Canadian data were collected from Statistics Canada at the provincial level. Chinese data were obtained from the CEIC Data at the provincial level. Brazilian data were taken from IBGE at the state level; data on Brazilian energy production and consumption were obtained from Ministério de Minas e Energia. Mexican data were collected from INEGI at the state level. South Korean data were from KOSIS, collected at the provincial level. Japanese data were collected from the Statistics Bureau of Japan at the prefecture level. The remaining data were obtained from the World Bank at the national level. For all regions, port data were from World Port Source, and tax rates were compiled from EY, Deloitte, KPMG, and the World Bank’s Doing Business report.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



10.4.2 Agglomeration Economies In addition to the location fundamentals of MP, agglomeration economies can also affect multinationals’ location choices. The advantage of proximity can differ dramatically between multinational and domestic firms and between MNC foreign subsidiaries and MNC headquarters. For instance, multinationals often incur substantial trade costs in sourcing intermediate inputs and reaching downstream buyers. They also face significant market entry costs when relocating to a foreign country because of, for example, limited supplies of capital goods. Further, given their technology intensity, technology diffusion from closely linked industries can be particularly attractive to MNCs. The agglomeration forces evaluated include (i) vertical production linkages, (ii) externality in labour markets, (iii) externality in capital-good markets, and (iv) technology diffusion. We describe in the following subsections how each of these four factors is measured in the empirical analysis.

10.4.2.1 Vertical Production Linkages Marshall (1890) argued that transportation costs induce plants to locate close to inputs and customers and determine the optimal trading distance between suppliers and buyers. This can be especially true for MNCs given their large volumes of sales and intermediate inputs.¹⁹ Compared to domestic firms, multinationals are often the leading corporations in each industry. Because they tend to be the largest customers of upstream industries as well as the largest suppliers of downstream industries, the input–output relationship between MNCs (e.g. Dell and Intel, Ford and Delphi) can be far stronger than that between average domestic firms.²⁰ To determine the importance of customer and supplier relationships in multinationals’ agglomeration decisions, we construct a variable, IOlinkagek~k , to measure the extent of the input–output relationship between each pair of industries. We use the 2002 Benchmark Input–Output (I–O) Data (specifically, the Detailed-Level Make, Use and Direct Requirement Tables) published by the Bureau of Economic Analysis (BEA), and define IOlinkagek~k as the share of industry k’s inputs that come directly from industry ~k; and vice versa. These shares are calculated relative to all input–output flows including those to non-manufacturing industries and final consumers. As supplier flows are not symmetrical, we take either the maximum or the mean of the input and output relationships for each pair of industries.

¹⁹ For FDI theoretical literature in this area, see, e.g., Krugman (1991); Venables (1996); and Markusen and Venables (2000). ²⁰ Head et al. (1995) note, for example, that the dependence of Japanese manufacturers on the ‘justin-time’ inventory system exerts a particularly strong incentive for vertically linked Japanese firms to agglomerate abroad.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

10.4.2.2 Externality in Labour Markets Agglomeration can also yield benefits through external scale economies in labour markets. Because firms’ proximity to one another shields workers from the vicissitudes of firm-specific shocks, workers in locations in which other firms stand ready to hire them are often willing to accept lower wages.²¹ Externalities can also occur as workers move from one job to another. This is especially true between MNCs which are characterized by similar skill requirements and large expenditures on worker training. MNCs can have a particularly strong incentive to lure workers from one another because the workers tend to receive certain types of training that are well suited for working in most multinational firms (business practices, business culture, etc.).²² To examine labour market pooling forces, we follow Ellison et al. (2010) in measuring each industry pair’s similarity in occupational labour requirements. We use the Bureau of Labor Statistics’ (BLS) 2006 National Industry-Occupation Employment Matrix (NIOEM), which reports industry-level employment across detailed occupations (e.g. Assemblers and Fabricators, Metal Workers and Plastic Workers, Textile, Apparel, and Furnishings Workers, Business Operations Specialists, Financial Specialists, Computer Support Specialists, and Electrical and Electronics Engineers). We convert occupational employment counts into occupational percentages for each industry, map the BLS industries to the SIC3 framework, and measure each industry pair’s labour similarity, labork~k , using the correlation in occupational percentages.

10.4.2.3 Externality in Capital-good Markets External scale economies can also arise in capital-good markets. This force has particular relevance to multinational firms given their large involvement in capital-intensive activities as shown in Alfaro and Chen (2014).²³ Geographically concentrated industries offer better support to providers of capital goods (e.g. producers of specialized components and providers of machinery maintenance) and reduce the risk of investment (due to, for example, the existence of resale markets).²⁴ Local expansion of capital intensive activities can consequently lead to an expansion of the supply of capital goods, thereby exerting a downward pressure on costs. To evaluate the role of capital-good market externalities, we construct a new measure of industries’ similarity in capital-good demand using capital flow data from

²¹ This argument was formally considered in Marshall (1890), Helsley and Strange (1990); and Krugman (1991). ²² The flow of workers can also lead to technology diffusion, another Marshallian force discussed below. ²³ See Alfaro and Hammel (2007) for evidence on capital flows and capital goods imports. ²⁴ Agglomeration can also induce costs by, for example, increasing labour and land prices. Like benefits, these costs can be potentially greater for industries with similar labour and capital-good demand, in which case the estimated parameters of the variables would represent the net effect of similar factor demand structures on agglomeration decisions.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



BEA. The capital flow table (CFT), a supplement to the 1997 Benchmark Input–Output (I–O) accounts, shows detailed purchases of capital goods (e.g. motors and generators, textile machinery, mining machinery and equipment, wood containers and pallets, computer storage devices, and wireless communications equipment) by industry. We measure each using industry pairs’ similarity in capital-good demand structure, denoted by capitalgoodk~k , using the correlation of investment flow vectors.²⁵ Industry pairs that exhibit the strongest correlation in capital-good demand include SIC 381 (Search, Detection, Navigation, Guidance, Aeronautical, and Nautical Systems) and SIC 387 (Watches, Clocks, Clockwork Operated Devices, and Parts); SIC 202 (Dairy Products) and SIC 206 (Sugar and Confectionery Products); SIC 326 (Pottery and Related Products) and SIC 328 (Cut Stone and Stone Products); and SIC 221 (Broadwoven Fabric Mills, Cotton) and SIC 228 (Yarn and Thread Mills).

10.4.2.4 Technology Diffusion A fourth motive relates to the diffusion of technologies. Technology can diffuse from one firm to another through the movement of workers between companies, interaction between those who perform similar jobs, or direct interaction between firms through technology sourcing. This has been noted by Navaretti and Venables (2006), who predict that MNCs may benefit from setting up affiliates in proximity to other MNCs with advanced technology (e.g. ‘so-called centers of excellence’). The affiliates can benefit from technology spillovers, which can then be transferred to other parts of the company. To capture this agglomeration force, we construct a proxy of technology diffusion frequently considered in the knowledge spillover literature (see, e.g., Jaffe et al. 2000; Ellison et al. 2010), using patent citation flow data taken from the NBER Patent Database. The data, compiled by Hall et al. (2001), include detailed records for all patents granted by the United States Patent and Trademark Office (USPTO) from January 1975 to December 1999. Each patent record provides information about the invention (e.g. technology classification, citations of prior art) and inventor submitting the application (e.g. name and city). We construct the technology diffusion variable, that is, technologyk~k , by measuring the extent to which technologies in industry k cite technologies in industry ~k, and vice versa.²⁶ In practice, there is little directional difference in technologyk~k due to the extensive number of citations within a single technology field. We obtain both maximum and mean for each set of pairwise industries. Constructing the proxies of agglomeration economies using the US industry account data is motivated by three considerations. First, compared to firm-level input–output,

²⁵ Note that this measure captures a different dimension of industry-pair relatedness than vertical production linkages. Unlike vertical production linkages, industry-pair correlations in capital-good demand reflect industry pairs’ similarity in capital-good demand and, thus, scope for externality in capital-good markets. ²⁶ The concordance between the USPTO classification scheme and SIC3 industries is adopted in the construction of the variable.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

factor demand, or technological information, which is typically unavailable, industrylevel production, factor, and technology linkages reflect standardized production technologies and are relatively stable over time, limiting the potential for the measures to endogenously respond to MNC agglomeration.²⁷ Second, using the USA as the reference country while our analysis covers multinational activity around the world further mitigates the possibility of endogenous production, factor, and technology linkage measures, even though the assumption that the US production structure carries over to other countries could potentially bias our empirical analysis against finding a significant relationship. Third, the US industry accounts are more disaggregated than most other countries’, enabling us to dissect linkages between disaggregated product categories. Table 10A.1 reports the summary statistics of industry-level control variables. Table 10A.2 presents the correlation matrix. For example, the correlation between industry-pair production linkages and similarity in capital-good demand is about 0.19, the correlation between production linkages and technology diffusion is about 0.29. Table 10.A.2 also shows the mean and maximum measures of production linkages and technology diffusion to be highly correlated.²⁸

10.5 D: T WB D

.................................................................................................................................. Our empirical analysis employs a unique worldwide establishment dataset, WorldBase, that covers more than 43 million public and private establishments in more than 100 countries and territories (see Alfaro and Chen 2014). WorldBase is compiled by Dun & Bradstreet, a leading source of commercial credit and marketing information since 1845, from a wide range of sources.²⁹ All information collected by Dun & Bradstreet is verified centrally via a variety of manual and automated checks.³⁰ Dun & Bradstreet’s WorldBase is, in our view, an ideal data source for the research question proposed in this study offering several distinct advantages over alternative data

²⁷ Concerns surrounding the endogeneity of agglomeration economies are further discussed and analysed in Section 10.7; see also the discussion in Alfaro and Chen (2014). ²⁸ We used the mean values in our analysis, but obtained similar results when we used the maximum measure. ²⁹ For more information, see: http://www.dnb.com/us/about/db_database/dnbinfoquality.html The data set employed in this chapter was acquired from Dun & Bradstreet with disclosure restrictions. ³⁰ Alfaro and Charlton (2009) use the data to study vertical and horizontal activities of multinationals, Alfaro and Chen (2012) MNCs’ reactions to the Global Financial crisis. The data have also been used to analyse vertical integration decisions in Alfaro et al. (2016) and global integration choices in Alfaro et al. (2018).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



sources. First, its broad cross-country coverage enables us to examine agglomeration on a global and continuous scale. Examining the global patterns of agglomeration allows us to offer a systematic perspective that takes into account nations at various stages of development. Viewing agglomeration on a continuous scale is important in light of the increasing geographic agglomeration occurring across regional and country borders. Second, the database reports detailed information for multinational and nonmultinational, offshore and headquarters establishments. This makes it possible to compare agglomeration patterns across different types of establishments and investigate how the economic geography of production evolves with forms of firm organization. Third, the WorldBase database reports the physical address and postal code of each plant, whereas most existing datasets report business registration addresses. Existing studies have tended to use distance between administrative units, such as state distances, as a proxy for the distance of establishments. In doing so, establishments proximate in actual distance but separated by administrative boundaries (e.g. San Diego and Phoenix) can be considered dispersed. Conversely, establishments far in distance but located in the same administrative unit (e.g. San Diego and San Francisco) can be counted as agglomeration. We obtain latitude and longitude codes for each establishment using a geocoding software (GPS Visualizer). This software uses Yahoo’s and Google’s Geocoding API services, well known as the industry standard for transportation data. It provides more accurate geocode information than most alternative sources. The geocodes are obtained in batches and verified for precision. We apply the Haversine formula to the geocode data to compute the great circle distance between each pair of establishments.³¹ The distance and the trade cost information is used to construct an index of agglomeration following the empirical methodology described in Section 10.4.

10.5.1 MNC Establishment Data Our main empirical analysis is based on MNC manufacturing establishments in 2005. WorldBase reports, for each establishment in the data set, detailed information on location, ownership, and activities. In this chapter we use industry information including the four-digit SIC code of the primary industry in which each establishment operates; ownership information including headquarters, domestic parent, global parent, and position in the hierarchy (branch, division, headquarters); detailed location

³¹ To account for other forms of trade barriers, such as border, language, and tariffs, we further estimated a more comprehensive measure of trade cost between each pair of plants. The results are available upon request.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

information for both establishment and headquarters; and operational information including sales, employment, and year started. An establishment is deemed as an MNC foreign subsidiary if it satisfies two criteria: (i) it reports to a global parent firm, and (ii) the headquarters or the global parent firm is located in a different country. The parent is defined as an entity that has legal and financial responsibility for another establishment.³² We drop establishments with zero or missing employment values and industries with fewer than ten observations.³³ Our final sample includes 32,427 MNC offshore manufacturing plants. Top industries include Electronic Components and Accessories (367), Miscellaneous Plastics Products (308), Motor Vehicles and Motor Vehicle Equipment (371), General Industrial Machinery and Equipment (356), Laboratory Apparatus and Analytical, Optical, Measuring, and Controlling Instruments (382), Drugs (283), Metalworking Machinery and Equipment (354), Construction, Mining, and Materials Handling (353), and Special Industry Machinery except Metalworking (355). Top host countries include China, the USA, the UK, Canada, France, Poland, the Czech Republic, and Mexico. More than 20 per cent of pairs of multinationals located within 200 km are in different countries. The percentage rises to 45 per cent at 400 km and 70 per cent at 800 km. This is not surprising given countries’ growing participation in regional trading blocs and rapid declines in cross-border trade costs.

10.5.2 Domestic Plant Data Conducting an empirical analysis of all domestic manufacturing plants is infeasible given the size of the entire WorldBase data set and computational intensity of the procedure. Consequently, to keep the analysis feasible, we adopt a random sampling strategy. For each SIC 3-digit industry with more than 1,000 observations, we obtain a random sample of 1,000 plants. For industries with fewer than 1,000 observations, we include all domestic plants. This yields a final sample of 127,897 domestically owned plants.

³² There are, of course, establishments that belong to the same multinational family. Although separately examining the interaction of these establishments is beyond the focus of this chapter, we expect the Marshallian forces to have a similar effect here. For example, subsidiaries with an input– output linkage should have incentives to locate near each other, independently of ownership. See Yeaple (2003b) for theoretical work in this area and Chen (2011) for supportive empirical evidence. One can use a similar methodology (estimating geographic distributions of establishments that belong to the same firm and comparing them with distributions of counterfactuals) to study intra-firm interaction (see Duranton and Overman 2008). ³³ Requiring positive employment helps to exclude establishments registered exclusively for tax purposes.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



10.6 A  R  L F  A E

.................................................................................................................................. We now examine the roles of location fundamentals and agglomeration economies in explaining the pairwise-industry agglomeration of MNCs and how the effects might differ across multinational foreign subsidiaries, domestic plants, and multinational headquarters. Formally, we estimate the following empirical specification: agglomerationk~k ðTÞ ¼ αK þ β1 fundamentalsk~k þ β2 IOlinkagek~k þ β3 labork~k þ β4 capitalgoodk~k þ β5 technologyk~k þ εij ;

ð7Þ

where agglomerationk~k ðTÞ is the agglomeration index of industry pairs k and ~k at threshold distance T (relative to the counterfactuals) and the right hand side includes (i) the agglomeration patterns predicted by MP location fundamentals (fundamentalsk~k ) based on the two specifications considered in Subsection 10.4.1, and (ii) proxies for agglomeration forces described in Section 4.2 consisting of input–output linkages (IOlinkagek~k ), labour- and capital-good market similarities (labork~k and capitalgoodk~k ), and technology diffusion (technologyk~k ). In addition to the location fundamentals and the agglomeration economies considered above, multinationals might also agglomerate because of factors like shared natural advantage (e.g. climate) and externality in institutional and physical infrastructure investment. We account for these factors with both the location fundamental measures and an industry fixed effect. Specifically, we include αk , a vector of industry dummies that takes the value of 1 if either industry k or ~k corresponds to a given industry, and zero otherwise. These industry dummies control for all industryspecific factors and agglomeration patterns. Summary statistics for MNC and domestic agglomeration indices are reported in Table 10.1.

10.6.1 MNC Offshore Agglomeration We consider first the agglomeration of MNC foreign subsidiaries. Table 10.2 reports the regression results based on measure 1 of location fundamentals. Agglomeration forces including vertical production linkages, capital-good market correlation, and technology diffusion all play a significant role and display the expected signs.³⁴ For

³⁴ In univariate regression results for each of our main variables, all the agglomeration variables were found to be highly significant across the different distance threshold levels. The estimated effects also exhibited expected signs. Across agglomeration forces, capital-good market correlation had the greatest

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

Table 10.1 Descriptive statistics for MNC and domestic agglomeration indices Obs.

Mean

Std. Dev.

Min.

Max.

Subsidiaries (percentage points) Threshold (T) = 200 km T = 400 km T= 800 km T= 1600 km

8004 8004 8004 8004

0.099 0.219 0.520 1.028

0.239 0.522 1.206 2.357

0.000 0.000 0.000 0.000

3.060 6.631 14.419 23.941

Domestic plants (percentage points) Threshold (T) = 200 km T = 400 km T= 800 km T= 1600 km

8004 8004 8004 8004

0.102 0.235 0.550 1.210

0.289 0.545 1.384 2.424

0.000 0.000 0.000 0.000

4.012 7.935 16.539 26.340

Subsidiaries workers (percentage points) Threshold (T) = 200 km T = 400 km T= 800 km T= 1600 km

8004 8004 8004 8004

0.095 0.194 0.418 0.742

0.274 0.528 1.038 1.853

0.000 0.000 0.000 0.000

2.997 5.553 10.139 17.211

Headquarters (percentage points) Threshold (T) = 200 km T = 400 km T= 800 km T= 1600 km

8004 8004 8004 8004

0.140 0.325 0.782 1.402

0.348 0.779 1.772 2.987

0.000 0.000 0.000 0.000

8.400 18.198 39.871 44.693

Notes: The agglomeration indices are constructed by comparing the estimated distance kernel function of each industry pair with the 95 per cent global confidence band of counterfactual kernel estimators at 200 km, 400 km, 800 km, and 1,600 km. All industry pairs (SIC3) are included. See text for detailed descriptions of the variables.

example, at 200 km a 100-percentage-point increase in the level of technology diffusion, that is, the percentage of patent citations between two industries, leads to a 0.6-percentage-point increase in the level of the agglomeration index between industries. This is equivalent to increasing the average (0.2) by a factor of 3. The location fundamental variable is significant at 1,600 km, influencing the spatial patterns of MNCs at a relatively aggregate geographic level. The lower panel of Table 10.2 reports the normalized beta coefficients.³⁵ Comparing the standardized coefficients of agglomeration forces, we find the effects of technology diffusion and capital-good market correlation to outweigh that of vertical production linkages, which suggests that, given the technology and capital intensive characteristics impact across all distance thresholds, followed by labour-demand correlation, technology diffusion, and production linkages. Tables showing univariate results are suppressed from the chapter due to space considerations but available upon request. ³⁵ Standardized coefficients enable us to compare the changes in the outcomes associated with the metric-free changes in each covariate.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Table 10.2 Location fundamentals, agglomeration economies, and MNC subsidiary agglomeration I IO linkages Capital good Labour Technology Location fundamentals Obs. R2 IO linkages Capital good Labour Technology Location fundamentals

T = 200 km

T = 400 km

T = 800 km

T = 1,600 km

0.265* (0.147) 0.038*** (0.014) 0.002 (0.016) 0.609** (0.293) 0.018 (0.025)

0.573* (0.306) 0.093*** (0.032) 0.015 (0.035) 1.178** (0.546) 0.019 (0.019)

1.331** (0.656) 0.241*** (0.066) 0.079 (0.068) 2.521** (1.117) 0.020 (0.022)

2.596** (1.296) 0.506*** (0.139) 0.231 (0.160) 4.395** (2.371) 0.021* (0.012)

7875 0.571

7875 0.600

7875 0.627

7875 0.631

Beta coefficients 0.014 0.014 0.035 0.039 0.002 0.007 0.031 0.027 0.266 0.264

0.014 0.043 0.015 0.025 0.279

0.013 0.046 0.023 0.022 0.333

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

of multinational firms, it is important to take into account not only vertical production linkages but also technology and capital-good market externalities in explaining MNCs’ offshore agglomeration. The parameter of labour-market correlation is insignificant in the multivariate regressions.³⁶ Comparing the estimates across distance thresholds, we find that at more aggregate geographic levels, the impact of technology diffusion diminishes and the effect of capital-good market externalities rises while the role of vertical production linkages remains mostly constant. The stronger effect of technology diffusion at shorter distance levels suggests that, compared to the other agglomeration economies, benefits from technology diffusion tend to be localized geographically.³⁷

³⁶ Excluding the capital-good market correlation variable, we found the technology diffusion and production linkage variables to remain positive and significant and the labour correlation coefficient to remain insignificant. This result suggests that the capital-good variable is capturing agglomeration incentives not represented by the other variables. ³⁷ When excluding the location fundamental variable, the coefficients and statistical significance of the agglomeration forces remain largely unchanged.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

Estimation results based on measure 2 of location fundamentals are reported in Table 10.3. The estimated parameters of agglomeration economies remain largely similar to Table 10.2. The location fundamental variable, obtained from the regionallevel specification, now exerts a significant effect on the agglomeration of multinational foreign subsidiaries at both 400 and 800 km. Comparing the relative importance of location fundamentals and agglomeration economies, we find the effect of location fundamentals to be outweighed by the cumulative effect of agglomeration forces in Table 10.3. At 400 km, a one-standard-deviation increase in location fundamentals leads to a 0.025-standard-deviation increase in the level of agglomeration, while the cumulative effect of agglomeration forces is 0.076 standard deviation.³⁸

Table 10.3 Location fundamentals, agglomeration economies, and MNC subsidiary agglomeration II IO linkages Capital good Labour Technology Location fundamentals (regional) Obs. R2 IO linkages Capital good Laboir Technology Location fundamentals (regional)

T = 200 km

T = 400 km

T = 800 km

T = 1600 km

0.249** (0.112) 0.037** (0.017) 0.001 (0.014) 0.573*** (0.161) 0.006 (0.007)

0.541* (0.302) 0.092*** (0.017) 0.001 (0.015) 1.101*** (0.458) 0.004*** (0.001)

1.252*** (0.222) 0.237*** (0.092) 0.045 (0.165) 2.330*** (0.343) 0.002* (0.001)

7875 0.570

7875 0.600

7875 0.626

7875 0.630

Beta coefficients 0.013 0.013 0.034 0.038 0.004 0.001 0.029 0.025 0.038 0.025

0.013 0.042 0.009 0.023 0.013

0.012 0.045 0.015 0.019 0.006

2.413*** (0.576) 0.499*** (0.153) 0.153 (0.135) 3.943* (2.560) 0.001 (0.003)

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

³⁸ Comparing Table 10.2 and Table 10.3, we also note the normalized parameter of the location fundamental variable to be significantly lower when the variable is constructed based on the regional estimation specification. One possible explanation is that measure 1, based on country-level location characteristics and actual regional share of multinational activity, represents an upper bound of location fundamentals, whereas measure 2, estimated and based on observable country and regional characteristics, serves as a lower bound.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



We have examined MNC offshore agglomeration thus far using the subsidiary as the unit of observation. We now take into account the different employment sizes of multinational subsidiaries, which essentially treats the worker as the unit of observation and measures the level of agglomeration among workers. This exercise, by differentiating the agglomeration incentives between individual establishments and workers, has implications for policy making targeted at influencing the geographic distribution of workers. Tables 10.4 and 10.5 reports the estimates based on the two measures of location fundamentals. Note that in contrast to Tables 10.2 and 10.3, in which labour market correlation does not exert a significant effect, multinational subsidiaries in industries with greater potential labour market externalities exhibit significantly higher level of employment agglomeration. Technology diffusion, another force of agglomeration that involves close labour interaction and mobility, also plays a significant role in explaining the agglomeration of MNC subsidiary workers between industries. In fact, technology spillover appears to be the strongest agglomeration factor at most distance thresholds. Further, at more aggregate geographic levels, the effects of labour market externalities and technology spillovers diminish, while capital-good market correlation exerts a significant and positive effect.

Table 10.4 Location fundamentals, agglomeration economies, and MNC subsidiary worker agglomeration I T = 200 km IO linkages Capital good Labour Technology Location fundamentals Obs. R2 IO linkages Capital good Labour Technology Location fundamentals

T = 400 km

T = 800 km

T = 1,600 km

0.145 (0.209) 0.041* (0.023) 0.048* (0.026) 2.262*** (0.516) 0.0004*** (0.0001)

0.256 (0.403) 0.109** (0.044) 0.088* (0.048) 3.957*** (0.867) 0.0004*** (0.0001)

0.272 (0.683) 0.315*** (0.089) 0.120 (0.104) 6.243*** (1.613) 0.0004*** (0.0001)

0.750 (1.160) 0.557*** (0.144) 0.128 (0.162) 9.333*** (2.356) 0.0004** (0.0002)

7875 0.327

7875 0.327

7875 0.363

7875 0.402

0.006 0.045 0.039 0.091 0.349

0.003 0.066 0.027 0.073 0.390

0.005 0.065 0.016 0.061 0.435

Beta coefficients 0.007 0.033 0.042 0.100 0.315

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

Table 10.5 Location fundamentals, agglomeration economies, and MNC subsidiary worker agglomeration II IO linkages Capital good Labour Technology Location fundamentals (regional) Obs. R2 IO linkages Capital good Labor Technology Location fundamentals (regional)

T = 200 km

T = 400 km

T = 800 km

T = 1,600 km

0.151 (0.120) 0.040*** (0.014) 0.057*** (0.022) 2.228*** (0.508) 0.002 (0.010)

0.269 (0.212) 0.106*** (0.018) 0.107** (0.049) 3.885*** (0.326) 0.004** (0.002)

0.299 (0.482) 0.308*** (0.087) 0.162* (0.077) 6.083*** (1.390) 0.007* (0.004)

0.801 (0.835) 0.544*** (0.176) 0.212 (0.050) 9.013*** (2.815) 0.009* (0.001)

7875 0.326

7875 0.326

7875 0.363

7875 0.402

Beta Coefficients 0.007 0.006 0.032 0.044 0.049 0.047 0.100 0.089 0.011 0.027

0.003 0.064 0.036 0.071 0.054

0.005 0.065 0.026 0.058 0.086

Notes: Bootstrapped standard errors in parentheses, *** p < 0.01, ** p < 0.05, * p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

10.6.2 MNC Headquarters Agglomeration We next examine the determinants of MNC headquarters clusters relative to MNC clusters overseas. To control for the role of location fundamentals in explaining the agglomeration of MNC headquarters, we follow the procedure described in Section 10.4.1, but obtain the level of MNC activities predicted for each MNC home country, and construct the expected distribution and agglomeration of MNC headquarters following the rest of the procedure. Table 10.6 reports the estimation results. All variables except vertical production linkages exert a significant effect. A one-standard-deviation increase in the location fundamental variable is associated with a 0.21 standard-deviation increase in MNC headquarters agglomeration, which suggests an important role for the characteristics of headquarter countries including market size, skilled labour endowment, and access to host countries. At 200 km, both technology diffusion and labour market correlation play a positive and significant role, with a cumulative effect of about 0.06. Beyond 200 km, the effect of the labour market becomes insignificant. Again, this result is consistent with the localized feature of labour markets and lower mobility of labour.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Table 10.6 Location fundamentals, agglomeration economies, and MNC headquarters agglomeration IO linkages Capital good Labour Technology Location fundamentals Obs. R2 IO linkages Capital good Labour Technology Location fundamentals

T = 200 km

T = 400 km

T = 800 km

T = 1,600 km

0.090 (0.174) 0.026 (0.019) 0.043** (0.021) 0.793*** (0.241) 0.022** (0.009)

0.156 (0.406) 0.084** (0.040) 0.064 (0.044) 1.727*** (0.477) 0.023*** (0.009)

0.127 (0.815) 0.261*** (0.088) 0.019 (0.104) 3.870*** (1.153) 0.024* (0.013)

0.457 (1.254) 0.459*** (0.164) 0.085 (0.180) 6.935*** (1.735) 0.019 (0.018)

7875 0.639

7875 0.65

7875 0.664

7875 0.667

0.003 0.024 0.020 0.027 0.212

0.001 0.032 0.003 0.027 0.208

0.002 0.033 0.007 0.028 0.213

Beta coefficients 0.003 0.017 0.030 0.028 0.212

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

Comparing Table 10.6 with Table 10.2, we find that location fundamentals and capital-good market externality exert a stronger effect on MNCs’ offshore agglomeration than on the agglomeration of MNC headquarters and, further, input–output relationships affect MNC subsidiaries but not headquarters. These results suggest that MNC subsidiary agglomeration, with their market-seeking and input-sourcing focuses, is more influenced by market access and comparative advantage motives, capital-good market externalities, and vertical production linkages, whereas agglomeration of headquarters, with their specialization in R&D, management, and the provision of other services, is more influenced by technology diffusion than by production linkages.

10.6.3 Comparing the Agglomeration of MNC Offshore and Domestic Plants Having established the agglomeration patterns of MNC foreign subsidiaries, we now investigate how the role of agglomeration forces varies systematically between multinational and non-multinational plants. Specifically, we evaluate how the role of

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

location fundamentals and agglomeration economies affects MNCs relative to domestic plants by estimating the following equation: ðTÞ  agglomerationdk~k ðTÞ agglomerationm k~k d m d m d ¼ ðβm 1  β1 Þfundamentalsk~k þ ðβ2  β2 ÞIOlinkagek~k þ ðβ3  β3 Þlabork~k d m d þðβm 4  β4 Þcapitalgoodk~k þ ðβ5  β5 Þtechnologyk~k þ εij ;

ð8Þ

where agglomerationm ðTÞ  agglomerationdk~k ðTÞ represents the difference between the k~k MNC and domestic pairwise-industry agglomeration indices, and the coefficient vector βm  βd represents the difference in the effects of the covariates on multinational foreign subsidiaries and domestic plants. The results based on the two measures of location fundamentals are reported in Tables 10.7 and 10.8. We find that proxies for capital-good market externalities and technology diffusion exert a stronger effect on multinationals than on domestic plants in same industry pairs. The role of the input–output relationship is not significantly different between the two at disaggregated geographic levels, but is significantly stronger for multinationals at more aggregate geographic levels (e.g. 800 km). Interestingly, potential externalities in the labour market, captured by industry-pair similarity in labour demand, exert a greater effect on the agglomeration of domestic plants than the agglomeration of multinational foreign subsidiaries. Location fundamental

Table 10.7 Comparing MNC subsidiaries with domestic plants I IO linkages Capital good Labour Technology Location fundamentals Obs. R2 IO linkages Capital good Labour Technology Location fundamentals

T = 200 km

T = 400 km

T = 800 km

T = 1,600 km

0.041 (0.599) 0.162*** (0.051) 0.110** (0.049) 1.214 (0.839) 0.047*** (0.003)

1.081 (1.306) 0.494*** (0.113) 0.443*** (0.112) 2.823* (1.706) 0.047*** (0.002)

5.447** (2.760) 1.335*** (0.220) 1.430*** (0.231) 24.272*** (3.409) 0.044*** (0.002)

10.876** (4.437) 2.383*** (0.366) 2.130*** (0.410) 62.572*** (6.220) 0.035*** (0.002)

7875 0.049

7875 0.053

7875 0.064

7875 0.073

0.008 0.067 0.065 0.021 0.217

0.020 0.085 0.099 0.086 0.219

0.023 0.086 0.084 0.126 0.228

Beta coefficients 0.001 0.047 0.034 0.020 0.213

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Table 10.8 Comparing MNC subsidiaries with domestic plants II IO linkages Capital good Labour Technology Location fundamentals (regional) Obs. R2 IO linkages Capital good Labour Technology Location fundamentals (regional)

T = 200 km

T = 400 km

T = 800 km

T = 1,600 km

0.023 (0.603) 0.183*** (0.048) 0.264*** (0.045) 0.943 (0.880) 0.011*** (0.001)

0.916 (1.285) 0.536*** (0.118) 0.774*** (0.102) 1.252** (0.602) 0.010*** (0.001)

5.014** (2.515) 1.421*** (0.217) 2.136*** (0.225) 20.632*** (3.346) 0.025*** (0.003)

10.094** (4.406) 2.533*** (0.375) 3.419*** (0.406) 55.824*** (6.314) 0.454*** (0.005)

7875 0.012

7875 0.016

7875 0.028

7875 0.034

Beta coefficients 0.0004 0.007 0.053 0.072 0.083 0.114 0.007 0.009 0.079 0.089

0.018 0.090 0.148 0.073 0.101

0.021 0.091 0.134 0.112 0.103

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

variables including market size, comparative advantage, and infrastructure also have a greater role in the agglomeration patterns of domestic plants. These findings are consistent with the characteristics of multinational firms: relative to their domestic counterparts, multinationals exhibit greater participation in knowledge- and capital-intensive activities and would thereby enjoy stronger agglomeration economies in technology and capital-good markets. Externalities such as technology diffusion and capital-good market scale economies thus provide multinational subsidiaries greater incentives to agglomerate with one another relative to domestic plants. Domestic plants, in contrast, place a greater emphasis on fundamental location characteristics such as market size, production cost, and infrastructure and labour market considerations.

10.7 A E A

.................................................................................................................................. A potential concern with our analysis thus far is that the agglomeration economy measures might endogenously reflect the agglomeration patterns of multinational firms. For example, the input–output linkage between the apparel and cotton industries

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

may reflect not just the inherent characteristics of apparel manufacturing, but also the agglomeration of the two industries due, for example, to availability of raw materials leading apparel manufacturers to favour cotton over other types of fabrics. Similarly, the technology spillover between the telecommunication and computer industries might be due not only to the intrinsic technological relationship between the two industries, but also to a historical factor that led the two industries to locate together and subsequently become familiar with each other’s technologies. This concern is mitigated in our work by three factors. First, our analysis controls for the role of location fundamentals and industry-specific characteristics. This enables us to separate industries’ geographic concentration due to location attractiveness from agglomeration activities driven by agglomeration economies. Second, our measures of agglomeration economies are constructed using US industry account data while the chapter examines global agglomeration patterns. US industries’ input–output linkages, factor market correlations, and technology spillovers are not very likely a result of agglomeration around the world. Third, the focus on MNCs reduces the possibility of reverse causation, as MNCs constitute a small subset of firms in each industry and the agglomeration economy measures are built with industry wide data that include information on domestic firms.³⁹ We nevertheless perform an additional exercise to further alleviate concerns about endogeneity. Because the global agglomeration patterns of multinational firms include the agglomeration of MNCs in the United States, we examine regional agglomeration for which the United States is excluded. If US domestic industry-pair relationships are affected by the agglomeration of MNCs in the United States, then one would expect the former to be less likely to be affected by the agglomeration of MNCs located in other regions such as Europe.⁴⁰ In this case, the agglomeration economy measures constructed with US industry account data are orthogonal to the agglomeration patterns observed in Europe.⁴¹ We proceed by repeating the procedure described in Section 10.4.1 to construct the agglomeration indices for MNCs located in Europe. These indices capture the degree to which MNCs in a given industry pair agglomerate in Europe at various threshold distances. The results are reported in Table 10.9. We find the estimates to be qualitatively similar to those reported in Tables 10.2 and 10.3.⁴² Multinational subsidiaries in industries with greater labour market correlation and technology spillover are found

³⁹ Alfaro and Chen (2014) further explore the process of agglomeration. ⁴⁰ On regional integration and the concentration of MNCs, see also Chen (2009). ⁴¹ Using another country’s data to construct the agglomeration economy variables would not alleviate the potential for endogeneity in our analysis because it would face issues similar to the US data. Using the US agglomeration economy measures to predict the agglomeration patterns in a non-US region would, however, mitigate the possibility of reverse causation and help identify the causal effects of agglomeration forces. ⁴² Because we are now examining regional, instead of global, agglomeration, we consider only threshold distances up to 800 km.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Table 10.9 The endogeneity of agglomeration economy measures: the agglomeration patterns of MNCs in Europe IO linkages Capital good Labour Technology Location fundamentals Obs. R2 IO linkages Capital good Labour Technology Location fundamentals

T = 200 kms

T = 400 kms

T = 800 kms

0.104 (0.079) 0.008 (0.010) 0.031*** (0.008) 0.335** (0.151) 0.001 (0.003)

0.248* (0.157) 0.031* (0.019) 0.032* (0.018) 0.514** (0.262) 0.004 (0.005)

0.454** (0.209) 0.044* (0.026) 0.036 (0.030) 0.715** (0.393) 0.003 (0.004)

7166 0.635

7166 0.717

7166 0.853

0.009 0.021 0.023 0.019 0.087

0.008 0.014 0.012 0.013 0.076

Beta coefficients 0.009 0.014 0.055 0.030 0.158

Notes: Bootstrapped standard errors in parentheses, ***p < 0.01, **p < 0.05, *p < 0.1. All regressions include industry fixed effect. Normalized beta coefficients in lower panel. See text for detailed descriptions of the variables.

to have a higher level of agglomeration, especially at the 200 and 400 km levels. Input– output production linkage and capital-good market correlation also exert a significant effect on the agglomeration of MNCs in Europe. Consistent with the earlier results, we find the effects of labour market externalities and technology spillovers to diminish at more aggregate geographic levels. Further, labour market externality appears to be the strongest agglomeration force at disaggregated distance levels.

10.8 C

.................................................................................................................................. The emergence of new multinational clusters is one of the most notable phenomena in the process of globalization. Multinationals follow distinctively different agglomeration patterns offshore than their domestic counterparts (Alfaro and Chen 2014). We examine in this chapter the relative importance of agglomeration forces versus location fundamentals in MNCs’ offshore as well as headquarter geographic patterns. Our analysis, using a worldwide plant-level dataset and a novel index of agglomeration, yields a number of new insights into the economic geography of multinational production.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

MP location fundamentals, although playing a significant and important role in explaining the agglomeration of multinational firms, are not the only driving force. In addition to market access and comparative advantage motives, multinationals’ location choices are significantly affected by agglomeration economies including not only vertical production linkages but also technology diffusion and capital-market externalities. Further, the importance of location fundamentals and agglomeration economies varies significantly between MNCs’ offshore agglomeration and the agglomeration of MNC headquarters and domestic plants. For example, MNCs’ offshore plants are significantly more influenced than non-MNC plants by capital-good market and technological agglomeration factors. Our results convey implications central to academic and policy debates on FDI. The agglomeration of economic activity, as long recognized by regional and urban economists and economic historians, is one of the salient features of economic development. An extensive body of research examines the distribution of population and production across space and the economic characteristics and effects of spatial concentrations. Understanding the emerging spatial concentrations of multinational production around the world and the driving forces behind these new concentrations in comparison to those of their domestic counterparts is crucial for designing and improving policies. Growing evidence suggests that multinationals play a significant role in the performance of local economies, resulting in local wage increases (see, for example, Aitken et al. 1996), spillovers (Javorcik 2004), reallocation (see, for example, Alfaro and Chen 2018a), and self upgrading (Bao and Chen 2018). Evidence has shown positive effects of FDI on the host country’s growth conditional on local conditions (Alfaro et al. 2004, 2010) and resilience to external shocks (Alfaro and Chen 2012). Recognizing these effects, many countries, including both FDI source and destination nations, have long offered lucrative incentives to MNCs in the hope of building and sustaining industrial clusters. Understanding the location interdependence of multinational firms and how they agglomerate with one another is critical to designing these economic policies.

 1

.................................................................................................................................. Table 10A.1 Descriptive statistics for agglomeration economies

APPENDIX # Obs. Mean Input–output (IO) linkages Capital good Labour Technology

7875 7875 7875 7875

0.003 0.476 0.333 0.007

1 Std. Dev. 0.012 0.209 0.227 0.012

Min.

Max.

0.000 0.004 0.014 0.000

0.193 1.000 1.000 0.179

Notes: Input–output (IO) linkages, capital good, labour, and technology correspond to the industry-level variables employed to proxy for the various agglomeration economies: vertical production linkages, externalities in factor markets including labour and capital goods, and technology diffusion. Same industry pairs (SIC3) are excluded. See text for detailed descriptions of the variables.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Table 10A.2 Correlation of agglomeration economies

IO linkages IO linkages (max.) Capital good Labour Technology Technology (max.)

IO linkages

IO linkages (max.)

Capital good

Labour

Technology

Technology (max.)

1.000 0.973 0.191 0.232 0.291 0.264

1.000 0.189 0.225 0.284 0.257

1.000 0.567 0.230 0.188

1.000 0.331 0.297

1.000 0.976

1.000

Notes: Obs = 7875. Both average and maximum measures are obtained for IO linkages and technology diffusion. See text for detailed descriptions of the variables.

A We thank participants in the Workshop on Structural Transformation and Jim Anderson, Bruce Blonigen, Gilles Duranton, James Harrigan, Keith Head, Tarun Khanna, Jim Markusen, Keith Maskus, Mike Moore, Henry Overman, John Ries, Roberto Samaneigo, and Tony Yezer for valuable comments and suggestions in various stages of the project and William Kerr for kindly providing us the patent concordance data. Hayley Pallan, Elizabeth Meyer, and Hillary White provided superb research assistance.

R Aitken, Brian, Ann Harrison, and Robert Lipsey, 1996. ‘Wages and Foreign Ownership: A Comparative Study of Mexico, Venezuela, and the United States’, Journal of International Economics, 40 (3–4), pp. 345–71. Alfaro, Laura and Andrew Charlton, 2009. ‘Intra-Industry Foreign Direct Investment’, American Economic Review, 99 (5), pp. 2096–119. Alfaro, Laura and Maggie X. Chen, 2012. ‘Surviving the Global Financial Crisis: Foreign Ownership and Establishment Performance’, American Economic Journal: Economic Policy, 4, pp. 30–55. Alfaro, Laura and Maggie X. Chen, 2014. ‘The Global Agglomeration of Multinational Firms’, Journal of International Economics, 99 (2), pp. 263–76. Alfaro, Laura and Maggie X. Chen, 2018a. ‘Selection and Market Reallocation: Productivity Gains from Multinational Production’, American Economic Journal: Economic Policy, 10 (2), pp. 1–38. (Also NBER Working Paper No. 18207.) Alfaro, Laura and Maggie X. Chen, 2018b. ‘Transportation Cost and the Geography of Foreign Investment’, in Bruce Blonigen and Wesley Wilson, eds, Handbook of International Trade and Transportation, Cheltenham: Edward Elgar Publishing. Alfaro, Laura and Eliza Hammel, 2007. ‘Capital Flows and Capital Goods’, Journal of International Economics, 72(1), pp. 128–50.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

Alfaro, Laura, Areendam Chanda, Sebnem Kalemli-Ozcan, and Selin Sayek, 2004. ‘FDI and Economic Growth: the Role of Local Financial Markets’, Journal of International Economics, 64 (1), 89–112. Alfaro, Laura, Sebnem Kalemli-Ozcan, and Vadym Volosovych, 2008. ‘Why Doesn’t Capital Flow from Rich to Poor Countries? An Empirical Investigation’, Review of Economics and Statistics, 90, pp. 347–68. Alfaro, Laura, Areendam Chanda, Sebnem Kalemli-Ozcan, and Selin Sayek, 2010. ‘Does Foreign Direct Investment Promote Growth? Exploring the Role of Financial Markets on Linkages’, Journal of Development Economics, 91 (2), 242–56. Alfaro, Laura, Paola Conconi, Harald Fadinger, and Andrew Newman, 2016. ‘Do Prices Determine Vertical Integration?’ Review of Economics Studies, 83 (3), pp. 855–88. Alfaro, Laura, Pol Antràs, Davin Chor, and Paola Conconi, 2018. ‘Internalizing Global Value Chains: A Firm-Level Analysis’, Journal of Political Economy (forthcoming). Antràs, Pol, 2003. ‘Firms, Contracts and Trade Structure’, Quarterly Journal of Economics, 118, pp. 1375–418. Antràs, Pol and Elhanan Helpman, 2004. ‘Global Sourcing’, Journal of Political Economy, 112 (3), pp. 552–80. Antràs, Pol and Elhanan Helpman, 2008. ‘Contractual Frictions and Global Sourcing’, in Elhanan Helpman, Dalia Marin, and Thierry Verdier, eds, The Organization of Firms in a Global Economy, Cambridge, MA: Harvard University Press. Bao, Cathy Ge and Maggie X. Chen, 2018. ‘Foreign Rivals are Coming to Town: Responding to the Threat of Foreign Multinational Entry’, American Economic Journal: Applied Economics (forthcoming). Bobonis, Gustavo J. and Howard J. Shatz, 2007. ‘Agglomeration, Adjustment, and State Policies in the Location of Foreign Direct Investment in the United States’, Review of Economics and Statistics, 89 (1), pp. 30–43. Carr L. David, James R. Markusen, and Keith E. Maskus, 2001. ‘Estimating the KnowledgeCapital Model of the Multinational Enterprise’, American Economic Review, 91 (3), pp. 693–708. Chen, Maggie X., 2009. ‘Regional Economic Integration and Geographic Concentration of Multinational Firms’, European Economic Review, 53 (3), pp. 355–75. Chen, Maggie X., 2011. ‘Interdependence in Multinational Production Networks’, Canadian Journal of Economics, 44 (3), pp. 930–56. Chen, Maggie X. and Michael Moore, 2010. ‘Location Decision of Heterogeneous Multinational Firms’, Journal of International Economics, 80 (2), pp. 188–99. Duranton, Gilles and William Kerr, 2015. ‘The Logic of Agglomeration’, NBER Working Papers No. 21452. Duranton, Gilles and Henry Overman, 2005. ‘Testing for Localization Using Micro Geographic Data’, Review of Economic Studies, 72 (4), pp. 1077–106. Duranton, Gilles and Henry Overman, 2008. ‘Exploring the Detailed Location Patterns of U.K. Manufacturing Industries Using Microgeographic Data’, Journal of Regional Science, 48 (1), pp. 213–43. Duranton, Gilles and Diego Puga, 2004. ‘Micro-Foundations of Urban Agglomeration Economies’, in J. Vernon Henderson and Jacques-François Thisse, eds, Handbook of Regional and Urban Economics, Vol. 4, Amsterdam: North-Holland, pp. 2063–117. Ellison, Glenn and Edward Glaeser, 1997. ‘Geographic Concentration in U.S. Manufacturing Industries: A Dartboard Approach’, Journal of Political Economy, 105 (5), pp. 889–927.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Ellison, Glenn, Edward Glaeser, and William Kerr, 2010. ‘What Causes Industry Agglomeration? Evidence from Coagglomeration Patterns’, American Economic Review, 100, pp. 1195–213. Grossman, Gene M. and Elhanan Helpman, 2002. ‘Integration Versus Outsourcing In Industry Equilibrium’, Quarterly Journal of Economics, 117 (1), pp. 85–120. Hall, Bronwyn, Adam Jaffe, and Manuel Trajtenberg, 2001. ‘The Geographic Concentration of Industry: Does Natural Advantage Explain Agglomeration?’ The NBER Patent Citation Data File: Lessons, Insights and Methodological Tools. NBER Working Paper No. 8498. Head, Keith and Theirry Mayer, 2004. ‘The Empirics of Agglomeration and Trade’, in J.V. Henderson, and J.-F. Thisse, eds, Handbook of Regional and Urban Economics, Vol. 4, Amsterdam: Elsevier, pp. 2609–69. Head, Keith and John Ries, 2008. ‘FDI as an Outcome of the Market for Corporate Control: Theory and Evidence’, Journal of International Economics, 74 (1), pp. 2–20. Head, Keith, John Ries, and Deborah Swenson, 1995. ‘Agglomeration Benefits and Location Choice: Evidence from Japanese Manufacturing Investments in the United States’, Journal of International Economics, 38 (3–4), pp. 223–47. Helpman, Elhanan, 1984. ‘A Simple Theory of Trade with Multinational Corporations’, Journal of Political Economy, 92 (3), pp. 451–71. Helpman, Elhanan, 2006. ‘Trade, FDI and the Organization of Firms’, Journal of Economic Literature, 44 (3), pp. 589–30. Helpman, Elhanan, Marc Melitz, and Stephen Yeaple, 2004. ‘Export versus FDI with Heterogeneous Firms’, American Economic Review, 94 (1), pp. 300–16. Helpman, Elhanan, Marc Melitz, and Yona Rubinstein, 2008. ‘Estimating Trade Flows: Trading Partners and Trading Volumes’, Quarterly Journal of Economics, 123 (2), pp. 441–87. Helsley, Robert W. and William C. Strange, 1990. ‘Matching and Agglomeration Economies in a System of Cities’, Regional Science and Urban Economics, 20 (2), pp. 189–212. Jaffe, Adam, Manuel Trajtenberg, and Michael Fogarty, 2000. ‘Knowledge Spillovers and Patent Citations: Evidence from a Survey of Inventors’, American Economic Review Paper and Proceedings, 90, pp. 215–18. Javorcik, Beata S., 2004. ‘Does Foreign Direct Investment Increase the Productivity of Domestic Firms? In Search of Spillovers through Backward Linkages’, American Economic Review, 94 (3), pp. 605–27. Krugman, Paul, 1991. ‘Increasing Returns and Economic Geography’, Journal of Political Economy, 99 (3), pp. 483–99. Markusen, James, 1984. ‘Multinationals, Multi-Plant Economies, and the Gains from Trade’, Journal of International Economics, 16 (3–4), pp. 205–26. Markusen, James, 2002. Multinational Firms and the Theory of International Trade, Cambridge, MA: MIT Press. Markusen, James and Anthony J. Venables, 1998. ‘Multinational Firms and the New Trade Theory’, Journal of International Economics, 46 (2), pp. 183–203. Markusen, James and Anthony J. Venables, 2000. ‘The Theory of Endowment, Intra-industry and Multi-national Trade’, Journal of International Economics, 52 (2), pp. 209–34. Marshall, Alfred, 1890. Principles of Economics, London: Macmillan and Co. Navaretti, Giorgio Barba and Anthony J. Venables, 2006. Multinational Firms in the World Economy, Princeton, NJ: Princeton University Press. Nunn, Nathan, 2007. ‘Relationship-Specificity, Incomplete Contracts, and the Pattern of Trade’, The Quarterly Journal of Economics, 122 (2), pp. 569–600.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



     

Ottaviano, Gianmarco I. P. and Diego Puga, 1998. ‘Agglomeration in the Global Economy: A Survey of the “New Economic Geography” ’, World Economy, 21 (6), pp. 707–31. Ottaviano, Gianmarco and Jacques-François Thisse, 2004. ‘Agglomeration and Economic Geography’, in J. Vernon Henderson, and Jacques-François Thisse, eds, Handbook of Regional and Urban Economics, Vol. 4, Amsterdam: Elsevier, pp. 2563–608. Overman, Henry and Diego Puga, 2009. ‘Labour Pooling as a Source of Agglomeration: An Empirical Investigation’. CEPR Discussion Papers No. 7174. Puga, Diego, 2010. ‘The Magnitude and Causes of Agglomeration Economies’, Journal of Regional Science, 50 (1), pp. 203–19. Redding, Stephen, 2010. ‘The Empirics of New Economic Geography’, Journal of Regional Science, 50 (1), pp. 297–311. Redding, Stephen, 2011. ‘Economic Geography: a Review of the Theoretical and Empirical Literature’, The Palgrave Handbook of International Trade, Basingstoke: Palgrave, ch. 16. Rosenthal, Stuart and William Strange, 2001. ‘The Determinants of Agglomeration’, Journal of Urban Economics, 50 (2), pp. 191–229. Rosenthal, Stuart and William Strange, 2004. ‘Evidence on the Nature and Sources of Agglomeration Economies’, in J. Vernon Henderson, and Jacques-François Thisse, eds, Handbook of Regional and Urban Economics, Vol. 4, Elsevier, Amsterdam, pp. 2119–71. Santos Silva, J. M. C. and Silvana Tenreyro, 2006. ‘The Log of Gravity’, Review of Economics and Statistics, 88 (4), pp. 641–58. United Nations Conference on Trade and Development, 2015. World Investment Report, New York: United Nations. Venables, Anthony, 1996. ‘Equilibrium Locations of Vertically Linked Industries’, International Economic Review, 37 (2), pp. 341–59. Yeaple, Stephen, 2003a. ‘The Role of Skill Endowments in the Structure of U.S. Outward Foreign Direct Investment’, Review of Economics and Statistics, 85 (3), pp. 726–34. Yeaple, Stephen, 2003b. ‘The Complex Integration Strategies of Multinational Firms and Cross-Country Dependencies in the Structure of Foreign Direct Investment’, Journal of International Economics, 60 (2), pp. 293–314. Yeaple, Stephen, 2009. ‘Firm Heterogeneity and the Structure of U.S. Multinational Activity: An Empirical Analysis’, Journal of International Economics, 78 (2), pp. 206–15.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

                      ......................................................................................................................

 

11.1 I

.................................................................................................................................. A has been the case for old industrialized countries and new industrialized countries, the economic development of poor countries today implies that GDP growth be accompanied by a ‘major transformation of their economies; and this raises the issue of the processes to guide these structural changes’ (UNECA 2013: preamble). This notion of structural transformation was central to the pioneering theory of economic development. It was subsequently relegated to the background of academic and strategic discussions that, beginning in the 1980s, turned their attention to financial issues and the objective of economic growth.¹ Productive transformation was then moved back to the status of a simple consequence of economic growth and the accumulation of capital, whereas it had until then been considered as its principal driver. During the first fifteen years or so of the twenty-first century, the analysis of structural transformation and the analysis of industrial policies adapted to generate it have once again become major themes of international institutions and an object of study for development economists (Hidalgo et al. 2007; McMillan and Rodrik 2011; Lin 2012; UNIDO 2013; IMF 2014). With his New Structural Economics theory, Lin (2009, 2012) is a major actor in this renewed interest of the industrial issue. Lin and Monga (2010) describe and prioritize policies that promote the emergency of ¹ The models of endogenous growth with structural change that appeared in the 1990s are an exception to this strong tendency (see Acemoglu 2009: ch. 21).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

new sectors and the productive transformation as a whole in developing countries. They particularly highlight the role of comparative advantages in this structural transformation process. A decisive element in the renewal of interest for a modernized conception of structural transformation was the emergence of a series of empirical works studying the modification of productive structure following seminal articles by Imbs and Wacziarg (2003) and Hausmann et al. (2007). The former demonstrated the existence of a quadratic relationship between productive diversification, measured by sectoral value added and employment, and economic development. The latter studied exports and developed an indicator of export sophistication making it possible to show empirically that the sophistication of exports is consubstantial with economic development. Since then the diversification and sophistication of exports have become established as the two indicators making it possible to measure and qualify the process of structural transformation. Recent literature is thus chiefly empiric; it does not propose a theoretical model of productive transformation but attempts to describe the relationship between these two variables and income and endeavours to study what determines them (Imbs and Wacziarg 2003; Klinger and Lederman 2004; Hausmann et al. 2007; Cadot et al. 2011b; Jarreau and Poncet 2012; IMF 2014). Thus, while pioneering development economists endeavoured to describe the evolution of sectoral production and employment structure at a high level of aggregation—they generally considered only three sectors—recent empiric literature is now focusing on the structure of exports and can therefore take up the question of structural change at a very high level of disaggregation. Speaking of productive transformation from export data, however, presupposes a hypothesis: changes in the structure of exports are assumed to be a pertinent indicator of modifications in the productive structure of the country as a whole. But this hypothesis, often ignored in empiric literature, is not confirmed in the current global economic context characterized by the global fragmentation of production. The global economy is indeed characterized by a vertical fragmentation of manufactured production into separate activities that can be carried out in different places by different companies. So there has been a dramatic turn from trade in goods to trade in tasks. Whereas economies are exchanging more and more tasks in the framework of global value chains (GVCs), the analysis of productive transformation by exports is focusing solely on the export of goods. Thus, despite undeniable advantages in terms of the availability of sectoral data, the characterization of productive transformation via exports can be problematic by introducing a statistical artefact linked to the fact that certain exports, as for example processing exports, often have no connection with the rest of the national productive structure. Moreover, while it is acknowledged that integration into GVCs facilitates the industrialization of developing economies in delimited segments of production, it can also be less meaningful in terms of long-term industrialization, since specialization in assembly tasks can force developing countries to amplify their comparative

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



advantages in unskilled labour-intensive activities (Baldwin 2012).² This ‘imported industrialization’ via direct foreign investments and GVCs is likely to generate an ‘immiserising specialization’ for developing economies whose initial capabilities are not very diversified (Gimet et al, 2010). Hence, value chains imply new issues, as much for the process of productive transformation itself as for its study. Only rarely, however, do macroeconomic studies put the study of structural change back into this context of global fragmentation of production and trade in tasks. The few empiric studies questioning the impact that multinational firms, integration into value chains or assembly activities have on productive transformation are carried out on the scale of the firms in a single country.³ As regards more macroeconomic analyses, only Lederman and Maloney (2012) emphasize the need to take into consideration the task actually carried out. The absence of data that would help identify positioning in the GVCs makes it difficult, however, to carry out macroeconomic analyses of the impact of GVCs on the process of structural transformation. In this chapter I propose to analyse the complex, multidimensional nature of structural transformation in a productive context marked by a strong international fractionalization of production. A first section offers a review of recent empiric literature concerning structural transformation and adopting an approach through exports. The usual two dimensions of productive transformation will be presented, that is, export diversification and sophistication. The new issues that GVCs represent for the study and process of structural transformation will be presented in a second section. These two facets of economic literature are as distinctive from each other as they are ignorant of each other. The former is chiefly empiric whereas the latter consists chiefly of affirmations confirmed in rare microeconomic studies. The confrontation between these two literatures will lead to the introduction of a third dimension in the discussion on structural change: the sustainability of the transformation process.

11.2 A  S T  E: D  S

..................................................................................................................................

11.2.1 Export Diversification The fundamental work on diversification focuses on the evolution of the sectoral distribution of employment, products, and exports. The relationship between diversification ² ‘Industrialization is easier and faster but industrialization is less meaningful’ in Baldwin (2012: 317–18). ³ See for example the work of Dai et al. (2016) or Koopman et al. (2008) on China and Paus and Gallagher (2008) on Mexico and Costa Rica.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

and income was first studied in 2003, in Imbs and Wacziarg’s seminal article ‘Stages of diversification’. Using sectoral employment and value added as a measure of production, the authors show that the diversification of production increases non-monotonically with income. Their relationship is illustrated by a quadratic U-shaped curve: economies tend to diversify to an income level assessed at $9,000 per capita in PPP, at which point they begin to specialize again. Klinger and Lederman (2006) and Cadot et al. (2011a) confirm this relationship with data from international trade. In these studies, the turning point is assessed at $20,000 to $25,000 per capita in PPP, showing a decidedly higher degree of export concentration for equivalent income levels.⁴ DeBenedictis et al. (2009) and Parteka (2007) find a negative monotonic relationship between export concentration and income. With economic development, exports tend to diversify but the intensity of the relationship weakens with income; thus there is a certain degree of non-linearity. So, while the various authors agree on the decrease of concentration in the first phases of economic development, there is no real consensus regarding the phase of export concentration of the most developed countries. When it emerges, however, export reconcentration appears at very high income levels. We can therefore conclude that, on average, developed economies have productive structures that are more diversified than those of developing economies. This diversification accompanying economic development can be the result of a more uniform distribution of exports or the introduction of new products. These mechanisms differ according to their positioning in relation to the technological frontier. On one hand, high-income economies delocalize production, necessitating factors of production which they no longer have in abundance, and specialize in activities that are intensive in technology and R&D. On the other hand, countries far from the technological frontier that have accumulated few endowments have few opportunities for diversification (Acemoglu and Zilibotti 1997). However, they have access to technologies already developed in high-income economies (Lederman and Maloney 2012). Thus, the nature of innovations changes with economic development (Klinger and Lederman 2011). Innovation consists of the introduction of both new goods that have never been produced and new goods previously not produced locally. As economies develop, they approach the technological frontier and the innovation process changes from imitation to the introduction of new goods. In economies on the technological frontier, innovation consists of the creation of globally new products. Thus changes in the nature of innovation (imitation or creation) that come from positioning with regard to the technological frontier impact the appearance of new products; they should be more frequent in imitative developing economies far from the technological frontier than in creative, developed economies on the frontier. In other words, developed countries that already have a diversified productive structure have fewer opportunities for diversification than developing economies (Cadot et al. 2011a). ‘Discovery episodes’ are therefore more frequent in developing economies (Lederman and Maloney 2012). In the first stages of economic development, diversification ⁴ Export data are much more disaggregated than production data, and the more disaggregated they are, the higher are the indices of concentration. What is more, exports are theoretically more concentrated than production since only a small proportion of firms export (Bernard and Jensen 2004).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



consists of the introduction of new products and, as development progresses, it changes dramatically towards a more uniform distribution of exports. In the first case, export growth occurs on the extensive margin; it consists of new exports to old or new markets or old exports to new markets. In the second case, export growth occurs on the intensive margin, resulting in an increase and a better distribution of existing exports. The extensive margin indicates the ability of countries to introduce new varieties onto the international market. On the whole, export growth is primarily explained by the intensive margin (Brenton and Newfarmer 2007; Amurgo-Pacheco and Pierola 2008; Besedes and Prusa 2011) and the extensive margin tends to decrease with economic development (Klinger and Lederman 2006; Cadot et al. 2013).⁵ These tendencies are therefore consistent with the evolution of the nature of innovation. Indeed, the extensive margin is particularly dynamic in the least developed economies, particularly in sub-Saharan Africa (Cadot et al. 2013). Entrepreneurship is thus seen as dynamic in low-income countries (Brenton and Newfarmer 2007), but these new exports have a very short life, rarely exceeding two years (Besedes and Prusa 2006). Diversification seems to be a process inherent to economic development (Cadot et al. 2011a); in fact, we see a break between high-income countries and others (Figure 11.1).

Export concentration (Theil index)

Low income

60

4

3.5

50

3.5

50

3

40

3

40

2.5

30

2.5

30

20

2

2 1995

2000

2005

2010

2015

60

20 1995

Lower middle income

2000

2005

2010

2015

Upper middle income

4

60

4

60

3.5

50

3.5

50

3

40

3

40

2.5

30

2.5

30

20

2

2 1995

2000

2005

2010

2015

Exports (% GDP)

High income

4

20 1995

2000

2005

2010

2015

Years Export concentration

Exports of goods and services (% of GDP)

 . Export concentration trend by income, 1995–2014 Source: Author’s calculation based on UN-COMTRADE; WDI.

⁵ The extensive margin is minor in construction: when a new export appears, it contributes little to export growth during the first year. But the following year, as it increases in scale, it is categorized as intensive margin. What is more, depending on the data’s aggregation level, identical products can be counted as intensive or extensive.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Their concentration, on the average, does not exceed 2.5, whereas it nears 3–3.5 in the other economies. Note, furthermore, that the concentration of exports of the poorest countries decreased over the period; it increased for the other three categories of countries with higher incomes. Over the past twenty years, and chiefly in middle-income economies, intensification of trade accompanied an increase in concentration. The economic development of these countries thus seems to be associated with a concentration of exports. Moreover, Chandra et al. (2007) and Parteka (2007) note large disparities between the countries. They find, among other things, that some countries have been able to diversify their productive structure without any real impact on their economic development. McMillan and Rodrik (2011) also introduce the notion of ‘wrong structural change’ as they discuss ‘growth-reducing’ or ‘productivity-reducing structural change’. They point out, for example, that in Latin America and sub-Saharan Africa the labour force migrated in the ‘wrong’ direction, that is, from more productive to less productive activities, in particular informal ones. Chandra et al. (2007) emphasize the existence of ‘something more’, to explain these disappointing results. They suggest the sophistication level of exports as a decisive impact factor for economic development.

11.2.2 Export Sophistication Economists of structural change describe it as the migration of the labour force from the primary sector to the manufacturing sector, thus asserting the superiority of the latter to the former as a driver of economic development. This poses the underlying question of products deemed to be ‘desirable’, because they generate economic development. While this idea of the superiority of certain products is old, Hausmann et al. (2007) empirically demonstrated, in their pioneering article ‘What you export matters’, that products have different impacts on economic development. Some products are therefore more meaningful than others. Over the years, indicators of ‘sophistication’ or ‘complexity’ and classifications ensuing from them have been proposed. These various classifications seek to measure—directly or indirectly—the complexity of production processes. The first classification put forward by the OECD in 1989, and updated in 1997, measures the complexity of products directly via their R&D intensity. It groups industries into four categories according to their technological intensity. However, the rarity of data limits measurement of the technological intensity of products, which makes for the very aggregated level of this first classification. More recent classifications propose indirect methods by assessing the complexity levels of products based on the characteristics of the exporters. The following subsections discuss an indirect approach through the characteristics of exporting countries.

11.2.2.1 Characteristics of Exporting Countries for Measuring Sophistication These classifications are currently the most widely used in development literature. They are based on two postulates: (i) the modernization of productive structure depends on

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



the accumulation of factor endowments and (ii) this accumulation can be assessed indirectly by the characteristics of exporting countries. It is ‘illusory’ to draw up an exhaustive list of all the direct and indirect factors that go into the production of a good (Lall et al. 2005). Besides the ‘traditional’ factor endowments, the production location can also be explained by logistics, the proximity of activities, natural resource needs, infrastructure, the fragmentation level of production, etc. Political factors are also decisive; this is, for example, the case with trade restrictions, tariff barriers, or trade agreements. The sophistication level of products is ‘an amalgam of these influences’ (Lall et al. 2005: 7). This is why, in the classifications of Lall et al. (2005) and Hausmann et al. (2007), the sophistication level of products is assessed via the average income of their exporters. Income is used as a proxy of the totality of the elements likely to impact productive structure; this gives an ‘outcome measure’ (Hidalgo et al. 2007). These classifications are thus truly constructed from the characteristics of the exporting countries, to be specific, on their income. The Sophistication Index of Lall et al. (2005) and the PRODY of Hausmann et al. (2007) are based on equivalent reasoning. They classify products according to their ‘implicit productivity/income level’ assessed by the income levels of the exporting countries. Mathematically, the Sophistication Index of Lall and his co-authors is the average, weighted by the country’s share of world exports, of the income of the countries exporting the good in question. The weight used to construct the PRODY is the revealed comparative advantage of each country in the product in question.⁶ Each product is associated with a productivity/income level: the PRODY.⁷ Exports and income are the two elements constituting the index. They are also its principal limits. Thus the very foundation of the index can be questioned. By definition the PRODY implies that ‘rich products are exported by the rich and poor products by the poor’. Felipe et al. (2012), among others, therefore criticize it for being circular. The PRODY implies that in order to develop, countries must manage to diversify toward products exported by rich countries. There is thus a certain fatality in the evolution of development levels. If productive specializations evolve, the products exported by rich countries will have the highest PRODY levels.

⁶ The PRODY level associated with product k is defined by:   X xjk =Xj Yj PRODYk ¼ X x jk j ð =Xj Þ

ð1:1Þ

j

Xj is the total exports of country j and xjk is the export of the good k by country j. Yj is the per capita GDP of country j. ⁷ A country’s EXPY associates an average income/productivity to the goods basket exported by the country under consideration. It is calculated by the average PRODY weighted by the share of each export in the total basket of the country in question.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

11.2.2.2 Characteristics of Countries and Products for Measuring Sophistication Subsequently, Hausmann’s team (2011) proposed a new indicator, the Product Complexity Index, presented in The Atlas of Economic Complexity.⁸ It falls within the framework of work presented earlier: it is based on the factor endowments necessary to production. The totality of factor endowments or ‘non-tradable inputs’ is called capabilities. According to Hausmann and Hidalgo (2011: 323), each capability represents a ‘specific infrastructure, regulations, norms, and other non-tradable activities, such as port or postal services, whose presence or absence can either facilitate or limit the production of these products’. On the scale of the firm, these capabilities include expertise and practices as well as organizational and managerial competence. Thus the notion of capability comprises all of the factors likely to affect trade.⁹ Whereas the preceding work of Lall et al. (2005) and Hausmann et al. (2007) assessed the entirety of these capabilities indirectly by the income of countries, Hausmann et al. (2011) assess them on the basis of two concepts: the ‘ubiquity’ of goods and the ‘diversity’ of the countries’ export baskets. This classification adopts an approach that is direct in terms of the characteristics of the ‘products’ and indirect in terms of the characteristics of the ‘exporting countries.’ Furthermore, the ‘country-characteristic’ used is based on the structure of the countries’ exports whereas the PRODY is constructed from only their income. So this methodology avoids the criticism of circularity directed at the PRODY. The Product Complexity Index thus seems better adapted than its predecessors for measuring the ‘sophistication’ dimension of a country’s structural transformation. The ubiquity of products makes it possible to judge the complexity of their production. Products with high ubiquity are exported by many countries and their production is thus assumed to be accessible regardless of development level and capabilities. On the contrary, those whose ubiquity is low are exported by few countries. Their production might necessitate a great variety of complex capabilities or else rare capabilities held by few countries. Low ubiquity can thus be explained by (i) the complexity of the production processes or (ii) the rarity of a factor of production. In the first case, the good in question is complex and few countries possess all the capabilities required to produce it. In the second case, its low ubiquity is explained by the rarity of a factor of production; it is not very sophisticated (this is, for example, the case of primary sector products). The authors make a distinction between these two possibilities by integrating the diversity of the export basket of the countries exporting the product in question, for the diversification level furnishes information on the accumulation of capability. Linked to the positive relationship established between diversification and income, the ⁸ The Atlas of Economic Complexity enables users to visualize world trade and its complexity. It results from collaboration between Harvard’s Center for International Development and MIT. It can be consulted at http://atlas.cid.harvard.edu/ ⁹ The concept of capability has been defined by many actors and for this reason is not the object of a general analytic framework. The absence of an analytical framework is a weakness in terms of scientific rigour and can give free rein to a multitude of interpretations.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



authors assume the existence of a positive relationship between capability accumulation and export diversification. Thus if a low ubiquity product is exported by countries whose exports are concentrated, then the low ubiquity is explained by the rarity of a factor of production. The good in question is therefore not very complex. On the contrary, countries able to produce a great diversity of goods have accumulated varied capabilities; the low ubiquity of the good is then explained by the complexity of its production process; the product in question is complex. The correction of the product’s ubiquity by the export diversity of the exporting countries enables the authors to construct the Product Complexity Index (PCI). In the end, the approach adopted to identify the complexity of products can be interpreted in the context of Vernon’s (1966) life cycle of products. In the first phase of the life cycle, economies which are R&D intensive, that is those most developed, introduce a new product on their domestic market. In a second phase, exportation develops, but developed economies still hold the monopoly of the item’s production. During these first two phases, its ubiquity is low. It increases during the phase of maturity marked by production delocalization and grows even more in the phase of decline. During the latter phase, production is concentrated in countries with an unskilled labour force. At the beginning of the cycle, the product’s ubiquity is low. At the end of the cycle, when its production has become accessible to the poorest economies, its ubiquity is high. In parallel, at the beginning of the cycle the product is exported by countries exporting a great diversity of goods whereas at the end of the cycle less developed exporting countries have concentrated exports. Table 11.1 presents the five most and the five least complex products. The very technology-intensive products are in the sectors of electronics, machinery, and communication. Surprisingly, in 2014 the most complex product was based on transformed metals. These metals are often transformed and gain value in industrialized economies. In the case of ‘lead tubes, pipes, and fittings’, 97 per cent of the global exports of this product are Japanese. So the ubiquity is very low and Japan has a diversified export basket; therefore in 2014 it was the most complex country. Similarly, 25 per cent of ‘cermets’ are exported by Germany and 17 per cent by Austria. In parallel, the least complex products are natural resources (agricultural and mining). Almost half, 44 per cent, of tin ores are exported by African countries and Burma, which exports 14 per cent; these economies have very concentrated exports. At the same time, the authors propose an indicator on the scale of the country, called the Economic Complexity Index (ECI), which is measured by the diversity of the export basket corrected by the ubiquity of the products composing it. Countries having diversified exports in low ubiquity products are countries that have accumulated diversified competencies. These countries have a high complexity level. On the other hand, if the export basket is composed of goods that are varied but have a high ubiquity, then the country has a low complexity level. In the country approach, the diversification level of the export basket is corrected by the ubiquity of the products composing it.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Table 11.1 The world’s most and least complex products, 2014 (classification HS-4 digit) HS4-code

Product

Top 5 products by complexity 7805 Lead tubes, pipes, and fittings 8444 Machines to extrude, cut man-made textile fibres 9204 Accordions and similar instruments 8457 Machining centres for working metal 8113 Cermets

PCI

Sector

6.6

Base metals and articles of base metal

5.7

Machinery and mechanical appliances, electrical equipment [ . . . ] Optical, photographic, cinematographic [...] Machinery and mechanical appliances, electrical equipment [ . . . ] Base metals and articles of base metal

5.6 5.1 5

Bottom 5 products by complexity 5303 Jute and other textile fibres 1801 Cocoa beans, whole

4.7 4.6

2615

3.8

Textiles and textile articles Prepared foodstuffs; beverages, spirit and vinegar; tobacco and manufactured tobacco substitutes Mineral products

3.8 3.8

Vegetable products Mineral products

0714 2609

Niobium (columbium), tantalum, vanadium or zirconium ores Manioc (cassava) Tin ores

Source: The Atlas of Economic Complexity.

Table 11.2 lists the ten most and ten least complex countries. Surprisingly, Hungary and the Czech Republic are more complex than the United States or France. This is also the case for the Slovak Republic and Slovenia. A look at the structure of exports of these four Eastern European countries clarifies this result. In 2014, their exports were diversified and their chief exports, ‘automobiles’ and ‘automotive parts and accessories’, (representing about 10 per cent of the total exports) had a low ubiquity. It turns out that five countries are responsible for more than half of the total exports of these products. In terms of complexity, automobiles are only in the 414th position and ‘automotive parts and accessories’ in the 155th position in a classification comprising 1,240 products. At the same time, the least complex countries are large exporters of natural resources. These economies are also conspicuous in Figure 11.2, illustrating the relationship between economic complexity level and per capita income. We see the existence of a non-linear relationship with a turning point above $40,000 per capita in PPP. The sophistication levels of countries with incomes of less than $20,000 per capita are strongly dispersed. Some low-income countries manage to reach a high complexity level; this is, for example, the case of the Philippines (whose sophistication level comes from exports of ‘electronic integrated circuits’ and ‘automatic data processing

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Table 11.2 The ten most and the ten least complex countries 10 most complex

10 least complex

Country

ECI

Country

ECI

Japan Germany Switzerland South Korea Sweden Austria Czech Republic Finland Hungary United Kingdom

2.2 1.9 1.9 1.8 1.7 1.7 1.6 1.6 1.5 1.5

Angola Nigeria Sudan Guinea Yemen Mauritania Libya Congo Papua New Guinea Malawi

2.3 2.1 1.7 1.7 1.7 1.6 1.6 1.5 1.5 1.4

Source: The Atlas of Economic Complexity.

Lowess smoother JPN

2

Economic Complexity Index

CZE HUN SVN SVK CHNMEX EST THA ROU POL BLRHRV MYS LVA LTU PRT BIH BGR PHL UKR TUR PAN IND TUN GRC JOR URY RUS SLVIDNZAF CRI BRA LBN VNM MDA EGYCOL MUS GEO ARG DOM JAM MKD GTM CHL SEN HND PRY MAR KAZ KHM ZMB ALB KEN CMR PAK NAM PER BWA ZWE NIC UGA CIV IRN BGD MDG TZA ECU TKM VEN MOZ BOL GAB ETH AZE GHA DZA MNG MWI PNG MRT GINYEM SDN

1

0

–1

–2

DEU SWE AUT

KOR

CHE

FIN GBR ITA FRA IRL USA HKG BELDNK ISR NLD ESP

SGP

NOR

CAN NZL

OMN

ARE AUS

SAU

KWT

NGA

0

20000

40000 60000 GDP per capita (constant 2010 US$)

80000

 . Economic complexity and GDP per capita (PPP 2010 US$), 2014 Source: Author’s based on Hausmann et al. (2011) and WDI.

equipment’). Similarly, we note a large income level dispersion in countries whose sophistication level is close to 1. On one hand, there are China, Mexico, and Romania whose per capita income (in PPP) is less than $20,000 and, on the other hand, Singapore, Holland, and Denmark whose incomes are over $40,000 per capita (PPP). The few

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

countries with high income and low sophistication levels are important exporters of natural resources. This is, for example, the case of Norway, Oman, and Australia. By combining the notions of diversification and sophistication, the PCI seems to be more suitable for measuring the sophistication level than the PRODY. This indicator has the advantage of going beyond the PRODY’s circularity limit and being an amalgam of the characteristics of countries and products. By attempting, however, to associate a sophistication level with products—the export approach—rather than with the production process carried out, all of these classifications face a certain number of limitations that must be emphasized. The different sophistication measures endeavour to characterize the product with the idea, following the example of Hausmann et al. (2007), that ‘what you export matters’. Lederman and Maloney (2012), in their article ‘Does what you export matter?’, contest the pertinence of a ‘product’ type approach. They defend ‘how’ in relation to ‘what’, that is, an approach by ‘tasks’ rather than ‘products’. According to Lederman and Maloney (2012: 2), ‘the externality argument is one of the strongest for asserting the superiority of some goods over others’. Nevertheless, depending on its production process (that is, ‘how’), the same good can generate different externalities. The production of a good can be carried out with different technologies and different combinations of factors of production. There are therefore varied production processes able to generate different externalities. What is more, by restricting the analysis to products, the notion of sophistication does not take into account the concept of innovation, although this is a driver of structural change (Lederman and Maloney 2012). According to Schumpeter (1949), innovation covers (i) the introduction of new merchandise, a new quality of existing merchandise, or new production methods, (ii) access to a new market, (iii) the appearance of a new supply source, or (iv) the installation of a new organization within the industry. These different innovations are the result of the accumulation of factor endowments and are not taken into account in current measures of sophistication. Indeed, the complexity measures (PRODY and PCI) are disconnected from the concept of innovation despite its being theoretically at the very origin of the process of modernizing productive structures. Finally, these data include re-exports that are, however, totally independent of the productive structure of economies. Lederman and Maloney (2012) cite, for example, the case of Singapore which is an important platform for the re-exportation of coffee although it does not produce any itself. In conclusion, an identical level of sophistication can come from very diverse export structures. The average sophistication of an export basket can derive from a structure that is sophisticated in its entirety, as is, for example, the case in highincome economies. It can also come from a few very sophisticated major exports. Thus, while sophistication adds to the analysis of export diversification, the reverse is also true. Economic literature often dissociates these two dimensions of structural transformation even though they are complementary. Taken alone, they can lead to distorted conclusions regarding the structural transformation process in progress.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



11.3 T A  E  G V C

.................................................................................................................................. Diversification and sophistication of the productive structure are the two principal dimensions of structural change. In order to study them, international trade data reporting exports of goods at a high disaggregation level are used. The fragmentation of the productive process, however, implies a distance between the export of the finished product as reported in this data and the creation of value added, for there are growing differences between countries exporting finished products and economies creating value added. We find, for example, that the growth of manufactured exports is clearly greater than that of value added. The share of value added in the growth of exports is thus decreasing. According to Johnson and Noguera (2016), this decline coincides with changes in the global economy. It reveals a double counting of exports and of the growing trade in intermediate input. The growth of GVCs thus automatically distorts the conclusions drawn from trade. Consequently, theoretical and empiric studies on structural transformation are somewhat upset in this new economic context. With the international division of labour, countries specialize in tasks within the production of a good, and therefore export and import a large amount of the same product. According to theories of international trade, this indicates a comparative advantage and disadvantage in the same sector. Countries have a comparative advantage in a task, but not in the final good, as trade analyses might predict. This leads us to reconsider conclusions drawn exclusively from trade in terms of the comparative advantage of countries and the sophistication level of exports (Schott 2008; Koopman et al. 2014). In this context the expression ‘kaleidoscopic comparative advantage’ can be used to emphasize the volatility of the factors determining the geographic location of the various activities. Bhagwati and Dehejia (1993), for example, emphasize that comparative advantages can be unstable since there can be a break between the factor content of the exported goods and the factor endowments of the economies that export these goods.¹⁰ The use of data from international trade is thus not appropriate for measuring task trade even though it is a major characteristic of the current economic environment. While studies focusing on structural change use the ‘product’ as object of study, there can be a great difference between what a country exports and its true participation in the process of production (Johnson and Noguera 2016). By reporting the gross flows rather than the values added, it is difficult to establish the kind of activity carried out in the exporting economy. A country’s exports therefore do not reflect the factor and

¹⁰ Recent work by international trade economists proposes an adaptation of the Heckscher-Ohlin (HO) model on the scale of the task carried out (Grossman and Rossi-Hansberg 2008; Baldwin and Robert-Nicoud 2014).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  SCENARIO 1 x

y

z Home Foreign x

Export

Export

Home Foreign

z

SCENARIO 2

y

 . Ordinary versus processing trade Source: Van Assche and Gangnes (2010).

technological endowments used during the domestic production activity. They can reflect the production technologies and factors imported through intermediate input. In this context, two kinds of export activities can be distinguished: ordinary and process. Ordinary trade includes the importation of goods for final consumption and exportation not based on imported inputs. Reciprocally, processing trade includes taxexempt imports that will be incorporated into exports and the exports that are intensive in these imported inputs (Feenstra and Wei 2010: 2). As illustrated in Figure 11.3, proposed by Van Assche and Gangnes (2010), the exported good z is produced from two inputs, x and y. Two scenarios are conceivable. In the first case, inputs x and y and exported good z are produced locally. In the second scenario, inputs x and y are imported to be assembled locally in order to produce z which will then be exported. The export structure is identical in the two scenarios (z is exported) but in the second case international trade data will overvalue the economy’s production activities and the driving effects of manufactured exports on the rest of the economy. They include x and y in the value of z even though they are not locally produced. The various sophistication measures are based on the hypothesis that economies’ export baskets reveal their factor endowments. Consequently, whatever the scenario (ordinary or process), the sophistication level of export z is equivalent. In the second scenario, the sophistication level of exported product z is actually imported. Chinese customs data distinguish ordinary from process type activities. Thus more in-depth work on the impact of value chains on Chinese structural transformation does in fact exist. Dai et al. (2016) have shown that process and ordinary activities are different. The former are less productive, pay lower wages, invest less in R&D, and are not very capital-intensive. They are intensive in unskilled labour and are concentrated in firms with foreign capital. Jarreau and Poncet (2012) and Poncet and Starosta de Waldemar (2013) show that economic growth is driven only by upgrading ordinary activities exports. There are no gains from processing activities or from foreign firms even though these are the chief ‘actors’ in the upgrading of Chinese exports. The origin of upgrading is therefore crucial in judging countries’ industrialization capacities. With the fragmentation of production processes, there is a growing gap between the task carried out and the exports reported in international trade data. The various tasks

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



involved in the process of producing a product are counted in a single product line although they can involve different capabilities. No matter what task is carried out, only the products listed in international trade data determine the sophistication level. Because of this, no distinction is made between an assembly task and a task that is more complex if both activities participate in making the same product. The more fragmented the productions, the more critically important is this observation. Lall (2000) introduced very early on the expression ‘statistical illusion’ to describe the specialization of developing countries in high technology products while they are specializing in labour-intensive process. Jarreau and Poncet (2012) also speak of an ‘artefact’ due to an unsuitable measurement of export sophistication. The expressions ‘statistical illusion’ and ‘statistical artefact’ describe sophistication levels resulting from the data used rather than from a real accumulation of capabilities. Hausmann et al. (2007) ignore this problem when they claim that exporting high-PRODY products unconditionally drives growth; this is not guaranteed in the case of processing exports. Srholec (2007), for example, shows that Asian economies which manage to join GVCs experience a meteoric increase in high-technology exports, but no improvement of their capabilities. So the high-tech product must be distinguished from the high-tech process. The sophistication level is in this case the result of participation in labour-intensive stages of the production processes of hightechnology goods, which mechanically increases the assessed sophistication level based on the export data of finished products. As stated in the report African Economic Outlook 2014, ‘The presence of high-tech goods in a country’s export basket therefore no longer implies the presence of a wide set of industrial capabilities, but merely the presence of the respective assembly operation’ (OECD, ADB, and UNDP 2014: 129). Thus the export of sophisticated products is no longer ‘the hallmark’ of a successful transformation (Baldwin 2012). Sophistication measures based on export baskets give a potentially distorted image of economies’ capabilities. If the global economic context turns the study of structural transformation upside down, it is not without consequence on its process. The effects of integration into value chains are hotly debated in development literature, and while some emphasize the opportunities they represent, others point out the considerable risks they carry. Joining GVCs offers new prospects to developing economies and some authors claim they have become a condition vital to economic development (Gereffi and Fernandez-Stark 2011). For example, they make it possible to create jobs, integrate into international trade, and attract direct foreign investment while driving the emergence of higher value added activities. The international division of production currently makes it possible for developing economies to fit directly into existing value chains. They can participate in world trade and become industrialized without having to develop an entire economic sector, and this despite limited capacities (Baldwin 2012). Indeed, in concentrating on a segment of the value chain, countries do not need to develop all the upstream and downstream competencies (Cattaneo et al. 2013). The stronger the product’s international divisibility, the more entry possibilities there are that in turn facilitate more industrialization possibilities at all levels of development. Value chains

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

make industrialization less complex and faster (Baldwin 2011).¹¹ By joining them, economies can become industrialized and export complex products (listed as such in international trade data) while maintaining their initial factor endowments. This is the direct effect. For there to be a profound structural transformation, however, countries must succeed in accumulating new capabilities in order to diversify and modernize their productive structure. The heart of the debate thus focuses on the externalities of industrialization by insertion into value chains. Whereas one side of the literature asserts the existence of a transfer of knowledge and technologies, many authors emphasize that only simple activities, which do not act as drivers, are delocalized. While international production fragmentation can imply a mobility of skills and managerial competencies, as a general rule only non-essential activities are delocalized, resulting in the transmission of skills needed for low added value processing tasks (OECD 2008). Thus there is in fact little knowledge transfer with few effects on the domestic industrial fabric. And if there are potential transfers, these economies have low absorption capacities that do not allow them to appropriate the foreign technologies. In this context, economies tend to specialize in very precise and simple tasks; so called task-based production. So there is an increase in the comparative advantage in intensive unskilled labour activities. The major risk lies in remaining confined to simple tasks without accumulating new capabilities. This will check future structural transformation. Indeed, according to Lin (2012), capital accumulation and the upgrading of factor endowments are the engines of a sustainable structural transformation. The report African Economic Outlook 2014 specifies that ‘without upgrading and the accumulation of new capabilities . . . GVC integration risks downgrading’ (OECD, ADB, and UNDP 2014: 131). In the end, fragmentation of the productive process tends to increase countries’ specialization in niches in the value chain, resulting in ‘lock-in situations’ (Srholec 2007), whereas diversification of the productive structure is a fundamental dimension of structural change. Thus, this form of production can stimulate industrialization, but the major issue is then to succeed in moving along on value chains while accumulating new capabilities.

11.4 C: T S   P  S T

.................................................................................................................................. I have emphasized in this chapter that the limits of an approach based on exports, generally ignored in literature, become particularly problematic in the context of value chains. While export diversification and sophistication are assumed to characterize ¹¹ This industrialization and integration into trade should drive socio-economic improvements, among other things, by job creation, particularly for populations previously excluded from the job market (women and migrants). It thus presents new income sources for households.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



productive transformation because they are the signal of the accumulation of new domestic capabilities, integration into GVCs can muddy this signal. Indeed, at this time exports are no longer necessarily the continuity of production; drawing conclusions concerning productive transformation only from the analysis of exports can therefore be risky. The use of data from international trade, however, seems inevitable. This is why, in order to correct the distortion caused by the approach through exports, I propose adding the dimension of sustainability to the usual definition of productive transformation. Sustainability comprises the notions of the continuity and depth of the transformation process. Consequently, I define sustainable productive transformation as a process of export diversification and sophistication resulting from the accumulation of capabilities, thereby enabling the country to enter a virtuous circle of lasting transformation. While the data available at this time lead us to use these indicators of sophistication calculated from international trade data, it seems necessary to integrate the limits associated with these indicators into the analysis of productive transformation. Export structure is no longer necessarily the mirror of factor endowments; and yet the indicators used to describe productive transformation rest on the assumed relationship between factor endowments and export structure. Depth, then is called upon to distinguish sophistication that is an illusion/artefact from a sophistication that is the result of an accumulation of capabilities. Recent studies agree on the fact that industrialization by insertion into value chains is faster and easier; Baldwin (2012) adds that it is also less meaningful. He makes it clear that ‘now, exporting sophisticated manufactured goods is no longer the hallmark of having arrived. It may simply reflect a nation’s position in a global supply chain’ (Baldwin 2012: 19). Entry into the era of industrialization thus does not necessarily mean entry into a virtuous circle of transformation. With insertion into value chains, manufactured exports can increase even as they concentrate, and without becoming more complex. They thus drive structural change, but do not always allow its continuity. In fact, industrialization via insertion into value chains can bring about a specialization in specific activities. In being confined to the simplest stages, developing countries increase their comparative advantage in intensive unskilled labour activities, which can, in the long term, block the continuity of the transformation process. Continuity, as I understand it, is entry into a virtuous circle of transformation (thus avoiding the specialization trap).

R Acemoglu, D., 2009. Introduction to Modern Economic Growth, Princeton, NJ: Princeton University Press. Acemoglu, D. and F. Zilibotti, 1997. ‘Was Prometheus Unbound by Chance? Risk, Diversification, and Growth’, The Journal of Political Economy, 105 (4), pp. 709–51.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Amurgo Pacheco, A. and M. D. Pierola, 2008. ‘Patterns of Export Diversification in Developing Countries: Intensive and Extensive Margins’. Policy Research Working Paper No. 4473, World Bank. Baldwin, R., 2011. ‘Trade and Industrialisation after Globalisation’s 2nd Unbundling: How Building and Joining a Supply Chain Are Different and Why it Matters’. NBER Working Papers No. 17716. Baldwin, R., 2012. ‘Global Supply Chains: Why They Emerged, Why They Matter, and Where They Are Going’. CEPR Discussion Papers No. 9103, Centre for Economic Policy Research. Baldwin, R. and F. Robert-Nicoud, 2014. ‘Trade-in-goods and Trade-in-tasks: An Integrating Framework’, Journal of International Economics, 92 (1), pp. 51–62. Bernard, A. B. and J. B. Jensen, 2004. ‘Why some Firms Export’, The Review of Economics and Statistics, 86 (2), pp. 561–9. Besedes, T., and T. J. Prusa, 2006. ‘Ins, Outs, and the Duration of Trade’, Canadian Journal of Economics, 39 (1), pp. 266–95. Besedes, T. and T. J. Prusa, 2011. ‘The role of extensive and intensive margins and export growth’, Journal of Development Economics, 96 (2), pp. 371–9. Bhagwati J. and V. Dehejia, 1993. ‘Freer Trade and Wages of the Unskilled: Is Marx Striking Again?’ Discussion Paper No. 672, Conference Paper, ‘The Influence of International Trade on US Wages’. Brenton, P. and R. Newfarmer, 2007. ‘Watching More than the Discovery Channel: Export Cycles and Diversification in Development’, Policy Research Working Paper No. 4302, World Bank. Cadot, O., C. Carrere, and V. Strauss-Khan, 2011a. ‘Export Diversification: What’s Behind the Hump?’, Review of Economics and Statistics, 93 (2), pp. 590–605. Cadot, O., C. Carrere, and V. Strauss-Kahn, 2011b. ‘Trade Diversification: Drivers and Impacts’, in M. Jansen, P. Ralf, and J. M. Salazar-Xirinachs, eds, Trade and Employment: From Myths to Facts, Geneva: ILO. Cadot. O., C. Carrère, and V. Strauss-Kahn, 2013. ‘Trade Diversification, Income, and Growth: What Do We Know?’ Journal of Economic Surveys, 27 (4), pp. 790–812. Cattaneo, O., G. Gereffi, S. Miroudot, and D. Taglioni, 2013. ‘Joining, Upgrading and Being Competitive in Global Value Chains: A Strategic Framework’. Policy Research Working Paper Series No. 6406, World Bank. Chandra, V., J. Boccardo, and I. Osorio Rodarte, 2007. ‘Export Diversification and Competitiveness in Developing Countries’. Working Draft, World Bank. Dai, M., M. Maitra, and M. Yu, 2016. ‘Unexceptional Exporter Performance in China? The Role of Processing Trade’, Journal of Development Economics, 121 (C), pp. 177–89. DeBenedictis, L. M. Gallegati, and M. Tamberi, 2009. ‘Overall Trade Specialisation and Economic Development: Countries Diversify’, Review of World Economics, 145 (1), pp. 37–55. Feenstra, R. C. and S.-J. Wei, 2010. China’s Growing Role in World Trade, Chicago: University of Chicago Press. Felipe, J., U. Kumar, A. Abdon, and M. Bacate, 2012. ‘Product Complexity and Economic Development’, Structural Change and Economic Dynamics, 23 (1), pp. 36–68. Gereffi, G. and K. Fernandez-Stark, 2011. Global Value Chain Analysis: A Primer, Durham, NC: Center on Globalization, Governance & Competitiveness. Gimet, C., B. Guilhon, and N. Roux, 2010. ‘Fragmentation and Immiserising Specialization: The Case of the Textile and Clothing Sector’, Working Paper GATE 2010-03.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Grossman, G. M. and E. Rossi-Hansberg, 2008. ‘Trading Tasks: A Simple Theory of Offshoring’, American Economic Review, 98 (5), pp. 1978–7. Hausmann, R., and C. Hidalgo, 2011. ‘The Network Stucture of Economic Output’, Journal of Economic Growth, 16 (4), pp. 309–42. Hausmann, R., J. Hwang, and D. Rodrik, 2007. ‘What You Export Matters’, Journal of Economic Growth, 12 (1), pp. 1–25. Hausmann, R., C. Hidalgo, S. Bustos, M. Coscia, A. Simoes, and M. A. Yildirim, 2011. The Atlas of Economic Complexity, Cambridge, MA: Harvard University Press. Hidalgo, C. A. et al., 2007. ‘The Product Space Conditions the Development of Nations’, Science, 317 (5837), pp. 482–7. Imbs, J. and R. Wacziarg, 2003. ‘Stages of Diversification’, American Economic Review, 93 (1), pp. 63–86. IMF, 2014. ‘Sustaining Long-Run Growth and Macroeconomic Stability in Low-Income Countries: The Role of Structural Transformation and Diversification’. IMF Policy Paper. Jarreau, J. and S. Poncet, 2012. ‘Export Sophistication and Economic Growth: Evidence from China’, Journal of Development Economics, 97 (2), pp. 281–92. Johnson, R. C. and G. Noguera, 2016. ‘A Portrait of Trade in Value Added over Four Decades’. NBER Working Papers No. 22974, National Bureau of Economic Research. Klinger, B. and D. Lederman, 2004. Discovery and Development: An Empirical Exploration of ‘New’ Products. Working Paper No. 3450, World Bank Policy Research. Klinger, B. and D. Lederman, 2006. ‘Diversification, Innovation, and Imitation Inside the Global Technology Frontier’, World Bank Policy Research Working Paper No. 3872, World Bank. Klinger, B. and D. Lederman, 2011. ‘Export Discoveries, Diversification and Barriers to Entry’, Economic Systems, 35 (1), pp. 64–83. Koopman, R., Z. Wang, and S.-J. Wei, 2008. ‘How Much of Chinese Exports is Really Made in China? Assessing Domestic Value-Added When Processing Trade is Pervasive’. NBER Working Papers No. 14109, National Bureau of Economic Research. Koopman, R., Z. Wang, and S.-J. Wei, 2014. ‘Tracing Value-Added and Double Counting in Gross Exports’, American Economic Review, 104 (2), pp. 459–94. Lall, S., 2000. ‘The Technological Structure and Performance of Developing Country Manufactured Exports, 1985–1998’. Working Paper No. 44, University of Oxford. Lall, S., J. Weiss, and J. Zhang, 2005. ‘The Sophistication of Exports: A New Measure of Product Characteristics’, QEH Working Papers QEHWPS 123, University of Oxford. Lederman D., and W. F. Maloney, 2012. Does What You Export Matter? Washington, DC: World Bank. Lin, J. Y., 2009. Economic Development and Transition, Thought, Strategy and Viability, Canbridge: Cambridge University Press. Lin, J. Y., 2012. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: The World Bank. Lin, J. Y. and C. Monga, 2010. ‘Growth Identification and Facilitation, The role of the State in the Dynamics of Structural Change’. Policy Research Working Paper No. 5313, The World Bank. McMillan, M. S. and D. Rodrik, 2011. ‘Globalization, Structural Change and Productivity Growth’, NBER Working Papers No. 17143, National Bureau of Economic Research, Inc. OECD, 2008. ‘Export Diversification and Global Value Chains: Lessons from Selected Case Studies’, Business for Development, Paris: OECD Publishing.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

OECD, ADB, and UNDP, 2014. African Economic Outlook 2014: Global Value Chains and Africa’s Industrialization. Thematic Edition, Paris: OECD Publishing. ONU, 2013. Diversification and Sophistication as a Lever for the Structural Transformation of North African Economies, Rabat: ECA/SRO-NA. Parteka, A., 2007. ‘Employment Export Specialization Patterns versus GDP per Capita Performance—Unifying Approach’, Quaderno di ricerca No. 302, Universita politecnica delle marche. Paus, E. A. and P. Gallagher, 2008. ‘Missing Links: Foreign Investment and Industrial development in Costa Rica and Mexico’, Studies in Comparative International Development, 43, pp. 53–80. Poncet, S. and F. Starosta de Waldemar, 2013. ‘Export Upgrading and Growth: The Prerequisite of Domestic Embeddedness’, World Development, 51 (C), pp. 104–18. Schott, P. K. 2008. ‘The Relative Sophistication of Chinese Exports’, Economic Policy, 23 (53), pp. 5–49. Schumpeter, J., 1949. ‘Economic Theory and Entrepreneurial History’, in R. R. Wohl, ed., Change and the Entrepreneur: Postulates and the Patterns for Entrepreneurial History, Research Center in Entrepreneurial History, Cambridge, MA: Harvard University Press. Srholec, M., 2007. ‘High-tech Exports from Developing Countries: A Symptom of Technology Spurts OR Statistical Illusion?’, Review of World Economics, 143 (2), pp. 227–55. UNECA, 2013. ‘Diversification and Sophistication as a Lever for the Structural Transformation of North African Economies’, ECA-NA/PUB/2013/2. UNIDO, 2013. Sustaining Employment Growth: The Role of Manufacturing and Structural Change, Industrial Development Report. Van Assche, A. and B. Gangnes, 2010. ‘Electronics Production Upgrading: Is China Exceptional?’, Applied Economics Letters, 17 (5), pp. 477–82. Vernon, R., 1966. ‘International Investment and International Trade in the Product Cycle’, The Quarterly Journal of Economics, 80 (2), pp. 190–207.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

                  Challenges and Opportunities ......................................................................................................................

    

12.1 I

.................................................................................................................................. S the first Industrial Revolution, infrastructure improvements have changed the boundary of production and reshaped economic geography. More recently, the information and communication technology (ICT) revolution has not only increased productivity, but also redefined the function of time and distance. The substantial decline of transaction costs enables finer specialization—in parts and activities, rather than the final products—to become an essential unit for trade. Global value chains (GVCs) redefine the mode of distribution of productive activity and the boundaries of trade. This opens new doorways for developing countries. Firms can seize potential opportunities in GVCs to specialize in the niches most consistent with their comparative and competitive advantages. Developing countries can jumpstart structural transformation and attain economic diversification at a lower level of development through knowledge transfers and product differentiation. People, as workers, owners of capital, and consumers, can have better opportunities for employment, investment, and consumption thanks to their access to global markets and can be more resilient to risks of a local nature. This chapter discusses the opportunities and challenges brought about by GVCs. It starts with an analysis of the ways in which they are (re)shaping geography and transforming the global economy (Section 12.2). It highlights some of the main risks and opportunities that firms and people face as they participate in GVCs (Section 12.3), with an examination of special challenges from the perspective of developing countries (Section 12.4). It ends with a discussion of some policy issues (Section 12.5).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

12.2 G V C (R) G  T  G E

.................................................................................................................................. GVCs have redefined the mode of production and the allocation of productive resources across the developed and developing world (Baldwin 2013). Progress all along the logistics chains, from containers to internet, enables the smooth flow of goods and services in a coordinated and inexpensive way, opens up opportunities for dispersed activities across national borders, and reshapes the geo-economic structure of the world economy. Infrastructure development includes two main stages: • Stage I: Development of transport infrastructure. The steam revolution in the 1800s lowered transport costs and made it feasible to spatially separate production and consumption. Production became specialized and concentrated in select areas. Proximity between firms clustering together lowered the coordination costs for complex production, and the heterogeneous patterns of production are reinforced (Florida 2005). • Stage II: Development of communication infrastructure. The ICT revolution in the 1980s lowered the coordination costs and separated the production stages previously performed in close proximity. The decline of the costs of coordination shifted the locus of globalization from sectors to stages of production, and transformed the pattern of activity in the global economy. Firm-specific knowledge and know-how became more internationally sharable, and technology became more mobile as the costs and risks of combining developed-economy technology with developing-nation labour are reduced. The large wage differences between developed and developing nations make it profitable to separate the stages of production across national borders, which strengthens this new international division of labour (Baldwin 2011a). The geographic unbundling of the stages of production opens new doors for developing economies to specialize in particular segments of the value chain where they have comparative advantages without the need of first building a large industrial base (Baldwin 2011b). The minimum effort required for a developing country to engage in global value chains is drastically reduced. In many cases, instead of saying a product is made in a certain country, it is more accurate to say it is made in the world with the final assembly taking place in a certain country. Developing economies can integrate into GVCs at a specific stage, starting from assembly in manufacturing and commodity production in agriculture, without having to build their entire value chain from scratch, as Japan and the Republic of Korea did in the late twentieth century (Baldwin 2013). As countries specialize in tasks and business functions rather than specific products, value chains tend to become increasingly global and spread through the developing

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



world. Countries no longer rely exclusively on domestic resources to produce goods and services, and thus become more interdependent. Products at different stages of value added may be imported, exported, and re-exported multiple times. More intermediate goods are traded across borders, and more imported parts and components are embodied in exports (Feenstra 1998). Intermediate goods, which are traded between countries and different value-adding activities that occur in various locations, have been the main drivers of the boom in trade since the 1990s. In 2009, world exports of intermediate goods exceeded the combined export values of final and capital goods for the first time, representing 51 per cent of non-fuel merchandise exports (WTO and IDE-JETRO 2011: 81). The heightened role of GVC trade in the global economy has led to a revolution in traditional trade statistics. In the past, most observers focused on the gross value of trade flows, but with the growing significance of imported intermediates in the exports of developing economies, especially for manufactured goods, the WTO, OECD, and other international organizations have developed new datasets based on value-added trade, or ‘net exports’ (gross exports minus imported intermediates) (WTO and IDEJETRO 2011; OECD 2011). This has dramatically changed our understanding of who captures the most value generated via export trade. In the case of Apple’s iPhone exports from China, for example, when a conventional measure is used that assigns the gross export value of the product to the exporting country, China incurs a $169.41 trade deficit with the United States for each unit shipped—that is, the final good factory price ($194.04) minus US inputs sent to China ($24.63 per phone). In value-added terms, however, the largest portion of the US trade deficit from its iPhone4 imports is incurred not with China, but via indirect exports from South Korea, Germany, and other high-value component suppliers. China’s actual value added is $6.54 per iPhone assembled and exported to the USA, which is just 3.4 per cent of the ex-factory final good export price (Gereffi and Lee 2012: 27). China’s trade deficit with the USA and other major trading partners is thus far smaller when calculated as net exports than gross exports.¹ After the 2008–09 economic crisis, global trade growth slowed significantly. This has led some to question whether the global economy has run into a ‘peak trade’ constraint—that is, the ratio of global trade to gross domestic product (GDP) has reached a limit (The Economist 2014). And if so, has the contribution of GVCs to structural transformation in the global economy also peaked? Indeed, the trade-toGDP ratio for the world as a whole has risen from about 25 per cent in the 1960s to 60 per cent today (Hoekman 2015: 5), which could turn out to be a peak for the world if the recent decline in trade is sustained, and in 2011–13, the volume of world trade expanded at an annual rate of just 3 per cent, which is less than half the average of roughly 7 per cent for the period 1985–2007 (Gangnes et al. 2015: 111). However, there ¹ Domestic content only accounts for about half of China’s manufacturing exports, and it is considerably smaller (18 per cent) in its processing trade exports, which are mostly done by foreignowned firms (Koopman et al. 2008).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

are both cyclical and structural factors involved in explaining the global trade slowdown, and the role of GVC participation is related to the structural factors. The most prominent cyclical (macroeconomic) explanation for the lack of trade dynamism since the economic crisis is weakness in aggregate demand, most notably in the Eurozone (intra-EU trade accounts for about one-third of global trade) but also more recently in China (which represents another 10 per cent of global imports) as well as sluggish trade in emerging economies (Hoekman 2015: 8). However, there is general agreement that GVC-based production strategies were a major factor behind the rapid growth of global trade relative to income in the 1990s and the first half of the 2000s.² GVC growth was a major feature of trade expansion in China as well as central and eastern Europe in the pre-crisis era, and the re-integration of these regions had a juggernaut effect on the volume and structure of world trade that will diminish as China in particular rebalances its economy towards the domestic market and away from the export-driven growth model. But there is still considerable growth potential for the GVC model elsewhere. Many developing countries in regions of the world that have lagged behind Europe, East Asia, and North America could increase their participation in GVCs, especially if international trade in services expands more rapidly. New technologies are also enhancing the ability of small firms to link up with the global economy, and GVCs can facilitate that process. In the past two decades, parts exports from developing countries rose much more than developed country exports. During the 2000s, major exporters of intermediate and final manufactured goods include China, South Korea, and Mexico, and major exporters of primary products include Brazil, Russia, and South Africa (Gereffi 2015). Consider the BRICs (Brazil, Russia, India, and China) as an example. The opening of the BRICs in the late 1980s added huge product and labour markets, which greatly accelerated the globalization process. These giant economies offered a seemingly inexhaustible pools of low-wage workers, increasingly capable manufacturing and trade infrastructures, abundant raw materials and huge underserved domestic markets with incipient middle classes. In the 1990s and 2000s, GVC activities grew exponentially: China became the ‘factory of the world’, India the world’s ‘back office’, Brazil focused on agricultural and primary commodities, and the Russian Federation on natural resources (Gereffi and Sturgeon 2013). As more South–South trade occurred between developing countries in recent years, this growth involving emerging economies has contributed to shifting end markets in GVCs (Staritz et al. 2011). As Porter argued in his well-known work (Porter 1985), which applied the Ricardian principle of comparative advantage to a firm’s value chains, firms can focus on what

² GVC trade, which can be operationalized as imported intermediates that are used for the production of another country’s exports, as a share of world exports grew by more than 10 percentage points from 40 per cent in 1995 to 52 per cent in 2008, before declining slightly in 2009 (Gangnes et al. 2015: 114–15).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



they do best and outsource the rest. As transaction costs and coordination costs are lowered, the activities along a value chain can be performed within the same firm or divided among different firms in diverse locations. The distribution of production activities across the geographic dimension reflects the balance of the costs and benefits of specialization and dispersion (Baldwin 2013). The gains from specialization and cost of dispersion fall as coordination technology improves and transportation costs decline. The gain from dispersion rises with the diversity of production conditions in various nations. Fractionalization of the GVCs is conditioned by the interplay between the gains from specialization and the costs of coordination and risk. From the spatial dimension, the decline of transportation costs and coordination costs shifts the balance of forces of dispersion and agglomeration (Krugman 1991; Puga and Venables 1996; Fujita et al. 1999). The finer division of labour allows firms to sort stages geographically according to the cost of labour, capital, and technology. A spatial sorting resulted in a pattern of North-to-South offshoring with skill-intensive stages concentrated in high-wage nations and labour-intensive stages in low-wage ones. The heightened international mobility of technology enables the separation of labourintensive stages of manufacturing to developing countries. Joining GVCs through specializing in goods and services that are most suitable to a country’s evolving comparative and competitive advantages instead of having to follow the old-fashioned ‘import substitution’ model offers new ways for developing countries to industrialize.³ Countries with different capabilities tend to concentrate on distinct segments of the value chain. Developing countries, such as Malaysia, China, and Vietnam, carry out the fabrication stages while moving up the value chains from garments and textiles to full-package production of a wide range of electronics and other consumer products; the advanced-technology nations, such as Finland, Japan, and Germany, are increasingly focusing on designing and making sophisticated components that are exported for final assembly elsewhere (Baldwin 2013). However, as the offshoring of the stages of production is largely cost driven, the value added of the offshored stage is often very limited. For the same reason, while industrialization has been made ‘easier’ in GVCs, specialization in a specific stage, even final-assembly manufacturing, does not automatically bring full benefits to the host country. For the N95 Nokia smartphone as well as for Apple iPhone4, the value incurred at the final assembly stage differs significantly using conventional measures and value-added measures (Ali-Yrkko et al. 2011; Draper et al. 2012). The decline in the share of the value-added of fabrication in the product’s total value relative to the pre- and post-fabrication stages—or, the deepening of the ‘smile curve’ in the literature—calls for re-thinking country strategies for industrialization. This highlights

³ There are still de facto ‘pre-conditions’ and facilitating factors for successfully joining and upgrading within GVCs. Other things being equal, relative political stability, stable institutions, an educated workforce, and good infrastructure are among the most important factors that favour beneficial participation in GVCs.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

the crucial need for developing countries to increase the share of value added and for diversification from pure assembly in manufacturing.⁴

12.3 R  O  P  GVC  F  P

.................................................................................................................................. National enterprise sectors are nested within larger regional and global enterprise sectors in GVCs. The interconnectedness across firms or sectors multiplies and intensifies. For developing countries, more specifically, opportunities include supplying much larger global demand, which reduces the scale and purchasing power limitations of the domestic market in developing economies, and obtaining advanced knowledge and know-how through the interactions with GVC partners. Along with opportunities, participation in GVCs brings risks (Gereffi and Luo 2015): • On the positive side: Participation in GVCs creates new opportunities for firms to profit and expands their market horizon, it provides employment and income sources for people, and widens their spectrum of consumption and investment. GVCs can also improve the resilience of individuals, households, and communities to cope with risks of a local nature. For example, in the 1990s, after the crises in Mexico, Thailand, and Indonesia, remittances increased sharply, which not only helped households to smooth consumption, but also provided the needed resources to overcome credit constraints for local entrepreneurs. In developing countries, contract production, with business made more predictable through ‘assured buy-back’ arrangements, increases entrepreneurship. Business outsourcing offers employment opportunities with higher and more stable income, compared to agricultural jobs and other alternatives. Inter-firm linkages within GVCs can also play a crucial role in transferring technological knowledge and promoting innovation (Gereffi 1999; Pietrobelli and Rabellotti 2009). Compliance with the quality standards of lead firms has the potential to improve the protection of workers, consumers, and the environment.

⁴ There are many opportunities to leverage local knowledge to add value in pre- and post-production services in natural resource-based GVCs in developing economies. Uruguay, for example, has developed a sophisticated livestock traceability system that it uses to monitor 100 per cent of the national cattle herd, which improves the marketability of its beef, the country’s leading export. Uruguay now exports these services to other countries with export-oriented livestock industries. Costa Rica is recognized worldwide as a leader in the field of environmental services, such as natural resources management, environmental impact studies, threatened and endangered species assessments, and environmental education and training, among others (Gereffi 2015).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



• On the negative side: Participation in GVCs increases the scale of information asymmetry and exposes firms and people to risks previously shielded by market boundaries and geographic distances. Shocks in one location can easily spread and magnify to the rest of the network, generating cascade effects and bringing new risks. For example, the 2011 earthquake in Japan affected the automobile industry worldwide. After the earthquake, due to the disruption of production of some automotive parts produced in Japan, many automobile companies such as Nissan, General Motors, and Ford closed plants in Mexico, the USA, China, and South Africa for several days. Risks can also originate from more stringent international standards for price, quality, production processes, and delivery schedules. As multinational lead firms set the rules in GVCs, if the risks are not well managed, less developed countries could be locked out from moving up to the higher value segments of the chain and therefore remain marginalized. There are two types of upgrading: economic upgrading and social upgrading (Barrientos et al. 2011). Economic upgrading is the process by which firms and workers move from low-value to high-value activities. Social upgrading refers to improvements in decent work and well-being within a specific enterprise (or associated group of enterprises) through better working conditions, pay, protection, and rights. How firms are positioned within GVC conditions the opportunities for economic upgrading and social upgrading.⁵ Common trajectories of social upgrading include situations where risks for workers are reduced as small-scale household work gets turned over into hightech and knowledge-intensive work. Other scenarios for social upgrading include certification of the overall standards observed in labour-intensive industries, thereby lowering risks to workers. The gains from ‘moving up the global value chain’ are not equally distributed. The distribution of costs and benefits varies widely across firms in different positions along the value chain, and among workers with different skills. Social upgrading can involve different combinations of: (a) economic upgrading: as enterprises move up value chains, the share of skilled workers typically increases; and (b) deliberate actions to introduce enforceable standards—minimum wages, paid time off, workplace safety, insurance, and so on—for those workers whose skill levels remain low, who are more easily replaced, and who for these reasons may be badly treated. In both developed and developing countries, the economic gains of participating in GVCs do not automatically translate into good jobs or stable employment. Workers do not necessarily benefit when the country or the firm in which they work moves up the

⁵ New mega-regional trade deals, such as the proposed Trans-Pacific Partnership (TPP) that seeks to link 12 Pacific Rim economies accounting for 40 per cent of world trade, seek to reinforce social and environmental upgrading through provisions related to issues such as labour standards and environmental protection. Social and environmental clauses are now becoming standard features of regional and multilateral trade agreements.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

value chain. On the one hand, unskilled workers in many developing countries can be excluded from the desirable job opportunities provided by technology-intensive or knowledge-intensive work, which tend to concentrate in more developed countries. On the other hand, there can be simultaneous social upgrading and downgrading for workers in the same enterprises: regular workers can have better statutory employment protection and benefit from labour standards, while irregular workers, often overrepresented among women, youth, minority, and other vulnerable groups, can suffer discrimination. The expansion of global production, especially in labour-intensive industries, has been an important source of employment generation (Lopez-Acevedo and Robertson 2016). However, economic upgrading does not always lead to social upgrading in the form of better wages and working conditions. For example, garment factories in Morocco led by fast-fashion buyers show that functional upgrading in GVCs can bring about social upgrading and downgrading simultaneously (Rossi 2011). In the worst case, economic upgrading typified by a number of successful export economies, especially in low-income countries, may be linked to a significant deterioration of labour conditions and other forms of social downgrading. Firms may choose to hire irregular workers in an effort to be responsive to buyers’ demands, and workers with casual contracts are often subjected to discrimination (Barrientos et al. 2011). Because GVCs may diverge in key respects from local value chains, the safety and quality benefits from industry standards that are enjoyed by global consumers may bypass domestic consumers.⁶

12.4 S C  L D C

.................................................................................................................................. The level of development of an economy and size of a firm condition the scope of benefits and vulnerability of its participation in GVCs. Less developed economies often suffer from their more limited capacity, and smaller firms are at a disadvantage in terms of technological capabilities, skills, and limited access to credit (Wignaraja 2015). Both less developed countries and smaller firms are penalized more when infrastructure gaps are severe.

⁶ The impact of Bangladesh’s 2013 Rana Plaza disaster, where more than 1,100 garment workers lost their lives in the collapse of a factory complex in Dhaka, has highlighted numerous efforts to identify the root causes of persistent corruption, sub-standard conditions and the exploitation of workers in large, labour-intensive export industries, typified by the garment sector. Because of the many well-known retailers and global brands linked to this kind of offshore production, new collective approaches to selfregulation, corporate social responsibility and social activism are being developed and analysed by researchers (Locke 2013; Appelbaum and Lichtenstein 2016).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



12.4.1 Positions and Nature of Value Chains GVCs typically consist of two types of firms: lead firms and supplier firms (Gereffi 1994). Lead firms, which often have their headquarters in developed countries with a large share of skilled- and knowledge-intensive jobs, define the structure of the industries in producer-driven chains, and accelerate the process of ‘global sourcing’ in buyer-driven chains (Gereffi 1999; Dicken 2011). Supplier firms, and their subcontractors, include small and medium-sized enterprises (SMEs). The vast majority of GVCs are led by large multinational enterprises (MNEs) from advanced countries. The role of the lead firms can range from being supportive and focusing on long-term ‘win– win’ to being predatory and focusing on reaping quick profits in the short term (Frederick and Gereffi 2009). The power relationship and governance patterns of GVCs depend mainly on three factors: (a) the complexity of the information involved in the transactions; (b) the possibility of codifying that information; and (c) the competence of the suppliers along the value chain (Gereffi et al. 2005). The ways for suppliers to acquire production capabilities and market access depend on the governance structure and parameters set by the lead firms.⁷ Relatedly, GVCs consists of two types of economies: ‘headquarter’ economies and ‘factory’ economies (Baldwin and Forslid 2013). Exports from ‘headquarter economies’ contain relatively few imported intermediates and often a high share of domestic value added, while exports from ‘factor economies’ contain a large share of imported intermediates and often a low share of domestic value added. For example, the gross value of Mexican exports and those of the USA show a sharp contrast: about 37 per cent of the gross value of Mexican exports consists of US intermediate inputs, while only 2 per cent of US exports consist of Mexican intermediate inputs (Lopez-González 2012). Most of the value is often created in upstream activities and that created downstream is limited. For example, countries that engage upstream, such as those focusing on innovation and design, are likely to reap more benefit from participation in a GVC through value creation and market power than countries that only operate at the lowest levels, such as assembling jeans or T-shirts for overseas markets in the apparel value chain. The latter are in a less advantageous position to create the expertise, institutions, or consumer markets needed to sustain or upgrade the production structure. The potential benefit that a country can reap from the participation of the GVC partly depends on its domestic value added in the chain. Whether the domestic value added in exports increases or not depends on the extent to which the country can

⁷ The power of lead firms in GVCs derives from the design, marketing, and commercialization assets they control, which allows them to exercise leverage over a large number of suppliers and workers, typically located offshore. In 2011, for example, Nike’s products were made in 930 factories in 50 countries, employing more than one million workers. However, Nike itself had just 38,000 direct employees, most of whom work in the USA. All of the other workers in Nike’s global supply chain were employed by subcontractors based in developing economies (Locke 2013: 48). Over 80 per cent of Wal-Mart’s more than 60,000 suppliers are located in China alone (Gereffi and Christian 2009: 579).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

provide the necessary materials and intermediate products domestically. While many countries that participate in GVCs have seen the share of their domestic value added in exports decline, China provides a successful example of moving up the GVC with increasingly competitive intermediate input sectors supporting the downstream processing export sectors. The share of domestic content in Chinese exports, driven by both the volume and varieties, increased from 65 per cent in 2000 to 70 per cent in 2007 (Kee and Tang 2016).

12.4.2 Size of Firms and Economies GVCs, as they provide opportunities to specialize in tasks within the chain, can operate to the benefit of smaller firms and firms in small or less developed countries. For small firms in developing countries, the participation in GVCs opens the doors for obtaining information about the type and quality of products demanded by consumers in global markets, as well as gaining access to those markets. Through unbundling the production processes in GVCs, less developed and smaller economies with less diversified production structures can find their niches in the global economy. They can specialize in the tasks in which they have expertise, rather than having to manage complex production processes in-house and competing along an entire line of activities in the GVCs. With participation in GVCs and the flourishing trade of intermediate goods and services, a country can no longer rely on its own endowment of productive factors, but it must also import intermediate goods and technology from other countries (Taglioni and Winkler 2016). Small countries may overcome their size disadvantage through the adoption of new policies, and by opening their markets and linking them more closely to other, larger markets. Research shows that SMEs have comparative advantages in sectors where specialization is important and economies of scale are not (OECD and World Bank 2015). SMEs in less developed countries can exploit high value-added niches in GVCs, such as organic production, where input costs are low. One example is in Rwanda, where organic and shade-grown coffee provides farmers of small-scale production a substantially higher premium over ordinary coffee (Nielsen 2008). Small size nonetheless plays a negative role in GVC participation. Market size affects lead firms’ decisions on where to base either their manufacturing and service operations or their innovation centres. Countries with a large local market can use it as a powerful attraction force for supply chain segments, and induce advanced-technology firms to employ and transfer technology. Particularly after the 2008 global recession, GVCs are becoming organizationally more consolidated and geographically more concentrated. Supply chain ‘rationalization’ shrunk the total number of suppliers in GVCs as consumption declined in most advanced industrial countries. GVC lead firms streamlined their supply chains to focus on fewer, larger, and more capable suppliers strategically located near dynamic nodes of GVCs (Gereffi 2014).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



GVCs are often geographically concentrated in a few countries—especially emerging economies—where there are large domestic markets and robust supplier bases, such as China, India, Brazil, and South Africa. For example, in the mobile phone sector, a hightech consumer product, the five largest exporters—China, South Korea, Hong Kong, Vietnam, and the USA—commanded 74 per cent of world exports in 2012, with China alone representing half of the total (Lee and Gereffi 2013). The higher concentration of production yields benefits in terms of large-scale clustering and agglomeration, but can be detrimental for the smaller and less developed economies not at the economic centre. Since the majority of firms in many developing countries are informal (Andrade et al. 2015), the challenges of small-size firms are often intertwined with informality. The participation of informal firms in GVCs is particularly challenging; many can benefit only indirectly as subcontractors of the suppliers of the lead firms, in part due to their lack of capacity. Most informal firms are concentrated in agriculture and the very low end of manufacturing and services activities. SMEs are even more vulnerable to supply chain inefficiencies compared with large firms. For SMEs, related to their small scale, logistics costs are disproportionately higher; industrial firms with fewer than 250 workers on average have logistics costs of about 15 per cent of overall revenue for the business unit, more than twice that of firms with over 250 workers (Straube et al. 2013). The analysis of GVCs provides several insights about how lead firms help to determine ‘winners’ in global supply chains in terms of country characteristics, capabilities of firms, the importance of international standards, and the key skills of workers. Some of the lessons from GVC analysis are: • Nationality matters—lead firms of different national origins behave differently. • Type of lead firm matters (i.e. whether they are retailers, brands, or manufacturers)—retailers and brands have very tough product and process standards, but they often permit more opportunities for upgrading among local suppliers than manufacturers, who tend to see local suppliers as competitors. • Position in supply chains matters—upgrading opportunities vary according to whether you are upstream or downstream in GVCs, and there is a growing role for global suppliers (e.g. first-tier suppliers like Foxconn in electronics, or in autos or aerospace) to capture more value in chains. • Size matters—larger countries and firms typically have more opportunities to achieve the scale needed to access global markets; smaller countries and firms need to rely on specialization and high-value activities to strengthen their position in GVCs. GVC analysis also can contribute to a critical task associated with the theory of new structural economics: identifying a country’s ‘latent comparative advantage’ (Lin and Monga 2010a, 2010b). The way this is handled in GVC studies is through the process of ‘competitive benchmarking’, whereby countries are identified that occupy structurally similar positions in the upgrading trajectories of specific GVCs in the country being analysed (Gereffi and Fernandez-Stark 2016). The first step in this methodology is value chain mapping, in which the main activities and value-added stages in a specific

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

industry of interest are highlighted. Then the upgrading stages in this value chain are laid out, according to the actual experiences of countries that occupy different positions in the value chain.⁸ Finally, the particular policies, institutions, international standards, and workforce development characteristics of each upgrading stage are analysed, according to where particular countries of interest would be positioned in the chain.⁹

12.5 P D

.................................................................................................................................. The extent to which firms and workers in a country can benefit from participation in GVCs depends on their competitiveness. In general, firms are in a better position to benefit from GVC participation if they are relatively large, technologically advanced, professionally managed, and can diversify their markets. Workers are likely to reap more benefits if their working conditions are relatively formalized and if they have higher skills that allow them to carry out better remunerated tasks. There is no magic bullet to improve competitiveness. Government policy makers may not know enough about the intricacies of global industries to spur specific forms of innovation in GVCs. However, as global lead firms typically do not pay supplier firms to undertake the upgrading required to become or remain competitive in GVCs, supportive government policies can be helpful for firms and workers. A better business environment can help attract investment, including foreign investment, with relevant technology components; a more effective education policy can help improve the skill level of the labour force; and a more inclusive social policy can help to support the vulnerable. This can jointly improve a country’s value-adding position in highly mobile segments of GVCs and the inclusive distribution of the benefits. For domestic value chains as well as for global value chains, lowering the transaction costs and uncertainty support economic upgrading, which can create more and better jobs; and facilitating investment in human capital and establishing and implementing sensible regulations can facilitate social upgrading, which can help people to benefit from participation in GVCs. In the context of GVCs, one area for policy intervention is to shift from creating fully blown, vertically integrated national industries to moving into higher-value niches. Integration in GVCs allows countries to follow their comparative advantage to focus on specific tasks and sub-sectors rather than producing all parts of the entire chain. ⁸ In the apparel industry, for example, relevant upgrading stages include: entry into the chain; assembly production (e.g. export-processing zones); full-package or OEM (original equipment manufacturer) production; ODM (original design manufacturing); and OBM (original brand manufacturing) (Gereffi 1999). ⁹ Illustrations from fruit and vegetables, apparel, and offshore services GVCs are provided in Gereffi and Fernandez-Stark (2016). Additional examples can be found on the website of the Duke University Global Value Chains Center (https://gvcc.duke.edu/).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



When early developers move up the value chain, it can encourage the next low-wage nation to step onto the development ladder. Crucial for the lower-income developing countries is to improve their human capital and strengthen the business environment to be ready to capture the opportunities in GVCs. One challenge that governments face is the dominant role that lead firms play in GVCs. Multinational lead firms can choose to work with certain global suppliers in multiple locations to reduce costs, so national governments often have less leverage to demand local content requirements or less scope to develop links to domestic suppliers (Gereffi and Sturgeon 2013). Diversification is crucial for managing risks in the participation in GVCs. Some large emerging economies choose to rely more extensively on regional production networks. Some companies, such as the garment firms in Bangladesh, use the shared supplier spillover of foreign direct investment (FDI) to enhance domestic firms’ performance.¹⁰ Others draw from global sourcing and value chain specialization to improve generic capabilities and become global suppliers to serve multiple customers instead of depending on a single lead firm. Governments can play a crucial role in supporting the provision of efficient and reliable infrastructure and logistic service to smooth the functioning of the GVCs. Timely access to inputs and final parts and products are increasingly important to link the different segments of the chains together (WEF 2013). In response to larger orders for more complex goods, suppliers need to upgrade their capabilities as developing countries acquire the infrastructure to sustain larger scale operations (Hamilton and Gereffi 2009). The stock of infrastructure is a key determinant of total factor productivity (Aschauer 1989). Improvement in infrastructure can contribute to increasing the value of other productive factors, including capital, labour, and land (Estache et al. 2002; Calderon and Serven 2008). Slow and unpredictable land transport keeps much of sub-Saharan Africa out of the higher value added segments of GVCs, including electronics (Christ and Ferrantino 2011). Most countries in sub-Saharan Africa remain at the start of the integration process into GVCs, while countries elsewhere have made significant progress in the past two decades. Supply chain development is crucial to increase the potential of trade and the profit margin, to remove bottlenecks and materialize latent comparative advantages, and to jump start and sustain structural transformation. In China, the prioritization of investment in infrastructure corresponding to its needs has facilitated rapid and inclusive growth in the past three decades through job creation and gradual and sustained upgrading in value chains. Along with the favourable institutional environment and strong macro and fiscal framework, the improvement of local infrastructure in China has played a key role in attracting the ¹⁰ A recent study (Kee 2015) indicates that the shared supplier spillovers of FDI explain one-fourth of the product scope expansion and one-third of the productivity gains within domestic firms in Bangladesh.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

inflow of foreign and domestic capital to establish and sustain industrial clustering, supporting technological innovation and structural upgrading. The development experiences in Africa and Asia, such as light manufacturing in Ethiopia, horticulture in Kenya, ready-made garments in Cambodia, and textiles in Pakistan, have also highlighted that, providing countries tackle the ‘hard’ and ‘soft’ infrastructure constraints to help realize their latent comparative advantages, there can be tremendous potential for productive job creation and structural transformation. Less developed countries can pragmatically address their poor infrastructure and business environment and weak human capital to proactively attract labour-intensive manufacturing to their countries,¹¹ tapping into the great potential to tie into GVCs and foster structural transformation, export diversification, and the possibility to absorb technology and skills. For the less developed countries with severe infrastructure gaps, industrial parks and special economic zones can provide the necessary conditions to lower the transaction costs that allow economies to develop their latent comparative advantages (Lin and Monga 2010a; Lin 2012a). Clustering and the agglomeration of firms could potentially lower the transaction costs and facilitate information sharing/learning, further improving the business environment in the industrial parks and special economic zones.¹² From importing intermediate goods to supporting the development of domestic supply chains, FDI can play an important role at different stages of development. Governments can consider a pragmatic approach, with targeted support to remove infrastructure bottlenecks, to jump-start the economy when overall capacity is still limited, and strategically develop latent comparative advantages in the value chain. The theory of the new structural economics argues for systematic industrial sector targeting and supportive public policies to help countries upgrade the production structure (Lin and Monga 2010b; Lin 2012b; Stiglitz et al. 2013). GVC-oriented industrial policies can go beyond the domestic economy focus of import-substitution policies, which try to recreate entire supply chains within a national territory, and target specialized higher-value niches that follow the country’s comparative advantages as well as provide a fertile ground to develop its latent comparative advantages. However, as more emerging economies try to move up the value chain, including carrying out additional processing in natural resource-based industries, increasing conflict is likely between emerging economies with clashing development strategies.

¹¹ While technology improvements have led to the increased utilization of automation and industrial robots, even in relatively low-cost countries such as China, this technology is typically utilized in industries where the pace of production and miniaturization of components makes human assembly less productive. For the vast majority of low-income developing economies, and most labour-intensive consumer-goods industries, automated production and robotization is not likely to compete with human assembly lines. ¹² A track-record of on-time delivery with consistently good quality is essential to win the confidence of international buyers, which is needed to participate in higher-value niches in GVCs.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



A We would like to thank Pieter Bottelier, Justin Lin, Célestin Monga, and participants in the authors’ workshop in preparation of this Handbook in Beijing (July 2016) for valuable comments. The findings, interpretations, and conclusions expressed in this chapter are entirely those of the authors. They do not necessarily represent the views of the International Bank for Reconstruction and Development/World Bank and its affiliated organizations, or those of the Executive Directors of the World Bank or the governments they represent.

R Ali-Yrkko, J., P. Rouvinen, T. Seppala, and P. Yl-Anttila, 2011. ‘Who Captures Value in Global Supply Chains? Case Nokia N95 Smartphone’, Journal of Industry, Competition and Trade, 11 (3), pp. 263–78. Andrade, G. H., M. Bruhn, and D. McKenzie, 2015. ‘A Helping Hand or the Long Arm of Law? Experimental Evidence on What Governments Can Do to Formalize Firms’, The World Bank Economic Review, pp. 1–31. Appelbaum, R. P. and N. Lichtenstein (eds), 2016. Achieving Workers’ Rights in the Global Economy. Ithaca, NY: Cornell University Press. Aschauer, D. A., 1989. ‘Is Public Expenditure Productive?’ Journal of Monetary Economics, 23 (2), pp. 177–200. Baldwin, R., 2011a. ‘21st Century Regionalism: Filling the Gap between 21st Century Trade and 20th Century Trade Rules’. CEPR Policy Insight No. 56. London: Center for Economic Policy Research. Baldwin, R., 2011b. ‘Trade and Industrialization after Globalization’s 2nd Unbundling: How Building and Joining a Supply Chain are Different and Why it Matters’. NBER Working Paper No. 17716. Cambridge, MA: National Bureau of Economic Research. Baldwin, R., 2013. ‘Global Supply Chains: Why They Emerged, Why They Matter, and Where They Are Going’, in Deborah K. Elms and Patrick Low, eds, Global Value Chains in a Changing World. Geneva: WTO Publications. Baldwin, R. and R. Forslid, 2013. ‘The Development and Future of Factory Asia’, manuscript for ADB, 28 June. Barrientos, S., G. Gereffi, and A. Rossi, 2011. ‘Economic and Social Upgrading in Global Production Networks: A New Paradigm for a Changing World’, International Labour Review, 150, pp. 319–40. doi: 10.1111/j.1564-913X.2011.00119.x Calderón, C. and L. Servén, 2008. ‘Infrastructure and Economic Development in Sub-Saharan Africa’, Policy Research Working Paper Series 4712, The World Bank. Calderón, C. and L. Servén, 2010. ‘Infrastructure in Latin America’, World Bank Policy Research Working Paper No. 5317. Washington, DC: The World Bank. Christ, N. and M. J. Ferrantino, 2011. ‘Land Transport for Exports: The Effects of Cost, Time, and Uncertainty in Sub-Saharan Africa’, World Development, 39 (10), pp. 1749–59. Dicken, P., 2011. Global Shift: Mapping the Changing Contours of the World Economy, 6th edn, New York: Guilford.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

Draper, P., U. Dadush, G. Hufbauer, J. Bacchus, and R. Lawrence, 2012. ‘The Shifting Geography of Global Value Chains: Implications for Developing Countries and Trade Policy’, Geneva: World Economic Forum. The Economist, 2014. ‘International Trade—A Troubling Trajectory’, 13 December. Estache A., V. Foster, and Q. Wodon, 2002. Accounting for Poverty in Infrastructure Reform— Learning from Latin America’s Experience, Studies in Development Series. Washington, DC: World Bank Institute. Feenstra, R., 1998. ‘Integration of Trade and Disintegration of Production in the Global Economy’, Journal of Economic Perspectives, 12 (4), pp. 31–50. Florida, R., 2005. ‘The World is Spiky’, The Atlantic Monthly, 296 (3), pp. 48–51. Frederick, S. and Gereffi, G., 2009 ‘Value Chain Governance’. United States Agency for International Development Briefing Paper. Available at: https://www.marketlinks.org/ sites/marketlinks.org/files/resource/files/vc_governance_briefing_paper.pdf Frederick, S. and G. Gereffi, 2011. ‘Upgrading and Restructuring in the Global Apparel Value Chain: Why China and Asia are Outperforming Mexico and Central America’, International Journal of Technological Learning, Innovation and Development, 4 (1–3), pp. 67–95. Fujita, M., P. Krugman, and A. J. Venables, 1999. The Spatial Economy: Cities, Regions and International Trade, Cambridge, MA: MIT Press. Gangnes, B., A. C. Ma, and A. Van Assche, 2015. ‘Global Value Chains and the Trade–Income Relationship: Implications for the Recent Trade Slowdown’, in Bernard Hoekman, ed., The Global Trade Slowdown: A New Normal? London: CEPR Press, pp. 111–26. A VoxEU.org eBook. Available at: http://voxeu.org/sites/default/files/file/Global%20Trade%20Slowdown_ nocover.pdf Gereffi, G., 1994. ‘The Organisation of Buyer-Driven Global Commodity Chains: How US Retailers Shape Overseas Production Networks’, in G. Gereffi, and M. Korzeniewicz, eds, Commodity Chains and Global Capitalism, Westport, CT: Praeger, pp. 95–122. Gereffi, G., 1999. ‘International Trade and Industrial Upgrading in the Apparel Commodity Chain’, Journal of International Economics, 48 (1), pp. 37–70. Gereffi, G., 2014. ‘Global Value Chains in a Post-Washington Consensus World’, Review of International Political Economy, 21 (1), pp. 9–37. Gereffi, G., 2015. ‘Global Value Chains, Development and Emerging Economies’, UNUMERIT Working Paper No. 2015-047. Background paper for UNIDO, Industrial Development Report 2016. Gereffi, G. and M. Christian, 2009. ‘The Impacts of Wal-Mart: The Rise and Consequences of the World’s Dominant Retailer’, Annual Review of Sociology, 35, pp. 573–91. Gereffi, G. and K. Fernandez-Stark, 2016. Global Value Chain Analysis: A Primer, 2nd edn. Durham, NC: Duke CGGC. Available at: http://www.cggc.duke.edu/pdfs/Duke_CGGC_ Global_Value_Chain_GVC_Analysis_Primer_2nd_Ed_2016.pdf Gereffi, G. and J. Lee, 2012. ‘Why the World Suddenly Cares about Global Supply Chains’, Journal of Supply Chain Management, 48 (3), pp. 24–32. Gereffi G. and X. Luo, 2015. ‘Risk and Opportunities in the Participation of Global Value Chains’, Journal of Banking and Financial Economics, 2 (4), pp. 51–63. Gereffi G. and T. Sturgeon, 2013. ‘Global Value Chain-Oriented Industrial Policy: The Role of Emerging Economies’, in D. K. Elms and P. Low (eds.), Global Value Chains in a Changing World. Geneva: World Trade Organization, Fung Global Institute and Temasek Foundation Centre for Trade & Negotiations, pp. 329–60.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Gereffi, G., J. Humphrey, and T. Sturgeon, 2005. ‘The Governance of Global Value Chains’, Review of International Political Economy, 12 (1), pp. 78–104. Hamilton, G. G. and G. Gereffi, 2009. ‘Global Commodity Chains, Market Makers, and the Rise of Demand-Responsive Economies’, in Jennifer Bair, ed., Frontiers of Commodity Chain Research, Stanford, CA: Stanford University Press, pp. 136–61. Hoekman, B., 2015. ‘Trade and Growth—End of an Era?’, in Bernard Hoekman, ed., The Global Trade Slowdown: A New Normal? London: CEPR Press, pp. 3–19. A VoxEU.org eBook. Available at: http://voxeu.org/sites/default/files/file/Global%20Trade%20Slowdown_ nocover.pdf Kee, H. L., 2015. ‘Local Intermediate Inputs and the Shared Supplier Spillovers of Foreign Direct Investment’, Journal of Development Economics, 112, pp. 56–71. Kee, H. L. and H. Tang, 2016. ‘Domestic Value Added in Exports: Theory and Firm Evidence from China’, American Economic Review, 106 (6), pp. 1402–36. Also available as World Bank Policy Research Working Paper No. 7491. Koopman, R., Z. Wang, and S.-J. Wei, 2008. ‘How Much of Chinese Exports Is Really Made in China? Assessing Domestic Value-Added When Processing Trade Is Pervasive’. National Bureau of Economic Research Working Paper No. 14109. Available at: http://www.nber. org/papers/w14109 Krugman, P., 1991. ‘Increasing Returns and Economic Geography’, Journal of Political Economy, 99, pp. 483–99. Lee, J. and G. Gereffi, 2013. ‘The Co-Evolution of Concentration in Mobile Phone Global Value Chains and Its Impact on Social Upgrading in Developing Countries’. Capturing the Gains Working Paper No. 25, March. Available at: http://www.capturingthegains.org/ pdf/ctg-wp-2013-25.pdf Lin, J. Y., 2012a. The Quest for Prosperity: How Developing Economies Can Take Off. Princeton, NJ: Princeton University Press. Lin, J. Y., 2012b. The New Structural Economics: A Framework for Rethinking Development and Policy. Washington DC: The World Bank. Lin, J. Y. and C. Monga, 2010a. ‘The Growth Report and New Structural Economics’ (June). World Bank Policy Research Working Paper No. 5336. Available at SSRN: http://ssrn.com/ abstract=1623714 Lin, J. Y. and C. Monga, 2010b. ‘Growth Identification and Facilitation: The Role of the State in the Dynamics of Structural Change’ (May). World Bank Policy Research Working Paper No. 5313. Available at SSRN: http://ssrn.com/abstract=1611526 Locke, R. M. 2013. The Promise and Limits of Private Power: Promoting Labor Standards in a Global Economy. New York: Cambridge University Press. Lopez-Acevedo, G. and R. Robertson (eds), (2016). Stitches to Riches. Washington, DC: World Bank. Lopez-González, J., 2012. ‘Vertical Specialisation and New Regionalism’, PhD Thesis, University of Sussex, UK, April. Nielsen, P. B. (ed.), 2008. ‘International Sourcing—Moving Business Functions Abroad’, Statistics Demark. OECD, 2011. Global Value Chains: Preliminary Evidence and Policy Issues. Paris: Organisation for Economic Co-operation and Development, DSTI/IND(2011) 3. Available at: http://www.oecd.org/dataoecd/18/43/47945400.pdf OECD and World Bank Group, 2015. ‘Inclusive Global Value Chains: Policy Options in Trade and Complementary Areas for GVC Integration by Small and Medium Enterprises

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



    

and Low-Income Developing Countries’. Report prepared for submission to G20 Trade Ministers Meeting Istanbul, Turkey, October. Available at: http://www-wds.worldbank. org/external/default/WDSContentServer/WDSP/IB/2015/12/11/090224b083c499d8/2_0/ Rendered/PDF/Inclusive0glob0developing0countries.pdf Pietrobelli, C. and R. Rabellotti, 2009. ‘The Global Dimension of Innovation Systems: Linking Innovation Systems and Global Value Chains’, B. A. Lundvall, J. Vang, and K. J. Joseph, eds, Handbook of Innovation Systems and Developing Countries. Cheltenham: Edward Elgar. Porter, M. E., 1985. Competitive Advantage, New York: Free Press. Puga, D. and A. J. Venables, 1996. ‘The Spread of Industry: Spatial Agglomeration in Economic Development’, Journal of the Japanese and International Economies, 10 (4), pp. 440–64. Rossi, A., 2011. ‘Economic and Social Upgrading in Global Production Networks: The Case of the Garment Industry in Morocco’, DPhil dissertation, Brighton: Institute of Development Studies, Sussex University. Staritz, C., G. Gereffi, and O. Cattaneo (eds), 2011. Special Issue. ‘Shifting End Markets and Upgrading Prospects in Global Value Chains’, International Journal of Technological Learning, Innovation and Development, 4 (1–3). Available at: http://www.inderscience. com/info/inarticletoc.php?jcode=ijtlid&year=2011&vol=4&issue=1/2/3 Stiglitz, J. E., J. Y. Lin, and C. Monga, 2013. ‘The Rejuvenation of Industrial Policy’ (September). World Bank Policy Research Working Paper No. 6628. Available at SSRN: http://ssrn.com/ abstract=2333944 Straube, F., R. Handfield, H.-C. Pfohl, and A. Wieland, 2013. Trends and Strategies in Lofistik und Supply Chain Management, Hamburg, Germany: Deutscher Verkehrs-Verlag. Taglioni D. and D. Winkler, 2016. Making Global Value Chains Work for Development, Washington, DC: The World Bank. WEF (World Economic Forum), 2013. Enabling Trade: Valuing Growth Opportunities, Geneva: WEF. Wignaraja, G., 2015, ‘Factors Affecting Entry into Supply Chain Trade: An Analysis of Firms in Southeast Asia’, Asia & the Pacific Policy Studies, 2, pp. 623–42. doi: 10.1002/app5.78 WTO and IDE-JETRO, 2011. Trade Patterns and Global Value Chains in East Asia: From Trade in Goods to Trade in Tasks’, Geneva: World Trade Organization and Tokyo, Japan, Institute for Developing Economies-Japan External Trade Organization. Available at: https://www.wto.org/english/res_e/booksp_e/stat_tradepat_globvalchains_e.pdf

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

      ......................................................................................................................

 

13.1 I

.................................................................................................................................. C and industrial parks are a worldwide phenomenon. Clusters and industrial parks in developed countries are the subject of a large body of literature. Michael Porter popularized the concept of ‘clustering’ in 1990 through his seminal book The Competitive Advantage of Nations, where he explained the advantages of industrial agglomeration in developed countries. Subsequent studies have primarily analysed economic agglomeration that spans regions and industries in the context of developed countries, where institutions and infrastructure are relatively well developed (Porter 1990; Saxenian 1994; Markusen 1996).¹ In fact, the ideas behind clustering have a long pedigree. Smith (1776), using the example of linen shirts, illustrated how the putting-out system was widely practised in the UK prior to the Industrial Revolution.² The putting-out system was also popular in Western Europe. Marshall devoted four chapters in his seminal book Principles of Economics (1920) to industrial districts, a term that preceded clusters. Similar arrangements have been observed in the Japanese garment industry during the nineteenth century (Nakabayashi 2006). ¹ See Ciccone and Hall (1996) and Ciccone (2002) for reviews of the clustering effect in the USA and Europe, respectively. See Fisher and Peters (2002) for a review of special economic zones in the UK and the USA. ² The putting-out system operated as follows: merchants acquired orders from the market and organized production by outsourcing incremental steps to skilled workers and farmers nearby, who finished the work in small-scale family workshops (Hounshell 1984). The concept of the putting-out system predates the concept of clusters. But, in essence, the two are similar.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Clusters are also ubiquitous in developing countries. For example, in Thailand, the ‘One Tambon, One Product’ programme has been widely promoted. The programme encourages each Thai tambon (subdistrict) to develop its industry centred around one key product. The Philippines also adopts a similar ‘One Town, One Product’ programme. Long and Zhang (2011) show that the cluster-based model has been a defining feature of Chinese industrial growth over the past several decades. Sonobe and Otsuka (2006) discuss both the pattern and the mechanism of cluster-based industrialization in Asian countries. Oyelaran-Oyeyinka and McCormick (2007) present nine case studies of clusters across seven African nations, suggesting that clusters are common across the world. While industrial clusters have been the focus of the cluster literature on developing countries, hometown-based clusters are another type which is of equal importance. While both types of cluster are related to geographic agglomeration, their specific linkages with geographic location are fundamentally different. The traditional concept of industrial clusters (or even a service cluster such as Silicon Valley) are characterized by entrepreneurs operating their businesses within a specific locality. In contrast, entrepreneurs in hometown-based clusters, who are bonded because they originate from the same place, do not necessarily operate physically close to each other. The phenomenon of hometown-based clusters is particularly relevant in China, where the concept of the hometown is deep-rooted. Theoretically, different sets of social network systems stemming from social categorizations in different developing countries may suggest that other definitions of clusters might be more relevant. For example, Indians are mainly categorized by castes instead of hometowns. In this case, a more relevant concept would be caste-based clusters.³ Yet, regardless of the definitions of clusters, be they industrial clusters, hometown-based clusters, or castebased clusters, they all share a similar set of advantages through the same mechanism and in this chapter they are analysed in the same framework. A hometown-based cluster in China, where entrepreneurs provide migratory harvesting services across provinces, will be discussed. Understanding the operation of this hometown-based cluster allows us to interpret the steadily growing agricultural sector in China despite numerous unfavourable conditions. It is an example of a wider observation—by examining clustering of production, many economic puzzles might be logically resolved. In this chapter, we review clusters and industrial parks in developing countries for two reasons. First, compared with the rich body of literature on clustering in developed countries, there is considerably lesson the phenomenon in developing countries. Second, the strategy of creating clusters and industrial parks fits particularly well with certain comparative advantages often found in developing countries. The absence of formal institutions, such as contract enforcement, is an endemic problem in the developing world. In addition, entrepreneurs face financial constraints when starting and running a business. The task of fixing institutional problems and ³ As an example, in India over 96 per cent of firms in the diamond industry belong to just three caste communities (Munshi 2011).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



developing sound financial systems overnight in developing countries is daunting. Despite such challenges, developing countries do have some comparative advantages, such as strong social capital in communities and abundant labour. In close-knit communities in developing countries, people often know each other well and develop strong social trust. Compared with the scarcer financial capital, labour is generally more readily availalble. Clustering offers an alternative way for developing countries to make better use of these existing strengths (abundant labour and strong social capital) to overcome the seemingly insurmountable financial and institutional constraints. Marshall (1920) highlighted the three major advantages of industrial districts (clusters): better access to suppliers and markets, labour market pooling, and spillovers of technological know-how. When final goods and intermediate-input markets are nearby, firms save on marketing and purchasing costs. When a large number of firms work in the same sector, workers are more willing to invest in their skills because they are portable across firms within the cluster. Proximity to other producers enables one to quickly learn the technologies prevalent in the cluster. All these advantages lower the transaction costs of operating a business in a cluster. Apart from the three major advantages Marshall identified, clusters have a few additional advantages. Within a cluster, a production process can be divided into many incremental steps, which are undertaken by various family workshops. Such a fine division of labour largely reduces the capital required to start a business at each step of production (Ruan and Zhang 2009; Long and Zhang 2011). In addition, due to strong social capital and proximity to each other, businesses in clusters make extensive use of inter-firm trade credit, which reduces their reliance on external funding for working capital. With a lower starting capital requirement and fewer working capital constraints, many previously financially restricted entrepreneurs can set up businesses in clusters, enabling them to create more employment opportunities, which developing countries desperately need. Mainstream economic theory suggests that the frequent subcontracting and fine division of labour within clusters would involve higher coordination costs (Williamson 1975: 26–30; Becker and Murphy 1992). However, in reality, it is widely observed that in clusters, thanks to repeated transactions, freely flowing information, and a strong social trust embedded in communities, entrepreneurs rely heavily on relational contracts to get around the problem of weak contracts (Greif 1993; Ruan and Zhang 2009; Long and Zhang 2011). Formal contracts are rarely signed in clusters. The transaction costs turn out to be much lower than previously thought in the literature. Private ordering becomes a major means to sort out contract disputes in the absence of formal institutions.⁴ China’s industrialization offers a good example in support of this narrative. China has become industrialized in just a few decades despite an initial lack of sound institutions or a well-developed financial system. The conventional wisdom in textbook ⁴ Private ordering is the process where the parties involved, instead of the State, set up social norms for the purpose of achieving various kinds of public goals, such as efficiency, market enhancement, and property rights protection.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

economics cannot explain the puzzles behind China’s rapid growth.⁵ Long and Zhang (2011) provide evidence that clustering plays a key role in driving China’s rapid industrialization by lowering starting capital barriers and reducing reliance on working capital. Clusters are often organically formed from existing industries as determined by historical legacy (Miller and Cote 1985). The role of government is normally limited at the initial stage. Yet governments, and in particular local governments, can help facilitate the growth of existing clusters. In some parts of the developing world, clusters are absent. Due to a lack of good infrastructure and sound institutions at the national level in developing countries, it is hard to create a new industry from scratch on a large scale. Instead, governments or business communities in developing countries often prefer to build industrial parks in a limited geographic area, where adequate infrastructure and an enabling business environment can be provided. They aim to attract foreign or domestic direct investment in the industrial parks in order to promote employment and facilitate technology transfer. Whereas clusters and industrial parks share the advantages of economic agglomeration, they differ fundamentally in terms of origin, entry barriers, composition of enterprises, and their entrepreneurship impacts on the local economy. The most prominent distinction is the degree of government intervention at the initial stage. In this chapter, we review the experiences of and lessons learned from building clusters and industrial parks.

13.2 B I C

.................................................................................................................................. Sonobe and Otsuka (2006) characterize the process of industrial development as taking place in three stages—namely, initiation, quantity expansion, and quality improvement. We can write this intuitively as 0→1→N→Q. The step of 0→1 stands for the initiation phrase; 1→N means the stage of quantity expansion; and the quality improvement step can be written as N→Q, where N and Q refer to quantity and quality, respectively. This section is organized according to the three stages.

13.2.1 0→1 Most clusters form organically. To illustrate how historical legacy determines cluster formation, we will explore the Chinese example. Many of today’s clusters in China originated from township and village enterprises (TVEs) or state-owned enterprises (SOEs). In the 1970s and 1980s, the Chinese constitution did not recognize and protect ⁵ In fact, developed countries faced the same problem in their early stages of development: small- and medium-sized enterprises in Northern and Western Europe and North America were rarely able to obtain credit from large national or regional financial institutions (Cull et al. 2006).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



private ownership. Largely due to the central government’s failure to protect property rights, TVEs blossomed (Xu and Zhang 2009). The township and village governments provided local de facto protection for the TVEs. By registering as a TVE, enterprises could circumvent the problem of weak institutions at the time, and they quickly expanded in response to rising market demand, which was a result of successful rural reform in the 1980s. Some workers who grasped the technology know-how in TVEs or SOEs began to set up workshops at home and sell the same product in the market. After observing the success of such private endeavours, other villages followed suit, triggering the birth of a cluster. The footwear cluster in Wenzhou (Huang et al. 2008) is a good example, and we describe its origins in Appendix A. Similar stories can be found in the cashmere sweater cluster in Puyuan (Ruan and Zhang 2009) and the children’s garment cluster in Zhili (Fleisher et al. 2010). The birth of clusters is primarily due to bottom-up responses to expanding market opportunities. Governments have played little role in initiating such clusters. More often than not, it was only after they had noticed the dynamics of the clusters that the local governments started to facilitate their growth. Most clusters in other countries also have a historical origin. For example, in Santa Catarina, Brazil, clustering is a prominent feature of industrial production. European immigrants started most of the clusters there. The textile cluster was developed by German immigrants who possessed experience in that trade and arrived in the state in the 1880s (Meyer-Stamer 1998).

13.2.2 1→N Because the barriers to entry are low, clusters often enjoy an initial period of rapid growth. However, the explosion in the number of businesses in a limited area quickly creates some bottlenecks, such as insecurity, lack of market places, and inadequate infrastructure. Because an individual enterprise will have trouble addressing large, external problems, collective action is needed. Compared with the limited role of individual firms, local governments and the local business community can play a more important role in leading collective action, as the case studies of a cashmere cluster (Ruan and Zhang 2009) and a potato production cluster (Zhang and Hu 2014) in China illustrate. Appendix B describes the establishment of a logistics centre in the Puyuan cashmere cluster in China. As the example shows, the local government in Puyuan responded to infrastructure bottlenecks by building a large logistics centre through a private– public partnership. As clusters evolve, bottlenecks arise successively at later stages. New constraints become binding and require continuous tinkering by governments. Indeed, government interventions should differ according to specific situations and be based on a bottom-up, demand-driven approach. Since clusters exist largely at the local level, it is the local

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

government, rather than the central government, that should play the key role in providing the required public goods and services, thanks to its informational advantage.

13.2.3 N→Q As clusters expand, the scale of production increases, depressing prices. Consequently, firms in clusters tend to engage in race-to-the-bottom price competitions It is a great challenge for an individual firm to upgrade product quality because others in the clusters can easily imitate its new product. Moreover, firms have no incentive to train workers because the trained workers can easily jump ship and go to work for competitors. Local governments can play an active role in shifting the equilibrium of competing for prices to competing for better quality (Sonobe and Otsuka 2006). For example, providing training to workers at the cluster level and encouraging enterprises to establish brand names are possible ways to improve the innovation capability of the cluster as a whole. In normal times, building up supporting institutions to encourage innovation is hard because the proposed changes are likely to produce losers, who will block the changes. Institutional reforms are more likely to occur after a crisis strikes. When a crisis emerges, the opportunity costs of producing high-quality goods, namely the profit of producing low-quality goods, fall. Based on surveys of clusters in China’s Zhejiang Province, Ruan and Zhang (2010) show that collective action related to quality upgrading is more likely to occur after a crisis. This appears to hold true in other developing countries as well: for example, a ban on the import of surgical instruments from developed countries led to upgrading the quality of surgical instruments in a cluster in Pakistan (Nadvi 1999). However, crisis is not a sufficient condition for a quality upgrade. Not all clusters can transform crises into opportunities and allow for quality upgrading—failures do happen. An insulated mug cluster in Yongkang, China, arose in 1995 and grew so quickly that the excess supply drove prices below production costs by 1996. The crisis came so quickly that collective action could not be taken promptly, which resulted in the collapse of the whole cluster (Ruan and Zhang 2010). Schmitz (1999) provides another example. In the 1980s, integration into the American footwear value chain allowed an export-oriented leather footwear cluster in Brazil’s Sino Valley to improve the quality of its products, its flexibility, and its speed of response. Yet the cluster’s exports and profits fell in the 1990s under global competitive pressure since it failed to upgrade in other areas that required coordination between stakeholders. There were two reasons: conflicting interests among entrepreneurial alliances and business associations, and the failure of leading enterprises to participate. This example again highlights how local governments could act as a coordinator to lower transaction costs that arise in complicated networks of businesses within clusters.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



The above-cited literature mainly focuses on upgrading processes that occur within localities, especially collaboration between local producers and the provision of public goods by governments. External linkages can also be used as a means to spur quality upgrades (Nadvi and Schmitz 1999). When working with global buyers, local producers must follow the often higher quality standards of foreign buyers. More stringent global standards make it imperative for local firms to improve their product quality. For instance, Taiwanese contract manufacturers in the electronics industry used the knowledge they had acquired working for their main global buyer for the purpose of supplying other markets. They even took over other lucrative functions such as process development and product design at later stages (Lee and Chen 2000). A similar story is found in the blue jeans industry in Torreon, Mexico, which performed functional upgrading in the 1990s (Bair and Gereffi 2001). Like crises, a connection with global buyers is not a sufficient condition for quality upgrades. Bazan and Navas-Alemán (2001) show that customized specification prevents Brazilian footwear suppliers of big US buyers from entering national or Latin American markets. Manufacturing to tight specifications for the main customers requires the whole production plant to be organized for that specific purpose. Enterprises wishing to participate have to build up highly developed but narrow capabilities. This hinders their ability to appropriately fine-tune product specifications to adapt to local markets.

13.3 T C   H- C: C S C  J, C

.................................................................................................................................. Despite small farm sizes and rising wages, the agricultural sector in China has been growing steadily in the past few decades.⁶ In this section, we examine how mechanization has made this possible and how clustering promotes the efficient use of machinery in agricultural production. In particular, an example of a combine service cluster will be carefully studied as an application of the 0→1→N→Q evolution framework introduced in the last section. While the example is taken from China, the findings and their implications are relevant in the setting of other developing countries such as those in sub-Saharan African as they face similar constraints as China in agricultural development (Collier and Dercon 2014). In China, rural industrialization and rural–urban migration following economic reforms in the 1980s pull labour away from farms. They together account for the

⁶ This section is largely adapted from Yang et al. (2013) and Zhang et al. (2015).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

substantial drop in employment levels in the agricultural sector. In 1978, the proportion of the Chinese population working on farms was over 92 per cent, compared to a significantly lower figure of 40 per cent in 2005 (Lin et al. 2003; McGregor 2005). The shortfall of labour logically suggests that the use of machinery complements agricultural production so as to maintain the growth of agricultural productivity. However, given the small farm size—averaging 0.5 hectares (ha) (compared to 150 ha in the USA)—many observers hold a relatively pessimistic view towards agricultural mechanization in China (Ruttan 2001; Pingali 2007; Otsuka 2013). They suggest that the tiny farm size inevitably leads to a limited use of machinery in different stages of agricultural production, and fragmentation of farmlands exacerbates the problem. Nonetheless, agricultural output and yields have been increasing in the past two decades. In 1978, yield was 2.5 tons/ha, compared to 3.5 in 2000 and 4.2 in 2010. Behind the growth of agricultural production is the increasing use of farm machinery, as well as other changes of input composition. According to the Chinese Statistical Yearbook published by the government in 2011, energy consumed by farm machinery use increased from 150 million kilowatts in 1985 to 950 million kilowatts in 2009. The rise of a farm mechanization outsourcing service industry could help to explain how China is achieving increasing mechanization given small and fragmented farmland. Using the 0→1→N→Q evolution framework, we consider the combine harvesting service cluster in Peixian County in the Chinese province of Jiangsu. Like industrial clusters, the role of government varies at different stages. The principle that local governments instead of central government should take the lead in facilitating the growth of the cluster remains true.

13.3.1 0→1 The combine service cluster is one of the oldest and largest providing an interprovince mechanization service. In the 1990s, farmers in Peixian County purchased tractors and combines to complement rice and wheat production. In this case, unlike industrial clusters which are usually formed as a bottom-up response, the local government of Peixian County played a determining role in the formation of the cluster. After returning from a study tour to learn about the mechanization experience in Weifang city in Shangdong province, Peixian Bureau of Agricultural Mechanization (PBAM) provided the necessary training and market information for machine-owning farmers. With the aim of recouping high investment costs, in 1998 these farmers started renting out their machines and providing harvesting services for farms in neighbouring areas. The convenient physical location is another reason why the cluster thrives and prospers in Peixian, but not elsewhere. Being surrounded by a dense transportation network, farmers in Peixian enjoy easy access to the outside market.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



13.3.2 1→N At the initial stage, the cluster started with fifty combines primarily supported by PBAM. Each combine is operated by three or four people. To recoup the investment cost of the machines, they began to provide service across provinces. On average, the owner of a combine could make 60,000 yuan profit, which was many times higher than on-farm income at the time. The news of this fruitful and profitable enterprise quickly spread across Peixian. It attracted others in Peixian to imitate it and become entrepreneurs specialized in providing a harvesting service, resulting in the exponential growth of the combine harvesting cluster. In a sense, the channel through which this hometown-based cluster expanded is identical to that of the Wenzhou footwear industrial cluster described in Appendix A, mainly by copying others. The role of government agencies evolved as the cluster expanded. Facilitation of the growth of the cluster became the key. Since the cost of a machine was prohibitively high, until 2004 only the wealthier families in Peixian were able to enter the cluster. In 2004, the government started providing subsidies to help less wealthy families participate in the expanding cluster. As the scale of the cluster makes it impossible for PBAM to escort all the entrepreneurs, it encouraged entrepreneurs, instead, to team up and ‘go-as-a-group’. For example, receiving a complaint that operators spent large sums on phone calls and messages in order to coordinate team production activities, PBAM set up a group messaging platform for the harvesting teams in collaboration with a telecommunications company. Unlike entrepreneurs in industrial clusters, not all the entrepreneurs in the hometown-based cluster are physically close to each other due to the migratory nature of the service. In this context, the strategy of ‘go-as-a-group’ has several advantages, including but not limited to greater bargaining power with local agents, pooling of spare parts for repair, sharing the client search cost and being more competent to cope with harassment and extortion from local gangs. As more entrepreneurs entered the cluster, they started to travel further away to provide harvesting services. The fact that China is a big country with varying harvesting seasons means it is physically possible to operate all year. Since the harvesting windows are generally narrow, combine harvesting service providers from Peixian compete with local service providers relying on provision of timely services. The distribution of nationwide harvest calendars by PBAM allows the entrepreneurs to catch the narrow harvesting windows. These factors together account for the rapid expansion of the clusters. In 2013, it had more than 1,000 combine harvesters operating in twelve provinces throughout the year.

13.3.3 N→Q At the initial stages, as operators were mainly providing harvesting services in neighbouring areas, they chose Futian (Chinese) combine harvesters, which were not geared

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

for travelling long distances. As operators of combine harvesters often do travel long distances when providing cross regional services, the frequent breakdowns of the Futian (Chinese) models led to a shift to Kubota (Japanese) tractor-combines, which were more durable and therefore suitable for long distance trips and the varying terrains of clients.

13.4 B I P

.................................................................................................................................. The major difference between building clusters and building industrial parks is the degree of government intervention at the initial stage. Whereas industrial clusters usually form organically without government intervention, industrial parks are initiated by governments to jump-start economic growth in a specific geographic region. Since improvement of the business environment of a country as a whole is neither economically nor politically viable, governments in developing countries often prefer to build industrial parks on a smaller scale. When constructing an industrial park, the government aims to attract investment by offering potential entrants geographically limited benefits of various kinds. UNIDO (1997) defined an industrial park, or the more general term special economic zone (SEZ),⁷ as ‘a tract of land developed and subdivided into plots according to a comprehensive plan with provision for roads, transport and public utilities with or without built-up (advance) factories, sometimes with common facilities and sometimes without them’. In addition to hard infrastructure, industrial parks often grant preferential policy and have different institutional arrangements from the rest of the country—such as tax and tariff reductions, looser labour regulations, different sets of laws, and many other practices that provide convenience and lower the costs of doing business.⁸ While the policy instruments that governments could use to lure investment are known, there are a few strategies they could follow to increase the chances of success: targeting international firms, targeting grouped businesses, incentivizing first movers, and adopting a step-by-step approach.

13.4.1 Targeting International Firms The high capital costs of investment suggest that firms operating in industrial parks in developing countries are generally large. Because domestic markets in developing countries are often in their infancy, they cannot absorb the production of firms in industrial parks. It is therefore more reasonable for industrial parks to target firms that bring in international market orders from abroad and finish the orders in the parks. ⁷ In this chapter the terms are used interchangeably for convenience. ⁸ See Aggarwal (2005) for a review of fiscal incentives provided by governments in India, Sri Lanka, and Bangladesh.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



Operating in this way has a few advantages. First, firms in the industrial parks can focus on their production without worrying too much about the thin domestic market. Second, many global buyers provide intermediate inputs for their orders and just put out the assembly step to domestic firms. Firms in industrial parks in developing countries are therefore less subject to the supply chain problems inherent in many developing countries. Over time, as domestic markets grow and as firms build their reputation locally, they can gradually expand their domestic market share and subcontract more tasks to other domestic firms, even outside the industrial parks. In doing so, they generate positive technological spillover to existing firms and contribute to overall economic growth (Glaeser and Gottlieb 2009; Greenstone et al. 2010).

13.4.2 Targeting Grouped Businesses Given the limited number of multi-establishment firms, the strategy of targeting only fully vertically integrated firms may not always be viable. As industrial production needs upstream and downstream supply chains, it is often hard for a small or mediumsized firm to survive in an isolated location. As a result, the ‘go-as-a-group’ model has come into being in recent decades: a powerful enterprise or business association takes the initiative to establish an overseas trade centre and industrial park, as a means of attracting domestic enterprises to go as a group. Some advantages of the ‘go-as-a-group’ strategy are mentioned in the hometown-based cluster example in Section 13.3, such as promoted security and lower search costs. In the context of industrial parks, the use of the strategy has some additional advantages. One is to maintain the original production connections overseas by investing as a group of upstream and downstream production enterprises and preserving the domestic industrial chain in the host country. Such a strategy has several advantages for enterprises in the group: achieving market internalization of intermediate products, formulating internalization advantages, reducing international market risk, reducing export tariffs, and optimizing the international investment environment. During trade dispute settlements, the grouped enterprises can negotiate and resolve trade quarrels with greater bargaining power. We offer the example of Yue Mei, a Chinese textile and garment company. In 2004, Nigeria banned the import of textile and garment products from China. Yue Mei in response planned to set up a processing plant in Nigeria but soon realized that the incomplete supply chain would make it hard to survive as an isolated business in the foreign business environment. In 2007 Yue Mei invited fifteen upstream and downstream enterprises, originally from China, and invested US$50 million to set up a textile and garment industrial park in a Nigerian free trade zone.⁹ Governments in developing countries might exploit the increasingly popular use of the go-as-a-group strategy by private enterprises and target such groups of foreign businesses.

⁹ ‘Yue Mei Group: From Product Export to Industrial Cluster Export’ (2009).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

13.4.3 Incentivizing First Movers Rodriguez-Clare (2007) and Lin (2011) emphasized the importance of ensuring that place-based development programmes are compatible with comparative advantages. Although the principle is clear, there are few clues as to which industry a government should support. Mapping from principle to action is not an easy task for a government. Instead of relying on governments to pick winners, an alternative strategy is to encourage private enterprises to discover a profitable business model. The process of cost structure discovery poses massive positive externality. Once first movers figure out a profitable business opportunity, others can easily imitate it. Therefore the first movers cannot capture the positive externality. Being aware of that, firms are often reluctant to be first movers, lowering their chances of discovering new business models and resulting in socially less-than-optimal outcomes. It makes economic sense to subsidize the first movers (Hausmann and Rodrik 2003; Lin 2010) by offering them special treatment—such as tax breaks and free land. However, to avoid rent-seeking behaviour, there needs to be a stick within the incentive programme: it ought to be designed with a predetermined exit strategy, and linked directly to individual company performance or a time window. While both elements of carrot and stick are present behind the success of cluster-based development in East Asia, Latin America has had ‘too much of the carrot and too little of the stick’ in its industrial policies. This could explain the discrepancy of industrial growth between the two (Rodrik 2004).

13.4.4 Step-by-Step Approach Xing and Zhang (2013) suggest that the successes of place-based policies in China are characterized by a gradual approach coupled with an experimental mentality. The development of China’s SEZs have followed a step-by-step approach: first came Shekou industrial park in 1979 (only 11 square kilometres), followed by the larger-scale Shenzhen SEZ (328 square kilometres) in 1980, followed by fourteen coastal opening-up cities in 1984, and culminating in China joining the World Trade Organization in 2001. But building industrial parks (or SEZs) is a new endeavour for many governments, and they are concerned about potential failures and negative spillovers. By starting small, governments can learn whether the idea of the industrial park works in local soil. If it fails, the negative spillover effect is limited to a narrow area. If it succeeds, it will boost governments’ confidence to scale up industrial parks to wider regions. Malaysia, Jamaica, Kuwait, and Jordan have adopted this gradual and experimental approach in testing the impacts of SEZs (Akinci and Crittle 2008). As an extreme example, Honduras in Central America has gradually increased the scale of its SEZs since the 1970s, and the government declared the whole country a ‘free zone area’ in 1998 (Farole and Akinci 2011).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



13.4.5 Industrial Park Failures Despite evidence that confirms the positive impacts of place-based policies in China and other countries, such as Indonesia, Malaysia, Sri Lanka, and South Korea (Jayanthakumaran 2003), such policies have failed in other locales. The political economy of land poses challenges to Indian’s SEZs. Legal restrictions remain and discourage private developers from assembling the necessary land for development. The Indian Zone Authority was not granted autonomy over zone development and approval clearance until 2005. Despite the launch of the 2005 SEZ Act, state governments and public-sector actors retain significant control over land procurement and transactions. Due to strict legal enforcement, including land ceiling and land use clauses, private developers need governmental patronage for land acquisition (Seshadri 2012). Even if they succeed, the size of a zone is limited to 5,000 hectares (Mitra 2007). Conflicting interests over land acquisition between citizens and firms operating in industrial parks is another hindrance.¹⁰ Issues of dislocation and rehabilitation, coupled with the fact that India is a democratic country, make the problem even more complicated. Similar to the Indian SEZs, which are managed by public-sector actors, the industrial zones in Egypt are managed by the central government. The failures of Egyptian industrial parks stem from information gaps between the central government and grassroots entrepreneurs. The bureaucratic SEZ system constitutes numerous layers and leads to a mismatch between government and SEZ firms. Worse still, as the zone policies evolve from time to time, SEZ firms must expend unnecessary energy understanding and dealing with repeated policy changes. In contrast, Zeng (2010) attributes the success of China’s SEZs to the active and pragmatic facilitation of the local governments and a strong commitment by the state. As demonstrated by Shenzhen, China’s first SEZ, fiscal decentralization incentivized better-informed provincial and municipal governments to tailor policies and regulations to local needs, such as providing a sound judicial system, constructing infrastructure, and granting preferential policies. A number of other initiatives failed for less complicated reasons. For example, an export processing zone in Senegal was unable to blossom because of high electricity costs, expensive labour, excessive bureaucracy, and a lack of transportation infrastructure (Cling and Letilly 2001). In general, the studies on the failure of industrial parks and SEZs are scant. More research is needed to understand such failures.

¹⁰ In 2008, with the aim of mass producing the Tata Nano, the world’s cheapest car at the time, the Indian vehicle manufacturer Tata Motors established a factory in an SEZ in Singur, West Bengal. The West Bengal government offered compensation to more than 10,000 farmers and acquired 1,000 acres of land for the project, while another 2,000 farmers refused. Their protest ultimately forced Tata Motors to abandon the plant. The protest in Singur is merely one of the many cases reflecting the wider problem in the implementation of place-based policies in India: industrialization needs land but local farmers are not willing to give it up (‘Tata Abandons Cheapest Car Plant’ 2008).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

13.5 G R

.................................................................................................................................. Clusters and industrial parks are location specific. Because of an informational advantage, local governments are in a better position than the central government to identify the bottlenecks that afflict clusters and industrial parks and work out solutions. As clusters and industrial parks evolve, new bottlenecks emerge, requiring new solutions. This in turn calls for continuous tinkering by local governments. It is important to place local governments and business communities in the driving seat of local economic growth so that they can watch out for and adjust to bumps in the road. However, it is challenging to strike a balance between autonomy and embeddedness (Rodrik 2004): to reduce corruption, the government needs to maintain its autonomy with regard to private interests, but to elicit information from the private sector the government should be embedded in a close relationship with it. China has used fiscal decentralization and the evaluation of officials’ performance as the major instruments to align local officials’ incentives with local economic development (Xu 2011). An essential element of fiscal decentralization in China is that career competition between regional officials at the same level is based on fiscal performance, which effectively mitigates the problem of incentive misalignment. However, the incentive design used in China may not apply to other countries. Due to differences in institutions, the forms of incentive mechanisms are likely to vary across countries and over time. In a country with strong state capacity, such as China, it is not an issue to earmark a certain area as an industrial park and provide it with favourable policies and infrastructure. But in some democratic countries, it may not be legitimate to offer special treatment to certain locations. The industrial park concept does not necessarily transmit well to all developing countries. One should bear in mind the limitations that apply to using industrial parks as a policy instrument to foster industrial development.

 

..................................................................................................................................

T O   W F C At the beginning of the 1980s, when China began its transition from a command economy to a market economy, footwear products were in seriously short supply. The strong market demand prompted many employees of state-owned or collectively owned footwear factories to set up their own footwear stalls or family workshops and produce whole shoes by themselves. Due to the highly technological requirements for whole-shoe production, most of the early newcomers to the industry were former technicians from the state or collective firms. A good example of technical diffusion may be seen in the state-owned Dongfanghong Leather Footwear Factory, which gave rise to three major enterprises, namely Jierda Footwear,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



China Aolun Shoes, and Wenzhou Dashun Footwear Machinery Manufacture, as well as many smaller enterprises, such as the Tailong Footwear Last Factory. Having apprenticeship experience was found to be a major asset in setting up a shoemaking business. The most prominent example is Yu Ashou, the founder of Jierda Footwear. Yu had sixteen apprentices, fifteen of whom set up their own companies, while the last one became his son-in-law and worked in Jierda Footwear. Copying and spin-offs further increased footwear production as did the rate of technological diffusion. Aokang and Hongqingting are two typical examples of spin-offs. Wang Zhentao and Qian Jinbo first worked as carpenters and later sold shoes together until 1988, when they co-founded a leather shoe factory. In 1995, the factory split into the Aokang Group and the Hongqingting Group, both of which still exist today. After the split, both groups grew into leading footwear companies. The formation of an industrial cluster is a process of production and technological diffusion through the copying of others. The success of one enterprise often lures others to imitate it, resulting in numerous enterprises being duplicated. As far as Wenzhou’s diffusion channels are concerned, this process was accomplished primarily through relatives and friends. Source: Adapted from Huang et al. (2008).

 

..................................................................................................................................

B  L C   P C C As production grew, so did the volume of transportation into and out of Puyuan. Initially many small, private logistics companies, each operating only one or two routes, served the cluster. It was not economical for each transport company to build separate loading docks and parking lots, meaning that trucks often blocked the streets when loading goods. Some of the companies even hired thugs to fight for the most lucrative routes. In 1995, to reduce chaos and improve efficiency, the local government intervened and organized twenty-seven private logistics and transport companies into a shareholding company with the local government as the largest shareholder. The company invested 40 million yuan to build a logistics business centre, a loading dock, a 150,000-square-metre warehouse, and a parking lot. The company has auctioned off 109 routes to more than 140 major Chinese cities to private investors. Although the company would seem to have a natural local monopoly, shipping costs through the Puyuan logistics centre have decreased since the company’s inception. This may be due to competition from the neighbouring Honghe Township’s logistics centre. Source: Adapted from Ruan and Zhang (2009).

A I am grateful for the support from China Natural Science Foundation (approval number 71874008) for this research; and for Chi Man Cheung’s excellent research assistance.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

R Aggarwal, A., 2005. ‘Performance of Export Processing Zones: A Comparative Analysis of India, Sri Lanka, and Bangladesh’, Indian Council for Research on International Economic Relations, 155, pp. 10–13. Akinci, G. and J. Crittle, 2008. Special Economic Zones: Performance, Lessons Learned, and Implications for Zone Development. Washington, DC: World Bank. Bair, J. and G. Gereffi, 2001. ‘Local Clusters in Global Chains: The Causes and Consequences of Export Dynamism in Torreon’s Blue Jeans Industry’, World Development, 29 (11), pp. 1885–903. Bazan, L. and L. Navas-Alemán, 2001. Comparing Chain Governance and Upgrading Patterns in the Sinos Valley, Brazil. Porto Alegre, Brazil: Federation of Industries of Rio Grande do Sul. Becker, G. and K. M. Murphy, 1992. ‘The Division of Labor, Coordination Costs, and Knowledge’, Quarterly Journal of Economics, 107 (4), pp. 1137–60. Ciccone, A., 2002. ‘Agglomeration Effects in Europe’, European Economic Review, 46 (2), pp. 213–27. Ciccone, A. and R. E. Hall, 1996. ‘Productivity and the Density of Economic Activity’, American Economic Review, 86 (1), pp. 54–70. Cling, J. P. and G. Letilly, 2001. ‘Export Processing Zones: A Threatened Instrument for Global Economy Insertion?’, Working Paper DT/2001/17, Paris: DIAL. Collier, P. and S. Dercon, 2014. ‘African Agriculture in 50 Years: Smallholders in a Rapidly Changing World?’, World Development, 63, pp. 92–101. Cull, R., L. E. Davis, N. R. Lamoreaux, and J. L. Rosenthal, 2006. ‘Historical Financing of Smalland Medium-Size Enterprises’, Journal of Banking and Finance, 30 (11), pp. 3017–42. Farole, T. and G. Akinci, eds, 2011. Special Economic Zones: Progress, Emerging Challenges, and Future Directions, Washington, DC: World Bank. Fisher, P. S. and A. H. Peters, 2002. State Enterprise Zone Programs: Have They Worked? Kalamazoo, MI: Upjohn Institute for Employment Research. Fleisher, B., D. Hu, W. McGuire, and X. Zhang, 2010. ‘The Evolution of an Industrial Cluster in China’, China Economic Review, 21 (3), pp. 456–69. Glaeser, E. L. and J. D. Gottlieb, 2009. ‘The Wealth of Cities: Agglomeration Economies and Spatial Equilibrium in the United States’, Journal of Economic Literature, 47 (4), pp. 983–1028. Greenstone, M., R. Hornbeck, and E. Moretti, 2010. ‘Identifying Agglomeration Spillovers: Evidence from Winners and Losers of Large Plant Openings’, Journal of Political Economy 118 (3), pp. 536–98. Greif, A., 1993. ‘Contract Enforceability and Economic Institutions in Early Trade: The Maghribi Traders’ Coalition’, American Economic Review, 83 (3), pp. 525–48. Hausmann, R. and D. Rodrik, 2003. ‘Economic Development as Self-Discovery’, Journal of Development Economics, 72 (2), pp. 603–33. Hounshell, D. A., 1984. From the American System to Mass Production, 1800–1932, Baltimore: Johns Hopkins University Press. Huang, Z., X. Zhang, and Y. Zhu, 2008. ‘The Role of Clustering in Rural Industrialization: A Case Study of the Footwear Industry in Wenzhou’, China Economic Review, 19 (3), pp. 409–20. Jayanthakumaran, K., 2003. ‘Benefit–Cost Appraisals of Export Processing Zones: A Survey of the Literature’, Development Policy Review, 21 (1), pp. 51–65.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



Lee, J. R. and J. S. Chen, 2000. ‘Dynamic Synergy Creation with Multiple Business Activities: Toward a Competence-Based Business Model for Contract Manufacturers’, in R. Sanchez and A. Heene, eds, Theory Development for Competence-Based Management, Advances in Applied Business Strategy, Stanford, CT: Jai Press, pp. 209–28. Lin, J. Y., 2010. ‘Six Steps for Strategic Government Intervention’, Global Policy, 1 (3), pp. 330–1. Lin, J. Y., 2011. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: World Bank. Lin, J. Y., F. Cai, and Z. Li, 2003. The China Miracle: Development Strategy and Economic Reform, Chinese University Press. Long, C. and X. Zhang, 2011. ‘Cluster-Based Industrialization in China: Financing and Performance’, Journal of International Economics, 84 (1), pp. 112–23. Markusen, A., 1996. ‘Sticky Places in Slippery Space: A Typology of Industrial Districts’, Economic Geography, 72 (3), pp. 293–313. Marshall, A., 1920. Principles of Economics, 8th edn, London: Macmillan. (Original work published 1890.) McGregor, R., 2005. ‘China Must Cut Farming Population, Says OECE’, Financial Times, 15 November. Meyer-Stamer, J., 1998. ‘Path Dependence in Regional Development: Persistence and Change in Three Industrial Clusters in Santa Catarina, Brazil’, World Development, 26 (8), pp. 1495–511. Miller, R. and M. Cote, 1985. ‘Growing the Next Silicon Valley’, Harvard Business Review, 63 (4), pp. 114–23. Mitra, S., 2007. ‘Special Economic Zones in India: White Elephants or Race Horses’. Available at SSRN: https://papers.ssrn.com/sol3/papers.cfm?abstract_id=969274 Munshi, K., 2011. ‘Strength in Numbers: Networks as a Solution to Occupational Traps’, The Review of Economic Studies, 78 (3), pp. 1069–101. Nadvi, K., 1999. ‘The Cutting Edge: Collective Efficiency and International Competitiveness in Pakistan’, Oxford Development Studies, 27 (1), pp. 81–107. Nadvi, K. and H. Schmitz, 1999. ‘Clustering and Industrialization: Introduction’, World Development, 27 (9), pp. 1503–14. Nakabayashi, M., 2006. ‘Flexibility and Diversity: The Putting-Out System in the Silk Fabric Industry of Kiryu’. Discussion paper, Osaka: Graduate School of Economics, Osaka University. Otsuka, K., 2013. ‘Food Insecurity, Income Inequality, and the Changing Comparative Advantage in World Agriculture’, Agricultural Economics, 44(s1), pp. 7–18. Oyelaran-Oyeyinka, B. and D. McCormick (eds), 2007. Industrial Clusters and Innovation Systems in Africa, Tokyo: United Nations University Press. Pingali, P., 2007. ‘Agricultural Mechanization: Adoption Patterns and Economic Impact’, Handbook of Agricultural Economics, 3, pp. 2779–805. Porter, M. E., 1990. ‘The Competitive Advantage of Notions’, Harvard Business Review, 68 (2), pp. 73–93. Rodriguez-Clare, A., 2007. ‘Clusters and Comparative Advantage: Implications for Industrial Policy’, Journal of Development Economics, 82 (1), pp. 43–57. Rodrik, D., 2004. ‘Industrial Policy for the Twenty-First Century’. Working paper. John F. Kennedy School of Government, Cambridge, MA: Harvard University. Ruan, J. and X. Zhang, 2009. ‘Finance and Cluster-Based Industrial Development in China’, Economic Development and Cultural Change, 58 (1), pp. 143–64.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Ruan, J. and X. Zhang, 2010. ‘ “Made in China”: Crisis Begets Quality Upgrade’. IFPRI Discussion Paper No. 1025, Washington, DC: International Food Policy Research Institute. Ruttan, V. W., 2001. Technology, Growth, and Development: An Induced Innovation Perspective, New York: Oxford University Press. Saxenian, A., 1994. Regional Advantage: Culture and Competition in Silicon Valley and Route 128, Cambridge, MA: Harvard University. Schmitz, H., 1999. ‘Global Competition and Local Cooperation: Success and Failure in the Sinos Valley, Brazil’, World Development, 27 (9), pp. 1627–50. Seshadri, T., 2012. ‘An Analysis of the Feasibility of Private Land Assembly for Special Economic Zones in India’, Urban Studies, 49 (10), pp. 2285–300. Smith, A., 1776. The Wealth of Nations, London: W. Strahan and T. Cadell. Sonobe, T. and K. Otsuka, 2006. Cluster-Based Industrial Development: An East Asian Model, London: Palgrave Macmillan UK. ‘Tata Abandons Cheapest Car Plant’, 2008. BBC News. Available at: http://news.bbc.co.uk/2/ hi/south_asia/7651119.stm UNIDO (United Nations Industrial Development Organization), 1997. Industrial Estates: Principles and Practice, Vienna: UNIDO. Williamson, O. E., 1975. Markets and Hierarchies, New York: Free Press. Xing, H. and X. Zhang, 2013. ‘The Logic of Adaptive Sequential Experimentation in Policy Design’. IFPRI Discussion Paper No. 1273. Washington, DC: International Food Policy Research Institute. Xu, C., 2011. ‘The Fundamental Institutions of China’s Reforms and Development’, Journal of Economic Literature, 49 (4), pp. 1076–151. Xu, C. and X. Zhang, 2009. ‘The Evolution of Chinese Entrepreneurial Firms: TownshipVillage Enterprises Revisited’. IFPRI Discussion Paper No. 854. Washington, DC: International Food Policy Research Institute. Yang, J., Z. Huang, X. Zhang, and T. Reardon, 2013. ‘The Rapid Rise of Cross-regional Agricultural Mechanization Services in China’, American Journal of Agricultural Economics, 95(5), pp. 1245–51. ‘Yue Mei Group: From Product Export to Industrial Cluster Export’, 2009. Economic Information Daily. Available at: http://jjckb.xinhuanet.com/gnyw/2009-07/01/content_166586. htm Zeng, D. Z. (ed.), 2010. Building Engines for Growth and Competitiveness in China: Experience with Special Economic Zones and Industrial Clusters, Washington, DC: World Bank. Zhang, X. and D. Hu, 2014. ‘Overcoming Successive Bottlenecks: The Evolution of a Potato Cluster in China’, World Development, 63, pp. 102–12. Zhang, X., J. Yang, and T. Reardon, 2015. ‘Mechanization Outsourcing Clusters and Division of Labor in Chinese Agriculture’. IFPRI Discussion Paper No. 1415, Washington, DC: International Food Policy Research Institute.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

        ......................................................................................................................

  Mobilizing Long-Term Liability Embedded Funds from International Institutional Investors to Emerging Markets .............................................................................................................

 -   

14.1 I

.................................................................................................................................. I needs for infrastructure services in emerging market and developing economies (EMDEs) are growing rapidly. Modern and high-quality infrastructure services are crucial for EMDE countries to improve competitiveness by increasing productivity and allowing greater access to the global market. At the same time, global trends such as urbanization, rapid economic and population growth in emerging countries, and rising expectations for service delivery are placing increasing pressure on the expansion and maintenance of infrastructure. Therefore, EMDE countries must seek additional stable investment sources in order to close the funding gap for infrastructure and stave off potential damages associated with insufficient provision of infrastructure services. In the past, infrastructure funding has depended largely on direct government funding and traditional sources of infrastructure finance, including international financial institutions (IFIs), development finance institutions (DFIs), and in some instances commercial banks. Governments in emerging economies are finding it increasingly difficult, however, to maintain the required levels of infra funding, particularly given strained budgets following the 2008 Global Financial Crisis (GFC), inherited regulatory changes such as IPSAS (International Public Sector Accounting Standards), and a legacy of committed support for underperforming and/or underdeveloped existing infrastructure. Furthermore, traditional sources of infrastructure finance must be supplemented by a wider range of long-term capital, particularly as

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Table 14.1 Two-track approach (current and additional approach) Current approach Target institutions

Mainly local investors Commercial banks, IFIs, DFIs, MDBs, NDBs Key Currency match characteristics Low government capacity for PPP Local investor familiar with local project environment Less strict regulation Limited asset size Short-term investment

Additional approach Mainly foreign investors Long-term investors (e.g. pension funds, life insurance) Currency mismatch Increased government capacity for PPP International investors unfamiliar with local project environment Strict regulation with changes Huge asset size Long-term investment

regulatory pressures (e.g. the Third Basel Accord on banking regulation (Basel III)) are placing greater constraints on the supply of long-term bank debt, and increasing importance is placed on matching the durations of projects with investment horizons to avoid refinancing risk. While the efforts of local governments and multilateral organizations to facilitate local funding and financing for EMDE projects undoubtedly remains critical to infra development, the magnitude of the funding gap and pervasiveness of challenges associated with raising sufficient funds suggests that a multidimensional approach is required. Moreover, this multidimensional approach demands integrating multiple strategies rather than implementing a host of strategies independently, in order to find better and more integrated solutions that expand funding sources, but with high flexibility and accessibility to capital markets. This chapter focuses specifically on the mobilization of external (extra-national) sources of private sector finance to support EMDE projects. To develop effective and bankable products, we must develop suitable structures that include new target institutions and also consider changing some characteristics of the infrastructure funding space (see Table 14.1). The proposed solutions can also play an important role in reducing the current fiscal burden on EMDE governments, enhancing the bankability of EMDE PPP projects, and bringing long-term institutional investors to the EMDE infrastructure market by offering financial products that serve their demands.

14.1.1 The Infrastructure Funding Gap According to the Global Infrastructure Hub (2018), by 2040 the population will grow by almost 2 billion people triggering a massive demand for infrastructure. To meet this demand, investment needs to reach US$94 trillion by 2040. Add to this the UN

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Sustainable Development Goals (SDG) of universal provision of water, sanitation, and electricity implies a total cost of US$97 trillion. Closing the gap and meeting the SDGs will require spending as a proportion of global GDP to grow from the current level of 3 per cent to 3.7 per cent. When looking only at EMDEs and the expenditure made through public–private partnerships (PPP), investment, as a share of GDP, grew solidly through the 1990s, from 0.1 per cent in 1991 to 1.1 per cent in 1997 (Ruiz-Nunez and Wei 2015). Following the crisis, however, this figure steadily declined over 7 years, from 1.1 per cent in 1997 to 0.2 per cent in 2004, as investments fell faster than GDP. Nonetheless, a wave of structural reforms, favourable growth policies, and a recovering global economy in the mid-2000s resulted in a second growth phase that culminated in record investments of US$158 billion in 2012. While total global investment in PPPs in absolute terms experienced a second expansion between 2005 and 2012, with higher investment levels, as compared to the previous growth wave, infrastructure investment as a percentage of GDP remained relatively low (between 0.2 per cent and 0.6 per cent), failing to meet or surpass pre-1997 levels (Ruiz-Nunez and Wei 2015). Additionally, the risks associated with infrastructure investments—especially, after the financial crises—have further impacted capital investments in the sector. In order to mitigate these risks and attract private financing, it is necessary to develop practical financing instruments whose design is based on a thorough analysis of the global financial market, as well as recent and expected changes in financial and accounting regulation and practice.

14.1.2 Responding to the Challenge Designing appropriate products to address the problems of infrastructure finance, particularly for EMDE countries, also requires addressing both public and private sector challenges. On the one hand, EMDE governments are faced with budget constraints, changing accounting standards (e.g. IPSAS 32), and inefficient pricing policies, as well as traditionally low levels of spending, operational inefficiencies, and low capacity in infrastructure asset management, which have led to degraded infrastructure. Private sector participation, on the other hand, is limited by an insufficient supply of bankable projects, as well as significant changes in regulation and accounting standards, including the International Financial Reporting Standards (IFRS), BASEL III, and the European Union’s Solvency II Directive (Solvency II), which will be discussed further. In response to these challenges, this chapter explores ways to lessen EMDE government fiscal burdens and to bring new funding sources to finance EMDE infrastructure projects by matching capital needs with appropriate investment candidates, while minimizing refinancing risk. By examining and diagnosing the primary challenges of the current environment for financing, we can engineer effective products for mobilization of international funds to these markets. The chapter reviews global financial market trends, including central bank monetary policy, global bond market trends, and global

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

equity market trends; major changes to regulatory and accounting standards; and impacts on the financial climate of infrastructure. The chapter also examines potential target institutions and countries, as well as cases of other product introductions, to propose a series of four innovative products that (1) consider global financial conditions and key stakeholder interests; (2) mobilize available resources to support EMDE infrastructure; and (3) promote infrastructure investment as a major asset class. The chapter concludes with proposals for four products, accompanied by introductory implementation notes: (i) ABS of existing infrastructure recycling; (ii) an infrastructure collateralized loan obligation (I-CLO) product; (iii) an infrastructure project puttable bond; and (iv) an infrastructure bond backed by Multilateral Investment Guarantee Agency (MIGA) nonhonouring of sovereign financial obligation bond.

14.2 C C  G I F  F

.................................................................................................................................. Understanding the current climate of infrastructure funding and finance worldwide requires an examination of the global financial market; the impacts of regulatory and accounting change on investments, particularly for infrastructure; and the array of demands, incentives, and constraints facing key stakeholders in infrastructure development, including EMDE countries and different types of investors.

14.2.1 Conditions of the Global Financial Market Regarding the overall conditions of the global financial market, we examine central banks’ key rates, as well as the conditions of global bond and equity markets. After the GFC, major advanced economies have been adopting accommodative monetary policies, such as quantitative easing and the lowering of key rates. More recently, some countries have adopted non-traditional negative central key rates to fight against deflation or currency appreciation, causing major global central bank key rates to fall to record-low levels, as shown in Table 14.2. The post-crisis world economy has faced low growth and overflowing liquidity, instigated by globally synchronized accommodative monetary policies, causing major market bond yields to fall over the last decade (see Figure 14.1). Within an environment of lowering interest rates, the low bond yield market is extremely challenging for institutional investors, especially long-term investors, due to low expected rates of return on the major asset classes of their portfolios. In addition, the global bond markets’ yield curve has flattened after the GFC due to low expected inflation, an abundance of liquidity, and increasing demand for longterm bonds, particularly from pension funds in ageing economies. This flattened yield

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Table 14.2 Major countries’ central bank key rates (as of 18 November 2016) Change after financial crisis Current rate (since Dec. 2007)

Country

Central bank Key interest rate

Last change

USA Eurozone UK Japan Canada Switzerland Sweden Australia New Zealand Norway Taiwan S. Korea

FED ECB BOE BOJ BOC SNB Riksbank RBA RBNZ Norges Bank BOT BOK

Federal Funds rate Refinancing tender Bank rate Overnight call rate Overnight lending rate 3-mo Libor Repo rate Cash rate Official cash rate Sight deposit rate Policy rate Central rate

16 Dec 2015 0.25–0.5% 10 Mar 2016 0.0% 5 Mar 2009 0.25% 29 Jan 2016 0.1% 15 Jul 2015 0.5% 15 Jan 2015 0.75% 11 Feb 2016 0.50% 2 Aug 2016 1.50% 10 Nov 2016 1.75% 18 Mar 2016 0.50% 30 Jun 2016 1.375% 9 Jun 2016 1.25%

7 6 5 4 3 2 1 0 –1 –2 31/12/2009

31/12/2010

31/12/2011

31/12/2012

31/12/2013

31/12/2014

# 375bp # 400bp # 525bp # 60bp # 375bp # 350bp # 450bp # 525bp # 650bp # 475bp # 200bp # 375bp

31/12/2015

USA

UK

CANADA

AUSTRALIA

SWISS

SWEDEN

NEW ZEALAND

CHINA

 . Major economy 10-year bond yields, December 2009–July 2016 Source: Bloomberg.

curve makes it far more difficult for long-term investors to meet their target rates of return, since long-term investors utilize term premiums of bond yields as one of the main sources of profits. Thus, it is highly likely that institutional investors will look for alternative investment opportunities to meet their target rates (Table 14.3). Lastly, global equity markets have recovered from the 2008 GFC, as major central banks have implemented intensive and proactive accommodative monetary policies, providing abundant liquidity to the global equity market. Nevertheless, after an initial rebounding from the Crisis, the global equity market has lost its growth momentum. In particular, emerging market equity has shown poor performance, while developed markets have achieved stable rates of return following the post-Crisis equity market

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Table 14.3 Major bond markets’ yield and yield curve (as of 4 August 2016) Tenors

USA Germany UK France Japan China Switzerland Sweden Australia Taiwan Singapore South Korea

2-year

5-year

10-year

30-year

2/10 yearterm spread

2/10 year-term spread change (since Dec. 2009)

0.68% 0.60% 0.18% 0.55% 0.19% 2.31% 1.07% 0.66% 1.47% 0.31% 0.90 1.21%

1.08% 0.48% 0.35% 0.37% 0.18% 2.55% 0.94% 0.37% 1.54% 0.51% 1.41 1.22%

1.55% 0.03% 0.80% 0.19% 0.09% 2.77% 0.58% 0.15% 1.94% 0.67% 1.84 1.40%

2.30% 0.45% 1.62% 0.94% 0.41% 3.30% 0.08% N/A N/A 1.33% 2.32 1.49%

87bp 57bp 61bp 74bp 10bp 45bp 50bp 80bp 47bp 35bp 94bp 19bp

270bp 205bp 272bp 239bp 114bp N/A 161bp 158bp 124bp N/A 210bp 102bp

Source: Bloomberg.

1250

450 430

1150

410 390

1050

370

950

350 330

850

310 290

750

270 250 31/12/2009

31/12/2010

31/12/2011

31/12/2012

MSCI ACWI (Left)

31/12/2013

31/12/2014

31/12/2015

650

MSCI EMERGING (Right)

 . Global equity market indices Source: Bloomberg.

rebound (see Figure 14.2). Overall, the global equity market has shown an 8.4 per cent annualized rate of return over the last 5 years to the end of June 2016. The rate of return of global equity has plunged over the past year, however, marking a 5.7 per cent rate of return from June 2015 to June 2016, led by the lowering growth rate of the world economy (MSCI AWCI). This is expected to put pressure on institutional investors to find alternative investment opportunities in order to meet target rates of return (see Table 14.4).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Table 14.4 Major equity market rates of return Indices

Average yield (July 2015 ~ June 2016)

Average yield (July 2013 ~ June 2016)

Average yield (July 2011 ~ June 2016)

5.7% 14.2% 1.7% 16.3% 0.3% 23.0% 31.5% 5.0%

4.1% 3.8% 10.2% 3.4% 1.5% 4.6% 16.0% 1.9%

8.4% 13.6% 11.8% 0.3% 4.7% 29.3% 3.0% 3.1%

MSCI AWCI MSCI EM S&P 500 Eurostoxx FTSE 100 Nikkei Shanghai South Korea Source: Bloomberg.

3.0% 2.0% 1.0% 0.0%

1997

1999

2001

2003

2005

2007

2009

2011

2013

2015

–1.0% –2.0% Greater China, Japan, Korea

US

Oil exporter (Major 10)

 . Current account as a percentage of world GDP, 1997–2015

14.2.2 Implications of the Global Market Trend Another driver of decreasing interest rates is attributable to the demand side. A global savings glut has resulted from growing imbalances of national current accounts and countries’ growing preferences to hold hard currency-based assets (mainly major countries’ bonds) to manage solvency risks (see Figure 14.3). Since the pace of investments has not kept abreast of the savings increase, the excess supply of savings has been transferred to the global bond market. The experience of the AFC has also spurred countries to hold foreign currency reserve assets to hedge the risk of financial system collapse due to a lack of foreign reserves (see Table 14.5). It is highly expected that the imbalance of current accounts and the savings glut will continue to grow, for three reasons. First, export-driven economies will hold the savings strategy to maintain low exchange rates in order to drive export-led growth, particularly because these countries have excess labour. They are also likely

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Table 14.5 Top 10 foreign currency reserve assets (as of April 2016) Country China EURO Saudi Arabia Hong Kong India

Official reserve asset (US$ billion) 3,520 820 581 380 366

Country

Official reserve asset (US$ billion)

Japan Switzerland Russia Federation South Korea Brazil

1,321 661 407 373 362

Source: International Monetary Fund (IMF) data.

Table 14.6 Global interest rates (as of 18 November 2016) Country

Yield

1 month

1 year

Country

Yield

1 month

1 year

USA Canada Brazil Mexico Germany UK France Korea

2.21% 1.53% 12.22% 7.01% 0.30% 1.38% 0.72% 2.10%

+41 bps +29 bps +86 bps +121 bps +25 bps +28 bps +39 bps +49 bps

+0 bps 12 bps 328 bps +96 bps 25 bps 60 bps 15 bps 17bps

Spain Netherland Portugal Greece Switzerland Japan Singapore Hong Kong

1.45% 0.45% 3.47% 7.22% 0.19% 0.01% 2.26% 1.24%

+33 bps +29 bps +20 bps 99 bps +34 bps +6 bps +40 bps +40 bps

33 bps 27 bps +78 bps +22 bps +17 bps 28 bps 30 bps 25 bps

Source: Bloomberg.

to purchase large quantities of US Treasury bonds to hedge currency-driven financial risks. Second, the demographic shifts associated with the ageing of many world economies will encourage increased saving. And third, the experiences of financial crises drive home the importance of savings in hard currency. Recently, global bond yields have been showing a rising trend (see Table 14.6) due to (i) a US monetary policy of normalization; (ii) expectation of a US expansionary fiscal policy with political change; and (iii) a recent upward trend for inflation rates. It is highly likely, however, that interest rates will not reach pre-GFC levels due to three factors: the global economy’s huge exposure to debt, the global economy’s slow recovery, and a global savings glut. Nevertheless, if global interest rates rise faster than market expectations or reveal a significant long-term upward trend, there will be limited space for international funds to consider new investments in emerging countries. Thus, there is a need to act swiftly with respect to developing products for mobilization, and approach potential investors before the current downstream window closes. While global markets have somewhat recovered from the GFC, due largely to lowered bond yields and the equity market rebound, investors are nevertheless facing

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



difficult and unprecedented market conditions with record-low bond yields, a flattened yield curve, and low expected returns on equity. Additionally, accommodative monetary policies, such as quantitative easing (QE) and the lowering of key rates, are no longer as effective as they have been in the past. Given these challenges, long-term institutional investors, such as pension and life insurance companies, are under pressure to achieve investment objectives of meeting their liability costs, which determine their target rates of return. Moreover, there is limited room for bond yields and bond yield curves to fall or flatten. As bonds have been the main source of profit for long-term institutional investors, these conditions limit the potential of traditional profit sources of cash flows or capital gains, and, in turn, push the institutional investor to look for alternative opportunities. Accompanied by an increasing risk appetite, it is an opportune time for emerging market projects to approach long-term liabilityembedded institutional investors, such as pension funds and life insurance companies.

14.3 C  R  A S

.................................................................................................................................. Four major shifts in the financial regulatory landscape have shaped a changing investor demand. These include the adoption of the second phase of the International Financial Reporting Standards fourth revision (IFRS 4 Phase II); the 32nd International Public Sector Accounting Standard (IPSAS 32); the European Union’s Solvency II Directive (Solvency II); and the Third Basel Accord on banking regulation (BASEL III). These are briefly sumarized below.

14.3.1 IFRS 4 Phase II (IFRS17) International accounting standards revisions most likely to affect insurance companies and pension funds relate to those imposed via the second phase of International Financial Reporting Standards Four, which is now referred to as IFRS17. The new financial reporting standard for insurance contracts is under preparation, with a likely effective date of 2021. The first phase of IFRS 4 (introduced in 2005) provided a universally accepted definition of an insurance contract, defined by the International Accounting Standard Board (IASB) as ‘a contract under which one party (the insurer) accepts significant insurance risk from another party (the policyholder) by agreeing to compensate the policyholder if a specified uncertain future event (the insured event) adversely affects the policyholder’ (CGFS 2011, 22). IFRS 4 Phase I additionally imposed disclosure requirements to explain insurers’ financial statements, particularly with respect to items arising from insurance contracts. This was aimed at helping consumers of the

OUP CORRECTED PROOF – FINAL, 10/1/2019, SPi



 -   

financial statements better understand expected amounts, timings, and uncertainty characteristics of future cash flows derived from insurance contracts (CGFS 2011). It is acknowledged, however, that this first phase was limited, as it did not address issues associated with the measurement and recognition of liabilities and, thus, did not rectify the problem of different valuation methodologies being applied by various local Generally Accepted Accounting Principles (GAAPs) (CGFS 2011). For instance, IFRS 4 Phase I allowed insurance companies to measure liabilities on an undiscounted basis, if they had applied this approach previously under national accounting standards prior to the adopting IFRS (CGFS 2011). As such, one insurer might assess and report a claim due tomorrow in the same way as a claim due in 20 years. Some insurance companies were using discount rates to reflect the characteristics of the assets backing the insurance liabilities, whereas others applied rates reflecting the characteristics of the liabilities themselves (CGFS 2011). As such, IFRS 4 Phase I was considered a temporary solution, pending the finalization of a second phase whose objective would be to set out a robust and relevant model for accounting for insurance liabilities. IFRS 4 Phase II (IFRS 17) rectified this issue with a standardized approach wherein insurers cannot measure insurance liabilities on an undiscounted basis. Rather, they must determine reasonable discount rates based on respective liabilities’ characteristics.

14.3.1.1 Impact of IFRS4 Phase II Insurance firms and pension funds are likely to feel the impact of IFRS 4 Phase 2 in three ways: (i) increased volatility of income statements due to marked-to-market valuation; (ii) changes associated with a general strategy shift from asset-only management to asset-liability matching; and (iii) additional capital increase requirements in preparation for expected inflation of liabilities resulting from revaluation. The first key impact will be increased volatility of income statements due to marked-to-market valuation. The insurance business is intrinsically long-term in nature, with liabilities often extending over many years (even decades in the case of life insurance). If insurance liabilities, under the final standard, are re-measured periodically through profit and loss statements, insurers’ reported annual performance will be influenced substantially by changes in financial and non-financial assumptions (based on financial markets) that are used for measuring those liabilities. For example, changes in interest rates could have a significant impact on earnings and capital. Changes in the financial features of insurance contracts largely reflect short-term market fluctuations that are mostly unrelated to the ultimate fulfilment of the insurance liabilities. Overall, to reduce volatility of their income statements, insurance companies and pension funds invest in assets that have similar characteristics with their own long-term liabilities. Thus, insurers are shifting their investment strategies from asset-only strategies to liability-driven investment strategies in order to lower the volatility of profits. This will also influence insurers to reduce their exposure to riskier equity-like assets and lower-rated long-term bonds. Pension funds are sometimes pressured by sponsor

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



companies to de-risk their asset sides, as the asset-valuation changes booked at the pension fund add volatility to the company’s per-share earnings. For insurers, increased accounting volatility can lead to direct market pressure to reduce their exposure to riskier asset classes or to regulatory pressure when solvency ratios come close to the regulatory minimum requirement. Insurers are also under pressure, however, to find long-term fixed incomes with high yields to meet their funding costs. Since a strategy of matching assets and liabilities tends to involve finding fixedincome financial instruments, these changes could open a window of opportunity to introduce new long-term high-yield investment instruments, despite the needs of lowering portfolios’ risk levels. Second, there is an apparent trend among insurance companies to shift from assetonly investment selection to asset-liability management (ALM), an approach to selecting investments with an eye to mitigating risks associated with mismatched assets and liabilities. ALM is now an essential part of insurers’ activities and a primary driver of their performance. When insurers apply the approach outlined in the exposure draft for IFRS 4 Phase 2, they may also choose fair valuation and match assets for all their liabilities in order to reduce accounting mismatches and profit volatility. Given the scarcity of long-term fixed income products with rates of return sufficient to meet the costs of liabilities in the market, infrastructure debt products with higher yield than highly rated traditional bonds will potentially face high demand. Lastly, IFRS4 Phase II does not permit measuring insurance liabilities on an undiscounted basis. Insurers must determine reasonable ways to discount liabilities, either by top-down or bottom-up approaches. Regardless, new discount rates for existing liabilities are expected to be lower than historical discount rates currently being used for valuation, thus significantly increasing liabilities. In order to maintain stable capital structures, insurance companies will need to supplement capital by issuing new capital or quasi-equity. In summary, IFRS4 Phase II will put pressure on insurance companies to reduce their exposure to riskier asset classes such as equity, non-investment grade bonds, or equity-like instruments in order to reduce the volatility of their portfolios. However, insurers will also face pressure to find long-term, fixed-income products to meet their market-to-market liability costs and to lessen the burden of additional capital supplements.

14.3.2 IPSAS 32 Both developed and developing countries have demonstrated widespread demand for improved public financial management in order to guarantee more accountability and transparency in the management and use of public funds. In view of this, the International Federation of Accountants (IFAC) establishes and promotes the application of International Public Sector Accounting Standards (IPSAS) by public sector entities around the world when preparing their General Purpose Financial Reports (GPFR).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Among many provisions set forth by IPSAS, IPSAS 32 deals specifically with service concession agreements, focusing on their governmental accounting impacts (APMG PPP Guide 2016). The IPSAS 32 guideline presents a comprehensive approach to accounting for public–private partnerships (PPPs) that includes most of the contracts defined as PPPs. IPSAS 32 describes service concession agreements as long-term contracts between a government and a private party whereby (a) the operator uses a public asset (such as a prison, airport, or water pipe) to provide a public service for a specified period of time on behalf of the government; and (b) the operator is compensated for its services over the period of the service concession arrangement (APMG PPP Guide 2016).¹ Both government-pay and user-pay PPP contracts are covered by IPSAS 32. Furthermore, IPSAS 32 states that contracts should be reflected on the government balance sheet reflecting the gross debt under the following circumstances: if government controls or regulates what services the operator must provide with the asset, to whom it must provide them, and at what price—and if the government also controls any significant residual interest in the asset at the end of the PPP contract term (APMG PPP Guide 2016).

14.3.2.1 Impact ofIPSAS 32 IPSAS 32 sets forth criteria for a project’s inclusion on the government balance sheet. Most PPPs do, indeed, meet these criteria, reflecting the notion that most PPPs are expected to have an actual impact on the aggregate public debt, particularly in the event of non-performance or default. Independent of the approach taken to determine the specific debt impacts, the effect of contracts on aggregate fiscal indicators (such as debt) will vary across different stages of project implementation. Accounting treatment also varies significantly across counties. In response, IPSAS 32 lays out an infrastructure contract’s effect on aggregate public debt in a standardized approach. The new standards require governments to more extensively account for PPPs on the balance sheet as ‘contingent liabilities’—that is, liabilities that could be incurred in the event of non-performance or default.² As such,

¹ The APMG International has addressed this issue. APMG International is an innovation of the Asian Development Bank (ADB), the European Bank for Reconstruction and Development (EBRD), the Inter-American Development Bank (IDB), the Islamic Development Bank (IsDB), the Multilateral Investment Fund (MIF), the Public–Private Infrastructure Advisory Facility (PPIAF), and the World Bank Group (WBG). ² According to the World Bank PPP Certification curriculum, ‘After the financial close of a PPP contract and while construction is underway, the government should include construction costs in the public balance sheet. Non-financial assets are also included, increasing gross debt but creating a null net balance sheet effect. Once the asset is operational, debt is reduced (debt amortization) by an amount equivalent to the value of each government payment that relates to the asset repayment (excluding interest and service costs). The value of a non-financial asset is also reduced based on its expected life (asset depreciation). When the asset is handed over at the end of the contract, there should be no debt remaining, and the residual value of the non-financial asset should continue to be depreciated accordingly. . . . For user-pay PPPs, the general outline of governmental debt under IPSAS 32 is very similar,

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



the inclusion of PPP contingent liabilities in government accounting generally increases recorded debt. The added fiscal burden limits a government’s ability to provide certain kinds of support for new infrastructure development, such as minimum revenue guarantees or termination fees. This added challenge requires new ways of attract private finance via innovative products.

14.3.3 Solvency II The European Union’s Solvency II Directive (Solvency II) is a comprehensive riskbased capital framework for insurance companies. The solvency capital requirement stipulates that insurance companies hold a certain level of eligible capital to ensure that they are able to meet policyholder obligations in preparation for market stress. Under the current Solvency I regime, insurance companies hold capital in proportion to technical provisions (life insurance) or gross premiums (non-life). Thus, the current regime is not able to take into consideration the risk sensitivities stemming from the asset side of the balance sheet. Solvency II, on the other hand, aims to reflect the full range of risks held by insurers on both their asset and liability sides. Capital requirements under Solvency II will reflect all the risks the firms are exposed to, including those related to insurance risks, such as longevity, morbidity, and catastrophic risks, as well as those arising from their holdings of financial assets. Thus, under Solvency II, capital requirements will be determined on the basis of the complete risk profile of the undertakings, as well as on the way in which such risks are managed (through financial hedging, reinsurance, or other risk-mitigating techniques).³ Among many changes, there are two significant alterations to Solvency I enshrined in Solvency II: (i) marked-to-market valuation of assets and liabilities, and (ii) solvency capital requirements. With respect to asset and liability valuation, in Solvency II, both assets and liabilities are marked-to-market. The present value of liabilities, or technical provision, is defined as the amount an insurer would have to pay to transfer its insurance obligations immediately to another willing buyer. It consists of the best estimate, the present value of the expected future cash flows (net payments to policyholders), calculated on a specified discount rate curve (term structure), and the risk margin, which is an additional premium above the best estimate. How the discount rate is determined is of considerable importance, given that the risk margin and the present value of liabilities will increase when this rate decreases. The current expectation of how this rate will be constructed is the swap curve (excluding credit risk), augmented

though amortization of the debt is based on the flow of the tariff revenues used for repayment of the principal.’ See https://ppp-certification.com/ppp-certification-guide/121-international-public-sectoraccounting-standards-ipsas-number-32 ³ Insurers can use either the standard formula or an internal model approach to calculate capital requirements, where the standard formula is prescribed in the Solvency II legislation and an internal model approach requires approval from the local regulator.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

by an illiquidity premium for those obligations coming due more than a year ahead. In addition, an extrapolation (technique) towards a fixed rate (ultimate forward rate) will be used to get discount rates for longer maturities than those available from market rates in different countries. With respect to solvency capital requirements (SCR) and risk modules, the SCR defines that an insurer has enough capital when it covers unexpected losses with a probability of 99.5 per cent over a one-year term. To meet the solvency capital requirement, the capital items must be recognized as eligible from a regulatory perspective. Capital items are divided into core capital (basic own funds) and contingent capital items (ancillary own funds), and classified into three tiers, depending on the lossabsorbency characteristics of the items. There are restrictions on the composition of capital for fulfilling the solvency requirement (Tier 1 should be at least 50 per cent of the SCR, Tier 3 less than 15 per cent of SCR) and the minimum requirement (Tier 1 and Tier 2 must comprise basic own-fund items only and Tier 1 should be at least 80 per cent of the MCR). The insurance undertaking can calculate SCR through an internal model, a standard model, or through a combination (a partial internal model).

14.3.3.1 Impact of Solvency II Solvency II is expected to impact the environment in which insurance companies operate, via risk-sensitive capital requirements and market-based valuation, which will incentivize insurance companies to de-risk their portfolios. In terms of impacts on investment strategies, while capital charges should not be the sole factor on which investment allocations of a portfolio are based under Solvency II, it is expected that a product engineered to lower the burden on insurers would enable increased financial commitments. An analysis of the standard formula of Solvency II, performed by Ernst and Young in 2015, utilized the Solvency II standard formula to provide an indicative comparison of return on capital and the potential efficiency of infrastructure loans, relative to other asset classes. The analysis presents scenarios with different asset types: (i) standard formula with matching asset portfolio and liabilities; and (ii) standard formula without matching asset portfolio and liabilities (Ernst and Young 2014, 2015). Based on the data shown here, infrastructure bonds with public ratings and matching liabilities have expected high returns and low burdens for capital charges. Thus, infrastructure bonds with a public rating structure would be an attractive proposal for insurance companies. Based on capital charges, an infrastructure bond with a credit rating has a capital charge similar to that of a corporate bond (see Table 14.7). However, when an infrastructure loan does not have a rating, the capital charge is far higher than corporate bonds, leading investors to prefer bonds with public ratings (Coughlan et al. 2015). Solvency II stipulates that insurers use standard formulas to calculate capital charges, unless they do receive prior approval from authorities. Insurers are also given guidelines for securing particular systems and human resources for the valuation of capital charges. Since it is extremely difficult to build internal models for EMDE infrastructure financing products, however, insurers may rely heavily on public credit ratings or turn

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Table 14.7 Fixed income instruments’ capital charges and return on equity (ROE)

Asset Type

Indicative capital Indicative capital charge (standard Annual ROE charge (standard Annual ROE (with matching formula without (without matching formula with matching liability) liability) matching liability) liability)

A-rated corporate bond BBB-rated corporate bond Infrastructure bond (A/BBB-rated) Infrastructure loan (unrated)

10.5%

2.87%

6.3%

2.98%

20.0%

2.83%

15.0%

2.96%

10.5%

3.18%

6.3%

3.30%

23.5%

3.17%

17.63%

3.32%

Source: Ernst & Young (2015).

down potential investments in projects without a public rating. As such, public ratings will be important for bond structures in order to attract insurers interested in longterm fixed income products without high capital charges.

14.3.4 BASEL III Similar significant changes are impacting the banking industry. Since the GFC, regulators have put in place reforms to strengthen banking resilience and financial stability. There has been good progress achieved in three key areas: capital requirements to manage a variety of risks; rules to promote funding stability; and rules to address potential liquidity mismatches. First, Basel III imposes stricter capital requirements and larger buffers to build a cushion against losses. Capital requirements have been enhanced for specific exposures, including market risk, complex securitization, and counterparty credit risk, all of which were major sources of losses during the GFC. Basel III rules require banks to hold additional capital, compared to Basel II, for these same risks. Traditionally, banks were obliged to hold 8 per cent of capital, with a minimum common equity ratio of only 2 per cent. Under the new framework, Common Equity Tier 1 (CET1) has to attain at least 4.5 per cent of risk-weighted assets. The minimum of Tier 1 capital is set at 6 per cent. Similar to Basel II, the sum of Tier 1 and Tier 2 capital must be at least 8 per cent (see Figure 14.4). Thibeault and Wambeke (2014) explained that the capital conservation buffer is a new instrument intended to safeguard banks. It consists of common equity of 2.5 per cent in risk-weighted assets, bringing the total common equity ratio to 7 per cent.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -    2.5% 2.5% 2.0%

4.0%

1.5%

2.0%

4.5%

2.0% BASEL II

Countercyclical buffer Capital conservation buffer Tier 2 Additional Tier1

Common equity

BASEL III

 . Capital ratios: Basel II versus Basel III

When a bank falls into this buffer range, supervisors can intervene gradually by restricting discretionary distributions (Thibeault and Wambeke 2014). The countercyclical buffer is a common equity tranche, ranging between 0 and 2.5 per cent, imposed by national supervisory authorities during periods of excessive credit growth and perceptions of increased systematic risk. Basel III also introduces more stringent requirements regarding the quality of capital instruments that compose the different tiers. Innovative step-up hybrid capital, previously included under Tier 1, will now be phased out. The classification of Tier 2 capital into higher and lower Tier 2 capital has been removed in order to simplify the framework. Tier 3 capital, consisting of short-term subordinated debt, is now eliminated. This capital tranche was previously only available to cover market risks (Visser and McEneany, 2015). Second, rules have been introduced to strengthen banks’ liquidity risk profiles and promote funding stability. In fact, new liquidity requirements are one of the most important changes introduced by Basel III. Capital requirements alone proved to be insufficient in preventing the past financial crisis. Basel II did not include any specific liquidity measures, despite the fact that liquidity risk can have a substantial impact during periods of financial distress. Therefore, two new ratios were adopted in Basel III: the liquidity coverage ratio (LCR), which is for short-term stress, and the net stable funding ratio (NSFR), which requires banks to map illiquid assets to long-term, stable funding sources. The LCR requires banks to hold a sufficient amount of high-quality, liquid assets (HQLA), which can be used to offset net cash outflows under a 30-day stress scenario. The LCR, the ratio of the stock of high-quality liquid assets (HQLA) to net cash outflows over a 30-day period, must meet or exceed 1.0. In other words, HQLA  100%: Net cash outflows over a 30-day period The numerator of the LCR includes unencumbered high-quality liquid assets (HQLA), which consists of assets that can be converted into cash at little or no loss of value during times of stress. Table 14.8 describes the main assets considered HQLA and the factor of the total amount of the respective assets that can be included as HQLA.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Table 14.8 High quality liquid assets (HQLA) Category

Factor

Level 1 assets: coins, bank notes, central bank reserves, securities from sovereigns, central banks, public sector entities, and multilateral development banks, sovereign, and central bank debt with a 0% risk weighting

100%

Level 2A assets: sovereign, central bank, multilateral development bank or public sector entity assets with a 20% risk weighting; corporate debt securities rated AA- or higher, covered bonds rated AA- or higher

85%

Level 2B assets: Residential mortgage-backed securities (RMBS), corporate debt securities rated between A+ and BBB-, common equity shares

50–75%

Finally, the new banking regulation addresses the ‘too-big-to-fail’ problem via the net stable funding ratio (NSFR). The NSFR is a longer-term structural ratio designed to address liquidity mismatches. The NSFR is set for more medium- and long-term funding for banks and aims to limit over-reliance on wholesale funding during periods of readily available market liquidity. Basel III imposed the requirement that the ratio of available stable funding to required stable funding meets or exceeds 1.0 In other words, Available stable funding  100%: Required stable funding The available stable funding (ASF) is determined by weighting the liabilities of the bank by the so-called ‘ASF Factor’.

14.3.4.1 Impact of BASEL III Under BASEL III, banks cannot decide how to allocate assets based only on regulatory standards. The impacts of Basel III will increase capital costs of lending activities and also affect liquidity requirements, which will, in turn, encourage banks to de-leverage and lower allocation weights on long-term investments. New liquidity requirements— most notably the LCR and NSFR—will increase costs through different channels. Banks will be required to increase allocation to high quality liquid assets (e.g. government bonds or high rated corporate bonds), or increase allocation to assets with a low required stable funding factor (i.e. high rated assets or assets with a residual maturity of less than one year). This will inevitably lower the yield of banks’ asset bases. The changes will also lead to increased funding from liabilities with low outflow assumptions (in order to optimize the LCR) or high available stable funding factors (in order to optimize the NSFR). Such liabilities include capital instruments, retail deposits, or generally, liabilities with residual maturities of one year or more. These are inevitably more expensive compared to short-term wholesale funding. Additionally, banks facing a shortfall in capital or stable funding will have to take significant mitigating steps in order to comply with Basel III requirements. These shifts

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

will lead to reduced lending capacity, deleveraging, and disintermediation. In combination, these trends will heavily burden banks and could limit their participation in long-term infrastructure financing.

14.3.5 Implications of Changes of Regulation and Accounting Standards The current low-interest-rate environment is putting a great burden on long-term liability-embedded institutions to meet their target rates of return. The concurrent changes to regulation and accounting standards increases pressure on financial institutions to control risks, manage portfolios with matching liabilities, and increase capital requirements for safety. More specifically, investors are challenged to (i) value assets and liabilities with fair value; (ii) enhance their capital structures with consideration of risk levels; and (iii) manage assets with due consideration of liabilities. These factors are expected to push financial institutions and governments to de-risk their portfolios and project support activities. Unlike commercial banks, insurance firms and pension funds must additionally match their portfolios with long-term liabilities. Thus, products developed specifically for these types of investors, which are also structured in such a way as not to increase the capital charge burden (e.g. a bond scheme with a public credit rating), hold promise with respect to mobilizing international long-term liability-driven assets to EMDE infrastructure. Among many conditions for enhancing bankability with regulation changes, credit ratings are most important as a pricing reference, as a tool to lower capital charges, and as a precondition of investment credit rating for internal investment policy.

14.4 T  EMDE I F

.................................................................................................................................. The main objective of this chapter is to identify potential institutions and countries that can mobilize funds effectively and efficiently to the emerging infrastructure market. Identifying appropriate institutions and countries requires the assessment of their resources, interests, and limitations, with respect to the global infrastructure financing market, and in light of the trends and changes described herein.

14.4.1 Target Institutions Finding suitable institutions to target, in order to successfully mobilize private funds to the emerging infrastructure market, depends on two criteria: (i) sufficient long-term liabilities with a sizeable asset base, and (ii) suitable risk appetites for EMDE projects.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

  PENSION/ INSURER

BANK 100% 90%

Others 26%

Others 22%

Others 22%

Composite Liability 18%

Equity/ Property 6%

Bonds 12%

Equity/ Property 19%

P&C Liability 14%

Loan 68% (short and medium term)

Deposits 66%

Bonds 64%

Life Liability 68%

80% 70% 60% 50% 40% 30% 20% 10% 0%

ASSET



Deposits 3%

LIABILITY

Short term (Asset & Liability)

ASSET

LIABILITY

Long term (Asset & Liability)

 . Insurer and bank balance sheets in comparison Source: ECB, Bank of England, Oliver Wyman analysis.

To the first, institutions must be able to make long-term investments from the early stages of a project until its closing in order to avoid the refinancing risks that present a major challenge to new projects. Thus, target institutions should possess sizeable asset bases and long-term liabilities. In addition, we must examine how institutions’ portfolios are constructed and whether the characteristics of their assets fit with infra finance (see Figure 14.5). Moreover, target institutions will have a higher-than-average risk appetite that spurs interest in new EMDE projects. To avoid refinancing risk—the main challenge for emerging market infrastructure projects—it is advisable to find institutions that can make long-term investments from the early project stages through to project conclusion, or at the very least until projects have been appropriately de-risked. These are typically institutions with long-term liabilities and sizeable asset bases. Target institutions will also have higher than average risk appetites and, preferably, a mandate for emerging market projects. Table 14.9 and Figure 14.5 summarize the typical asset bases, investment horizons, risk appetites, and investment objectives of key financial institutions. Life insurance, private and public pension funds, sovereign wealth funds, and endowments and foundations comprise the set of target institutions for which project periods may be best matched to investment periods. Considering investment horizons and risk appetite, it is advisable to first target sizeable life insurance and public pension institutions, and then extend product outreach to sovereign wealth funds, endowments, and foundations. With respect to how portfolios are constructed (Figure 14.5), pensions and insurance assets have higher proportions of long-term investments with capital markets—a good source of infrastructure financing.

Institution Commercial bank Infrastructure developer Non-life insurance Infrastructure & Public Employees Federation (PEF) Investment company Life insurance & private pension Public pension Sovereign wealth fund Endowment & Foundation

Assets under management (US$ Investment trillion) horizon

Risk appetite

Investment objective

40.2

ST

Low to medium Make net interest margins

3.4

ST

High

N/A

ST

Medium

Risks and constraints

Depends on fund High characteristics

ALM mismatch risk Intensifying regulatory environment: BASEL III Participate as project Limited capital to invest with participants long-term horizon Meet their liability funding ALM mismatch risk Intensifying regulatory cost calculated by actuaries environment: IFRS II, Solvency II Maximize beneficiary’s wealth Liquidity issue due to beneficiary redemption

29.0

ST to MT

Maximize company return

26.5

LT

10.9

LT

6.3 1.0

LT LT

2.7

Notes: ST: Short term, MT: Medium term, LT: Long term

Depends on funds’ mandates Medium

Meet their liability funding cost calculated by actuaries Medium Meet their liability funding cost calculated by actuaries Medium to high Maximize sovereign’s wealth High Maximize beneficiary’s wealth

Liquidity issue due to beneficiary redemption ALM mismatch risk I ntensified regulatory: IFRS II, Solvency II ALM mismatch risk Longevity risk is getting higher Government mandate approval issue Can have mandates which do not allow EMDEs investment

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

Table 14.9 Summary characteristics of relevant financial institutions

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



14.4.2 Target Countries Four criteria are set forth to identify target financing-source countries whose institutions would be more apt to demonstrate interest in EMDEs’ infrastructure investment. These are likeliest to be countries with (i) ageing societies; (ii) well-designed national pension systems and sovereign wealth funds; (iii) large life insurance companies; and (iv) low domestic interest rates. The first three criteria identify markets with significant and growing demand for long-term, fixed-income investments, while the fourth shows the attractiveness of infrastructure investments’ expected rate of return, compared to opportunities within the local investment environment, which currently account for the bulk of major investments in their portfolios. Based on the data provided in Tables 14.10 and 14.11, Japan, South Korea, Taiwan, Singapore, and some Chinese states are expected to have a growing demand for the long-term, fixed income projects. Table 14.11 also shows that Japan, Norway, South Korea, China, Singapore, Malaysia, and European countries have sizeable pension assets to invest in long-term projects. Table 14.12 shows that, with respect to the size of sovereign wealth funds, the USA; Asian countries, such as China, Hong Kong, South Korea, and Singapore; and countries with large oil reserves, such as the United Arab Emirates, Saudi Arabia, Qatar, and Russia, have accumulated sizeable sovereign wealth fund volumes. With respect to life insurance, the USA, East Asian countries including Japan and Korea, and European countries, such as the UK, Germany, and France, have large life insurance industries that may be suitable sources of long-term project investments (see Table 14.13). Assuming a project yield of 5.0 per cent per year, we can also determine the spread on local government bond yields, which reflects the attractiveness of projects (see Table 14.14).

Table 14.10 Largest percentage point change of the population aged 60 years or over Rank

Country of area

1 2 3 4 5 6 7 8 9 10

USA, Virgin Island Japan Malta Finland South Korea Aruba Martinique China, HK SAR China, Taiwan Curacao

Percentage change between 2000–15 10.9 9.9 9.3 7.3 7.2 7.0 6.9 6.9 6.7 6.6

Country of area

Percentage change between 2015–30

Cuba South Korea China, HK SAR China, Taiwan Curacao China, Macao SAR Thailand Martinique Brunei Darussalam Singapore

Source: United Nations (2015). World Population Prospects: The 2015 Revision.

12.8 12.7 12.3 12.1 11.7 11.4 11.2 11.0 11.0 9.9

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Table 14.11 Size of sovereign pension fund (in US$ billion) Rank

Fund

Countries

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

Government pension investment Government pension fund National pension National social security Canadian pension Central provident fund Employee provident fund GEPF Future fund Employees provident National wealth fund Labor pension fund Public Institute for Social Security Fondo de reserva security FRR

Japan Norway South Korea China Canada Singapore Malaysia South Africa Australia India Russia Taiwan Kuwait Spain France

Total asset 1,143.8 884.0 429.7 247.4 228.4 207.8 184.7 123.2 89.2 80.7 75.3 64.8 58.9 50.4 45.0

Table 14.12 Size of sovereign wealth fund (in US$ billion) Rank

Fund

Countries

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

CIC/SAFE/NCSSF/CADF ADIA/ADIC/EIC/ICD/IPIC/MDC/RIA GPF PIF/SAMA KIA GIC/TH HKMA QIA SKJSC/KNF/NIC RNWF/RRF/RDIF APF/NMSIC/PWMTF/PSF/PUF/ATF/NDLF/LEQTF/CSF/WVFF AFF/WAFF KIC LIA NDFI

China UAE Norway Saudi Arabia Kuwait Singapore Hong Kong Qatar Kazakhstan Russia United States Australia South Korea Libya Iran

Total asset 1,462 1,247 847 758 592 538 442 256 164 152 142 95 92 66 62

Of course, the investment attractiveness of projects will be evaluated by each projects’ risk profile, which includes country risk, construction risk, operation risk, and revenue risk, and the risk appetites of investors. That said, by examining spreads between local government bond yields and expected project rates of return, we can generally see which countries will demonstrate higher demand for infrastructure investment opportunities. Considering EMDE infrastructure project characteristics,

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 

Table 14.13 Size of life insurance industry (US$ billion) Rank

Countries

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15

United States Japan United Kingdom Germany Korea France Netherland Sweden Denmark Switzerland Ireland Australia Spain Norway Luxemburg

Total asset 3,711.0 3,329.4 2,108.7 1,272.4 566.1 551.1 534.4 461.2 432.7 365.2 246.8 234.4 184.4 174.4 174.4

Source: OECD Global Insurance Statistics.

Table 14.14 Yield spread, compared to domestic 10-year bond rates Category

Country

10-year Government bond yield

Spread

America

USA Canada Brazil Mexico

1.58% 1.12% 11.76% 5.97%

342bp 388bp 676bp 97bp

Europe

Germany UK France Netherland Sweden Norway Italy Switzerland

0.017% 0.83% 0.22% 0.10% 0.15% 1.00% 1.24% 0.57%

502bp 417bp 478bp 490bp 485bp 400bp 376bp 557bp

Asia Pacific

Japan Australia Hong Kong Singapore South Korea India China

0.24% 1.92% 0.90% 1.76% 1.41% 7.27% 2.82%

524bp 308bp 410bp 324bp 359bp 227bp 218bp

Source: Bloomberg.



OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

such as long-term credit risk and operational risk, countries with more than 300bp spread are the most promising potential investors. In conclusion, based on the four criteria outlined above, the most promising target countries by region (and in no order of preference) include: • North America: the USA and Canada; • Europe: the UK, Germany, France, Switzerland, Netherlands, Sweden; and • Asia Pacific: Japan, South Korea, Taiwan, and Singapore. Among these countries, Japan, South Korea, Taiwan, and Singapore are first-tier targets, since the issue of ageing in these countries is most urgent. Moreover, they have demonstrated stable growth in their pension funds, life insurance company assets, and sovereign wealth funds, which have also expressed urgency to support their countries’ ageing populations.

14.4.3 Promoting Infrastructure Projects as an Asset Class 14.4.3.1 Asset Allocation and Asset Class Concepts in Portfolio Management In modern portfolio fund management, asset allocation plays a pivotal role in establishing the systematic risk exposure an investor wants. A classic and frequently cited empirical study is Brinson et al. (1986). These authors interpreted the importance of asset allocation as the fraction of the variation in returns over time attributable to asset allocation, based on regression analysis. Brinson et al. concluded that asset allocation explained an average 93.6 per cent of the variation of returns over time for ninety-one large US-defined benefit pension plans from 1974 to 1983.⁴ An updated study found the average percentage variation explained was 91.5 per cent for US plans for the period 1977 to 1987 (Brinson et al. 1991). Blake et al. (1999) investigated asset allocation in the UK. Examining more than 300 medium-sized to large, actively managed UK-defined benefit pension schemes for the period 1986–94, they concluded that asset allocation accounted for approximately 99.5 per cent of the variation in plan total returns. These studies’ results concerning the relative importance of strategic asset allocation are aligned with pension funds’ typical investment emphasis. Thus, research and historical evidence show that financial institutions build portfolios based on strategic asset allocation, which is often the most important investment decision institutional investors make, since it determines most of the risks and returns. Since asset allocation is based on asset classes, it is important to build an emerging infrastructure project asset class for institutional investors to consider when portfolio building. Because infrastructure debt has not been widely considered a main asset class by investors, due to unfamiliar structures, risks associated with illiquidity, and difficulties in due diligence, we turn our attention to how infrastructure might be promoted in the global financial market. ⁴ The range was 75.5 per cent to 98.6 per cent.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



14.4.3.2 Increasing the Attractiveness of Infrastructure as an ‘Asset Class’ Building interest in infrastructure debt depends on highlighting the inherent benefits of infrastructure investment, namely, generally stable ratings and low credit losses, as well as boosting the profile and attractiveness of infrastructure debt within portfolio management. According Moody’s (2015) research, based on the agency’s rated infrastructure over the period from 1983 to 2015, infrastructure debts are less likely to incur losses than non-financial corporate issues, especially over longer investment horizons. On average, a total infrastructure debt security loses 0.3 per cent of its face value over a 5-year horizon and 0.4 per cent of its face value over a 10-year horizon, compared to 6.0 per cent and 8.9 per cent, respectively, for a typical non-financial corporate (NFC) debt. Infrastructure ratings are also more stable than NFC ratings, driven in large part by the US municipal infrastructure sector. On average, ratings on total infrastructure securities have been only 40 per cent as volatile as NFC ratings. The corresponding figures for ratings on US municipal infrastructure securities and corporate infrastructure and project finance securities are 19 per cent and 84 per cent, respectively (Moody’s 2015). Credit loss rates for single A- and Baa-rated corporate infrastructure and project finance debt securities are similar to NFC loss rates up through a 5-year horizon but diverge quickly thereafter (see Figure 14.6). Single A- and Baa-rated corporate infrastructure and project finance debt securities, which together comprise 75 per cent of corporate infrastructure and project finance ratings, have higher default rates but lower losses, given default (LGD) than like-rated NFC issuers. With respect to increasing demand for infrastructure debt in portfolio management, it is germane to note the important diversification effect that the inclusion of infrastructure debt can have on a portfolio. Given the observed correlations between the price of unlisted infrastructure and other asset classes (Figure 14.7), allocations to the infrastructure asset class can help investors manage risks in their portfolio via diversification. Moreover, like real estate, infrastructure is a real asset class, capable of generating cash flow indexed to inflation. Compared to real estate, however, the quasimonopolistic nature of infrastructure assets and the imposition of regulation can protect returns from market risk and volatility, limiting exposure to the economic cycle. This feature also differentiates infrastructure from other asset classes. Lastly, unlisted infrastructure has demonstrated strong performance with higher Sharpe ratios, a measure for calculating risk-adjusted returns, which can enhance overall portfolio returns based on risk level (Figure 14.8). As shown in Figure 14.9, annualized returns based on 1-year, 3-year, and 5-year infrastructure assets have demonstrated stable and outstanding performance. Overall, for those portfolio managers with real liabilities, the current market environment is challenging. Many bond yields, particularly inflation-linked yields, are at historic lows and the impact of this is to drive expected returns down and future liabilities up. Furthermore, the increasing correlation across many listed markets has made it difficult to achieve real returns. This is where infrastructure debt could be a

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

0.9% 0.8% 0.7% 0.6% 0.5% 0.4% 0.3% 0.2% 0.1% 0.0% Year 1

Year 2

Year 3

Year 4

Year 5

Year 6

Year 7

Year 8

Year 9

Year 10

Corporate infrastucture and project finance debt securities Non-financial corporate issuers

 . Credit losses for single A- and Baa-rated infrastructure debt and NFC Source: Moody’s Infrastructure Default and Recovery Rates, 1983–2015 Research report.

Correlation with unlisted infrastructure listed equity

listed infrastructure

global bond –0.3

–0.25

–0.2

–0.15

–0.1

–0.05

0

 . Correlations between select asset classes with unlisted infrastructure Source: Deutsche Asset & Wealth Management (2015).

4.0 3.5 3.0 2.5 2.0 1.5 1.0 0.5 0.0 Unlisted infrastructure

Listed equities

Global bonds

Listed infrastructure

 . Sharpe ratio comparison, 5-year results for the quarter to December 2014 Source: Deutsche Asset & Wealth Management (2015).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Annualised return (%)

20% 18% 16% 14% 12% 10% 8% 6% 4% 2% 0%

1Y Unlisted infrastructure

3Y Listed equities

5Y Global bonds

Listed infrastructure

 . Performance comparison, results for the quarter to December 2014 (annualized return) Source: Deutsche Asset & Wealth Management (2015).

productive asset class for investors. It gives pension schemes the opportunity to invest in non-correlated real assets with an attractive risk return profile. In addition to benefiting from enhanced diversification, investors can also use infrastructure to help match their liability profiles with a reasonably predictable and partly inflation-linked distribution stream. The infrastructure debt opportunity provides investors with the ability to exchange lower yielding, long-term government bond investments for infrastructure debt investments with similar ratings and long-term, stable cash flow characteristics, but higher yields.

14.4.3.3 Infrastructure Investment in a New Environment of Liability-Driven Investment As discussed in Chapter 13, institutional investors are placing more emphasis on asset allocation rather than security selection and market timing. Moreover, insurers, defined pension plans, and other institutional investors face streams of significant future liabilities to which assets must be appropriately matched. For pension funds, in particular, significant amounts of wealth are growing over long time horizons. It is, therefore, important that pension funds manage their assets to ensure that future liabilities can be met. Controlling the risk related to funding future liabilities is a key investment objective for such investors, who frequently take an asset–liability management approach to strategic asset allocation. To add further impetus, recent changes to regulation and accounting standards, such as IFRS and Solvency II, are accelerating the adoption of liability modelling practices and optimal asset allocation in relationship to funding liabilities. In this context, liability-driven investing (LDI) strategies are increasingly considered an integral part of asset management for pension funds and life insurers. ‘Liabilitydriven investment’ is a framework for understanding the nature of liabilities and

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

adopting an investment strategy with an overall objective linked to these liabilities. For example, in the Korean pension and life insurance sectors, many firms are setting up new teams or task forces to introduce LDI technologies. As LDI techniques are adopted by asset managers, long-term, fixed-income investment instruments are becoming more important, particularly since the liabilities of pensions and insurers are extremely long-term and undergoing a trend of expanded duration.

14.4.4 Facilitating the Emergence of the Infrastructure Asset Class To mobilize sizeable pension and life insurance assets for the EMDE infrastructure market, infrastructure project investments must be promoted as a major asset class (Blanc-Brude 2013). Because long-term infrastructure project investments have highly attractive characteristics, with high Sharpe ratios and long-term stable cash flows, it is important to engineer and introduce products that align with investor strategies, account for existing regulatory challenges, and are easily tradable. To create an easily tradable product, the products should have investment grading and clear, easily understandable terms of structure. For instance, mortgage-backed securities (MBS) are easily understood as a type of asset-backed security, secured by a mortgage or collection of mortgages with prepayment risk. Facilitating this shift requires the attainment of several goals. First, a pilot product with attractive terms should be developed and marketed to first-tier target institutions in identified target funding source markets to generate interest. In the absence of established market recognition, this will require the support of governments and development institutions, such as the World Bank. Once products are tested and refined, and once demand is established, these products should be extended to second-tier institutions and target countries. Last, once markets clearly understand the products, multilateral and government support should be gradually decreased. The final goal is to build EMDE as an asset class traded in both primary and secondary markets.

14.4.4.1 Drawing on Past Experience in Product Initiation Some past World Bank experiences with the design and initiation of innovative financial products bear lessons that apply to the proposed effort to introduce new products to mobilize international sources of finance into the infrastructure sector. Here, we summarize some key lessons from the introduction of a World Bank guarantee loan product and a convertible bond with project product. First, we draw on a World Bank guarantee loan product introduced in 2016 to inject finance into a Brownfield concession in South Asia. The experience with this product, which did not ultimately succeed in garnering sufficient investor interest, nevertheless revealed two important lessons. First, while the

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



Cross currency swap Sponsor Government (9) Milestone and availability payments

(3) WB pays PV UYU and receives PV in US$

(6) Equity (4) With [100% - PV] WB buys ‘project guarantee’ (7) Debt

SPV (8) Construction security package

Recipient commercial bank

World Bank (5) RCB sells option ‘project guarantee’ if NO completion as per the terms agreed at T0 + LC from standby bank

(2) 100% UYU bond proceeds

Pension funds (1) At Year 0, WB issues a 6 year nointerest paying bond (in hard currency) with a project conversion at maturity

 . Structure of Uruguay convertible bond with project funding

World Bank was willing to provide a 95 per cent guarantee, investors were not comfortable with purchasing a product without a public credit rating due to regulatory concerns.⁵ Second, most investors lacked the capacity to perform a sufficiently rigorous feasibility analysis of the project itself. As such, they revealed a preference for projects benefiting from confirmed government support, such as a termination fee or minimum revenue guarantee, in order to protect their principal. A second set of lessons may be drawn from experiences with a convertible bond issued to fund a transportation project in Uruguay. The design of the project involved the World Bank issuing a 5-year, no-interest paying bond to Uruguayan pension funds, with key terms and conditions similar to a World Bank bond (see Steps 1 and 2 in Figure 14.10). Thereafter, proceeds would be used to fund a transport infrastructure special purpose vehicle (SPV) (Steps 3 to 7 in Figure 14.10). The proceeds of the no-interest paying bond issued by the World Bank (WB) would be retained to ensure minimum payment obligations to the institutional investors after a 5-year period, and would be used by the Regional Commercial Bank(s) (RBC). The RCB would blend this money with its own resources to on-lend to the SPV at a more competitive cost. Through the project guarantee, the RCB would commit to repay the WB at a designated time. The repayment amount would differ if technical completion were achieved.

⁵ EMDE infrastructure loans without credit ration would incur higher charges due to BASEL and Solvency II.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

The flows can be summarized as follows, assuming 6 per cent interest rate in Uruguayan Peso (UYU) and a 5-year term: • World Bank receives UYU 100 million from pension funds through the bond; • World Bank buys the put option from the RCB. The premium to be paid equals the face value of the bond, minus the present value of UYU US$100 million (considering a discount factor in Year 5 of 6 per cent, UYU US$100 million— UYU US$75 million = US$25 million). Figure 14.11 summarizes the process intended after construction and ramp-up of the infrastructure project. The SPV would be tasked with constructing and rehabilitating roads against preset criteria and standards set by the Government. The SPV’s financing structure and construction security package would include equity from the project sponsor (UYU US$50M) and debt from the RCB (UYU US$100M). The SPV would issue a sponsor support agreement to the RCB against construction cost overruns and delays and would provide, as part of the security package, an on-demand and unconditional performance bond, to be issued by the EPC contractor. Government, on the other hand, would be obliged to make payments against prescribed construction milestones and availability payments, thus given revenue certainty to the project. This product, too, was not able to generate interest from targeted investors for several reasons. In addition to the regulatory ratings and due diligence issues outlined in the previous example, it was discovered that there was a limited market for currency cross swaps (CCS) for that particular emerging market currency. Moreover, it was recognized that refinancing in the middle of projects are difficult, and the complicated product structure deterred potential investors. Few local or regional commercial banks

Cross currency swap Sponsor Government

Equity

Milestone and availability payments

Contingent payments

(15) If NO Completion, SPV repays debt or penalties

SPV r Se

Customers

(10) Cross currency swap extinguishes 100% US$

e

vic

(12) At Year 6, SPV issues a 15 year project bond via RCB&WB to pension funds (Technical completion) (18) Pension funds may decide to refinance SPV using UYV proceeds

Recipient commercial bank

Pension funds

(16) RBC pays “project guarantee”

(13) Transfer of 15 year project bond

(14) Transfer of project bonds (11) 100% UYU (17) 'project guarantee'

 . Process after project construction and ramp-up

100% UYU

World Bank

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



(RCBs) were willing to assume the risks of the project with the selling put option, and sponsors recognized that they would still bear refinancing risks after the 5-year zero coupon bond repayment. Lastly, Uruguay pension funds were not interested, since local interest rates were high (central key rate of 9.25 per cent).

14.5 P  I F P

.................................................................................................................................. This section proposes some structures to be used to engage both public and private sectors in developing infrastructure. The suggestions presented here have been designed with consideration of the aforementioned financial market and regulatory factors as well as lessons from past experience. These proposals are intended as a starting point and will undoubtedly need to be modified and improved with inputs from relevant actors in the market, reflecting current financial market conditions and market demand.

14.5.1 Product Proposal #1: Asset-backed Security for PPPThis proposed asset-backed security (ABS) product is designed to lessen EMDE country fiscal burden and, thus, expand national capacity to offer support to additional PPPs. The main concept is to convert operating facilities without concession agreements (or other PPP type contractual agreements) to projects with specified contractual regimes such as concession agreements. At the core of the structure, is to provide operating and other rights such as provision of service, construction, etc. as set out in a PPP contract, to a special purpose vehicle (SPV). This SPVs are established in the form of companies to acquire and develop infrastructure assets and associated services. These rights could then be sold to long-term investors to generate additional financial resources for new infrastructure financing.

14.5.1.1 Profile of Target Projects and Private Financing Institutions and Expected Benefits a. Existing non-PPP infrastructure facility with high demand. b. Target investors: pension fund, life insurance, and sovereign wealth fund. c. Listing on major global market to facilitate trading in secondary market. By investing in project ABS, investors could secure long-term investments with profitable operating assets. In addition, investors could choose tranches based on their risk appetites. Government could procure additional sources to lesson fiscal burdens (Figure 14.12).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -    Rating agency Moody’s, S&P, Fitch

Government Cash

Infrastructure operation right#1

Operational right

Insurance

MIGA

Concession contract (termination fee, MRG etc.)

Purchase cash

Principal & interest

Infrastructure operation right#2

Special

Infrastructure operation right#3

Purpose Co Profit

Rating assessment Tranche AAA Tranche AA Tranche A

Excess spread

Equity Equity investment

Trustee

GIF

 . Proposed structure of existing infrastructure ABS

Expected benefits are summarized below: • Government: Lessening the fiscal burden and therefore increasing resources for other infrastructure developments. • Institutional investors: New players with long-term liabilities, such as the pension funds or life insurance firms, will be able to secure a long-term, fixed income project with credit enhancement from IFIs such as the World Bank Group or the African Development Bank. • By investing in projects that benefit from this structure, investors can secure longterm investments with a profitable operating asset, allowing investors to choose tranches based on their respective risk appetite.

14.5.2 Product Proposal #2: Infrastructure CLO Product (I-CLO) The main concept is to pool a variety of non-investment grade infrastructure loans that benefit from a public credit rating through a colaterized debt obligation. Each infrastructure loan would consist of different tranches and those tranches would be sold to investors with different risk appetites. IFIs could participate as equity investors or in arranging other equity investors to facilitate financing (Figure 14.13).

14.5.2.1 Profile of Target Projects and Private Financing Institutions and Expected Benefits a. Sound and economically viable infrastructure loans. b. Target investors: pension fund, life insurance, and sovereign wealth fund.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

  Asset management corporation Manage pool of loan portfolio Diversified infrastructure loan portfolio Loan Portfolio A (IBRD/IDA) . Loan Portfolio B (IFC) . . Infrastructure loan Portfolio C (commercial banking loan)

Swap bank

World Bank (IBRD/IDA)

Cross currency swap (Local currency/US$)

Principal & interest payment by seniority Special

Purchase cash

Rating agency Moody’s, S&P, Fitch Rating assessment

Guarantee of principal and interest (optional)

Cash flow



Class A Notes/AAA Class B Notes/AA

purpose Co. Remained cash flow

Class C Notes/A Equity Equity investment

Trustee

GIF

 . Proposed structure of infrastructure CLO

c. Available institutional investors base with a mandate to invest in infrastructure. d. Listing on major global market to facilitate trading. Expected benefits are summarized for key stakeholders below: • Institutional investors: New players with long-term liabilities, such as pension funds or life insurance firms, would be able to secure long-term, fixed-income project debt in accordance with each investor’s appetite for public credit rating. • Developers: The main benefit for a project sponsor would be in building a source of capital from local and international pension funds and life insurance firms. • IFIs: The use of a CLO could be effectively used as a way to recycle financial assets of their balance sheets thus allowing a more efficient use of IFI capital base.

14.5.3 Product Proposal #3: Project Puttable Bond The main idea behind this structure is to finance long-term investment throughout the life of the project minimizing refinancing risk for the project developers and to bring to market a bond that has been appropriately de-risked and thus is a suitable investment for long-term institutional investors. In this structure the SPV formed by the developers (and its associated lenders) would issue a project bond which would benefit from an IFI put option. To make the structure bankable, the bond should be investment graded. For example, suppose an institutional investor with an appetite for long-term investments buys a 20-year project bond (assuming 3 years construction, 2 years ramp-up, and 15 years operation). The bond’s principal and interest would be paid by the SPV as a result of the revenues it derives from the PPP contract (say through a government

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -    Contracting authority

World Bank IBRD Put option premium

Concession contract

Guarantee by selling put option

Purchase bond Sponsor A

Special purpose Co

Equity

EPC contract

EPC

Issue bond

Debt investors

O&M contract

O&M

 . Proposed structure of puttable project bond

milestone payment during construction and an availability payment during operation). The project bond would benefit from a put option against pre-defined specific minimum conditions (triggers), such as successful construction completion, minimum coverage ratios, and minimum credit rating conditions. The purchasing of the bond by the IFI would not necessarily have to be fully done by the IFI, but it could be partial, encouraging local and international institutional investors to co-invest with the IFI. Gradually, as the project develops, the IFI could sell its position to exisiting or other institutional investors (Figure 14.14).

14.5.3.1 Profile of Target Projects and Private Financing Institutions and Expected Benefits Sound and economically viable infrastructure projects, which are aligned with IFI’s country assistance strategy: a. Credit enhancement by an IFI put option and as appropriate other forms of project guarantees such as MIGA insurance. b. Target investors: pension funds, life insurance, and sovereign wealth funds. c. Listing on major global market to facilitate trading. Expected benefits are summarized for key stakeholders below: • Government: Main benefit to governments is promoting international institutional long-term investors’ funds to domestic infrastructure market. • Developers: Main benefit is building up a pipeline to international pension funds and life insurance firms to the domestic infrastructure market without concern about refinancing. Through this structure a sponsor/developer can finance long term, with lower rates than traditional financing. In addition, the puttable bond

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



structure is familiar to investors allowing portfolio managers to trade such bonds in the secondary market. • Institutional investors: New players with long-term liabilities (such as pension funds or life insurance firms) would be able to fulfil their mandate of investing in investment-grade, long-term, fixed income infrastructure without taking the risks of construction and early stage operational risks. It would also be meaningful for these institutions to advertise their organizations as supporting emerging countries’ development.

14.5.4 Product Proposal #4: MIGA Non-Honouring of Sovereign Financial Obligation Bond The main concept is to finance long-term investment funds using MIGA’s NonHonouring of Sovereign Financial Obligations programme. Part of a guaranteed bond could be issued benefiting from a AAA rating, equivalent to MIGA’s credit rating, and the remaining part would be issued with a low rating or without a rating. We further propose that MIGA expands its modality of engaging with EMDE governments on such guarantees to a programmatic basis. Currently, MIGA gets involved in a project-specific discussion on using its guarantees at the request of an investor, developer, or government. MIGA’s underwriting is done only at the project level. This modality is highly inefficient and does not provide the market and investors with the predictability they want while they plan for their investments. If MIGA engages with a country on a programmatic basis, and pre-underwrites this pipeline, based on considerations of sector economics, social and environment Contracting authority

World Bank IBRD Put option premium

Concession contract

Sponsor A

Purchase bond

Special purpose Co

Equity

Guarantee

MIGA covered Tranche (AAA) Purchase bond

EPC contract EPC

O&M contract

Uncovered Tranche (No rating)

O&M

 . Proposed structure of MIGA non-honouring of sovereign financial obligation bond

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

factors, and risk analysis, then MIGA would be in a position to issue a ‘covered tranche’ for the future pipeline of infrastructure projects for a country. Investors would have visibility of pricing early on and could make a choice of whether to participate in the ‘covered tranche’ or finance projects without such guarantees if they deem risk as acceptable. The main benefits of such a programmatic approach on the part of MIGA are: (i) it would allow MIGA to maximize its impact from its limited balance sheet and human resources; (ii) it would bring forward significant discussions on projects upstream and allow EMDE governments to prioritize and to decide whether specific projects should fall under the ‘MIGA-covered tranche’ of a multi-year PPP programme; (iii) it provides the market predictability and offers investors a choice to go covered or uncovered, based on MIGA’s early programmatic underwriting and corresponding pricing decisions for the guarantees (Figure 14.15).

14.5.4.1 Target Projects and Private Financing Institutions a. Sound and economically viable infrastructure projects which are aligned with World Bank and MIGA country assistance strategy. b. Credit enhancement by MIGA guarantee programme. c. Target investors: pension fund, life insurance, and sovereign wealth fund. d. Listing on major global market to facilitate trading.

14.5.5 Key Challenges and Areas for Further Development The structures that have been presented still require greater analysis. Naturally, their development needs to be linked to a specific project and must match the current needs of the infrastructure sector, its national characteristics, and the demands of international investors, particularly given the current financial and regulatory environments. Some of these are summarized below and are intended to act as a starting point for the further development of the four different structures proposed. • Need for robust legal and regulatory structures: the different proposals made all require an enabling environment that is conducive for such structures. • Need for careful calibration of social benefits derived from infrastructure: Undoubtedly, infrastructure development has an impact on society and as such critical issues such as affordability, ability to pay, keeping the public informed, etc. become critical for the success of such schemes. Education, public awareness, and articulation of benefits will be necessary accompaniments for the success of such schemes. • Project vs National Credit Ratings: In some instances, the role of IFIs to provide credit wraps or credit enhancements will be necessary so as to achieve a credit rating higher than the local government rating which in some instances may not be investment grade.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

 



• Linking to credit rating agencies: There is a need to actively engage with credit agencies to develop specific methodologies that appropriately rate the proposed structures. Their active participation to help design and to offer early opinions on proposed structres will be critical in developing sound and replicable models. • Appropriate capacity of international and local bank teams: Highly skilled international and local bank management teams are needed to manage the proposed structures. For example, for the CLO product, each infrastructure loan in a portfolio will involve numerous covenant issues, which will need to be evaluated and their implications understood. Furthermore, project design teams will need to start building statistically meaningful portfolio structures that effectively consider risks and industry characteristics. • Project pipeline: Appropriate and effective deal sourcing continues to be a recurring challenge and there needs to be a concerted global effort to look at developing projects across different countries and regions. Furthermore, within these efforts there should be special attention to ensuring that lessons learnt within different locations and regions are not lost and are transposed to other legislations, thus avoiding the re-invention or re-interpretation of infra related project risks. • Appropriate regulatory incentives: The proposed structures require the active transfer of ownership between, for example, banks that may have good performing loans to institutional investors. Appropriate incentives will need to be devised at both national and international levels to create a market where assets are appropriately allocated and recycled. Addressing these challenges is a worthwhile endeavour, particularly given that current resources are insufficient to address the growing infrastructure funding deficit, despite the fact that investors are actually interested in identifying opportunities to invest in the infrastructure sector. This situation belies a mismatch of currently available infra finance products and the demands of international investors. The proposed products attempt to deal directly with the current financial environment and recent regulatory impacts in order to promote infrastructure as an emergent and viable asset class for institutional portfolio building, with the aim of benefiting developing economies and their potential investors alike.

R APMG PPP Guide. 2016. 12.1 International Public Sector Accounting Standards (IPSAS) Number 32. Available at: https://ppp-certification.com/ppp-certification-guide/about-ppp-guide Bielenberg, A., M. Kerlin, J. Oppenheim, and M. Roberts, 2016. ‘Financing changes: how to mobilize private-sector financing for sustainable infrastructure’. McKinsey Center for Business and Environment, January, McKinsey & Co. Available at: http://www. indiaenvironmentportal.org.in/files/file/Financing_change_How_to_mobilize_privatesector_financing_for_sustainable-_infrastructure.pdf

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 -   

Blake, D., B. Lehmann, and A. Timmermann, 1999. ‘Asset Allocation Dynamics and Pension Fund Performance’, Journal of Business, 72 (4), pp. 429–61. Blanc-Brude, F., 2013. ‘From Asset Allocation to Infrastructure Investment: A Roadmap for the Development of Institutional Investment in Infrastructure’. Presentation to NATIXIS European Infrastructure Day, EDHEC-Risk Institute, Paris, 17 October. Brinson, G., L. R. Hood, and G. Beebower, 1986. ‘Determinants of Portfolio Performance’, Financial Analysts Journal, 42 (4), pp. 39–44. Brinson, G., B. Singer, and G. Beebower, 1991. ‘Determinants of Portfolio Performance II: An Update’, Financial Analysts Journal, 47 (3), pp. 40–8. CGFS (Committee on the Global Financial System), 2011. ‘Fixed Income Strategies of Insurance Companies and Pension Funds’. CGFS Paper No. 44, July 2011. Committee on the Global Financial System, Bank for International Settlements, Basel, Switzerland. Coughlan, A., D. Wright, and T. Silverman, 2015. ‘Just When You thought You Could Relax, It Looks Like IFRS4 Phase II Will Happen’. Presentation of the Financial Reporting Group, Institute and Faculty of Actuaries, 20 November. Available at: https://www.actu aries.org.uk/documents/f7-just-when-you-though-you-could-relax-it-looks-ifrs-4-phase2-will-happen Deutsche Asset & Wealth Management, 2015. ‘Why Invest in Infrastructure?’ Research report, May. Available at: http://infrastructure.deutscheam.com/content/_media/Research_ Deutsche_AWM_Why_Invest_in_Infrastructure_May_2015.pdf Ernst and Young. 2014. ‘IASB IFRS 4 and IAS 39: Insurance and Investment Contracts’. Ernst and Young. 2015. ‘Infrastructure Investments: An Attractive Option to Help Deliver a Prosperous and Sustainable Economy’. Available at: http://www.ey.com/Publication/ vwLUAssets/EY-infrastructure-investments-for-insurers/$FILE/EY-infrastructure-in vestments-for-insurers.pdf Global Infrastructure Hub, 2015. ‘Infrastructure Outlook, Global Infrastructure Hub and Oxford Economics’. Moody’s, 2015. ‘Infrastructure Default and Recovery Rates’. Available at: http://www.moodys. com Ruiz-Nunez, F., and Zichao Wei, 2015. ‘Infrastructure Investment Demands in Emerging Markets and Developing Economies’. Policy Research Working Paper No. 7414, World Bank Group, Washington, DC. Thibeault, A. and M. Wambeke, 2014. ‘Regulatory Impact on Banks’ and Insurers’ Investments’. Vlerick Centre for Financial Services, Vlerick Business School. UN (United Nations), 2015. World Population Prospects: The 2015 Revision, Department of Economic and Social Affairs, Population Division, New York: United Nations. Visser, E. and D. McEneany, 2015. ‘IFRS 4 Phase II Comparison with Solvency II and MCEV’. Milliman Briefing Note, Milliman, Inc. Available at: http://www.milliman.com/ uploadedFiles/insight/2015/ifrs-4-phase-2-comparison.pdf?lng=1048578

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  .............................................................................................................

E M P I R I C S OF STRUCTURAL CHANGE .............................................................................................................

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

   ......................................................................................................................

´     

15.1 I

.................................................................................................................................. S change is the foundation for sustained economic growth; only rarely has a country evolved from a low- to a high-income status without sustained structural transformation from an agrarian or resource-based economy towards an industry- or services-based economy (Monga 2013). Yet, while industrialization is incredibly powerful at lifting people out of poverty, it can also come at a significant environmental or social cost. The fact that the industrial sector emits almost 30 per cent of greenhouse gasses worldwide is but one example. Over the past decades, recognition has grown that structural transformation can no longer be pursued without considering its distributional, social, and environmental impact. This is exemplified in the almost universal acceptance of the Millennium Development Goals and their successor, the Sustainable Development Goals (SDGs). Increased attention to the idea of sustainable and inclusive growth has raised the question of how to measure and monitor such a multidimensional concept. This should necessarily start with the monitoring of individual goals in a way that is comparable across countries and continents. While most of the economic indicators are well established, there is far less consensus on how to measure the environmental and social aspects (Jerven 2014). This lack of progress was reflected in the call for a ‘comprehensive program of action on data’ made by UN Secretary-General Ban Ki-moon in December 2014—the final year of the MDGs. While this call has not gone unheeded, estimates of the cost of tracking the 196 targets of the SGDs still vastly exceed the available funding (Jerven 2014). As a result, both the availability and the quality of data will remain a concern for the foreseeable future.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

However, the focus of this chapter lies not on the measurement of the individual indicators of structural change, but on the question of how to subsequently summarize the wide set of indicators (both in terms of their number and in the breadth of topics) into an index of structural transformation. The usefulness of creating such an index vastly exceeds the simple ranking of countries they are typically reduced to. A well-constructed aggregate helps to distil the overall state of progress from a vast collection of indicators, summarizing them into something that can be more easily interpreted. Indexes point analysts towards countries whose score does not conform to expectations as well as help the policy makers to identify the best practices to emulate. Such a reduction in the dimension of the data set is, however, not without risk. A badly constructed index can lead to the wrong conclusions and policy recommendations. Moreover, many indexes are presented in a way that fails to accurately portray the underlying uncertainty of the information they are aggregating, creating a false sense of confidence that can be equally misleading. Researchers who work predominantly with qualitative analyses often question the primacy of indexes and similar types of data-driven quantitative studies. Economic historians in particular are critical of the quality of many of the data sources used. It is hard to dispute that the reliability of data drastically changes depending on the quality, independence, and neutrality of the statistical institution collecting them. Nevertheless, data sources are never impeccable and in order to make large cross-country comparisons one is forced to use imperfect data. Analyses using these data should therefore not be dismissed, but neither should this criticism be ignored as a utopian dream. While indexes allow one to quickly compare a large number of countries and years, they remain imperfect summaries. Users should be aware of their limitations and when possible complement them with qualitative analyses. This chapter sets out to highlight the specific problems that arise when composing an index of structural change and development, and suggests (improvements in the) ways to address them. For more detailed instructions on the construction of a composite index we refer to the OECD handbook on this topic (OECD 2008). Specifically, we discuss the issues that arise when composing both policy and outcome indexes. The former captures the existence of a policy environment that is conducive to structural transformation, while the latter measures the progress made towards an inclusive and sustainable economy with a modern manufacturing and services sector. To that end, we will refer to a number of existing indexes that track (aspects of ) structural change as examples, but this is by no means an exhaustive list. Some of the indexes we discuss measure the outcome, like the Quality of Growth Index (Mlachila et al. 2014) or the Economic Complexity Index (Hausmann et al. 2011). Examples of policy indexes include the Competitiveness Score of Delgado et al. (2012) and the Worldwide Governance Indicators (Kaufman et al. 2010). We will also refer to indexes that mix both policy and outcome such as the World Economic Forum’s Global Competitiveness Index.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



15.1.1 The Steps to Constructing an Index Broadly speaking, there are four steps to creating an index, and these are outlined in Figure 15.1. The first step is to precisely define what the index is trying to measure. This definition is key, as it will determine all subsequent steps and serve as the yardstick by which the construction of the index should be judged. It is important in this first step to make a distinction between indicators that track outcomes, and those that track the policy environment, as mixing both would cloud the interpretation of the final index results and might cause endogeneity problems in subsequent analyses (OECD 2008). You cannot correctly test whether having a high policy score leads to an improvement in outcome, if that policy index also includes indicators that track outcome and vice versa. Ravallion (2010) also argues that in order to create an objective index, its definition and its construction need to be determined by a theoretical model. However, we will argue that this objectivity is to a large extent illusionary. The second step is to identify suitable indicators that track (parts of ) the definition decided upon in step one. The OECD handbook discusses the various characteristics these indicators should have. The foremost concern with (multidimensional) indicators of structural change is, however, the quality and availability of the data. In order to avoid reducing the data set to only a handful of countries, the imputation of missing data is generally necessary. Yet, most indexes remain silent on how exactly this is done, how many observations were affected, let alone on how the final results are influenced by this crucial step. Overall, the way in which missing data are addressed needs to be accounted for with much more transparency, ideally by using the same method or rule for all indicators. Its effect on the final index value subsequently has to be determined and clearly communicated. In the third step, the individual indicators are normalized and aggregated into the final index. However, in the case of multidimensional indicators of structural change, 1. Define index 2. Collect indicators

…

Multiple imputation of missing values

3. Compose index

…

Normalize indicators Aggregate indicators

4. Analyse and report results Sensitivity analysis Reliability of the results

 . The four steps to constructing a composite index

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

the lack of an overall accepted theoretical framework hinders the construction of an objective index. For example, there is no standard way of weighting environmental degradation against an increasing manufacturing sector, let alone an answer to the question of whether one is allowed to compensate for the other. Most outcome indexes address this by assigning equal weights to each objective, but this is no more or less objective than other weighting schemes. In the absence of a theoretically motivated weighting scheme, a simple solution is to present the index along with its constituent parts. In this way, users can immediately see how the different scores on each component are combined into the final index score. A related issue that is discussed in this chapter concerns the way in which the structural characteristics of high-income countries are treated relative to those of lowincome countries. New Structural Economics tells us that the feasible and desired structure of an economy crucially depends on the level of development (Lin 2012). Nevertheless, most indexes work on an absolute scale that treats all countries in the same way. This is not a problem when the goal is to get a high-level overview of the state of development, but not when measuring specific characteristics of countries. To enable users to draw the right conclusion from the index, it might be necessary to incorporate differentiation in terms of level of development into the construction of the index. The final step is to analyse the index and report on the results. The key concern is to gauge the sensitivity of the index to the different modelling choices, including the way in which missing values were dealt with and the impact of (not) including outlying observations. These and other quality controls will ensure that unexpected patterns in the data are not due to conceptual or accidental errors in the production process and help get a clear insight into how reliable the results are. The presentation of the index needs to accurately represent this reliability in an intuitive way that even those without a strong statistical background can quickly comprehend and work with. The remainder of this chapter consists of two parts. Section 15.2 surveys the traditional approaches to measuring structural transformations in general terms. Section 15.3 then discusses more specific issues like the introduction of New Structural Economics and is focused on practical solutions to these problems. Section 15.4 concludes.

15.2 T A   S

..................................................................................................................................

15.2.1 To theorize or not to Theorize In his 2010 paper, Ravallion draws a distinction between theory-based and ‘mash-up’ indexes of development. As the name suggests, the former category consists of indexes

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



whose construction is based on (economic) theory, like the Gross Domestic Product. In contrast, the construction of the latter is at the discretion of the analyst, who first chooses the individual indicators that capture various aspects of an often-unobserved overall concept, like human development, governance, or structural transformation. Those indicators are then normalized and aggregated frequently without the guidance of theory or other scientific literature. Examples of these ‘mash-up’ indexes include the Worldwide Governance Indicators, the Human Development Index and the Global Competitiveness Index (Ravallion 2010). While Ravallion’s distinction seems like an elegant way of categorizing the different indexes of development and structural transformation, it is to a large extent an artificial one. First, the contention that GDP is a model index that is legitimated by its theoretical underpinnings, is invalidated by the large body of academic research highlighting its many shortcomings. In fact, many of the ‘mash-up’ indicators such as the Gender and Human Development Indexes have arisen to compensate for its shortcomings. More generally, the expectation of developing an index of structural transformation that passes a strict theoretical test is unrealistic. Both the theoretical model and the regressions used to test it are host to numerous assumptions and while there exist statistical procedures for testing some of these assumptions, the tests often cannot detect substantial failures. Furthermore, as pointed out by Freedman (2010), model testing may become circular, whereby breakdowns in assumptions are detected and the model is then redefined to accommodate these. There are cases where it might be necessary to restrict the analysis of structural transformation to singular aspects like GDP/capita or the size of manufacturing and services in GDP, for example when analysing structural transformation over long periods of time. Nevertheless, the consensus both in and outside of academia is that structural transformation cannot be studied in these narrow terms, but should encompass environmental, distributional, and social concerns as well. This multidimensionality is reflected in the Sustainable Development Goals that, in addition to the development of a modern economy, also look at the strength of environmental protection and of gender equality. Unfortunately, there does not exist a single theoretical framework that is broad enough to encompass all of these concerns while at the same time being specific enough to identify which indicators should be included or how the different dimensions should be weighted and aggregated. That is not to say that there should not be a clear and systematic way of choosing the indicators of structural transformation. This system will depend on the precise definition of structural transformation that the index is trying to measure and the extent to which it should be informed by existing theory and literature (OECD 2008). Lacking a formal multidimensional theoretical framework as a base, the system employed to choose or reject the indicators should be clearly communicated and applied indiscriminately in order to minimize the spectre of subjectivity.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

15.3 A N A  M S T

..................................................................................................................................

15.3.1 Dealing with Missing Data Once we move beyond the indexes that capture the structural change at a general level and start analysing its specific components, the lack of comparable qualitative data on a worldwide scale quickly becomes apparent. Even when the data exist, the more intricate indicators are often only available with a one- or two-year delay, making it hard to get an overview of current progress. While estimates of economic growth are often produced on a quarterly basis, other hard-to-measure concepts or survey data might only be available every couple of years. As a result, tracking structural change over a long time-span or for a large group of countries inevitably requires finding a way of dealing with missing data. Because of the size of this problem, simply ignoring differences in availability between countries and computing the index with whatever data are available is not a feasible option. Instead, the most often used approach is to limit the number of countries and years that are covered (i.e. case deletion). However, the availability of data is positively linked to the level of development of the countries, meaning that the countries left out of the index are those for which insight into the process of structural transformation is most essential. At the same time, the analysis is often restricted to only those variables that have high availability, which runs the risk of creating an index that measures what is available, rather than what is intended. The trade-off between the availability and relevancy of the indicators should be undertaken carefully and with a transparent and clearly communicated rule that is applied to all data sources. However, even when the set of countries, years, and indicators is limited in this way, the availability problem of indicators of structural transformation is such that any data set that aims to cover a significant fraction of countries will inevitably have gaps. While the way in which the missing data are treated will have a drastic impact on the final index or analysis, many of the existing indicators remain relatively vague on this subject. The data section might mention that some observations were imputed, but without mentioning how this was done, or the number of variables and observations for which this was necessary. Most importantly, the effect on the final index value and its reliability is often unknown, or not made clear. Multiple imputation is a straightforward way in which missing data can be dealt with that also allows a precise determination of how it has affected the reliability of the resulting index. The idea is that the gaps in the data set are filled with a whole range of possible values, giving us multiple complete data sets. This can be seen in step 2 of Figure 15.1: the initial data set (the grey circle) is complemented with different possible values for each missing observations, resulting in many possible version of the dataset (the black circles). The index can subsequently be computed for each of the imputed

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



data sets, culminating in hundreds of index values (the small black circles in the third step). The index that is reported is typically the average of all imputed values, and the variance over all imputations gives a clear indication of how reliable this estimate is. The fewer data points have to be filled in and the smaller the range of possible values, the smaller this variance will be. To that end, the final step in Figure 15.1 brings together the different index values computed in the previous step, resulting in one final index and its confidence interval. There are different ways in which the data can be imputed. Single imputation often uses the time-pattern to fill in gaps: for example by computing the average of the preceding and successive observations or even simply by repeating the preceding observation. Alternatively, Multiple Imputation using Chained Equations (MICE) uses the information from the variables that are available (Van Buuren and Groothuis-Oudhoorn 2011).¹ The optimal procedure will depend on the characteristics of the data. For example, because their data set had both an important time and crosssectional dimension, Lin et al. (2018) augmented the MICE approach by using a statespace model to impute the most likely missing values. This allowed them to combine the information from both dimensions, thereby significantly decreasing the uncertainty of their imputed values.

15.3.2 Composing the index 15.3.2.1 Normalization Before the individual indicators can be aggregated, their values first have to be made comparable. In other words, each of the indicators has to be transformed such that they express what is a good versus a bad score. As the OECD handbook (2008: 27–9) gives an overview of many different ways in which this can be done, we will limit this section to a few complementary remarks. Our first comment is to point out an important downside of using discrete methods of normalization, namely that it creates discontinuities in the index value. Take the example of cut-off points that distinguish poor performance from intermediate and good performance. While choosing the correct cut-off has the potential to add a lot of information to the index, it also creates these discontinuities. Small changes in an indicator will have a completely different effect depending on how close the initial value was to the cut-off point. Similarly, minute differences between countries can result in diverging index values, simply because one country managed to just exceed the cut-off point. In addition, the use of cut-off points significantly increases the risk of ¹ Often the multiple imputation algorithm can be significantly improved by including the information from variables that are not used to construct the index, but that are good predictors for the variables whose observations are missing. The former are typically very similar to the latter, but that were not selected because they are constructed using a different methodology or because they are only available for a select group of countries.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

rank-seeking behaviour, where countries only seek to improve their performance until it just exceeds the cut-off value (Ravallion 2010). Rankings are often used to normalize the indicators, probably because they offer a very clear interpretation: the best performing country has a score of one, the secondbest score has rank two, etc.² However, using rankings to normalize indicators suffers from the same problems outlined above: they can enlarge small differences between countries whose performance is essentially the same. When computed over many years, this creates a highly unstable index as meaningless changes in the underlying indicators keep perturbing the index values (Høyland et al. 2012). Alternatively, the (empirical) Conditional Density Function (CDF) is a continuous way of normalizing the indicators that avoids this instability problem, while keeping the clarity of interpretation. A CDF expresses the probability that you find a similar or lower value in the sample. This means that the top/worst performing country gets a score of one/zero, while any score in between can be interpreted as the fraction of countries that perform worse. Unlike standardization, empirical CDF estimators can handle both continuous and discrete data (e.g. binary indicators) and unlike rankings it has no problem with countries having the same value (Henderson and Parmeter 2015).

15.3.2.2 Aggregation After normalization, the indicators are combined into the final index value. However, as noted earlier, the lack of an overall accepted formal (theoretical) framework means there is a ‘right’ of aggregating multidimensional indicators of development and structural transformation. When constructing a policy index, you can regress the indicators on a relevant outcome variable and use the estimated coefficients as weights. For example, Delgado et al. (2012) regress output per potential worker on three indexes capturing the micro and macro policy environment as well as the social infrastructure and political institutions. The (standardized) coefficients are then used as weights in order to combine the sub-indexes into their overall index of competitiveness. An obvious caveat when using this approach is that these weights ultimately depend on the appropriateness of the outcome variable (in this case the outcome per potential worker) and cannot be used when there is more than one outcome variable. Moreover, the parameters that are estimated should not be confused with exact values, and their uncertainty needs to be taken into account in the computation of the weights and presentation of the index. Unlike their policy counterparts, outcome indexes have no clear answer to the question of how to weigh, for example, environmental concerns relative to the size of the manufacturing sector. While social choice theory formally rules out an ideal solution to such a multi-criteria problem, it does offer tools to help make a trade-off between the different goals (see e.g. Giard and Roy 1985). Nevertheless, many outcome indexes simply assign equal weights to each indicator (e.g. the Human Development ² Unlike for example standardization (i.e. setting the mean to zero and standard deviation to one), which do not have a pre-determined maximum or minimum score.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



Index), or category of indicators (e.g. the Quality of Growth Index constructed by Mlachila et al. 2014). However, the simplest weighting scheme is no more or less objective than any of the other possible weighting schemes. In addition, the way in which the indicators are normalized will still change the index values and alter the implied trade-offs between the different components in the final index score. In addition to the choice of weighting scheme, the aggregation step also requires that a choice be made regarding the substitutability of the different goals. For example, can a country with a low gender equality score compensate with a high score on the environmental or manufacturing component? And should this country receive a similar score to a country that has an intermediate outcome on all components? If substitutability is zero, the index is reduced to the minimum score over all components, which undermines the idea of a multidimensional index. Yet, if neither of the extremes is chosen there are an infinite number of values that could be assigned to the substitutability parameter. Geometric aggregation³ is a straightforward option in-between these two extremes, and is, for example, used as a robustness check in the Quality of Growth Index (Mlachila et al. 2014). That being said, most indexes end up imposing perfect substitutability. In lieu of an objective way of summarizing a multidimensional set of indicators, the simplest way (perfect substitutability and equal weights) is not without its merits. The most straightforward way to compensate for the naivety of this approach is to subsequently report the index in conjunction with its constituent indicators. For example, in the presentation of the index composed by Kroll (2015) that compares the readiness of developed countries for the SDGs, the author mainly focuses on performance on the different components. To start, he summarizes his findings using a coloured matrix that shows each country’s score on the thirty-four components. The index itself is presented along with a radar graph that shows exactly how each score came about. While this does not resolve the idiosyncratic nature of the weighting scheme, it gives users a clear idea of the implied trade-offs in the final index value, and allows them to redraw the conclusion to better reflect those aspects they are focused on.

15.3.3 Insights from New Structural Economics One of the key insights from New Structural Economics is that a country’s best strategy for development depends on its level of development. The sectors that best enable a country to grow are contingent on its comparative advantage, and the latter evolves over time and will change as per capita income increases (Lin 2012). At the outset, it is not feasible for relatively poor countries like Ethiopia or Afghanistan to have the same environmental, institutional, and economic characteristics as rich countries such as ³ Assuming equal weights, the geometric average of n indicators is

p ffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi n x1 x2 ⋯xn :

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

Japan or Switzerland. More importantly, it is also not in their best interest to emulate these characteristics, as their comparative advantage and optimal development strategy will lie elsewhere. Figure 15.2 illustrates this using the Economic Complexity Index (ECI) of Hausmann et al. (2011), which captures the knowledge embedded in a country’s population as expressed in their industrial composition. Using a scatterplot in panel a and by plotting the joint density in panel b, these graphs link the ECI to the level of development. The latter is captured by the log of the Gross National Income per capita (GNIpc). It clearly shows that as the level of development increases, the range of ECI values shifts upwards. What New Structural Economics argues is that this shift is not just due to a better score of the developed countries, but also reflects differences in the feasibility and the desirability of a knowledge-intensive manufacturing sector. (a) Scatterplot

(b) Joint density

2 0.10

1

0.06 0.04 0.02 0.00

ty ensi

−1

6

−2

2 1

6

8

GNIpc

10

c Ip

−3 12

0

10

–1 12

(c) CDF per development level

EC I

GN

8

–2

(d) Conditional CDF

0.6

0.8

ion but stri l Di ona

diti Con

1.0

0.8

ion but stri l Di ona

diti Con

1.0

0.6 0.4 0.2

–2

6 8 10

–1

c Ip

4.0

2

GN

t en m lop ve

–1

0

1

EC I

De

0.4 0.2 0.0 1.0 1.5 2.0 2.5 3.0 3.5

12

0

1

2

EC I

ECI

tD Join

0.08

0

–2

 . Economic Complexity Index and GNI per capita Notes: Panel a shows the scatterplot of the Economic Complexity Index (ECI) and the level of development as measured by the log of GNI per capita, and panel b shows their joint density. Panel c shows the Cumulative Density Function of the ECI computed for each level of development from low-income (1) to high-income (4), while panel d shows the conditional CDF of ECI given the level of development (the log of GNI/cap).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



Most indexes of structural transformation and development, on the other hand, compare countries on an absolute scale, regardless of their level of development. Obviously, this is not a problem when the aim of the index is to provide an overall assessment of the level of development using only one or a select group of general indicators. Think, for example, of the GNI per capita or the Human Development Index that focus respectively on one or three key aspects of human development. However, in the case of indexes that focus on specific characteristics (such as the development of the financial industry) and especially those that combine these specific characteristics in an overall index (such as the World Economic Forum’s Global Competitiveness Index), the underlying assumption is that the desirability of these characteristics of countries does not depend on the level of development. Experienced analysts might be able to take the level of development into account when using such an index with an absolute scale. However, this is a lot to ask of the average user, as it requires her to know which countries have a similar level of development and take their performance into account when assessing the score of any one country. Instead of constructing the index on an absolute scale and relying on the individual users to correct for the level of development, we can also adjust the construction of the index itself to ensure that this is automatically taken into account. A straightforward way in which this is sometimes accomplished is by dividing countries into groups with similar levels of development (or other relevant economic characteristics such as being landlocked, or the abundance of natural resources) and comparing the structural characteristics within these groups. Kroll (2015), for example, only looks at the characteristics of the high-income countries when assessing their readiness for the SGDs. However, when the scope of the index is expanded to cover multiple groups over a longer period of time, using these fixed groups creates discontinuities. This is illustrated in panel c of Figure 15.2, which shows the CDF transformation of the economic complexity index for low-income (1), lower- and upper middle-income (2 and 3) and high-income (4) countries. Within each income category, the transformed variable is smooth, but the CDF function jumps between the different categories. As a result, a country that improves its level of development might see its score decreased, simply because the set of countries it is compared to has changed. While this problem can be avoided by keeping the countries in each group fixed, this means that the relevance of the set of ‘comparable’ countries decreases over time. Moreover, it wrongly implies that the level of development cannot be changed. A solution to this problem is to use a continuous way of taking the level of development into account: all observations are taken into account, but countries that are closer in terms of level of development receive a higher weight. To that end, Lin et al. (2018) use a conditional Cumulative Distribution Function to normalize the individual indicators before aggregating them in their index of Inclusive and Sustainable Transformation. The conditional CDF indicates the probability of finding a country with a similar level of development that scores worse on that component. A CDF score of 0 (1) means that the country performs worse (better) than all other countries with a similar level of development. The final panel of Figure 15.2 shows what

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

this looks like in the case of the ECI. Overall, this graph mirrors that of panel c in which fixed groups were used, but without the discontinuities. Instead, the group of countries that one is compared to slowly changes as the level of development increases, allowing us to better track each country’s progress over time.

15.3.4 Computing and Presenting the Reliability of the Index More often than not, index values are presented as facts and treated similarly to an observed truth. However, as has been shown throughout this chapter, index values are always uncertain, regardless of the way in which they were constructed. The sources of uncertainty are plethoric, starting with measurement errors in the underlying indicators and differences in the reliability in the data of some countries. The construction of the index is another important source of uncertainty, particularly the way in which missing data are treated. Therefore, the final stage of constructing an index should be to test the sensitivity of the results to the various modelling choices and accurately determine the reliability of the results. A first step in correctly presenting the index results is to consistently include the estimated standard deviation and confidence intervals of each index value. While more and more indicators of development are presented in this way, the subsequent analyses often completely ignore the fact that the index was constructed. For example, changes in the ranking of countries are reported on without taking the (un)reliability into account, but rankings can be particularly misleading when used on indexes that measure absolute progress. Small changes can lead to completely different rankings when scores lie close together, regardless of whether the change was significant or not (Høyland et al. 2012). However, there are many ways in which index results can be presented that give an appropriate sense of the uncertainty of the result and avoid misleading the user. Kaufman et al. (2010) suggest a simple rule of thumb when comparing their Worldwide Governance Indicators, which is to see whether the 90 per cent confidence intervals overlap. Høyland et al. (2012) compute the 95 per cent confidence interval of rankings, while Standaert (2015) groups together countries for which the difference in the index value is not statistically significant. Other possibilities are to build an online platform where users can query the significance of differences between countries or changes over time. In addition to these straightforward examinations that compare the index values over time or countries, many indexes are also used in regression analyses. Policy indexes like the Doing Business index or the Worldwide Governance Indicators, for example, are used in countless estimations as response, explanatory, or control variables. However, these studies often ignore the uncertainty of these index values and, as Desbordes and Koop (2015) show, this can significantly bias the result. This uncertainty can easily be accounted for by using multiple imputation or Bayesian Gibbs sampling. The only requirement is a data set containing a few hundred sampled values

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  



of the index, which is precisely what is generated when the missing values are addressed using multiple imputation (Desbordes and Koop 2015). Solt (2016), for example, publishes the Standardized World Income Inequality database in a ready-made format that allows the users to completely take uncertainty into account. While these methods require additional time and effort to compute, they ensure that even those without a strong statistical background can use and interpret the index correctly.

15.4 C

.................................................................................................................................. This chapter discussed specific issues and questions that arise when amalgamating the different indicators of structural transformation into a single index. To enhance the objectivity of such a multidimensional index in the absence of a theoretical model, the importance of singular rules that are indiscriminately applied when selecting, imputing, normalizing, and finally aggregating the different indicators cannot be overstated. The impact of these modelling decisions on the final index should then be tracked and openly communicated, particularly the way in which missing data are treated. This chapter also brings to the fore a central tenet from New Structural Economics stating that the desired structural characteristics of countries are determined by their comparative advantage, which in turn depends on their level of development. While this is no problem for those indicators that measure the overall level of development, the same does not hold for indexes that summarize specific structural characteristics of countries. Rather than relying on individual users to compare the performance of countries given their level of development, we argue that this conditionality should be embedded in the construction of these indexes of structural transformation.

R Buuren, S. van and Karin Groothuis-Oudshoorn, 2010. ‘Mice: Multivariate imputation by chained equations in R’, Journal of Statistical Software, pp. 1–68. Delgado, Mercedes, Christian Ketels, Michael E. Porter, and Scott Stern, 2012. ‘The Determinants of National Competitiveness’. National Bureau of Economic Research Working Paper No. 18249. Desbordes, Rodolphe and Gary Koop, 2015. ‘Should We Care About the Uncertainty Around Measures of Political-Economic Development?’, Journal of Comparative Economics, 44 (3), pp. 752–63. Freedman, David A., 2010. Statistical Models and Causal Inference: A Dialogue with the Social Sciences, Cambridge: Cambridge University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´     

Giard, Vincent (ed.), and Bernard Roy, 1985. Méthodologie multicritère d’aide à la décision, Paris: Editions Economica. Hausmann, Ricardo, César A. Hidalgo, Sebastián Bustos, Michele Coscia, Alexander Simoes, and Muhammed A. Yildirim, 2011. Atlas of Economic Complexity: Mapping Paths to Prosperity, Cambridge, MA: Harvard Center for International Development. Henderson, Daniel J., and Christopher F. Parmeter, 2015. Applied Nonparametric Econometrics, Cambridge: Cambridge University Press. Høyland, Bjørn, Karl Moene, and Fredrik Willumsen, 2012. ‘The tyranny of international index rankings’, Journal of Development Economics, 97 (1), pp. 1–14. Jerven, Morten, 2014. ‘Benefits and Costs of the Data for Development Targets for the Post2015 Development Agenda’. Data for Development Assessment Paper Working Paper, Copenhagen: Copenhagen Consensus Center. Kaufman, Daniel, Aart Kraay, and Massimo Mastruzzi, 2010. ‘The Worldwide Governance Indicators: A Summary of Methodology, Data and Analytical Issues’. World Bank Policy Research Working Paper No. 5430. Kroll, Christian, 2015. Sustainable Development Goals: Are the Rich Countries Ready? Gütersloh: Bertelsmann Foundation. Lin, Justin Yifu, 2012. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: World Bank. Lin, Justin Yifu, Célestin Monga, and Samuel Standaert, 2018. ‘The Inclusive Sustainable Transformation Index’, Social Indicators Research. Available at: http://doi.org/10.1007/ s11205-018-1977-1 Mlachila, Montfort, René Tapsoba, and Sampawende Tapsoba, 2014. A Quality of Growth Index for Developing Countries: A Proposal, Washington, DC: International Monetary Fund. Monga, Célestin, 2013. ‘Winning the Jackpot: Jobs Dividends in a Multipolar World’, in Joseph E. Stiglitz, Justin Yifu Lin, and Ebrahim Patel, eds, The Industrial Policy Revolution II—Africa in the 21st Century, New York: Palgrave Macmillan, pp. 135–71. OECD and Joint Research Centre-European Commission, 2008. Handbook on Constructing Composite Indicators: Methodology and User Guide, Paris: OECD Publishing. Ravallion, Martin, 2010. ‘Mashup Indices of Development’. Policy Research Working Paper No. 5532, Washington, DC: World Bank. Solt, Frederick, 2016. ‘The Standardized World Income Inequality Database’, Social Science Quarterly, 97 (5), pp. 1267–81. Standaert, Samuel, 2015. ‘Divining the Level of Corruption: A Bayesian State-Space Approach’, Journal of Comparative Economics, 43 (3), pp. 782–803.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

        ......................................................................................................................

    ......................................................................................................................

 .    . 

16.1 I

.................................................................................................................................. O the past half-century, while the world’s population more than doubled, the quantity and real value of agricultural output more than trebled, even though land in agriculture increased by only about one-tenth (Alston and Pardey 2014). This remarkable accomplishment belied prophecies of doom in the early 1960s, when many pundits foresaw nothing but trouble in the world food equation, and hopeless hunger for generations to come in much of the world. These dark Malthusian prophecies reflected a perspective based on traditional agriculture and conventional inputs; they did not anticipate the transformation of agriculture that was to come. In Transforming Traditional Agriculture T.W. Schultz (1964) envisioned a crucial role for investments in ‘non-traditional’ inputs such as knowledge and education, and improvements in the quality of material inputs and people, to help shift agriculture to a firmer footing and capitalize on agriculture as an engine of economic growth. He began the book: The man who farms as his forefathers did cannot produce much food no matter how rich the land or how hard he works. The farmer who has access to and knows how to use what science knows about soils, plants, animals, and machines can produce an abundance of food though the land be poor. Nor need he work nearly so hard and long. He can produce so much that his brothers and some of his neighbors will move to town to earn their living. Enough farm products can be produced without them. (Schultz 1964: 3)

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

And he ended it: The knowledge that makes the transformation possible is a form of capital, which requires investment—investment not only in material inputs in which this knowledge is embedded but importantly also in people. (Schultz 1964: 206)

Schultz emphasized the importance of incentives, and no doubt he was also conscious of the importance of infrastructure and institutions, including well-functioning markets and a peaceful, law-abiding society. At the time of that writing the world was a much different place, and much of the substantial transformation of agriculture worldwide that has taken place since then can be attributed to the types of investments in knowledge and people Schultz envisioned. In 1961, of the world’s 3.0 billion people an estimated 770.8 million (25.1 per cent) were directly engaged in farming, and agriculture represented 13.4 per cent of overall income (measured in terms of GDP). In the half century since, the world’s total population has increased by a factor of 2.4, to a total of 7.4 billion, and agriculture’s share of the global economy has shrunk. In 2014, of the world’s 7.4 billion people, an estimated 1.3 billion (18 per cent) were directly engaged in farming, but agriculture represented just 3.9 per cent of overall income (World Bank 2017). Still, in the middle- and low-income countries, where most of the world’s farmers are to be found, agriculture accounts for a much greater share of national income and employment. A great many more people still depend on agriculture for their livelihoods and nearly half of the world’s population still lives in rural areas, mostly in agriculturally based households (Alston and Pardey 2014). Many are subsistence farmers, operating very small farms using very little in the way of marketable inputs other than the land they farm and their own family labour. A majority of the world’s poor can be found among these rural residents—indeed, for this reason in his Nobel Prize Lecture Schultz (1979) famously said ‘ . . . if we knew the economics of agriculture, we would know much of the economics of being poor’. Many countries have barely begun the transition process that others have undergone extensively since Transforming Traditional Agriculture was first published. In this chapter, using data largely at the level of countries, we take a detailed look at the changes in the structure of agricultural production around the world over those five and a half decades. The patterns of change have been uneven. Where and how the world’s food is produced today is very different from where and how it was produced over fifty years ago. The changes have also been systematic. The high-income countries like the USA represent a declining share of global agricultural output, while middleincome countries are becoming dominant, and the poorest of the poor countries continue to struggle. We explore the past productivity patterns and their sources as drivers of this transformation, and raise concerns about future prospects given various shifts now taking place in the natural and political climates facing agricultural

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



producers and the public and policy institutions that serve them—particularly as they pertain to agricultural science, technology, and innovation.¹

16.2 A   B E

.................................................................................................................................. The transformation of worldwide agriculture over the past half century entailed changes in the structure of agriculture—the input and output mixes and the nature of farms and farming—and changes in the role of agriculture in the economy, while the transformation of agriculture itself contributed to broader economic growth and poverty reduction. As per capita income rises in a country, the agricultural share of GDP typically falls. In the USA, for example, agriculture’s share of GDP was around 60 per cent in 1865 but in just fifty years had fallen to about 15 per cent of GDP (Carter et al. 2006). In Figure 16.1, in 1960 the high-income countries as a group were well below this 15 per cent threshold while the LAC regional average was just below it (14.7 per cent). Every other regional average was far above this threshold but, with one exception trending towards it and successively passing it in 1992 (EE&FSU), 1973 (MENA), and 2004 (A&P); one region (SSA) remains stubbornly above the 15 per cent line. Alternatively, using country-specific data, in 1960, 39.3 per cent of the world’s population lived in countries in which agriculture produced less than 15 per cent of GDP. This share increased to 49.5 per cent in 1980, 62.8 per cent in 2000, and 69.9 per cent in 2014. Figure 16.2 plots country-specific measures of the share of GDP from agriculture against GDP per capita. The individual bubbles represent country-specific average observations for 2010–14. The relationship is clearly negative in general—a larger share of total national income from agriculture is associated with a lower per capita income—although this relationship is not always smooth and monotonic in the time-series plots for particular countries. Around the world today can be found countries at every stage of the transition that is now largely complete in the high-income countries. In the USA, for example, the total farm population peaked at 32.5 million people, 31.9 per cent of the total US population in 1916. Since then the US population has continued to grow while the farm population declined to 2.9 million in 2006, just 1 per cent of the total population of 300 million. ¹ Throughout this chapter we group countries by geographic region or based on per capita Gross National Income in 2010: in 2012 according to the World Bank (2012), low-income countries had per capita income of $1,035 or less; lower-middle-income, $1,036–$4,085; upper-middle-income, $4,086– $12,615; and high-income, $12,616 or more. The high income ‘region’ includes Western Europe, the USA and Canada, Japan, Australia, and New Zealand. These and other high-income countries are excluded from other geographical regions, which comprise Asia and Pacific (A&P), Latin America and the Caribbean (LAC), Eastern Europe and the former Soviet Union (EE&FSU), the Middle East and North Africa (MENA), and sub-Saharan Africa (SSA). For example, A&P excludes Japan and Singapore, and MENA excludes Qatar and United Arab Emirates.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    .  45

50 40 Percentage

40

Asia&Pacific

35

Percentage

25 20

10

High income 1960 1969 1978 1987 1996 2005 2014

SSA

EE&FSU

15 10

Middle

20 0

30

Low

30

MENA

LAC

5 0 1960

High income 1969

1978

1987

1996

2005

2014

 . Agricultural GDP as share of GDP (by region), 1960–2014 Notes: Countries grouped based on their GNI per capita in 2010 according to World Bank’s (2012) classification, which designated countries to be low-income if their average per capita income was $1,005 or less; lowermiddle-income, $1,006–$3,975; upper-middle-income, $3,976–$12,275; and high-income, $12,276 or more. The high income ‘region’ includes primarily Western Europe, the USA and Canada, Japan, Australia and New Zealand. These and other high-income countries are excluded from other geographical regions, which comprise Latin America & the Caribbean (LAC), Eastern Europe & the former Soviet Union (EE & FSU), Asia & Pacific (A&P), the Middle East & North Africa (MENA), and sub-Saharan Africa (SSA). For example, A&P excludes Japan and Singapore, and MENA excludes Qatar and United Arab Emirates. GDP and agricultural GDP data are available from the UN (2015a) from 1970 onwards. We extended the series to 1960 using estimates from the World Bank (2014) and other sources (e.g. Gapminder, national statistical agencies) when no data were available from the World Bank (2014). Source: Authors’ calculations based on UN (2015a) and World Bank (2014).

In India, 46.4 per cent of the 2014 population were still in agriculture and 58.0 per cent earned less than $3.10 per day (2011 PPP$). Many low-income countries are like Mali where over 73 per cent of the population live on farms and about 78 per cent earn less than $3.10 per day; these countries have not really begun the transformation process. Timmer (2009) observed that the decline in the agricultural share of labour generally lags the decline in the agricultural share of GDP, reflecting some ‘stickiness’ of adjustments in farm labour. The coming decades may see monumental changes in the structure of the farm sector in the countries that have barely begun the process of transition. But the evidence from the past suggests that this transition may not be smooth and may involve very substantial costs of adjustment. Although it seems clear that economic growth entails a reduction in the relative importance of agriculture in an economy, the causal mechanisms in the process by which this transformation takes place are not fully clear. The hard empirical challenge is to sort out the relative roles of the push from agriculture versus the pull from growth in the rest of the economy as contributors to a drift from farm work to part-time farming or

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



Agricultural share of GDP (logarithms) 102

Liberia

51

Indonesia

Nigeria

Niger Brazil

Percentage

26 13

Malaysia USA

Congo D.R. Somalia China

6 3 2 1 0 400

800

1,600

3,200

6,400

12,800

GDP per capita (logarithms)

25,600

51,200 2005 PPP$

 . Agriculture’s share of GDP versus GDP per capita Notes: The bubbles represent country-specific average observations for 2010–14. The downward sloping straight line represents an OLS linear best fit. Other line plots represent the time path of annual values of the respective output shares plotted against GDP per capita (both on a logarithmic scale) for the period 1961–2014 for Brazil, China, Indonesia, Nigeria, and the United States. 2005 PPP$ are purchasing power parity dollars denominated in 2005 prices. Source: Authors’ calculations based on UN (2015a, 2015b) and FAOSTAT (2015).

non-farm employment when both push and pull processes are at work, synergistically, occurring in conjunction with changes in educational status, per capita incomes and other changes, as elaborated by Timmer (2009) and others. Sorting out these issues is a classic and still largely unresolved question at the intersection of agricultural economics and growth theory (see, for example, Gollin et al. 2002 and Herrendorf et al. 2013).

16.3 T S L  G A P

.................................................................................................................................. Global agricultural production has been dominated for a long time by a short list of relatively large and populous countries. In 2011–13, just ten countries accounted for 55.7 per cent of the world’s cropland, and five (India, the USA, China, the Russian Federation, and Brazil) had 41.4 per cent of the total. In contrast, the 100 countries with the smallest shares made up only 0.78 per cent of the world’s cropland area. Production is even more spatially concentrated, with more than half the world’s 2011–13

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

agricultural output coming from only five countries and almost three-quarters of the total output produced by just twenty countries.² Four of the top five countries in global agricultural output, including the top one (China), are not high-income countries. The growth in agricultural output has been very uneven (Table 16.1). Today’s highincome countries produced 43.5 per cent by value of total agricultural output in 1961, and, although production by the high-income countries almost doubled by 2013, their share of the global total shrank to 24.0 per cent. The EE&FSU region produced 13.8 per cent of global food output in 1961, but by 2013 it was producing only 6.3 per cent of the global total. Conversely, the global share of agricultural production increased for all other regions. Notably, the A&P region increased from 24.2 per cent of global agricultural output in 1961 to 44.8 per cent in 2013. Unlike every other industrial production process, agriculture is distinguished by its intensive use of land and other natural resources as inputs, the relevant properties of which vary markedly over space and time. Farming is enormously diverse around the world, with major differences in farming systems, technologies, farm sizes, the mixture of outputs produced, the types of inputs used to produce them, and input proportions. Some of these differences reflect differences in soils and climate or infrastructure that influence agricultural possibilities. Others reflect differences in the relative prices of inputs and outputs and other factors that determine comparative advantage, as well as government policies that dampen its relevance. Some places can grow bananas and

Table 16.1 Regional value of agricultural production, 1961 and 2013 1961 Output

2013 Share

Output

Share

(2005 PPP$ billion) (percent) (2005 PPP$ billion) (percent) High income Eastern Europe & Former Soviet Union Asia & Pacific Latin America & Caribbean Middle East & North Africa Sub-Saharan Africa

325 103 180 69 28 42

43.5 13.8 24.2 9.2 3.7 5.6

605 159 1,128 325 131 171

24.0 6.3 44.8 12.9 5.2 6.8

World

746

100

2,519

100

Notes: See notes to Figure 16.1 for definition of regional groupings. 2005 PPP$ are purchasing power parity dollars denominated in 2004–06 average agricultural prices. Source: Authors’ calculations based on FAOSTAT (2016).

² Crop production is also distributed unevenly within as well as among countries, reflecting climatological and other influences (see Joglekar et al. 2016).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



pineapples, others can grow lettuce and strawberries, and some can at best graze cattle at less than one beast per square mile. As well as affecting what can be grown, and what it is economic to grow, location affects yield and quality of production, and susceptibility to biotic (pests and diseases) and abiotic (climate and soils) constraints (Beddow et al. 2014). Abstracting from geopolitical boundaries, Figure 16.3 displays the distribution of global crop production, expressed as shares of total value, differentiating between irrigated and rainfed agriculture across temperate and tropical latitudes. The bulk of global crop production (86.0 per cent) occurred north of the equator—comprising 62.9 per cent in the temperate north and 23.1 per cent in the tropical north—with only 9.1 per cent of production taking place in the tropical south and just 4.9 per cent in the temperate south (Joglekar et al. 2016). This north–south pattern reflects the fact that only 16.7 per cent of the total harvested area is located south of the equator. Like crop production in aggregate, irrigated agriculture is concentrated in the temperate northerly latitudes—as is population (58.8 per cent is in the temperate north) for similar reasons.³ The discussion so far has emphasized supply conditions, but demand matters too. Indeed, we can account for much of agriculture’s economic geography crudely using Engel curves and data on income distribution and population. The logic is simple.

Share of Crop Production Value (per cent)

Tropic of Capricorn

Equator

Tropic of Cancer

3

2

4.9% of Total

9.1% of Total

23.1% of Total

62.9% of Total

1

0 –40

0 Latitude (degrees)

40

Irrigated

Rainfed

Production System:

 . The latitudinal geography of rainfed versus irrigated crop production, 2005 Source: Joglekar et al. (2016) based on You et al. (2016).

³ This estimate is based on rastered (i.e., pixelated) population data sourced from CIESIN (2016).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

Farm outputs tend to be economically heavy, fragile, or perishable, and significant quantities are consumed within the subsistence households that produced them, or nearby. Consequently, food commodities are predominantly produced close to where they will be consumed.⁴ Since per capita food consumption patterns are driven to a great extent by Engel relationships, the distribution of income has a great deal to say about a country’s national bundle of food production. The country-specific data on food production patterns support this idea, which to some extent belies simplistic notions of (Ricardian) comparative advantage based entirely on resource endowments. In Figure 16.4, each bubble represents a country, sized according to the country’s share of global population in 2013 and shaded by geopolitical region, treating ‘high-income’ countries as a region for this purpose. Looking across countries, calories produced from staple crops as a share of calories from all crops has a visibly negative relationship with average per capita income (on a logarithmic scale)—an Engel effect on the national agricultural output mix! As incomes grow both within and outside agriculture, we would expect to see the mixture of agricultural production shifting over time in the direction of commodities that have larger income elasticities of demand—away from staple food grains to feed grains (i.e. towards livestock) and horticulture, and within those categories towards individual commodities that have larger income elasticities of demand. Indeed, globally, the mix of production has shifted significantly in the direction of commodities used as inputs to produce food eaten by people with higher incomes, especially in the places where incomes are higher, implying shifts in the importance of staple food grains in total agricultural production and in the importance of staple grains and animal protein as sources of food calories produced. Our conception of Engel effects on production, and the data we have presented, suggest that this happens not only globally but also country by country as per capita incomes grow. Historically, growth in demand for food, driven by growth in both population and per capita incomes, was met mainly by expanding the resource base for agriculture, in particular land. But during the past fifty years, in most regions of the world agricultural production expanded mainly by increasing the output per unit of land against a relatively slowly growing land base. Over the period 1961 to 2013, agricultural land use grew at a slow and shrinking rate of less than 0.23 per cent per year, while population grew at about 1.7 per cent per year and the real output from agriculture grew by about 2.3 per cent per year. These increases in land productivity were accomplished by intensifying the use of ‘modern’ inputs—in particular machinery, fertilizers, and irrigation—combined with improved genetic material of

⁴ Some clear exceptions must be made for specific farm products that are (in some cases, at least, of necessity) shipped from other areas. For example, soybeans and bananas, are examples of commodities for which international trade is comparatively important and, conversely, rice is an example of a commodity for which international trade is comparatively thin. Like other staple food crops, much of the world’s rice is produced and consumed within the same household, and some more is consumed very nearby.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



2013

100

Share of calories from staple crops (per cent)

90 80 70 60 50

Asia & Pacific (A&P)

40

Eastern Europe and Former Soviet Union (EE&FSU)

30

High income

20

Latin Am. & Caribbean (LAC)

10 0 148 –10

Middle East & Nth Africa (MENA) Sub-Saharan Africa (SSA) 403

1,097

2,981

8,103

22,026 59,874 GDP per capita (2005 PPP$, log scale)

Low and lower-middle income Upper-middle and high income

 . Engel effects on the country-specific composition of agricultural output, 2013 Notes: Countries are grouped into regions and income classes using the classification schema from the World Development Indicators 2015 report. Available at: https://openknowledge.worldbank.org/handle/10986/21634 See notes to Figure 16.1 for details on regional aggregates. The area of each country bubble is proportional to that country’s share of global population. Staple crops include the following commodities: maize, cassava, sorghum, rice, millet, yams, sugar, pulses, wheat, other cereals, sweet potatoes, groundnuts (shelled equivalent), plantains, palm oil, and bananas. The horizontal line represents the world weighted average of the share of total calories from staple crops. Source: Authors’ calculation based on FAOSTAT Commodity Balances and Food Supply databases downloaded on 21 December 2016 from http://www.fao.org/faostat/en/#data (which represents data last updated on 20 December 2016).

agricultural plants and animals and improved methods of production, derived from organized scientific research, itself a relatively recent innovation. Along with increases in the quantities of land, labour, irrigation, and fertilizer inputs, this growth reflected improvements in input quality, including new and better machines, new varieties of crops and livestock, and better-educated farmers, as well as institutional change and other changes in technology not embodied in inputs. While per capita agricultural output has grown generally, the growth has been uneven among regions and over time (Figure 16.5). In today’s high-income group of countries, per capita agricultural output has been essentially flat since 1980. However, in many of today’s middle-income countries production per capita has grown very rapidly, even with reasonably rapid population growth. The picture is comparatively dismal for sub-Saharan Africa, with a decline in the real value of agricultural output per capita of 0.12 per cent per year from 1961 to 2013.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

Value of agricultural output per capita (200 PPP$)

600

High income

500

EE&FSU

400

LAC 300

MENA

200

Asia&Pacific

100 0 1961

1971

1981

SSA

1991

2001

2011

 . Per capita agricultural production by region, 1961–2013 Notes: Countries are grouped according to World Bank classifications. See notes to Figure 16.1 for details on regional aggregates. 2005 PPP$ are purchasing power parity dollars denominated in 2004–06 average agricultural prices. Source: Authors’ calculations based on FAOSTAT (2016).

16.4 A  O G

.................................................................................................................................. The observed changes in production over the past 50–100 years have been enabled by changes in the composition, quality, and use of conventional inputs—land, labour, materials, and so on—and by the application of unconventional inputs that increased the observed productivity of the conventional inputs. Patterns of conventional input use vary systematically among countries according to their stage of development as measured by per capita income (Table 16.2). In 1961, today’s high-income countries accounted for 43.5 per cent of total global agricultural output but only 24.1 per cent of global population, 27.4 per cent of global agricultural land use, and 8.4 per cent of global agricultural labour. However, the high-income countries accounted for 75.2 per cent of the world’s use of fertilizer and 81.1 per cent of the world’s stock of tractors used in agriculture. By 2014, today’s high-income countries accounted for just 25.5 per cent of global agricultural output, and even further reduced shares of global population, global agricultural land use, and global agricultural labour. As economies become richer they substitute purchased inputs and machinery for primary inputs like land and especially labour. Indeed, the high-income countries have increased their use of fertilizer by 94.5 per cent, but even so their share of the global total use of fertilizer has shrunk considerably; they almost doubled their stock of tractors, holding their global share almost constant. High-income agriculture continues to make significantly greater use of modern land- and labour-saving inputs compared with agriculture in middle- and especially low-income countries, reflecting responses to relative prices (Hayami and Ruttan 1971). High-income countries also invest more intensively in unconventional inputs, in particular science, technology, and education.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



Table 16.2 Conventional and unconventional inputs used in agriculture, 1961 and 2014 Countries by income class Variable

Unit

High

Upper middle

Lower middle

Low

World

1961 Agricultural laboura Agricultural land Fertilizer Tractors Animal traction Cropland per agricultural labour

million million ha million ton million million HP ha per person

65.0 1,107.2 24.5 9.2 16.2 6.3

386.8 1,660.3 6.7 2.0 61.8 1.4

228.4 807.5 1.2 0.1 63.9 1.4

90.5 473.0 0.1 0.1 11.1 1.2

770.7 4,047.9 32.5 11.3 152.9 1.8

2014 Agricultural laboura Agricultural land Fertilizer Tractorsb Animal traction Cropland per agricultural labor

million million ha million ton million million HP ha per person

15.3 1,082.5 47.6 16.7 15.3 23.9

588.1 1,974.2 98.7 6.7 62.8 1.1

478.4 938.6 43.6 3.7 130.3 0.9

251.8 555.3 4.1 0.2 28.2 0.7

1,333.6 4,550.6 193.9 27.3 236.6 1.2

Notes: Countries are grouped based on per capita income in 2010 according to the World Bank’s (2012) classification, which designated countries to be low-income if their average per capita income was $1,005 or less; lower-middle-income, $1,006–$3,975; upper-middle-income, $3,976– $12,275; and high-income, $12,276 or more. Agricultural land is the sum of permanent pasture and harvested area; cropland is the sum of arable and permanently cropped land; fertilizer represents nitrogen, phosphate, and potash in tons of plant nutrients consumed; tractors is the number of agricultural tractors in use. According to FAOSTAT (2016) agricultural tractors ‘generally refers to total wheel, crawler, or track-laying type tractors and pedestrian tractors used in agriculture’. Animal traction represents the stock of buffaloes, horses, asses, mules, and camels converted to horsepowers (HP) units using conversion factors from Craig et al. (1997). ‘ha’ denotes ‘hectares’. Source: Authors’ calculations based FAOSTAT (2015, 2016). a

Agricultural labour represents economically active population in agriculture. As of October 2016, FAOSTAT no longer reports this series. Therefore, we used the series downloaded in August 2015 from FAOSTAT. b 2009 estimate given this was the last year FAOSTAT reported tractor data.

16.4.1 Uneven Agricultural Productivity Growth Especially in the more recent period, the lion’s share of the growth in crop production is attributable to growth in yield, with a comparatively static area of land in production. These yield gains can be attributed to various causes, but principally an increased use of modern inputs (including chemical fertilizers, pesticides, and irrigation) and improved cultural practices applied to modern, higher-yielding varieties. However, the measured global growth rate of cereal yields—that is, the total quantity of cereals produced

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

per unit of land area harvested per year worldwide—seems to have been slowing over time, especially when comparing the rate of growth of the past decade or two against the growth rates witnessed in the 1960s and 1970s. In particular, since 1960 worldwide aggregate cereal yields have been growing approximately linearly, which implies diminishing proportional growth rates. But the patterns of yield growth vary across crops, and among countries, as well as over time within countries, in ways that preclude confident generalizations. In particular, because weather-induced year-to-year variation in yields is large relative to the trend growth rates, we cannot conclusively reject the hypothesis of constant proportional growth in the global average yield for cereals; however the data seem generally more consistent with linear (diminishing proportional) growth in yields. We also see some evidence of a slowdown in broader measures of agricultural productivity growth in many parts of the world. Measures of land and labour productivity growth exhibit mixed patterns among countries and over time. As shown by Alston and Pardey (2014), China is an important exception, representing a large part of the total, and having sustained high productivity growth rates. Taking China out of the world picture, both land and labour productivity growth rates in rest-of-world agriculture, in aggregate, were slower after 1990 than before. Among the top ten agricultural producers—accounting for almost two-thirds of the world’s 2011–13 average value of agricultural output—only China (ranked first in terms of agricultural production), India (ranked second), and Brazil (ranked fourth) had higher rates of land productivity growth after compared with before 1990. Among the top twenty agricultural producers (77 per cent of the world’s 2011–13 average value of agricultural output), the slowdown was more pronounced in land productivity (55 per cent of the top countries) than labour productivity (35 per cent). As Alston and Pardey (2014) discuss in some detail, the available evidence also points to a slowdown in the more comprehensive measures of multifactor productivity (MFP, which some call total factor productivity, TFP) that are available for some of the richer countries whose economies have largely completed the agricultural transformation noted above. However, the evidence on TFP or MFP is much less complete and the measures that do exist are much more open to question—especially those based on FAO data.⁵

⁵ Using TFP measures derived largely from FAO data for the period 1961–2009, Fuglie (2012: 356) reached the opposite conclusion noting ‘there does not seem to be a slowdown in sector-wide global agricultural productivity growth. If anything, the growth rate in global agricultural TFP accelerated, in no small part because of rapid productivity gains achieved by developing countries, led by Brazil and China, and more recently because of a recovery of agricultural growth in the countries of the former Soviet Union and Eastern Europe.’ (See also Fuglie 2008.) There are reasons to be cautious in drawing such conclusions on the basis of these TFP (total factor productivity) estimates given the nature of the source data and the potential for systematic biases in the input (and output) measures used to construct these TFP estimates, as discussed in Alston et al. (2010b: ch. 15). In addition, in several (high-income) countries, including the United States, the more complete MFP measures that are available point to a slowdown (Alston et al. 2010a; and Andersen et al. 2018).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



16.4.2 Investing in Innovation Public- and private-sector organized agricultural science was a primary driver of the growth in productivity of land and labour in agriculture that has been a central element in and consequence of the transformation of agriculture. Slowing rates of productivity growth raise concerns about whether the world is spending enough on agricultural science, technology, and innovation, as well as other unconventional inputs to agriculture such as educating individuals and building institutions and infrastructure. As documented by Pardey et al. (2016), in 2011, a total of about $38.1 billion (in 2009 dollars converted at purchasing power parity exchange rates) was spent on public-sector agricultural and food research worldwide. While our empirical handle on private investments in food and agricultural research is less certain, the available evidence indicates that the private sector spent $31.1 billion (2009 PPP$) on agricultural and food research in 2011. The largest share of that private research, 64 per cent, took place in the high-income countries and, for the high-income countries at least, almost one-half of that research was concerned with producing off-farm innovations, primarily those related to food processing. The potential for spillovers of agricultural technologies is mitigated by differences in climate and other aspects of the natural resource stocks that govern agricultural potential and the structure and intensity of agricultural input use. Even so, spatial movement of agricultural technologies has played important roles in the development of agriculture, although typically involving some significant investment in adaptive research (Alston 2002; Pardey et al. 2006). Both through informal and marketmediated mechanisms and through direct involvement in international agricultural research programmes (see, e.g., Pardey and Beddow 2013), the high-income countries have sponsored innovation in agriculture throughout the world, including in the world’s poorest countries that do not invest much on their own behalf. This source of global public R&D goods may be drying up, and that might have contributed to the observed slowdown in productivity growth. In the high-income countries, in spite of compelling evidence of high rates of return and a significant productivity slowdown, public support for agricultural science has broadly waned. In 1960, $6.19 billion (2009 PPP$) was spent on public agricultural and food research and today’s high-income countries accounted for 55.9 per cent of the world’s total. Some fifty years later, in 2011, that high-income country share had dropped to 47 per cent, with the US share dropping over the same period from 20.2 per cent to just 11.5 per cent of global public spending on agricultural and food research (Pardey et al. 2016). Real rates of annual public agricultural and food research spending have begun to decline in many countries, including the USA. Moreover, of the amounts being spent on ‘agricultural science’ in high-income countries, an ever-increasing share is being directed towards off-farm issues—such as health and nutrition, food safety, biofuels technology, and the environment—leaving less for research directed at maintaining or increasing farm productivity (Pardey et al. 2013).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

Public sector research capacity in the agricultural sciences in many (especially high-income) countries has been run down over decades, infrastructure has been depreciated, and the majority of the scientists in many countries are close to retirement age. On the salutary side, agricultural research is on the rise in the large, populous middle-income countries: Brazil, India, and China together provided 27.1 per cent of the world’s public agricultural research in 2011. These countries have amongst the largest total numbers of farmers and the ‘food-poor’ whose lives can be very substantially improved through agricultural innovation leading to more abundant and cheaper food. In 2011, China spent more than any other country on public-sector agricultural R&D, with a budget of $4.7 billion (compared with the USA at $4.4 billion). But we also see the continuation of a large and growing global divide, with the world’s poorest countries falling even farther behind. Notably, the nations of sub-Saharan Africa spent just 6.2 per cent of the world’s total public-sector agricultural research in 2011, down from 11.7 per cent in 1960.

16.5 P   F Y  C

.................................................................................................................................. The past 50–100 years have witnessed dramatic changes in agricultural production and productivity, driven to a great extent by public and private investments in agricultural research, with profound implications especially for the world’s poor. A trend of slowing growth or real reductions in public spending by the high-income countries on agricultural productivity-enhancing research has already begun to contribute to a slowdown in their agricultural productivity growth—and this can only make the current trends worse. But over time and among countries the developments in agricultural production and productivity have been uneven, resulting in seismic shifts in the world table of agricultural production over the past few decades, and prospects for continuing shifts over the decades to come. A half-century ago, today’s high-income countries dominated agricultural production and public agricultural research. In the fifty years since then, these countries have shrunk in relative global importance both as agricultural producers and in terms of agricultural research. In counterpoint, the middle-income countries—especially China and Brazil—have grown in importance both as agricultural producers and as performers of agricultural research. These countries have significantly reduced the relative role of agriculture in their own economies while rising to a position of dominance within the global agricultural economy—mirroring the status of today’s high-income countries a half century ago. Meanwhile, many of the world’s poorest countries continue to lag behind in agricultural production and productivity, in agricultural research, and in making their economic transition away from agriculture.

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



These uneven developments, and the associated systemic seismic shifts in agricultural production, productivity, and spending patterns for agricultural research mean that the world, especially the world’s poor, will increasingly depend on the middle-income countries for agricultural innovations and abundance. The shifted shares of the world’s agricultural science—as documented by Pardey et al. (2016)—will have implications over decades to come for the balance of research undertaken, global patterns of productivity and prices, competitiveness and comparative advantage, the mix and quality of food and other agricultural products produced, and the livelihoods of farmers and their families. Even if we see a reversal of research investment trends in the high-income countries, both toward higher rates of agricultural research spending and toward a renewed focus on sustaining and increasing crop yields and other dimensions of agricultural productivity—which does not seem very likely—it does seem likely that today’s middle-income countries, and especially China, India, and Brazil will increasingly determine the future path of poverty and hunger in the world and the vulnerability of the poor to food price shocks of the kind experienced in 2008 and 2012.⁶ These countries are now poised potentially to play a role in the coming fifty years that was played by today’s high-income countries in the past fifty years. One of the major global challenges in the years ahead will be getting the relevant agricultural innovations into the hands of the world’s poor farmers, such as those in south Asia and sub-Saharan Africa. Even with the rise of some middle-income countries, agricultural and food research continues to be concentrated in just a handful of nations. In 2011, the top ten countries ranked by spending on agricultural R&D accounted for 70 per cent of the total investment worldwide; the bottom 100 countries contributed just 9 per cent of that year’s total. Yet these 100 countries are home to 22 per cent of the world’s population. A sustained increase in government funding is imperative, along with robust and agile institutional innovations that foster public and private investment in poorcountry agriculture. Without efforts to improve the global spread and adaptation of locally relevant technologies, it is likely to get much harder for poor farmers to feed themselves, let alone their nations’ increasingly urbanized populations. In those countries that are currently responsible for most of the world’s agricultural production, the innovation challenges are also pressing, if different. Without sufficiently supported research and innovation in agriculture, crop yields are bound to grow ever-more slowly, and ultimately to decline absolutely as environmental changes (including changes in weather patterns and crop pests and diseases driven in part by climate change) and increasingly restrictive technology policies undermine past productivity gains. Achieving ever-higher productivity to feed a growing, increasingly wealthy and ⁶ Uncertainties about this future are underscored by recent developments in Brazil, whose current status and prospects for agricultural R&D funding are conditioned by both the 40 per cent reduction in total R&D funding during 2013–16 as a consequence of reduced public spending, and the prospects of a freeze on total federal government spending for the next two decades (Angelo 2016).

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

more urbanized population, while sustaining or rehabilitating fragile natural resources, is going to require considerably more investment in agricultural R&D. It will also require both public and private investment, because the two tend to support different, often complementary, types of R&D.

16.6 C

.................................................................................................................................. If recent trends continue, global agricultural R&D in the middle of the twenty-first century will look very different from how it looked at the dawn of the century. It is encouraging to see the rise of agricultural R&D in the rapidly growing middle-income countries, and the increase in private-sector participation in various regions. But the retreat from public agricultural R&D by rich countries and the continued comparatively low rates of investment in many poorer countries, are nonetheless concerning. Also concerning is the rising political opposition to the use of modern technologies, including GMOs (genetically modified organisms) and the newer technologies that will be enabled by gene editing techniques, the use of which will be essential if the already-too-scarce research resources are to be used to maximum good effect and enable a completion of the agricultural transformation envisioned by Schultz over half a century ago.

A We are grateful for research assistance provided by Connie Chan-Kang and Ali Joglekar. The work for this project was partly supported by the University of California; the University of Minnesota; the HarvestChoice initiative, funded by the Bill and Melinda Gates Foundation; and the Giannini Foundation of Agricultural Economics.

R Alston, J. M., 2002. ‘Spillovers’, Australian Journal of Agricultural and Resource Economics, 46 (3), pp. 315–46. Alston, J. M. and P. G. Pardey, 2014. ‘Agriculture in the Global Economy’, Journal of Economic Perspectives, 28 (1), pp. 121–46. Alston, J. M., M. A. Andersen, J. S. James, and P. G. Pardey, 2010a. Persistence Pays: U.S. Agricultural Productivity Growth and the Benefits from Public R&D Spending, New York: Springer. Alston, J. M., B. A. Babcock, and P. G. Pardey eds, 2010b. The Shifting Patterns of Agricultural Production and Productivity Worldwide, CARD-MATRIC Electronic Book, Ames, IA: Center for Agricultural and Rural Development. Available at: from https://lib.dr.iastate. edu/cgi/viewcontent.cgi?article=1001&context=card_books

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi

   



Andersen, M. A., J. M. Alston, P. G. Pardey, and A. Smith, 2018. ‘A Century of U.S. Farm Productivity Growth: A Surge Then a Slowdown’, American Journal of Agricultural Economics, 100 (4), pp. 1072–90. Angelo, C., 2016. ‘Brazil’s Scientists Fight Funding Freeze: Proposed Law Could Restrict Research Spending for 20 Years’, Nature, 439, pp. 480. Beddow, J., T. Hurley, P. Pardey, and J. M. Alston, 2014. ‘Rethinking Yield Gaps’, in Neal Van Alfen, ed., Encyclopedia of Agriculture and Food Systems, Vol. 3, San Diego: Elsevier, pp. 352–65. Carter, S. B., S. S. Gartner, M. R. Haines, A. L. Olmstead, R. Sutch, and G. Wright, eds, 2006. Historical Statistics of the United States: Earliest Times to the Present, Millennnial Edition, Vol. 4, Cambridge and New York: Cambridge University Press. CIESIN (Center for International Earth Science Information Network). 2016. Gridded Population of the World, Version 4 (GPWv4): Year 2005 Population Count Adjusted to Match 2015 Revision of UN WPP Country Totals, Palisades, NY: NASA Socioeconomic Data and Applications Center (SEDAC). Available at: http://sedac.ciesin.columbia.edu/data/collect ion/gpw-v4/sets/browse accessed 12 July 2016. Craig, B. J., P. G. Pardey, and J. Roseboom, 1997. ‘International Productivity Patterns: Accounting for Input Quality, Infrastructure and Research’, American Journal of Agricultural Economics, 79 (4), pp. 1064–76. FAOSTAT, 2015. ‘A Database of the Food and Agriculture Organization of the United Nations (FAO)’. Available at: http://faostat.fao.org accessed August 2016. FAOSTAT, 2016. ‘A Database of the Food and Agriculture Organization of the United Nations (FAO)’. Available at: http://faostat.fao.org accessed October 2016. Fuglie, K., 2008. ‘Is a Slowdown in Agricultural Productivity Growth Contributing to the Rise in Agricultural Prices?’ Agricultural Economics, 39, pp. 431–44. Fuglie, K., 2012. ‘Productivity Growth and Technology Capital in the Global Agricultural Economy’, in K. O. Fuglie, S. L. Wang, and V. E. Ball, eds, Productivity Growth in Agriculture: An International Perspective, Wallingford: CAB International. Gollin, D., S. Parente, and R. Rogerson, 2002. ‘The Role of Agriculture in Development’, American Economic Review, 92(2), pp. 160–4. Hayami, Y. and V. W. Ruttan, 1971 (reprinted 1985). Agricultural Development: An International Perspective, Baltimore, MD: Johns Hopkins University Press. Herrendorf, B., R. Rogerson, and Á. Valentinyi, 2013. ‘Two Perspectives on Preferences and Structural Transformation’, American Economic Review, 103 (7), pp. 2752–89. Joglekar, A. B., P. G. Pardey, and U. Wood Sichra, 2016. ‘Where in the World are Crops Grown?’, HarvestChoice Brief. St. Paul, MN: International Science & Technology Practice and Policy (InSTePP) Center, University of Minnesota and Washington, DC: International Food Policy Research Institute (IFPRI). Pardey, P. G. and J. M. Beddow, 2013. Agricultural Innovation: The United States in a Changing Global Reality, CCGA Report, Chicago: Chicago Council on Global Affairs. Pardey, P. G., J. M. Alston, C. Chan-Kang, E. C. Magalhães, and S. A. Vosti, 2006. ‘International and Institutional R&D Spillovers: Attribution of Benefits Among Sources for Brazil’s New Crop Varieties’, American Journal of Agricultural Economics, 88(1), pp. 104–23. Pardey, P. G., J. M. Alston, and C. Chan-Kang, 2013. Public Food and Agricultural Research in the United States: The Rise and Decline of Public Investments, and Policies for Renewal. Washington, DC: AGree Policy Report. Available at: http://www.foodandagpolicy.org/sites/ default/files/AGree-Public%20Food%20and%20Ag%20Research%20in%20US-Apr%202013.pdf

OUP CORRECTED PROOF – FINAL, 29/12/2018, SPi



 .    . 

Pardey, P. G., C. Chan-Kang, S. P. Dehmer, and J. M. Beddow, 2016. ‘Agricultural R&D is on the Move’, Nature, 15 (537), pp. 301–3. Schultz, T. W., 1964. Transforming Traditional Agriculture, New Haven, CT: Yale University Press. Schultz, T. W., 1979. ‘The Economics of Being Poor’. Nobel Prize Lecture, republished at: http://www.nobelprize.org/nobel_prizes/economic-sciences/laureates/1979/schultz-lecture.html Timmer, C. P., 2009. A World without Agriculture: The Structural Transformation in Historical Perspective, Washington, DC: American Enterprise Institute Press. United Nations, Department of Economic and Social Affairs, Population Division, 2015b. World Population Prospects: The 2015 Revision, New York: UN. Available at: http://esa.un. org/unpd/wpp/Download/Standard/Population/ accessed 4 January 2016. United Nations Statistics Division, 2015a. UN National Accounts Main Aggregates Database, New York: United Nations. Available at: http://unstats.un.org/unsd/snaama/introduction .asp accessed 24 October 2016. Last update December 2015. World Bank, 2012. World Bank list of economies (April 2012), Washington, DC: World Bank. World Bank, 2014. World Development Indicators Online (last updated 22 July 2014), Washington, DC: World Bank. Available at: http://data.worldbank.org/data-catalog/ world-development-indicators accessed September 2014. World Bank, 2017. World Bank Development Indicator Database, Washington, DC: World Bank. Available at: http://databank.worldbank.org/data/reports.aspx?source=worlddevelopment-indicators accessed January 2017. Last updated on 3 January 2017. You, L., U. Wood-Sichra, S. Fritz, Z. Guo, L. See, and J. Koo, 2016. Spatial Production Allocation Model (SPAM) 2005 v2.3. HarvestChoice Data Product, Washington, DC: International Food Policy Research Institute (IFPRI) and St. Paul: International Science and Technology Practice and Policy (InSTePP) Center, University of Minnesota. Unpublished data (accessed 20 January 2016).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

                   ......................................................................................................................

 

17.1 I

.................................................................................................................................. S on changes in the structure of production gained attention over half a century ago, especially with the work of Kuznets (1957), which depicts an increase in manufacturing with rising per capita income. Among the most well-known studies on structural change in addition to Kuznets’ are those of Fisher (1939), Clark (1957), Chenery and Syrquin (1975) and Kader (1985). Although Chenery (1960) and Chenery and Taylor (1968) focused on development patterns in the manufacturing sector, structural change within the manufacturing sector has subsequently not been studied in depth. As argued by Lin (2010), the optimal industrial structure differs according to stage of development and country, given features, and, consequently, countries in different development stages have comparative advantages in different industries. Identifying latent comparative advantages and understanding their evolutions can help countries pursue welfare-enhancing industrial structural change, something many developing countries have been struggling to achieve (McMillan and Rodrik 2011). Haraguchi and Rezonja (2011) elucidate the patterns of manufacturing structural change and indicate how comparative advantages, technological capability, and country specific conditions together influence manufacturing development. This chapter incorporates employment aspects into structural change analysis to highlight the interaction between the output, productivity, and employment growth rates of large countries and to illustrate manufacturing industries’ different

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

development stages associated with periods of employment enhancement and contraction.¹ While a ‘fundamental identity’ at macro-level linking employment, labour productivity, and aggregate output suggests the following relationship, there is no clear evidence of causality or patterns of change in the relationships among the three variables (Landmann 2004). In the short term, Landmann argues that both the rate of employment growth and productivity growth are pro-cyclical, meaning that both drop in times of recession and surge during a boom. However, in the long-term, there is no clear trend in the relationship. Output growth
employment growth rate þ productivity growth rate At the level of manufacturing industries, the relative importance of the rates of employment growth and productivity growth in terms of their contribution to output performance may differ from industry to industry and change within an industry along the stages of a country’s development. The key issues addressed in this chapter are the growth patterns of manufacturing industries identified in past studies (Haraguchi and Rezonja 2010, 2011) with regard to output (value added per capita), how the relative significance of the two right hand side variables changes. Based on the findings, the chapter suggests a pro-employment path of manufacturing development. In the period of a given industry’s expansion, that is, when its value added increases, the growth rate of labour productivity is usually positive as it is a function of learning, capital accumulation, and economies of scale. Hence, the key questions we would like to address are when, to what extent, and for how long employment increases during the industry’s expansion, and how the contribution of employment and productivity growth in relation to that of value added changes along the path of a country’s development. Based on the above equation, Rada and von Arnim (2011) identified three different relationships among the variables. In the case of profit-led growth, it can be shown that output, productivity, and employment all grow strongly. As a result of the investment of rising profits, the industry’s output and employment expand, leading to greater efficiency and competitiveness and hence resulting in an increasing volume of demand for the industry’s products. In the case of wage-led growth, both employment and output decline while only labour productivity expands. In such a situation, wages rise as a result of productivity growth, but the decline of the employment rate is faster than the rate of productivity growth, which inevitably translates into output decline. Essentially, wage-led industries are in a declining stage of development where even efficiency gains from labour productivity increases cannot prevent the fall in the industry’s output (Ocampo et al. 2009). Finally, Rada and von Arnim speak of a weakly profit-led growth when output ¹ Due to differences in manufacturing development patterns among large, middle, and small countries and space limitations, this study focuses on the analysis of large countries with a population of more than 12.5 million.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



growth is at least in a positive range, even though it is not growing as fast as the rate of labour productivity (Rada and von Arnim 2011). The above categorization is useful for identifying three different stages of an industry based on its output, employment, and productivity relationships. However, it does not provide any clues on how such relationships are likely to evolve for different manufacturing industries as countries develop. This study looks at the changes in the relationships from the perspective of manufacturing structural change and links such changes with different stages of an industry’s development in large countries. Analyses of the industrial transformation that takes place across industries and a comparison thereof provides insights for developing policies which support the expansion of and employment generation in manufacturing industries at different stages of a country’s development.

17.2 D, V,  E

.................................................................................................................................. To illustrate the structural change of manufacturing industries, this study uses GDP per capita² as the independent variable while the dependent variable is represented by one of the following three—value added per capita, employment–population ratio (EP ratio), and labour productivity.³ To measure the size of an industry or its output level, value added rather than gross output is a better indicator, as the former reflects only what the industry actually produced and excludes purchased inputs. Furthermore, to estimate the patterns of the changes in the variables using industrial panel data of different countries, we normalized value added and employment by expressing both in per capita terms. Each of the three dependent variables is examined in relation to the changes in GDP per capita. Our approach differs from that of Rada and von Arnim (2011), who used the elasticities of output and employment in terms of labour productivity. In our study, the elasticity of each variable is calculated in terms of GDP per capita, and the three elasticities are compared at different levels of GDP per capita. The analysis is conducted for the manufacturing industries at the two-digit level of the International Standard Industrial Classification (ISIC) revision 3. There are twentythree industrial categories in total. However, as countries often report industries 18 and 19, 29 and 30, 31 and 32, and 34 and 35 together, we combined each pair into one industrial category to have a consistent data set across countries. Furthermore, we dropped industry 37, recycling, as it has only been reported by a very limited number of countries. The following table presents the industrial classifications used in this study. Ideally, real value added should be calculated for the two dependent variables used in ² GDP per capita is PPP adjusted and constant at 1995 price. ³ EP ratio, as it is, will be very small numbers. Thus, in this chapter we express it as a percentage by multiplying by 100.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

the analysis—value added per capita and labour productivity—as an output in constant price excluding various purchases from other industries valued in constant prices. However, such price-adjusted data are not available for a large number of countries, in particular for developing countries, to reliably estimate the development patterns of manufacturing industries. Alternatively, to adjust changes in price, we use the Index of Industrial Production (IIP) which is available at the two-digit level of the ISIC. Some countries have already begun reporting their industrial data based on the latest ISIC revision (revision 4); however, we use the IIP based on revision 3 of the ISIC, which has been widely used since the mid-1980s. To obtain a longer time series data UNIDO has combined the IIP of ISIC revision 2, which goes back to the early 1960s, with revision 3 to arrive at an IIP that covers the years 1963 to 2004 based on revision 3 of the ISIC. By multiplying such a series of volume indices by the value added of a given base year, we are able to approximate real value added for a time series.⁴ However, the IIP is only available for around seventy countries; hence, when using this approach, approximately fifty countries which do not have an IIP, but for which the nominal value added data for their manufacturing industries is available, cannot be included in the regressions to estimate manufacturing development patterns. Since many countries without an IIP are developing and emerging countries, it is important to also reflect their development trajectories in the estimations of manufacturing structural change. Manufacturing sector-wide value added (MVA) deflators are available for most of the countries without an IIP. However, applying an MVA deflator across manufacturing industries might produce biases, as inflation rates from one industry to another may differ significantly (e.g. between the food and beverages industry and the petrochemical industry) for given years.⁵ To reflect the industry-specific inflation trend, we decompose the respective country’s manufacturing-wide deflation using an inflation structure based on the same year’s IIP of another country located in the same region and at a relatively similar development stage. Using this approach, we try to reflect industry-specific inflation trends by equalling the sum of the nominal value added divided by the sum of the real value added of manufacturing industries with the country’s MVA deflator. This approach allows us to include around seventy countries with and fifty countries without an IIP in our estimations. A UNIDO working paper (Haraguchi 2012) explains this procedure in detail in its Appendix A. ⁴ Depending on the given country, changes in the weight of quality and products in an industry may not necessarily be regularly updated in IIP. The gradual changes in the value added share in output may not be appropriately reflected in the IIP despite regular adjustments. ⁵ The authors first determined whether a manufacturing value added deflator (MVA deflator), i.e. a manufacturing sector-wide deflator, could be used for the 70 countries with an IIP. Where this was found to be suitable, a country’s MVA deflator could be used to deflate the valued added across manufacturing sub-sectors within a country for all 120 countries with MVA deflators. To check this, the manufacturing development patterns were estimated for the 70 countries with an IIP and MVA deflators, using both their IIP and MVA deflator. The two estimated patterns based on the IIP and MVA deflator approaches were compared to determine whether the differences between the two were statistically significant. The two patterns significantly varied for many industries, and we were therefore not able to adjust nominal values by using MVA deflators, which were available for 120 countries.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



Table 17.1 Manufacturing data classification used in this study ISIC description

Abbreviation

ISIC code

Food and beverages Tobacco products Textiles Wearing apparel, and fur & leather products, and footwear Wood products (excluding furniture) Paper and paper products Printing and publishing Coke, refined petroleum products, and nuclear fuel Chemicals and chemical products Rubber and plastics products Non-metallic mineral products Basic metals Fabricated metal products Machinery and equipment n.e.c. & office, accounting, computing machinery Electrical machinery and apparatus & radio, television, and communication equipment Medical, precision, and optical instruments Motor vehicles, trailers, semi-trailers, & other transport equipment Furniture; manufacturing n.e.c.

Food and beverages Tobacco Textiles Wearing apparel

15 16 17 18 & 19

Wood products Paper Printing and publishing Coke and refined petroleum Chemicals Rubber and plastics Non-metallic minerals Basic metals Fabricated metals Machinery and equipment

20 21 22 23 24 25 26 27 28 29 & 30

Electrical machinery and apparatus Precision instruments Motor vehicles

31 & 32

Furniture, n.e.c.

36

33 34 & 35

Source: Created by the authors.

Past studies acknowledge that country size has an overarching influence on economic structural change (Chenery and Taylor 1968; Perkins and Syrquin 1989) with effects on both the intercepts as well as the slope of the estimated patterns. Thus, instead of including population in the equation as an additional explanatory variable, many studies resort to dividing countries into size groups, applying a given population size as a threshold. The problem related to this approach in past studies has been that this threshold was often arbitrarily used without determining whether such groups statistically differ in terms of their development patterns. To classify countries into three groups of different sizes, we apply thresholds to divide them into small, medium, and large countries, and examine at which threshold level the maximum number of manufacturing industries is obtained, whose development patterns statistically differ from one another in terms of value added per capita. This is achieved by applying the Wald test. Based on our test results, we use thresholds of 3 million and 12.5 million to divide countries into small, medium, and large countries. In accordance with these thresholds, medium-sized countries with a population from 3 million to 12.5 million have different development patterns than small-sized countries with a population of less than 3 million for thirteen out of eighteen manufacturing industries. The

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

development patterns of all industries in large-sized countries with a population of over 12.5 million differ from those in medium-sized countries. It does not suffice to divide countries into three groups using the above method to unequivocally claim that a distinct pattern emerges for each group. Ideally, countries in the same group should at least have statistically equal coefficients for the slopes. To determine whether countries within the same group have similar development patterns, we examine the statistical significance of both the individual country intercepts and slopes of the explanatory variables used in the equations to estimate the value added per capita. Individual country intercepts are significant across most of the countries and industries, therefore, it can be inferred that countries differ in terms of intercept levels. Individual slopes are statistically insignificant for the majority of countries across all industries, which indicates that countries in the same size group do not significantly differ from each other in terms of slope. Given the three size groups of countries classified by the aforementioned measures, this study focuses on the large country group due to the space required for the analysis of manufacturing industries at a disaggregated level. In the long term, it is assumed that industries undergo three development stages— pre-takeoff, growth, and decline—following a pattern of a cubic function. However, those industries which can sustain growth over a long period of time may have a more linear development trajectory, while other industries which experience growth from a very early stage of development and only decline at a later stage, may indicate a more quadratic pattern. Hence, in addition to GDP per capita, we include cubic and square terms of GDP per capita in the equation in order for the results to denote possible patterns of manufacturing development, depending on the statistical significance of these GDP per capita terms. To control for the effect of unobserved countryspecific conditions, we apply the fixed effect estimation procedure. For this purpose, the following equations are used for each manufacturing industry in the group of large countries. In RVAict ¼ α1 þ α2 ∗ In RGDPct þ α3 ∗ In RGDP2ct þ α4 ∗ In RGDP3ct þ αc þ eict ð1Þ In EMPcti ¼ α1 þ α2 ∗ In RGDPct þ α3 ∗ In RGDP2ct þ α4 ∗ In RGDP3ct þ αc þ eict ð2Þ In LPcti ¼ α1 þ α2 ∗ In RGDPct þ α3 ∗ In RGDPct2 þ α4 ∗ In RGDPct3 þ αc þ eict ð3Þ where: -

RVA indicates real value added per capita EMP represents real employment–population ratio LP means labour productivity RGDP stands for real GDP per capita RGDP² denotes real GDP per capita square RGDP³ signifies real GDP per capita cubic αc is country fixed effect ecti refers to unexplained residual.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



Both dependent and explanatory variables are expressed in logarithmic terms to measure the elasticity of each variable. The regression results are presented in Appendix A.

17.3 R

..................................................................................................................................

17.3.1 Value Added Growth as a Function of the Labour Productivity and Employment Growth Rates Based on the identity discussed in the introduction, this section will look into the relationships between the rate of labour productivity growth and employment growth and how they together influence value added growth (Figures 17.1–17.3 and Appendix Tables 17B.1 and 17B.2). The eighteen manufacturing industries studied in this chapter are classified into early, middle, and late industries depending on whether an industry reaches its highest share in total manufacturing value added before a GDP per capita of US$5,000, between US$5,000 and US$20,000 or after US$20,000, respectively. Table 17.2 shows the changes in the shares of manufacturing industries every $1,000 of GDP per capita. The highest shares of the industries are indicated by a box. The early industries include food and beverages (ISCI code: 15), tobacco (16), textiles (17), wearing apparel (18), wood products (20), printing and publishing (22), coke and refined petroleum (23), non-metallic minerals (26), and furniture, n.e.c. (36). The middle industries are paper (21), basic metals (27), fabricated metals (28), and precision instruments (33). The late industries comprise chemicals (24), rubber and plastics (25), machinery and equipment (29), electrical machinery and apparatus (31), and motor vehicles (34).

17.3.1.1 The Early Industries Up to around US$10,000 GDP per capita, the early industries, especially the food and beverages, textiles, wearing apparel, and non-metallic mineral industries, have significant weight in total manufacturing value added (Table 17.2). However, early industries increase their value added per capita (industry size) through different combinations of both the labour productivity and employment growth rates (Figure 17.1). The food and beverages industry, usually the largest manufacturing industry in this early stage, increases production on account of the steady growth of the productivity and a slow decline in the employment growth rate. The non-metallic mineral industry displays a similar pattern, but experiences both a faster productivity growth rate and a faster decline of the employment growth rate. The textiles industry may experience fast growth up to US$5,000, mainly due to decelerating but fast productivity growth. Given the declining trend of labour productivity, the rapid drop in the employment growth rate causes the fast decline of the value added per capita growth rate in the textiles industry from around US$6,000 per capita. In the case of the wearing apparel industry, the very high

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Table 17.2 Value added shares in total MVA for large countries GDP per capita (in thousands) ISIC 15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

1

2

3

4

5

6

7

8

9

10

11

12

13

14

15

16

20.2 21.2 20.3 19.6 18.9 18.4 17.9 17.5 17.0 16.6 16.3 15.9 15.6 15.2 14.9 14.6 6.6 6.1 5.0 4.2 3.6 3.1 2.7 2.4 2.2 1.9 1.8 1.6 1.5 1.4 1.3 1.2 13.4 9.7 8.8 8.5 8.3 8.1 7.9 7.6 7.3 6.9 6.6 6.2 5.9 5.6 5.2 4.9 1.3 5.0 7.1 7.8 7.8 7.5 7.0 6.4 5.9 5.4 4.9 4.5 4.1 3.8 3.4 3.2 6.6 4.3 3.2 2.6 2.2 1.9 1.7 1.5 1.4 1.2 1.1 1.0 1.0 0.9 0.8 0.8 3.2 3.4 3.4 3.4 3.4 3.4 3.4 3.4 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.3 4.1 4.1 4.0 3.8 3.6 3.5 3.4 3.2 3.1 3.0 2.9 2.8 2.7 2.7 2.6 4.7 4.1 3.8 3.7 3.6 3.6 3.5 3.5 3.4 3.3 3.3 3.2 3.2 3.1 3.0 2.9 6.2 8.8 9.7 10.2 10.6 11.0 11.3 11.5 11.7 11.9 12.1 12.3 12.4 12.6 12.7 12.9 2.7 3.7 4.0 4.2 4.2 4.3 4.4 4.4 4.5 4.5 4.5 4.6 4.6 4.6 4.7 4.7 4.0 6.9 7.5 7.4 7.2 6.9 6.7 6.4 6.2 5.9 5.7 5.6 5.4 5.2 5.1 5.0 5.7 4.6 4.7 5.1 5.6 6.0 6.5 6.8 7.2 7.5 7.7 7.9 8.1 8.2 8.3 8.3 6.7 4.0 3.7 3.9 4.1 4.3 4.6 4.8 5.0 5.1 5.2 5.3 5.4 5.4 5.4 5.4 4.3 3.1 2.9 3.0 3.2 3.5 3.7 4.0 4.3 4.6 4.8 5.1 5.4 5.7 5.9 6.2 1.0 2.3 3.2 4.0 4.7 5.3 5.9 6.5 7.1 7.6 8.1 8.7 9.2 9.7 10.2 10.7 0.8 0.7 0.7 0.8 0.8 0.9 0.9 0.9 1.0 1.0 1.0 1.0 1.0 1.0 1.0 1.0 6.0 3.5 3.5 3.9 4.5 5.1 5.7 6.3 7.0 7.6 8.2 8.7 9.2 9.7 10.1 10.5 3.2 4.5 4.2 3.8 3.4 3.0 2.8 2.6 2.4 2.2 2.1 2.0 1.9 1.8 1.8 1.7

17

18

19

20

14.4 1.1 4.6 2.9 0.7 3.5 2.5 2.9 13.0 4.7 4.8 8.4 5.4 6.4 11.2 1.0 10.8 1.7

14.1 1.0 4.3 2.7 0.7 3.4 2.4 2.8 13.2 4.8 4.7 8.4 5.3 6.7 11.7 1.0 11.1 1.6

13.8 1.0 4.1 2.5 0.6 3.4 2.4 2.7 13.3 4.8 4.6 8.3 5.3 6.9 12.2 1.0 11.4 1.6

13.6 0.9 3.8 2.3 0.6 3.4 2.3 2.7 13.4 4.8 4.5 8.3 5.2 7.2 12.7 1.0 11.6 1.5

21

22

23

24

25

26

27

28

29

30

31

32

33

34

35

36

37

38

39

40

13.4 0.8 3.6 2.1 0.6 3.4 2.3 2.6 13.6 4.9 4.5 8.3 5.1 7.4 13.3 0.9 11.8 1.5

13.1 0.8 3.3 2.0 0.5 3.4 2.2 2.6 13.7 4.9 4.4 8.2 5.0 7.6 13.8 0.9 12.0 1.5

12.9 0.8 3.1 1.9 0.5 3.4 2.2 2.5 13.8 4.9 4.3 8.1 4.9 7.9 14.3 0.9 12.1 1.5

12.7 0.7 2.9 1.7 0.5 3.4 2.1 2.4 14.0 5.0 4.2 8.0 4.8 8.1 14.8 0.9 12.2 1.4

12.5 0.7 2.7 1.6 0.4 3.4 2.1 2.4 14.1 5.0 4.2 7.9 4.7 8.3 15.3 0.9 12.3 1.4

12.3 0.6 2.5 1.5 0.4 3.4 2.1 2.3 14.2 5.0 4.1 7.8 4.6 8.5 15.9 0.9 12.3 1.4

12.1 0.6 2.4 1.4 0.4 3.4 2.0 2.3 14.4 5.1 4.1 7.7 4.5 8.7 16.4 0.8 12.4 1.4

11.9 0.6 2.2 1.3 0.4 3.3 2.0 2.2 14.5 5.1 4.0 7.6 4.4 8.9 16.9 0.8 12.4 1.4

11.8 0.6 2.1 1.3 0.4 3.3 1.9 2.2 14.6 5.1 4.0 7.5 4.2 9.1 17.5 0.8 12.3 1.4

11.6 0.5 1.9 1.2 0.3 3.3 1.9 2.1 14.7 5.2 3.9 7.3 4.1 9.3 18.0 0.8 12.3 1.4

11.4 0.5 1.8 1.1 0.3 3.3 1.9 2.0 14.9 5.2 3.9 7.2 4.0 9.5 18.6 0.8 12.2 1.4

11.3 0.5 1.7 1.1 0.3 3.3 1.8 2.0 15.0 5.3 3.8 7.1 3.9 9.7 19.1 0.7 12.2 1.4

11.1 0.5 1.6 1.0 0.3 3.3 1.8 1.9 15.1 5.3 3.8 6.9 3.8 9.8 19.6 0.7 12.1 1.3

10.9 0.5 1.5 1.0 0.3 3.3 1.8 1.9 15.2 5.3 3.8 6.8 3.7 10.0 20.2 0.7 12.0 1.3

10.8 0.4 1.4 0.9 0.3 3.2 1.8 1.9 15.3 5.4 3.7 6.6 3.5 10.2 20.8 0.7 11.8 1.3

10.6 0.4 1.3 0.9 0.3 3.2 1.7 1.8 15.4 5.4 3.7 6.5 3.4 10.3 21.3 0.7 11.7 1.3

10.5 0.4 1.2 0.8 0.2 3.2 1.7 1.8 15.6 5.4 3.7 6.3 3.3 10.5 21.9 0.7 11.6 1.3

10.3 0.4 1.1 0.8 0.2 3.2 1.7 1.7 15.7 5.4 3.6 6.2 3.2 10.6 22.4 0.6 11.4 1.3

10.2 0.4 1.1 0.7 0.2 3.2 1.7 1.7 15.8 5.5 3.6 6.0 3.1 10.7 23.0 0.6 11.2 1.3

10.0 0.4 1.0 0.7 0.2 3.2 1.6 1.6 15.9 5.5 3.6 5.9 3.0 10.9 23.5 0.6 11.1 1.3

Notes: For the descriptions of ISIC codes, refer to Table 17.1. ISIC codes 15, 16, 17, 18, 20, 22, 23, 26, and 36 are early industries. ISIC codes 21, 27, 28, and 33 are middle industries. ISIC codes 24, 25, 29, 31, and 34 are late industries. Source: Created by the authors based on INDSTAT 2012 data.

Growth elasticity –1

–1 1 2

0

–2

1

0

–2

3

2

1

0

–2 Real GDP per capita ($US)

Employment-population ratio 3

2 2

–1

3 3

2 2

–1 0

0

–1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

–1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

2

Growth elasticity

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

3

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity 3

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

3

Growth elasticity

0

Growth elasticity

1

Growth elasticity

–1 1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

–1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity –2

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    Food and beverage

Real GDP per capita ($US)

Textile

Real GDP per capita ($US)

Wood products

Real GDP per capita ($US) Real GDP per capita ($US)

Coke and refined petroleum Non-metallic mineral

–2

Labor productivity



Tobacco

1

0

–2 Real GDP per capita ($US)

Wearing apparel

1

0

–2 Real GDP per capita ($US) Printing and Publishing

1

–2

2

1

Real GDP per capita ($US)

3 Furniture, etc.

2

1

0

–2

Real GDP per capita ($US)

Value added per capita

 . Value added as a function of the labour productivity and employment growth rate—early industries

Source: Created by the authors based on INDSTAT 2012 data.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

value added per capita growth rate at an early stage of a country’s development (from US $1,000 to US$4,000 GDP per capita) is attributable to the high employment growth rate. Thus, the rapid fall of the value added per capita growth rate coincides with the precipitate decline of employment growth, although the former’s decline is eventually tempered by a gradual increase in the rate of labour productivity growth. The wearing apparel industry’s growth rate, therefore, largely depends on labour’s cost competitiveness, as the wearing apparel industry does not seem to render much room for the substitution of labour for capital.

17.3.1.2 The Middle Industries Among the four middle industries, our estimation indicates that the basic metals and fabricated metals industries have the potential to occupy a sizeable share in the manufacturing industry following the decline of the early industries, and up to around US $20,000 GDP per capita (Table 17.2). Between the two, the fabricated metals industry is a more labour intensive and employment-generating industry while the basic metals industry is capital intensive and oriented towards adding value without adding much labour (Appendix D). The two industries have a quite similar development pattern as shown in Figure 17.2, but the relationship between the labour productivity and the employment growth rate which underlie the pattern differ between the two industries. The fabricated metals industry decreases both the rate of labour productivity as well as the employment growth rate, but a slower decline in the employment growth rate Basic metals

2

2

0 –1 –2

1 0 –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

1

Growth elasticity

3

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

Paper 3

–2

Real GDP per capita ($US)

Real GDP per capita ($US)

Precision instrument

2

2

0 –1 –2

Real GDP per capita ($US) Employment-population ratio

1 0 –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

1

Growth elasticity

3

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

Fabricated metals 3

–2

Labor productivity

Real GDP per capita ($US) Value added per capita

 . Value added as a function of the labour productivity and employment growth rate—middle industries Source: Created by the authors based on INDSTAT 2012 data.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



prolongs the growth of value added per capita. The basic metals industry, in turn, tends to experience a relatively rapid decline in terms of employment growth rate; however, the fast increase in labour productivity helps sustain the growth of value added faster than the rate of the economy-wide average up to around US$20,000 (the point where the growth elasticity of value added per capita reaches one).

17.3.1.3 The Late Industries After reaching US$15,000 GDP per capita, apart from a high value added share of the food and beverages industry, the manufacturing industry of large countries is usually dominated by late industries such as the electrical machinery and apparatus, chemicals, motor vehicles, and machinery and equipment industries (Table 17.2). Among these, the electrical machinery and apparatus and the chemicals industries, in particular, are expected to become the two largest manufacturing industries at a high income level after US$30,000 GDP per capita as shown in Table 17.2. The main reason behind the rapid growth of the electrical machinery and apparatus industry is its unrivalled capacity to sustain the fast growth rate of labour productivity while the factors underlying the steady growth of the chemicals industry seem to be a slower decline of the employment growth rate, combined with a relatively high growth of labour productivity, especially during the high income stage of a country’s development. As shown in Figure 17.3, the value added of the motor vehicle industry could expand as rapidly as that of the electrical machinery and apparatus sector—the industry with the fastest growth potential—up to around US$15,000, and could thereafter continue to grow at a speed similar to that of the chemicals industry—the industry with the second fastest growth potential—until the country reaches a GDP per capita of around US$23,000. However, the growth rate of the motor vehicles industry slows down relatively quickly, and the difference in the value added of the motor vehicles and of the two fastest growing industries continue to increase as the country’s income level rises further. This slowdown of the motor vehicles industry is mainly attributable to its limited capacity to increase labour productivity at higher GDP levels relative to those of the electrical machinery and apparatus and chemicals industries. Another late and potentially major industry in terms of value added share in total MVA is the machinery and equipment industry. The growth trend of the industry’s labour productivity and employment are similar to that of the chemicals industry, but compared with the latter, the initial development of the machinery and equipment industry is much more likely to be driven by the employment growth rate than by that of productivity growth. The labour productivity growth rate starts to play an important role in sustaining the value added growth rate at a later development stage than in other late industries, but its relatively high growth rate after reaching the lowest point helps to sustain the rapid growth of value added, which is expected to eventually surpass the motor vehicle industry’s value added level. The above analysis elucidates the role of both the employment and labour productivity growth rate in the development paths of major manufacturing industries associated with the different stages of a country’s development. Each manufacturing industry has its unique combination of employment and productivity growth rate patterns,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  Rubber and plastic

2

2

1 0 –1 –2

1 0 –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

3

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

Chemical products 3

–2 Real GDP per capita ($US)

Real GDP per capita ($US)

Machinery and equipment

Electrical machinery and apparatus 2

1 0 –1 –2

1 0 –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

2

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

3

–2

Real GDP per capita ($US)

Real GDP per capita ($US)

Motor vehicles

2 1 0 –1

1,000 3,000 5,000 7,000 9,000 11,000 13,000 15,000 17,000 19,000 21,000 23,000 25,000 27,000 29,000 31,000 33,000 35,000 37,000 39,000

Growth elasticity

3

–2 Real GDP per capita ($US) Employment-population ratio

Labor productivity

Value added per capita

 . Value added as a function of the labour productivity and employment growth rate—late industries Source: Created by the authors based on INDSTAT 2012 data.

which underlie the level of the given manufacturing industry’s value added. To return to the main theme of this chapter, the following sections will focus on the employment potential of manufacturing industries based on our estimates.

17.3.2 Employment 17.3.2.1 The Early Industries Among the early industries, the major contributors to employment are the food and beverages, textiles, and wearing apparel industries.⁶ Indeed, none of the other ⁶ UNIDO’s industrial statistics usually only include those engaged in formal enterprises and, depending on reporting country, small enterprises and employment therein are excluded from the statistics.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



0.6 0.5

EP ratio

0.4 0.3 0.2 0.1

00 7,0 00 9,0 00 11 ,00 0 13 ,00 0 15 ,00 0 17 ,00 0 19 ,00 0 21 ,00 0 23 ,00 0 25 ,00 0 27 ,00 0 29 ,00 0 31 ,00 0 33 ,00 0 35 ,00 0 37 ,00 0 39 ,00 0

00

5,0

3,0

1,0

00

0

Real GDP per capita ($US) Food and beverages Wearing apparel Coke and refined petroleum

Tobacco Wood products Non-metallic minerals

Textiles Printing and publishing Furniture, n.e.c.

 . Employment growth elasticity—early industries Source: Created by the authors based on INDSTAT 2012 data.

industries reach the employment level of these three industries. As shown in Figure 17.4, both the food and beverages and the wearing apparel industries reach their highest levels of employment with EP ratios of 0.5056 and 0.5063, respectively, when the country reaches a GDP per capita of around US$12,000. The difference between the two industries is that the food and beverages industry always has a large rate of employment, while the wearing apparel industry employs a large number of workers during a relatively short period of a country’s development only. The EP ratio of the food and beverages industry is 0.1575 at a GDP per capita of US$1,000. The EP ratio gradually rises, and the level of employment becomes three times larger at the highest level to subsequently slowly decline, yet still maintaining an EP ratio of 0.3159, that is, 62 per cent of its highest level at a GDP per capita of US$40,000. In contrast, the wearing apparel industry starts with a very small EP ratio of 0.0146, and the level of employment rapidly increases 35 times before it reaches its highest level at 0.5063—a level other industries do not usually reach. Reaching this level at a GDP per capita of US$12,000, the level of employment of the wearing apparel industry declines rapidly and shrinks by 30 per cent by the time the country reaches a GDP per capita of US$20,000. As for the food and beverages industry, which also experiences its highest level of employment at US$13,000, the same scale of employment reduction only occurs when a country’s GDP per capita rises to around US$37,000. This indicates that the wearing apparel industry could experience a substantial employment growth rate at a low income stage, but once a country loses its comparative advantage at a GDP

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

per capita level of around US$8,000, the industry’s employment level reaches its peak and then rapidly declines. The textiles industry, another major contributor to total manufacturing employment at an early stage of a country’s development, combines the trends of both the food and beverages and the wearing apparel industries. From a very early stage of a country’s development, the textiles industry, like the food and beverages industry, employs a relatively large number of workers, as the EP ratio of 0.2605 at a GDP per capita of US $1,000 indicates (Appendix C). Subsequently, the level of employment increases slowly and reaches its peak with an EP ratio of 0.3676 at a GDP per capita level of around US $5,000. After reaching this peak level, the industry, to a large extent, follows the trend of the wearing apparel industry. The level of employment declines rapidly, and by the time the country reaches a GDP per capita level of US$13,000, the level of employment reduces by 30 per cent compared with the peak level. Besides the three above-mentioned industries, the non-metallic minerals industry could, as an early industry, make a sizeable contribution to the level of employment. It starts with an EP ratio of 0.0635 at a GDP per capita level of US$1,000, about one quarter of the initial textile industry’s level of employment. At that stage of a country’s development, the non-metallic minerals industry is the third largest employment source after the food and beverages and the textiles industries. From this initial level, the level of employment increases three-fold to reach its peak with an EP ratio of 0.1768 at a GDP per capita of US$9,000. By then, the country has developed other industries to increase manufacturing employment, and the non-metallic mineral industry is no longer the third but eighth out of eighteen industries in terms of employment size. As is the case for the food and beverages industry, the employment size of the nonmetallic minerals industry also only declines gradually. Its employment size therefore surpasses that of the textiles and wearing apparel industries at a GDP per capita of US $22,000 and US$35,000, respectively, as the latter two industries’ employment size reduces much faster than the non-metallic minerals industry’s. This industry produces bricks, cement, and glass, which are primarily used for construction. Compared with many other industries, yet similar to the food and beverages industry, the non-metallic minerals industry is more domestically oriented and serves demands that are relatively income inelastic. These factors seem to contribute to the stability of the industry’s employment size. Other early industries employ a small number of workers. At a GDP per capita of US $19,000, the printing and publishing industry reaches its highest level of employment with an EP ratio of 0.0996, which is around one-fifth and only 60 per cent of the highest level of employment for the wearing apparel and non-metallic minerals industries, respectively. The unique feature of the printing and publishing industry, however, is the stability of its employment. The industry only doubles its employment size from its initial and lowest size at a GDP per capita level of US$1,000 by undergoing a long period of development, reaching the highest employment level at a GDP per capita of US$19,000. Thereafter, the employment level decreases very slowly and at a GDP per capita of US$40,000 sustains 80 per cent of its highest level of employment.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



The printing and publishing industry shares the characteristics of the non-metallic minerals industry, but the former is likely to be even more domestically oriented and income inelastic, making it the most stable manufacturing industry in terms of employment change. The coke and refined petroleum industry employs the smallest number of workers in the early and in fact among all manufacturing industries. The industry’s employment size is only 4 per cent of that of the wearing apparel industry when comparing the two industries’ highest levels of employment. The coke and refined petroleum industry is a very capital intensive industry. The industry’s production process only requires a small number of workers relative to the output it produces. For example, at a GDP per capita of US$28,000, both the coke and refined petroleum and the textiles industries produce about the same amount of value added per capita, but the employment size of the former is only 14 per cent of the latter’s. The coke and petroleum industry could make a substantial contribution to the economy in terms of value added when it reaches its peak, which is comparable to that of other early industries such as the textiles and wearing apparel industries, but in terms of employment, its contribution is very limited (Appendix B). The tobacco industry, which is the second lowest employment source after the coke and refined petroleum industry, reaches its highest level of employment when a country’s GDP per capita is only US$2,000. The difference between the two industries is that the tobacco industry’s contribution is limited in terms of both value added and employment, while only employment is noticeably low for the coke and refined petroleum industry. Finally, the wood products industry is the smallest industry based on estimations of the peak value added per capita a manufacturing industry can reach. However, due to its relatively labour intensive production, the level of employment is approximately two and three times larger, respectively, than the highest levels of employment of the tobacco and the coke and refined petroleum industries. Although the wood products industry’s contribution to both value added and employment are relatively small, they are stable throughout a country’s development. For example, due to the rapid decline of the level of employment in the textiles industry, the wood products industry’s level of employment may actually exceed that of the textile industry.

17.3.2.2 The Middle Industries The industries in the middle group are defined here as those whose share of valued added in total manufacturing reaches its highest level when a country’s GDP per capita is between US$5,000 and US$20,000. The paper, basic metals, fabricated metals, and precision instruments industries fall into this category (Figure 17.5). Among the four industries, the fabricated metals industry makes the largest contribution to a country’s level of employment. At a low income level of US$1,000 GDP per capita, the industry already employs a relatively large number of workers with an EP ratio of 0.0533, the largest among the middle and late industries and the fifth largest among all eighteen manufacturing industries at that income level. The level of

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  0.6 0.5

EP ratio

0.4 0.3 0.2 0.1

00 7,0 00 9,0 0 11 0 ,00 0 13 ,00 0 15 ,00 17 0 ,00 19 0 ,00 0 21 ,00 23 0 ,00 0 25 ,00 0 27 ,00 0 29 ,00 0 31 ,00 33 0 ,00 0 35 ,00 37 0 ,00 0 39 ,00 0

00

5,0

3,0

1,0

00

0.0

Real GDP per capita ($US) Paper Fabricated metals

Basic metals Precision instruments

 . Employment growth elasticity—middle industries Source: Created by the authors based on INDSTAT 2012 data.

employment continues to rise until the country reaches a GDP per capita of around US $19,000. At its peak with an EP ratio of 0.2164, the level of employment of the fabricated metals industry is around half of that of the highest level reached by the food and beverages or the wearing apparel industries, or around 20 per cent higher than the peak level of employment of the non-metallic minerals industry. Although a comparison of the peak levels of employment among manufacturing industries places the fabricated metals industry sixth in terms of employment size, the industry’s weight in total manufacturing employment increases as a country’s income rises because of the relative stability of the fabricated metals industry’s level of employment and the rapid decline in the employment levels of other industries (e.g. textiles and wearing apparel) which employ more workers than the fabricated metals industry. Consequently, as a country’s GDP per capita moves beyond US$25,000, the fabricated metals industry becomes the second largest employer after the food and beverages industry. The basic metals industry is the next largest source of employment among the middle industries. At its peak, the industry employs around 65 per cent and a quarter of the peak levels of employment of the fabricated metals and the wearing apparel industries, respectively. The industry reaches its highest level of employment at a relatively early stage of a country’s development at a GDP per capita of US$9,000, the earliest among the middle and late industries. As the country’s income level increases further, the level of employment of the basic metals industry shrinks rapidly down to half of the highest employment level at a GDP per capita of US$24,000, and to one-third at a GDP per

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



capita of US$31,000. The main reason why the basic metals industry has a lower level of employment than the fabricated metals industry—another metals-related industry—is the capital intensiveness of the former industry relative to the latter one. To produce the same amount of value added, the basic metals industry only needs around half of the workers required for the fabricated metals industry. During the course of a country’s development, the basic metals industry can be twice the size of the fabricated metals industry in terms of value added, but the former industry reaches only half of the latter’s level of employment when comparing their peak levels. The paper industry has limited potential to absorb a country’s labour force. At its peak, the industry employs slightly more workers than the wood products industry or around one-sixth of the wearing apparel industry, which is the largest industry in terms of level of employment. Although the paper industry’s output could grow ten times larger than the highest output level the wood products industry could ever attain, in terms of employment growth the paper industry follows a pattern similar to that of the wood products industry, which is slow in both periods of employment growth and reduction. Due to the paper industry’s capital intensive production process, it can achieve a much higher level of output than the wood products industry; however, the two industries generally face similar demand characteristics which are domestically oriented and income inelastic, making the employment trends of both industries relatively stable. The precision instruments industry is the smallest among the middle industries and the third smallest among all manufacturing industries after the coke and petroleum and the tobacco industries, when comparing the highest level of employment they can reach throughout the course of a country’s development. The precision instruments industry starts with a very small level of employment at a GDP per capita of US$1,000, far smaller than any other industry, and increases the level of employment relatively rapidly for a long period of a country’s development, reaching its highest level of employment at a late development stage only, namely at US$24,000. The industry seems to be a skills and technology intensive industry, which could keep output and employment growing even after a country’s income rises to a fairly high level, although both of them are very small in terms of share in total manufacturing due to the specialized nature of the precision instruments industry.

17.3.2.3 The Late Industries The industries in the late group reach their highest value added shares in total manufacturing after a country’s GDP per capita surpasses US$20,000. There are no industries in this category which have a very high level of employment comparable to that of some of the early industries, such as the wearing apparel, food and beverages and textiles industries. However, they cannot be considered small either because at their peaks they employ around one-third to half of the peak level of employment of the wearing apparel industry which attains the highest level of employment among all manufacturing industries.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  0.6 0.5

EP ratio

0.4 0.3 0.2 0.1

1,0

00 3,0 00 5,0 00 7,0 00 9,0 0 11 0 ,00 0 13 ,00 0 15 ,00 0 17 ,00 0 19 ,00 0 21 ,00 0 23 ,00 25 0 ,00 27 0 ,00 0 29 ,00 0 31 ,00 0 33 ,00 0 35 ,00 0 37 ,00 0 39 ,00 0

0.0

Real GDP per capita ($US) Chemicals Rubber and plastic Electrical machinery and apparatus

Machinery and equipment Motor vehicles

 . Employment level (EP ratio)—late industries Source: Created by the authors based on INDSTAT 2012 data.

Among the late industries, the electrical machinery and apparatus industry is expected to make the largest contribution to a country’s level of employment and be the fourth largest employment source of all manufacturing industries, after the wearing apparel, food and beverages, and textiles industries (Appendix C). The industry’s level of employment at an early stage of a country’s development is low and does not expand much until a country’s GDP per capita rises to around US$3,000 (Figure 17.6). Subsequently, employment expands rapidly, reaching the highest level at a GDP per capita of US$16,000. At its peak, the industry’s level of employment represents around half of the peak level that the wearing apparel and food and beverages industries can reach in their developments. Although the output (in terms of value added) of the electrical machinery and apparatus industry continues to grow much faster than the economy’s average, even at a high income stage of a country’s development, the level of employment decreases rather quickly, more rapidly than the levels of other late industries (Appendix B). The motor vehicle industry reaches its peak level of employment at the same GDP per capita level as the electrical machinery and apparatus industry, namely US$16,000, with a slightly lower level of employment. The difference between the two industries is that employment in the motor vehicle industry is more stable and changes more slowly than the electrical machinery and apparatus industry. Therefore, even though the electrical machinery and apparatus industry could reach a higher level of employment than the motor vehicle industry, the latter is likely to have a higher level of employment than

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



the electrical machinery and apparatus industry at income levels lower than US$10,000 and higher than US$25,000 GDP per capita. The machinery and equipment industry employs 12 per cent and 17 per cent less workers than the electrical machinery and apparatus and the motor vehicle industries, respectively, based on a comparison of their highest levels of employment (Appendix C). After the peak level of employment is reached in the machinery and equipment industry, employment decreases even more slowly than that of the motor vehicle industry. At a GDP per capita income of US$40,000, the level of employment of the equipment and machinery industry will be 47 per cent lower than at its highest level, while the level of employment of the motor vehicle and electrical machinery and apparatus industries drops by 60 per cent and 70 per cent, respectively, after reaching their peaks. In terms of employment, the chemicals industry and rubber and plastics industry are the two smallest of the five late industries. Among all manufacturing industries, however, their levels of employment are medium sized, ranking 8th and 9th out of eighteen industries. The chemicals industry employs slightly more workers than the rubber and plastics industries at their highest levels. After reaching the highest level of employment at a GDP per capita of US$11,000, earlier than other late industries, the chemicals industry gradually reduces its level of employment. The chemicals industry differs from other late industries on account of its high employment size at a relatively early stage of a country’s development. At a GDP per capita of between US$6,000 and US$7,000, the industry’s level of employment becomes the fourth highest after that of the food and beverages, textiles, and wearing apparel industries. Finally, the rubber and plastics industry is notable for its contribution to a country’s level of employment at very high income levels. After reaching its peak level of employment at US$24,000, employment decreases very slowly and remains at around 90 per cent of its highest level at US$40,000. Thus, even though other late industries may employ more workers than the rubber and plastics industry at its peak level, the rubber and plastics industry becomes the largest employment source among the late industries at higher income levels, and the third largest among all manufacturing industries after the food and beverages and fabricated metals industries. In terms of value added, the rubber and plastics industry is not likely to grow as large as other late industries. Thus, the rubber and plastics industry is the most labour intensive late industry, similar to the fabricated metals industry in the middle industry group (Appendix D).

17.4 A

.................................................................................................................................. Throughout a country’s development, the food and beverages industry is a major source of employment. Except for a brief period of a country’s early development stage in which the industry’s level of employment maybe surpassed by that of the textiles and wearing apparel industries, the food and beverages industry usually

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

employs the largest number of workers at all income levels. As noted above, at a relatively early stage of development, that is, less than US$10,000 GDP per capita, the wearing apparel and textiles industries together with the food and beverages industry constitute the three largest contributors to a country’s manufacturing employment. Other manufacturing industries usually do not reach the level of employment of any of these three industries during the course of their development. Therefore, from a very early stage of a country’s development up to a GDP per capita of around US$10,000, a country may experience a rapid increase in manufacturing employment, as the three largest employment sources in manufacturing industry develop during that income range. Starting from a very low income level, the textiles and food and beverages industries employ a relatively large number of workers. These are industries related to basic needs, that is, even before industrialization takes off; for example, at a GDP per capita of US $1,000 or less, a country usually has a non-negligible number of firms in these industries, employing more workers than other manufacturing industries. While the number of workers in the food and beverages and textiles industries increases 3.2 times and 40 per cent, respectively, from their initial levels at a GDP per capita of US$1,000 by the time they reach their peak levels, the wearing apparel industry could increase the number of employees 35 times, thus surpassing the level of employment of the food and beverages sector at its peak to become the largest industry in terms of employment. This development pattern of large countries indicates that the wearing apparel industry is the first major industry which basically needs to be developed from scratch (at least for formal enterprises) in order to increase manufacturing employment and to absorb surplus labour from agriculture to put a country on a steady path of industrialization. Considering the sizeable contribution of the wearing apparel industry to a country’s value added, whether a country’s wearing apparel industry can experience rapid and steady growth is likely to have a significant influence on the country’s economic and social progress at a relatively early stage of development. With regard to the characteristics of employment trends described in Section 17.3, the level of employment in both the textiles and wearing apparel industries declines rapidly after their highest level of employment has been reached. As Appendix B indicates, the end of employment growth in the three major employment sources of a country’s early development occurs when a country’s per capita income rises to around US$12,000, when both the wearing apparel and food and beverages industries reach their highest levels of employment.⁷ In the case of the wearing apparel industry, the beginning of the decline in the level of employment roughly coincides with the decline of industry size in terms of value added, because there is little potential to substitute capital for labour. In contrast, there is some room in the textiles industry for sustaining the rapid growth of the value added through factor substitution, which may take place when the employment increase comes to an end, that is, around a GDP per capita of US$5,000 to a GDP per capita of US$9,000 (Appendix B). ⁷ The textiles industry tends to reach its highest level of employment at a much lower income level, at approximately US$5,000.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



The increase and decline of the food and beverages, textiles, and wearing apparel industries in terms of employment and value added denote that the middle and late industries have to be ready to assume the role of the engine of manufacturing growth before the early industries begin to decline, which may occur around US$12,000 for employment and at a somewhat higher GDP level for value added per capita. Due to the rapid decline in particular of the level of employment in the early industries after reaching its peak, the steady growth of the middle and late industries before the level of employment begins to decline in the early industries is essential for maintaining the momentum of industrialization. Because there are no middle and late industries which are comparable to the wearing apparel and textiles industries among the early industries in terms of labour absorption capacity, a country needs at least four or five middle and late industries to develop continually in order to compensate for the decline of the major employment sources of the early industries. For example, if a country can follow the typical development patterns of large countries as identified in this study, the combined level of employment of the four middle and late industries—fabricated metals, machinery and equipment, electrical machinery and apparatus, and motor vehicle industries—could surpass total employment from the wearing apparel, electrical machinery and apparatus, and textiles industries at a GDP per capita of around US$11,000. Which industries a country should pay particular attention to and possibly support to facilitate growth depends on the balance the country aims to strike between economic and social progress and its demographic and geographic conditions, which have positive or negative effects on certain industries’ growth (Haraguchi and Rezonja 2011). Based on this study’s results, the following Section 17.4.1 suggests possible paths of manufacturing development in favour of employment creation and economic growth.

17.4.1 Pro-employment Path of Manufacturing Development Ideally, rapid economic growth leads to broad-based employment generation, raising the level of employment in all industries to create quantitatively high and qualitatively diverse jobs, most likely not proportionally across industries, but in accordance with a country’s shifting endowment structures. This will open job opportunities for a country’s labour force, requiring different skills and experiences which correspond to the given stage of development to reduce structural unemployment. However, countries that face difficulties reducing high unemployment or that need to create more productive jobs (e.g. the majority of manufacturing activities), may have to pursue a more strategic approach to job creation. Most countries fall into this category and we address them in this section. Three conditions need to be considered for illustrating a pro-employment path of development—growth potential, labour intensiveness, and patterns of structural change within manufacturing industries. On the one hand, if we only look at industries’

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

labour intensiveness in terms of job creation without giving due consideration to the growth potentials of these industries, we might erroneously direct a country to pursue industries which use relatively more labour, but might have limited growth potential. These industries might have high labour intensity in terms of production, but their absolute volume of employment may remain small if the industry size remains small. On the other hand, the consequence of the reverse situation, that is, only taking growth potential into consideration, is apparent. It is equivalent to the situation that no consideration is given to pro-employment manufacturing development. Finally, neither industries’ growth potential nor their labour intensiveness is static. They change as a country develops and its comparative advantages shift. When a country increases its per capita income, it may be futile to attempt sustaining the growth of early industries, although they may be more labour intensive. Moreover, within an industry, labour intensiveness changes because a factor substitution of capital with labour might occur as a country develops. Hence, as both the structure within the manufacturing industry and the structure of production within industries changes, the relative importance of industries for job creation shifts from one industry to another. At a very early stage of a country’s development, the food and beverages and textiles industries should be the two major sources of manufacturing employment as well as of value added, because their activities relate to the basic needs of citizens. At an income level of less than US$2,000, a limited number of formal enterprises will be engaged in other industries. At a very low income level, there are two other industries that could substantially contribute to employment, namely the chemicals and the non-metallic minerals industry. In the early development of a country, the chemicals industry produces basic materials, such as soap and essential supplies, including fertilizers, required by the dominant sector of the early economy, agriculture. The non-metallic industry also develops from an early stage of development because it produces building materials, such as glass, cement, and bricks, which face a certain level of demand regardless of income levels. The chemicals industry’s labour intensity is not high, but due to the high output level from an early stage of development, the number of workers employed in the industry is relatively large. These four industries could support the expansion of manufacturing employment at a very low income level, providing jobs of higher productivity than those in the agrarian sector. There is a relatively high domestic demand for these industries, as they relate to citizens’ basic needs; therefore, countries seek to meet such demands through domestic production without too much dependence on imports. The food and beverages and the chemicals industries, in particular, could grow in terms of both employment and value added over longer periods of development. Building a solid foundation for these industries early on is important for a country’s sustainable development. The four industries mentioned above are, in a sense, ‘original’ industries, because they have commonly always existed in most countries to meet citizens’ basic needs, even before industrialization took place, although their production processes and products most likely differed from modern ones. Therefore, the first challenge for countries aiming to follow the pro-employment path of manufacturing development is to diversify manufacturing from the four ‘original’ industries and establish a new

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



labour intensive industry while promoting the further growth and development of those initial industries. From this perspective, the successful establishment of the wearing apparel industry is crucial and demonstrates whether a country can actually create a vibrant manufacturing industry which absorbs a substantial number of workers. The wearing apparel industry (including fur, leather products, and footwear in our data) is not part of the ‘original’ industries and must therefore be developed by the country, but, if successful, has the potential for tremendous growth in terms of value added and employment in a relatively short period of time. The development of this industry might therefore represent a first experience of how industrialization can make a difference in a country’s development due to its capacity to employ a large number of workers in higher productive activities than those required by the agricultural and subsistence sectors, which could consequently lead to a reduction in poverty. Until around US$4,000, the rapid growth of the level of employment of both the wearing apparel and the food and beverages industries can contribute to the increase in manufacturing employment. From US$4,000 to US$9,000, the continued rapid employment generation in the wearing apparel industry is likely to be the single most important source of labour absorption in highly productive manufacturing jobs. At their peak, the wearing apparel or the food and beverages industry has the potential to employ at least twice as many workers as the highest level of employment of any other manufacturing industry, except for the textiles industry which, at its peak, can reach two-thirds of the highest employment levels of the two industries. For low income countries, successful development of the new wearing apparel industry could provide an initial boost at the start of the industrialization process, providing productive and higher-income jobs to a large number of people and, hence, increasing personal income, firm profits. and government revenue for investment in better education, training, and infrastructure. In this sense, the continual development of the wearing apparel industry, including leather and footwear production, is important not only for the country’s industry on the whole, but also in terms of initial institutional, policy, and managerial learning from this early industrialization experience, and laying the foundation for continuous industrial upgrading through increasing investment in human and capital resources. For most countries, the rapid growth of the wearing apparel industry in a country’s early stage of development could become the first test of whether the process of industrialization can take off and achieve continual progress. As Appendix B indicates, one unique characteristic of the wearing apparel industry is the limited possibility for the substitution of production factors. In other words, the technology in the wearing apparel industry will always be labour intensive, offering little opportunity for the substitution of capital with labour. Therefore, when the increase in wage level makes the industry uncompetitive, the decline of the industry’s value added soon follows because the growth cannot be sustained by using more capital in place of labour. This is the only industry in which the decline of employment nearly coincides with that of value added per capita. Even though the wearing apparel industry has the potential to become a major source of manufacturing employment, especially at a relatively low income level, there

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

are two reasons why countries must prepare alternative sources of employment long before the wearing apparel industry begins to decline. First, the speed of the industry’s employment slowdown and decline tends to be faster than that of any other industry. The industry maintains a fast growth rate of employment until it nearly reaches its peak. Therefore, once the industry’s employment generation starts slowing down, it does so quickly and, after the peak level of employment has been reached, it declines rapidly. It would be too late to start developing alternative manufacturing sources once the wearing apparel industry starts showing signs of a slowdown in employment growth. Second, other industries with higher labour productivity and, hence, wage levels, could become larger than the wearing apparel industry in terms of value added starting at around US$7,000 GDP per capita, even though the level of employment in the wearing apparel industry will continue to be one of the largest up to a fairly high income level. Thus, to attain personal income growth and sustained economic development, countries need to increase the shares of employment in higher productivity industries, even when the level of employment of the wearing apparel industry is still rapidly increasing. Due to the impending slowdown of the labour intensive industries and the possibility of providing higher value-adding jobs, the middle and late industries need to play an increasingly important role as alternative employment sources from around US $7,000 onward. As discussed above, it is estimated that no industry in the middle and late industries will generate as many jobs as the food and beverages, textiles, or wearing apparel industries at any income level. Therefore, to maintain the level of manufacturing employment or to slow the rate of its decline, several middle and late industries have to develop simultaneously and start providing alternative and higher paying jobs to replace declining labour intensive industries. From US$7,000 to US$16,000, the motor vehicle and electrical machinery and apparatus industries may be likely to provide more job opportunities than other middle and late industries and could compensate for the reduction in employment of the early industries, which, taken as a whole, would start from around US$11,000 (Appendix C). From around US$16,000, the level of employment of the motor vehicle and electrical machinery and apparatus industries reduces not due to output decline, but due to the substitution of capital for labour. Of the two industries, the electrical machinery and apparatus industry offers greater opportunities for factor substitution, as seen in the continued rapid growth of value added up to a very high income level, despite the reduction in employment (Appendix B). Two other industries, the fabricated metals and machinery and equipment industries could emerge as major employment sources at a relatively late stage of a country’s development. Our results indicate that they would reach their highest levels of employment at around US$20,000 GDP per capita after the level of employment of the electrical machinery and apparatus and motor vehicles industries begins to decline, hence contributing to the sustainment of manufacturing employment. Finally, the rubber and plastics industry could be an important source of manufacturing employment at a very high income level, that is, over US$30,000. Although the industry’s employment level would probably not reach the levels of employment of the electrical

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



machinery and apparatus, motor vehicles, and fabricated metals industries, the significance of the rubber and plastics industry as a source of manufacturing employment is likely to increase due to its relatively high labour intensive production and very slow decline of employment during the high income stage when most of the manufacturing industries’ level of employment is rapidly decreasing (Appendix D and Appendix B).

17.5 C

.................................................................................................................................. Based on the fundamental identity discussed in the introduction, the chapter first illustrated the relationships between the growth of value added, employment, and labour productivity at the sub-sector level in the manufacturing sector using our estimation results. The growth of employment and productivity play different roles in the rise and fall of value added across manufacturing industries and throughout the stages of development within an industry. Although the value added per capita of manufacturing industries usually follows a pattern of rapid increase, slowdown, and decline, the unique characteristics of each industry are revealed in terms of underlying changes in the growth rates of and relationships between employment and labour productivity. After setting employment growth within the context of its relationship with value added and labour productivity growth, this study focused on the patterns of employment changes and, based on these patterns, delineated the pro-employment path of manufacturing development of large countries taking the changing structures of industries into consideration. Our analysis highlights three critical developments that need to take place during manufacturing development in order to create manufacturing jobs, increase the wage levels, and sustain the manufacturing employment or slow its pace of decline. The initial challenge is to develop the wearing apparel industry (including fur and leather products and footwear), which has great potential to increase manufacturing employment at an early stage of a country’s development, but is often not part of the ‘original’ industries which have existed since the pre-modern period. A successful development of the wearing apparel industry along with the modernization of the food and beverages, textiles, and non-metallic minerals industries would help absorb surplus labour from the agricultural and subsistence sectors and contribute to poverty reduction. The next challenge a country must tackle is the development of multiple middle and late industries to offset the decline of employment in labour intensive industries from around US$7,000 GDP per capita onwards. A middle or late industry with the largest employment potential can only employ around half of the workers that either the wearing apparel or food and beverages industries can. Thus, in terms of the number of jobs, the decline in employment of the early industries has to be offset by the concurrent development of several middle and late industries. From the viewpoint of welfare, this shift in manufacturing employment from labour intensive to more capital or skill intensive industries would help increase the country’s income level and

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

economic growth due to the rise in domestic demand. Finally, at a very high income level (i.e. above around US$25,000 GDP per capita), when most of the manufacturing industries’ level of employment declines, a country needs to know which industries are likely to maintain the given level of employment or slow the pace of employment decline. The production processes inherent in the food and beverages, rubber and plastics, fabricated metals, machinery and equipment, motor vehicle, and chemicals industries seem to provide countries with the possibility of retaining employment relatively longer than other industries. To realize this, each of these industries must undergo structural changes to sustain their expansion, which partially compensates for the reduction in labour arising from the increasing substitution of labour with capital.

 17A

..................................................................................................................................

R R Table 17A.1 Value added per capita ISIC codea 15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

GDPpc

(GDPpc)2

(GDPpc)3

Constant

N

R2 (overall)

3.41 2.34 34.00*** 24.02** 11.37 5.53 3.56 15.32** 3.61 4.83 14.79** 31.54*** 41.19*** 20.40** 8.02 26.12*** 45.21*** 21.58**

0.81 0.20 4.46*** 1.83 1.70** 1.02* 0.06 2.18** 0.00 0.14 1.18* 4.04*** 5.11*** 2.56** 0.44 3.45*** 5.49*** 2.06*

0.04* 0.02 0.19*** 0.04 0.08** 0.05** 0.01 0.09*** 0.01 0.00 0.03 0.16*** 0.20*** 0.10** 0.01 0.14*** 0.21*** 0.07

0.88 18.66 83.60*** 93.83*** 22.30 3.77 23.76 31.72 22.75 26.94 57.10*** 77.91*** 106.88*** 50.16* 39.48* 59.75** 119.47*** 74.02**

835 726 863 760 787 789 763 574 849 818 837 682 804 783 828 538 794 661

0.84 0.59 0.69 0.65 0.64 0.91 0.84 0.70 0.88 0.86 0.87 0.84 0.87 0.82 0.84 0.79 0.84 0.80

Notes: *p < 0.10; **p < 0.05; ***p < 0.01. ISIC descriptions are as follows: 15—Food and beverages, 16—Tobacco, 17—Textiles, 18—Wearing apparel, fur &leather products, and footwear, 20—Wood products, 21—Paper, 22—Printing and publishing, 23—Coke and refined petroleum, 24—Chemicals, 25—Rubber and plastics, 26—Non-metallic minerals, 27—Basic metals, 28—Fabricated metals, 29—Machinery and equipment n.e.c. & office, accounting, computing machinery, 31—Electrical machinery and apparatus & radio, television, and communication equipment, 33—Precision instruments, 34—Motor vehicles, trailers, semi-trailers, & other transport equipment, 36—Furniture, n.e.c. Source: Calculations by the authors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



Table 17A.2 Labour productivity ISIC code 15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

GDPpc

(GDPpc)2

(GDPpc)3

Constant

N

R2 (overall)

10.39* 2.91 1.98 4.57 15.19* 13.41** 32.90*** 21.20* 22.18*** 7.01 28.68*** 11.61 19.13** 30.02*** 66.89*** 23.41* 4.99 3.58

1.00 0.42 0.02 0.44 2.04** 1.40* 3.32*** 2.64* 2.35*** 0.74 3.04*** 1.20 2.62*** 3.21*** 7.54*** 2.37* 0.68 0.32

0.04 0.01 0.00 0.01 0.09** 0.05* 0.11*** 0.10* 0.09*** 0.03 0.11*** 0.05 0.11*** 0.12*** 0.29*** 0.08 0.02 0.01

28.39* 12.55 6.79 22.93 43.28* 36.04* 100.00*** 62.16* 63.10*** 15.25 83.03*** 31.68 50.52** 86.56*** 191.46*** 68.82* 16.87 21.19

830 720 858 758 785 782 750 572 846 815 832 682 808 766 813 541 794 653

0.55 0.47 0.54 0.10 0.25 0.54 0.45 0.46 0.65 0.49 0.61 0.58 0.48 0.37 0.52 0.25 0.41 0.03

Notes: *p < 0.10; **p < 0.05; ***p e  -0.5, -0.5 > e -1, and less than -1, respectively. The numbers in the table show the highest actual level (not growth rate) of the corresponding variables that industries are estimated to reach during the course of a large country’s development from US $1,000 to US$40,000 GDP per capita. Source: Created by the authors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

25 25 25 26 26 26 27 27 27 28 28 28 29 29 29 31 31 31 33 33 33 34 34 34 36 36 36

ISIC code

Variable 21,000 22,000 23,000 24,000 25,000 26,000 27,000 28,000 29,000 30,000 31,000 32,000 33,000 34,000 35,000 36,000 37,000 38,000 39,000 40,000

15 15 15 16 16 16 17 17 17 18 18 18 20 20 20 21 21 21 22 22 22 23 23 23 24 24 24 25 25

VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP

ΔΔ  ΔΔ   ++   +   ΔΔ    ΔΔ  + ΔΔ  ΔΔ ΔΔ  + +  ++ + Δ

ΔΔ  ΔΔ   ++   ΔΔ   ΔΔ    ΔΔ  + ΔΔ  ΔΔ Δ  + +  ++ + Δ

ΔΔ  +   ++   ΔΔ   ΔΔ    ΔΔ  + ΔΔ  ΔΔ Δ  + +  ++ + 0.1807

ΔΔ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  + +  ++ + 

ΔΔ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  + +  ++ + 

ΔΔ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  + +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  + +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  + +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  ΔΔ +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  ΔΔ +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ Δ  ΔΔ +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  + Δ  ΔΔ 53.67  ΔΔ +  ++ + 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ +  ++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ +  ++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ +  ++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ ΔΔ  +++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ ΔΔ  +++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    ΔΔ  ++ Δ  ΔΔ   ΔΔ ΔΔ  +++ ΔΔ 

Δ  +   ++   ΔΔ   ΔΔ    Δ  ++ Δ  ΔΔ   ΔΔ ΔΔ  +++ ΔΔ 

318.45  +   ++   ΔΔ   ΔΔ    100.32  ++ 51.66  ΔΔ   ΔΔ 502.98  +++ 174.52 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

Table 17B.2 Changes in the growth rate of value added per capita, employment–population ratio, and labour productivity (from US$21,000 to US$40,000 GDP per capita)

LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP VAPC EP LP

+ ΔΔ  + ΔΔ  ++ ΔΔ  Δ ++  + ++  +++ ΔΔ Δ ΔΔ +  ++ ΔΔ  ΔΔ

+ ΔΔ  + ΔΔ  ++ ΔΔ  Δ ++  + ++  +++ ΔΔ Δ ΔΔ +  ++ ΔΔ  ΔΔ

+ ΔΔ  + ΔΔ  ++ ΔΔ   ++  + ++  +++ ΔΔ Δ ΔΔ +  ++ ΔΔ  ΔΔ

+ ΔΔ  + ΔΔ  ++ Δ   ++  + ++  +++ ΔΔ 0.0559 ΔΔ +  ++ ΔΔ  ΔΔ

+ ΔΔ  + ΔΔ  ++ Δ   ++  + ++  +++ Δ  ΔΔ +  ++ ΔΔ  ΔΔ

+ ΔΔ  + ΔΔ  ++ Δ   ++  + ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ ΔΔ  + Δ  ++ Δ   ++  + ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ ΔΔ  + Δ  ++ Δ   +  + ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ ΔΔ  + Δ  ++ 105.18   +  + ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ ΔΔ  + Δ  ++    +  ++ ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ Δ  + Δ  ++    +  ++ ++  +++ Δ  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ Δ  + Δ  ++    +  ++ ++  +++ 20.11  ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ Δ  + Δ  ++    +  ++ ++  +++   ΔΔ ΔΔ  ++ ΔΔ  ΔΔ

+ Δ  ++ 190.73  ++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ Δ  ++   +++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ Δ  ++   +++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ Δ  ++   +++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ Δ  ++   +++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ Δ  ++   +++    +  ++ ++  +++   ΔΔ Δ  ++ ΔΔ  ΔΔ

+ 113.84  ++   +++    344.43  ++ 746.23  +++   ΔΔ 350.77  ++ 42.2  ΔΔ

Notes: ‘+’ indicates that growth elasticity is greater than 1, and +, ++, and +++ correspond to elasticity (e), 1  e < 1.5, 1.5  e < 2, and are greater than 2, respectively. ‘Δ’ means that the elasticity is between 0 and 1, and Δ and ΔΔ signifies that the elasticities lie within the range of 0  e < 0.5 and 0.5  e < 1, respectively. ‘-‘indicates the range in which the elasticity is negative (negative growth). ,, and  denote that the elasticities are 0 > e  0.5, 0.5 > e  1, and less than 1, respectively. The numbers in the table show the highest actual level (not growth rate) of the corresponding variables that industries are estimated to reach during the course of a large country’s development from US$1,000 to US$40,000 GDP per capita. Source: Created by the authors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

25 26 26 26 27 27 27 28 28 28 29 29 29 31 31 31 33 33 33 34 34 34 36 36 36

A C

.............................................................................................................................................................................................................................

Table 17C.1 Changes in the level of employment (EP ratio) (US$1,000 to US$20,000 GDP per capita) ISIC code 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 9,000 10,000 11,000 12,000 13,000 14,000 15,000 16,000 17,000 18,000 19,000 20,000 15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

0.158 0.029 0.261 0.015 0.031 0.022 0.046 0.013 0.061 0.015 0.064 0.034 0.053 0.049 0.029 0.002 0.048 0.018

0.241 0.034 0.298 0.042 0.055 0.031 0.046 0.014 0.093 0.035 0.095 0.060 0.068 0.040 0.033 0.003 0.054 0.027

0.308 0.034 0.336 0.091 0.067 0.041 0.052 0.016 0.118 0.054 0.121 0.084 0.087 0.051 0.050 0.005 0.074 0.040

0.360 0.032 0.358 0.155 0.073 0.050 0.059 0.018 0.138 0.072 0.140 0.104 0.105 0.066 0.072 0.008 0.097 0.055

0.401 0.030 0.368 0.227 0.076 0.058 0.065 0.019 0.153 0.088 0.154 0.119 0.122 0.083 0.097 0.011 0.121 0.071

0.432 0.027 0.367 0.298 0.077 0.065 0.071 0.020 0.164 0.102 0.164 0.129 0.138 0.101 0.124 0.015 0.144 0.086

0.456 0.025 0.359 0.362 0.077 0.070 0.076 0.021 0.172 0.114 0.170 0.136 0.152 0.118 0.149 0.019 0.165 0.100

0.474 0.023 0.347 0.416 0.077 0.075 0.081 0.021 0.178 0.125 0.174 0.140 0.164 0.135 0.173 0.023 0.184 0.113

0.487 0.021 0.331 0.457 0.076 0.078 0.085 0.021 0.181 0.134 0.176 0.141 0.174 0.150 0.195 0.027 0.201 0.125

0.496 0.019 0.313 0.485 0.074 0.080 0.088 0.021 0.183 0.142 0.177 0.140 0.184 0.164 0.213 0.030 0.214 0.135

0.502 0.017 0.295 0.501 0.073 0.081 0.091 0.021 0.184 0.149 0.176 0.137 0.191 0.177 0.229 0.034 0.225 0.143

0.505 0.016 0.276 0.506 0.071 0.081 0.093 0.021 0.184 0.155 0.174 0.134 0.198 0.187 0.240 0.037 0.234 0.150

0.506 0.014 0.258 0.503 0.069 0.081 0.095 0.020 0.183 0.160 0.172 0.129 0.203 0.196 0.249 0.041 0.239 0.155

0.505 0.013 0.240 0.491 0.068 0.081 0.097 0.020 0.182 0.165 0.168 0.124 0.207 0.203 0.255 0.043 0.243 0.159

0.502 0.012 0.223 0.474 0.066 0.080 0.098 0.019 0.180 0.168 0.165 0.119 0.211 0.208 0.258 0.046 0.245 0.162

0.498 0.011 0.207 0.453 0.064 0.078 0.099 0.019 0.177 0.172 0.161 0.113 0.213 0.212 0.258 0.048 0.245 0.164

0.493 0.010 0.192 0.429 0.063 0.077 0.099 0.018 0.174 0.174 0.157 0.108 0.215 0.214 0.257 0.050 0.244 0.165

0.487 0.009 0.177 0.403 0.061 0.075 0.100 0.017 0.171 0.176 0.153 0.102 0.216 0.215 0.253 0.052 0.241 0.165

0.481 0.008 0.164 0.377 0.060 0.073 0.100 0.017 0.168 0.178 0.148 0.097 0.216 0.215 0.248 0.053 0.237 0.164

0.474 0.008 0.151 0.350 0.058 0.071 0.100 0.016 0.164 0.179 0.144 0.091 0.216 0.214 0.242 0.054 0.232 0.162

early middle late

0.633 0.851 1.064 1.250 1.410 1.543 1.648 1.726 1.778 1.808 0.112 0.162 0.217 0.267 0.311 0.347 0.377 0.401 0.419 0.433 0.203 0.255 0.347 0.445 0.542 0.634 0.719 0.795 0.861 0.917

1.818 0.443 0.964

1.812 0.450 1.000

1.792 0.454 1.028

1.761 0.455 1.047

1.721 0.455 1.059

1.675 0.453 1.064

1.625 0.449 1.063

1.572 0.445 1.056

1.518 0.439 1.046

1.463 0.433 1.032

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

C  

Table 17C.2 Changes in the level of employment (EP ratio) (from US$21,000 to US$40,000 GDP per capita) 21,000 22,000 23,000 24,000 25,000 26,000 27,000 28,000 29,000 30,000 31,000 32,000 33,000 34,000 35,000 36,000 37,000 38,000 39,000 40,000

15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

0.467 0.007 0.140 0.323 0.057 0.069 0.099 0.015 0.161 0.180 0.139 0.086 0.216 0.212 0.235 0.055 0.227 0.160

0.459 0.007 0.129 0.297 0.056 0.067 0.099 0.015 0.157 0.180 0.135 0.081 0.215 0.210 0.227 0.055 0.221 0.158

0.451 0.006 0.119 0.273 0.054 0.064 0.098 0.014 0.153 0.181 0.131 0.076 0.214 0.206 0.218 0.056 0.214 0.155

0.443 0.006 0.110 0.249 0.053 0.062 0.098 0.014 0.150 0.181 0.126 0.072 0.212 0.202 0.209 0.056 0.207 0.152

0.435 0.005 0.102 0.227 0.052 0.060 0.097 0.013 0.146 0.181 0.122 0.067 0.210 0.198 0.200 0.056 0.200 0.148

0.427 0.005 0.094 0.206 0.051 0.058 0.096 0.012 0.142 0.180 0.118 0.063 0.208 0.193 0.190 0.056 0.193 0.144

0.418 0.005 0.087 0.187 0.049 0.055 0.095 0.012 0.139 0.180 0.114 0.060 0.206 0.188 0.181 0.055 0.186 0.140

0.410 0.004 0.080 0.170 0.048 0.053 0.094 0.011 0.135 0.179 0.110 0.056 0.203 0.182 0.171 0.055 0.179 0.136

0.401 0.004 0.074 0.153 0.047 0.051 0.093 0.011 0.131 0.178 0.106 0.052 0.200 0.177 0.162 0.054 0.171 0.132

0.393 0.004 0.068 0.138 0.046 0.049 0.092 0.010 0.128 0.177 0.103 0.049 0.198 0.171 0.153 0.053 0.164 0.128

0.385 0.003 0.063 0.125 0.045 0.047 0.091 0.010 0.125 0.176 0.099 0.046 0.195 0.165 0.145 0.053 0.157 0.124

0.377 0.003 0.059 0.112 0.044 0.045 0.089 0.009 0.121 0.175 0.096 0.043 0.192 0.159 0.136 0.052 0.150 0.120

0.369 0.003 0.054 0.101 0.044 0.043 0.088 0.009 0.118 0.174 0.092 0.041 0.189 0.153 0.128 0.051 0.143 0.116

0.361 0.003 0.050 0.091 0.043 0.041 0.087 0.009 0.115 0.173 0.089 0.038 0.185 0.148 0.120 0.050 0.137 0.112

0.353 0.003 0.046 0.082 0.042 0.040 0.086 0.008 0.112 0.171 0.086 0.036 0.182 0.142 0.113 0.049 0.130 0.108

0.345 0.003 0.043 0.074 0.041 0.038 0.084 0.008 0.109 0.170 0.083 0.033 0.179 0.136 0.106 0.047 0.124 0.104

0.338 0.002 0.040 0.066 0.040 0.036 0.083 0.007 0.106 0.168 0.080 0.031 0.176 0.131 0.099 0.046 0.118 0.100

0.330 0.002 0.037 0.059 0.039 0.035 0.082 0.007 0.103 0.167 0.077 0.029 0.173 0.125 0.092 0.045 0.112 0.096

0.323 0.002 0.034 0.053 0.039 0.033 0.080 0.007 0.100 0.165 0.074 0.028 0.169 0.120 0.086 0.044 0.107 0.092

0.316 0.002 0.032 0.048 0.038 0.032 0.079 0.006 0.097 0.164 0.072 0.026 0.166 0.115 0.081 0.043 0.102 0.089

early 1.408 middle 0.425 late 1.014

1.354 0.418 0.994

1.301 0.410 0.972

1.250 0.402 0.949

1.200 0.393 0.924

1.153 0.384 0.899

1.107 0.376 0.873

1.064 0.367 0.846

1.022 0.358 0.820

0.983 0.349 0.793

0.945 0.340 0.767

0.909 0.331 0.742

0.875 0.323 0.716

0.843 0.314 0.692

0.813 0.306 0.668

0.784 0.298 0.644

0.756 0.290 0.622

0.730 0.282 0.599

0.705 0.274 0.578

0.681 0.266 0.558

Notes: EP ratio here is calculated by level of employment divided by population times 100. The numbers in the boxes are the highest employment levels that industries are estimated to reach during the course of a large country’s development from US$1,000 to US$40,000 GDP per capita. Source: Created by the authors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

ISIC code

C    Table 17D.1 Changes in labour intensity (employment per value added) (from US$1,000 to US$20,000 GDP per capita) ISIC code 1,000

2,000

3,000

4,000

5,000

6,000

7,000

8,000

9,000

10,000 11,000 12,000 13,000 14,000 15,000 16,000 17,000 18,000 19,000 20,000

15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

0.0267 0.0130 0.0725 0.0198 0.0303 0.0213 0.0262 0.0080 0.0247 0.0222 0.0327 0.0306 0.0399 0.0304 0.0338 0.0105 0.0361 0.0140

0.0162 0.0072 0.0407 0.0137 0.0226 0.0129 0.0136 0.0045 0.0130 0.0145 0.0173 0.0190 0.0249 0.0186 0.0166 0.0077 0.0226 0.0102

0.0115 0.0048 0.0263 0.0124 0.0178 0.0093 0.0093 0.0030 0.0084 0.0109 0.0118 0.0127 0.0171 0.0136 0.0114 0.0065 0.0156 0.0092

0.0089 0.0035 0.0186 0.0122 0.0147 0.0072 0.0073 0.0023 0.0061 0.0087 0.0090 0.0090 0.0127 0.0108 0.0088 0.0057 0.0115 0.0089

0.0073 0.0027 0.0140 0.0124 0.0126 0.0059 0.0061 0.0018 0.0046 0.0073 0.0073 0.0066 0.0099 0.0090 0.0072 0.0053 0.0088 0.0088

0.0061 0.0022 0.0110 0.0125 0.0111 0.0049 0.0053 0.0014 0.0037 0.0063 0.0062 0.0051 0.0080 0.0076 0.0061 0.0050 0.0070 0.0087

0.0053 0.0018 0.0089 0.0126 0.0099 0.0042 0.0047 0.0012 0.0030 0.0055 0.0053 0.0040 0.0067 0.0066 0.0052 0.0047 0.0057 0.0087

0.0047 0.0016 0.0074 0.0127 0.0090 0.0037 0.0043 0.0010 0.0025 0.0049 0.0047 0.0032 0.0057 0.0057 0.0045 0.0045 0.0047 0.0086

0.0042 0.0013 0.0063 0.0126 0.0084 0.0032 0.0040 0.0009 0.0022 0.0044 0.0042 0.0026 0.0050 0.0050 0.0039 0.0043 0.0040 0.0085

0.0672 0.0381 0.1682 0.0979 0.0408 0.0592 0.1206 0.0231 0.0842 0.0499 0.1381 0.0521 0.0685 0.0969 0.2455 0.0235 0.0698 0.0483

0.0038 0.0012 0.0055 0.0125 0.0078 0.0028 0.0037 0.0008 0.0019 0.0040 0.0037 0.0022 0.0045 0.0045 0.0034 0.0041 0.0034 0.0083

0.0034 0.0010 0.0048 0.0122 0.0074 0.0025 0.0035 0.0007 0.0016 0.0037 0.0034 0.0018 0.0040 0.0040 0.0030 0.0040 0.0029 0.0081

0.0032 0.0009 0.0043 0.0119 0.0070 0.0023 0.0033 0.0006 0.0014 0.0034 0.0031 0.0016 0.0037 0.0035 0.0026 0.0039 0.0025 0.0079

0.0029 0.0008 0.0038 0.0116 0.0067 0.0021 0.0031 0.0006 0.0013 0.0031 0.0029 0.0013 0.0034 0.0032 0.0023 0.0038 0.0022 0.0077

0.0027 0.0008 0.0035 0.0112 0.0064 0.0019 0.0030 0.0005 0.0011 0.0029 0.0026 0.0012 0.0032 0.0028 0.0020 0.0037 0.0020 0.0075

0.0025 0.0007 0.0032 0.0107 0.0062 0.0017 0.0029 0.0005 0.0010 0.0027 0.0024 0.0010 0.0030 0.0026 0.0018 0.0036 0.0018 0.0072

0.0024 0.0006 0.0029 0.0102 0.0061 0.0015 0.0028 0.0004 0.0009 0.0026 0.0023 0.0009 0.0028 0.0023 0.0016 0.0035 0.0016 0.0069

0.0023 0.0006 0.0027 0.0098 0.0059 0.0014 0.0027 0.0004 0.0008 0.0024 0.0021 0.0008 0.0027 0.0021 0.0014 0.0034 0.0014 0.0067

0.0021 0.0005 0.0025 0.0093 0.0058 0.0013 0.0026 0.0004 0.0008 0.0023 0.0020 0.0007 0.0025 0.0019 0.0012 0.0033 0.0013 0.0064

0.0020 0.0005 0.0023 0.0088 0.0057 0.0012 0.0025 0.0003 0.0007 0.0021 0.0018 0.0006 0.0024 0.0017 0.0011 0.0033 0.0012 0.0061

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

A D

..........................................................................................................................................................................................................................................

Table 17D.2 Changes in labour intensity (employment per value added) (from US$21,000 to US$40,000 GDP per capita) ISIC code 21,000 22,000 23,000 24,000 25,000 26,000 27,000 28,000 29,000 30,000 31,000 32,000 33,000 34,,000 35,000 36,000 37,000 38,000 39,000 40,000 0.0019 0.0005 0.0022 0.0083 0.0056 0.0011 0.0024 0.0003 0.0007 0.0020 0.0017 0.0006 0.0023 0.0016 0.0010 0.0032 0.0011 0.0058

0.0018 0.0004 0.0020 0.0078 0.0055 0.0010 0.0023 0.0003 0.0006 0.0019 0.0016 0.0005 0.0022 0.0014 0.0009 0.0031 0.0010 0.0056

0.0017 0.0004 0.0019 0.0073 0.0054 0.0009 0.0023 0.0003 0.0006 0.0018 0.0015 0.0005 0.0022 0.0013 0.0008 0.0031 0.0009 0.0053

Source: Created by the authors.

0.0017 0.0004 0.0018 0.0069 0.0054 0.0009 0.0022 0.0003 0.0005 0.0017 0.0014 0.0004 0.0021 0.0012 0.0007 0.0030 0.0008 0.0051

0.0016 0.0004 0.0017 0.0064 0.0054 0.0008 0.0021 0.0003 0.0005 0.0017 0.0014 0.0004 0.0021 0.0011 0.0006 0.0029 0.0008 0.0048

0.0015 0.0003 0.0016 0.0060 0.0053 0.0008 0.0021 0.0002 0.0004 0.0016 0.0013 0.0004 0.0020 0.0010 0.0005 0.0029 0.0007 0.0046

0.0015 0.0003 0.0016 0.0056 0.0053 0.0007 0.0020 0.0002 0.0004 0.0015 0.0012 0.0003 0.0020 0.0009 0.0005 0.0028 0.0006 0.0043

0.0014 0.0003 0.0015 0.0053 0.0053 0.0007 0.0020 0.0002 0.0004 0.0015 0.0011 0.0003 0.0019 0.0009 0.0004 0.0028 0.0006 0.0041

0.0014 0.0003 0.0014 0.0049 0.0053 0.0006 0.0019 0.0002 0.0004 0.0014 0.0011 0.0003 0.0019 0.0008 0.0004 0.0027 0.0006 0.0039

0.0013 0.0003 0.0014 0.0046 0.0053 0.0006 0.0019 0.0002 0.0003 0.0013 0.0010 0.0003 0.0019 0.0007 0.0003 0.0027 0.0005 0.0037

0.0013 0.0003 0.0013 0.0042 0.0053 0.0005 0.0018 0.0002 0.0003 0.0013 0.0010 0.0002 0.0019 0.0007 0.0003 0.0026 0.0005 0.0035

0.0012 0.0002 0.0013 0.0039 0.0053 0.0005 0.0018 0.0002 0.0003 0.0012 0.0009 0.0002 0.0018 0.0006 0.0003 0.0026 0.0005 0.0033

0.0012 0.0002 0.0012 0.0037 0.0053 0.0005 0.0018 0.0002 0.0003 0.0012 0.0009 0.0002 0.0018 0.0006 0.0002 0.0025 0.0004 0.0031

0.0012 0.0002 0.0012 0.0034 0.0053 0.0005 0.0017 0.0002 0.0003 0.0012 0.0008 0.0002 0.0018 0.0005 0.0002 0.0025 0.0004 0.0030

0.0011 0.0002 0.0012 0.0031 0.0053 0.0004 0.0017 0.0002 0.0003 0.0011 0.0008 0.0002 0.0018 0.0005 0.0002 0.0024 0.0004 0.0028

0.0011 0.0002 0.0011 0.0029 0.0054 0.0004 0.0017 0.0001 0.0002 0.0011 0.0008 0.0002 0.0018 0.0004 0.0002 0.0024 0.0004 0.0026

0.0011 0.0002 0.0011 0.0027 0.0054 0.0004 0.0016 0.0001 0.0002 0.0010 0.0007 0.0002 0.0018 0.0004 0.0002 0.0023 0.0003 0.0025

0.0010 0.0002 0.0011 0.0025 0.0054 0.0004 0.0016 0.0001 0.0002 0.0010 0.0007 0.0002 0.0018 0.0004 0.0001 0.0023 0.0003 0.0024

0.0010 0.0002 0.0010 0.0023 0.0055 0.0003 0.0016 0.0001 0.0002 0.0010 0.0007 0.0001 0.0018 0.0004 0.0001 0.0023 0.0003 0.0022

0.0010 0.0002 0.0010 0.0021 0.0055 0.0003 0.0015 0.0001 0.0002 0.0009 0.0006 0.0001 0.0017 0.0003 0.0001 0.0022 0.0003 0.0021

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 33 34 36

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

A The author is grateful to Mr Jesus Felipe, Advisor in the Economics and Research Department of the Asian Development Bank, Ms. Cecilia Ugaz Estrada, Director of Policy, Research and Statistics Department of UNIDO, and Mr. Ludovico Alcorta, former Director of the Department, for their helpful comments and feedback, to Gorazd Rezonja and Charles Fang Chin Cheng, former and current UNIDO consultants, for data processing and graphic support, to Ms. Niki Rodousakis, UNIDO staff member, for editing, and to Ms. Iguaraya Saavedra, also UNIDO staff member, for formatting and Mr Ludovico Alcorta, former Director of the Department, for their helpful comments and feedback, to Gorazd Rezonja and Charles Fang Chin Cheng, former and current UNIDO consultants, for data processing and graphic support, to Ms Niki Rodousakis, UNIDO staff member, for editing, and to Ms Iguaraya Saavedra, also UNIDO staff member, for formatting.

D This chapter has been produced without formal United Nations editing. The designations employed and the presentation of the material in this document do not imply the expression of any opinion whatsoever on the part of the Secretariat of the United Nations Industrial Development Organization (UNIDO) concerning the legal status of any country, territory, city, or area or of its authorities, or concerning the delimitation of its frontiers or boundaries, or its economic system or degree of development. Designations such as ‘developed’, ‘industrialized’, and ‘developing’ are intended for statistical convenience and do not necessarily express a judgement about the stage reached by a particular country or area in the development process. Mention of firm names or commercial products does not constitute an endorsement by UNIDO.

R Chenery, H. B., 1960. ‘Patterns of Industrial Growth’, The American Economic Review, 50 (4), pp. 624–54. Chenery, H. B. and M. Syrquin, 1975. Patterns of Development 1950–1970, Oxford: Oxford University Press. Chenery, H. B. and L. Taylor, 1968. ‘Development Patterns: Among Countries and Over Time’, The Review of Economics and Statistics, 50 (4), pp. 391–416. Clark, C., 1957. The Conditions of Economic Progress, New York and London: Garland Publishing. Fisher, A.G.B., 1939. ‘Production, Primary, Secondary and Tertiary’, Economic Record, 15, pp. 24–38. Haraguchi, N., 2012. ‘Unravelling Manufacturing Development: The Role of Comparative Advantage, Productivity Growth and Country-Specific Conditions’. Development Policy and Strategic Research Branch Working Paper No. 16/2011, UNIDO.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   



Haraguchi, N. and G. Rezonja, 2010. ‘In Search of General Patterns of Manufacturing Development’, Development Policy and Strategic Research Branch Working Paper No. 02/2010, UNIDO. Haraguchi, N. and G. Rezonja, 2011. ‘Emerging Patterns of Manufacturing Structural Change’. Development Policy and Strategic Research Branch Working Paper No. 04/2010, UNIDO. Kader, A., 1985. ‘Development Patterns among Countries Reexamined’, The Developing Economies, 23 (3), pp. 199–220. Kuznets, S., 1957. ‘Quantitative Aspects of the Economic Growth of Nations: II. Industrial Distribution of National Product and Labor Force’, Economic Development and Cultural Change, 5 (4) Supplement, pp. 1–111. Landmann, Oliver, 2004. ‘Employment, Productivity and Output Growth’. ILO Employment Strategy Department Working Paper No. 2004/17. Available at: http://citeseerx.ist.psu.edu/ viewdoc/download?doi=10.1.1.554.5378&rep=rep1&type=pdf accessed 23 August 2018. Lin, J., 2010. ‘New Structural Economics: A Framework for Rethinking Development’. Policy Research Working Paper No. 5197. Washington, DC: World Bank. McMillan, M. and D. Rodrik, 2011. ‘Globalization, Structural Change, and Productivity Growth’, in M. Bachetta and M. Jansen, eds, Making Globalization Socially Sustainable, New York: International Labor Organization and World Trade Organization. Ocampo, Jose A., Codrina Rada, and Lance Taylor (eds), 2009. Growth and Policy in Developing Countries: A Structuralist Approach, New York: Columbia University Press. Perkins, D. and M. Syrquin, 1989. ‘Large Countries: The Influence of Size’, in H. Chenery and T. N. Srinivasan, eds, Handbook of Development Economics, Volume II, Amsterdam: Elsevier Science Publishers. Rada, Codrina and Rudiger von Arnim, 2011. ‘Structural Transformation in China and India: The Role of Macroeconomic Policies’, Structural Change and Economic Dynamics, 23 (3), pp. 264–75.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

       ......................................................................................................................

 ,  -,   

18.1 I

.................................................................................................................................. I the coming decades, a number of global megatrends will shape societies and economies considerably. Some of these trends have been at play already for a few decades but all of them are set to continue or even intensify in coming years. In particular, the four largely exogenous forces examined in this chapter—demographic shifts, globalization, technological progress, and climate change—are projected to shape labour markets, economic development, health outcomes, and the distribution of income and wealth. • In countries with ageing populations and a shrinking labour force, demographic trends could exert a dampening effect on economies’ growth potential. But countries where populations are still youthful hold great potential to reap the demographic dividend. • Technological change and globalization have had an overwhelmingly positive impact on welfare at the global level, but could widen income inequality. Higher levels of income inequality, in turn, could impinge on growth and stability. • Climate change poses major risks particularly to the poorest economies which are still highly dependent on agricultural production but creates opportunities for the development of new technologies and new avenues for growth.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Not surprisingly, combating the negative effects of these megatrends and protecting living standards are high on the agenda of policy makers around the globe, and country experiences and previous studies have put forward economic policy options, including deploying fiscal policy to combat income inequality and climate change. This chapter highlights a novel solution to contribute to mitigating the negative implications of these megatrends—increasing gender equality, providing more equal access to economic opportunities for women and actively seeking to boost female economic participation. Section 18.2 describes the megatrends in detail and ways in which greater gender equality could alleviate their deleterious effects. Section 18.3 outlines policies to promote more equal gender-enabling conditions and outcomes.

18.2 N  M—W R  G E?

.................................................................................................................................. In this section, we describe the most important of the megatrends—demographic change, globalization, technological progress, and climate change—and then lay out how the main challenges posed by these trends could be offset by greater gender equality.

18.2.1 Demographic Change 18.2.1.1 Shifting Populations Two major demographic trends are likely to dominate the world in the years ahead. The first is changes in the age structure of the population; and second, where most people in the world will live.

18.2.1.1.1 Ageing On the first demographic factor, two related trends are shaping the world’s population—the number of people in the world is growing at a much slower rate than in the past, and as a group, they are ageing at an unprecedented rate. Falling fertility is at the heart of the slower rate of growth of the world’s population as well as the rise in the median global age. For the first time in history, by 2020, children under the age of 5 will be outnumbered by those aged 65 and above. In high-income countries, dependency ratios could increase from slightly above 50 per cent in 2015 to almost 75 per cent in 2060 and above 80 per cent in 2100 (Figure 18.1). With a lower share of the population in the labour force, real GDP per capita growth in these countries could decline.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &   90 80

Dependency ratio

70 60 50 40 30 20 10 0

15 20 25 30 35 40 45 50 55 60 65 70 75 80 85 90 95 00 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 21

High-income countries

Middle-income countries

 . Dependency ratios, 2015–2100 (Population younger than 16 and older than 64 in percent of population aged 15–64) Source: United Nations Population Division.

However, population trends are geographically asynchronous. More than 20 countries—mostly in advanced Europe, Japan, and China—will experience population declines in the coming decades. In contrast, there is currently a demographic ‘youth bulge’ in emerging market and developing countries (EMDCs) where almost 3 billion people—about half of global population—are currently below the age of 25. For example, by 2035, the number of people reaching working age in sub-Saharan Africa is projected to exceed that of the rest of the world combined and, by 2050, the region’s population is expected to more than double, from currently about 800 million to about 2 billion people (IMF 2015). But even in EMDCs, notwithstanding the current youth bulge, ageing will intensify at an unprecedented rate in future. Ironically, therefore, given rising life expectancy, the most rapid increase in the 65-and-older population is occurring in developing countries, which will see a jump of 140 per cent by 2030. For middle-income countries, this implies that the dependency ratio could rise from almost 50 per cent in 2015 to more than 61 per cent in 2060 and almost 70 per cent in 2100 (Figure 18.1).

18.2.1.1.2 Urbanization In addition, population growth will largely take place in urban areas going forward. Today, roughly 50 per cent of the world’s 7 billion people live in urban areas. Virtually all of the projected 1.3 billion additional people in the world between now and 2030 will be concentrated in the urban areas of the developing world, rising fastest in Asia and Africa. Growing urbanization will bring potential gains from knowledge spillovers and conglomeration effects, but reaping those will require investments in infrastructure and education.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Could increasing gender equality play a role in addressing the economic problems of ageing and contribute to reaping the ‘demographic dividend’? In turn, how will gender inequality be impacted by urbanization?

18.2.1.2 Reaping the Demographic Dividend In many developing countries, the recent period during which the number of workers has been growing more rapidly than the number of dependents provided a substantial boost to economic growth, as the resources saved from having fewer dependents provided a ‘demographic dividend’. For the case of India, Aiyar and Mody (2011) estimate that 40–50 per cent of per capita growth has been attributable to the demographic dividend since the 1970s. In East Asia, the demographic transition has likely contributed one-fourth to two-fifths to GDP per capita growth rate of around 6 per cent between 1965 and 1990 (Bloom et al. 2003). In some countries, the demographic dividend will be contingent on declining fertility rates. For example, according to UN population projections, the West African Economic and Monetary Union (WAEMU) will undergo a demographic transition over the next few decades, characterized by declines in infant mortality rates (Hooley and Newiak 2016). In addition, if fertility rates in the WAEMU decline from their current level of almost six children per woman to below four— the UN’s most optimistic scenario—the share of the working age population will increase from 52 per cent to 58 per cent by 2035, implying a higher share of the population that is potentially economically productive. However, if fertility rates in the region do not decline from their current elevated levels, the working age population share will stay constant and the demographic transition and associated growth dividend will remain elusive in the medium term. In addition, the rapid increase in population would put enormous pressure on public services and infrastructure which are inadequate even at current population levels (Guengant and May 2013). In other words, even with an increasing share of the working age population, growth effects of the demographic dividend are not automatic. The shift in demographics needs to be complemented by investments in education and health care to ensure the entrance of the growing workforce into the labour market at higher wages and higher productivity employment (IMF 2013), and via lower infant mortality and thereby to a lower birth rate. Studies show that increasing women’s access to education will be an important part of the solution. While gender gaps in primary education are closing worldwide, literacy rates are still lower for women than men, especially in South Asia, the Middle East, and North Africa, and sub-Saharan Africa. For example, in low-income countries, for every ten men only nine women are enroled in secondary education, and female tertiary enrolment rates are 35 per cent lower than male ones (Fabrizio et al. 2015). Closing these gaps would give a tremendous boost to human capital, providing strong tailwinds for the demographic dividend.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

18.2.1.3 Mitigating the Impact of Population Ageing Absent other changes, ageing and increases in life expectancy would lower per capita GDP growth by shrinking labour input and putting strains on pension and health care systems by raising dependency ratios. Increasing education levels for women and raising female participation in the labour force would help to alleviate these pressures. There is already considerable evidence of the benefits in countries in northern Europe but also more recently in Japan (Steinberg and Nakane 2012). Using an occupational choice model which accounts for dependency ratios as well as various restrictions to women’s participation in economic activities, Cuberes et al. (2016) show that even relatively slow decreases in gender gaps in the labour force could significantly reduce the negative effect on GDP from population ageing. In most countries, continuous steps to eliminate gender gaps in the long term (fifty years) could more than compensate for the negative effects of a declining labour force by 2035, leading to overall GDP gains in countries such as Chile, Czech Republic, Japan, Lebanon, Macedonia, Malta, and Mauritius. In the vast majority of other countries, the effect of rising dependency ratios can be reduced by more than 50 per cent, and policies which would accomplish faster declines in gender gaps would, of course, yield higher gains.

18.2.1.4 Harnessing Urbanization for Growth and Gender Equity Urbanization offers significant opportunities such as the benefits arising from good infrastructure and agglomeration—the so-called ‘urban advantage’. The jobs created in urban areas are more likely to be in the services sector and favour ‘brain’ over ‘brawn’, and therefore are likely to create greater opportunities for women than men. To reap the benefits from increasing urbanization, countries should invest in upgrading infrastructure and improving educational quality. Doing so will potentially create economic opportunities, particularly for women. Indeed, there is ample research linking better infrastructure to greater participation in the labour force by women, leading to higher growth. Das et al. (2015), using household-level survey data, find that, women living in Indian states with better access to roads and electricity are more likely to be in the labour force. Agénor and Canuto (2013), in an overlapping generations model, argue that infrastructure improves women’s time allocation and bargaining power, with substantial positive implications for long-run growth in Brazil, results which are consistent with Khera (2016) who finds similar effects in a dynamic stochastic general equilibrium model calibrated to India.

18.2.2 Globalization Globalization—the process of international integration through, inter alia, trade and financial flows—has been a transformational force for the past half century. It has resulted in massive increases in trade, including, importantly, through the growth of

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



global value chains—or the full range of activities that are required to bring a product from its conception, design, raw materials and intermediate inputs, marketing, distribution, and finally to sale to the final consumer. Globalization has presented tremendous opportunities for women but also some challenges (Pieters 2015). • Globalization underlies the nearly universal increase in women’s share of the nonagricultural labour force in many EMDCs, as witnessed by Bangladesh and Vietnam. The rapid growth of global value chains has benefited women as it has created employment opportunities in sectors such as garments and textiles, which tend to favour women. For example, in Mauritius, the development of the textile industry coincided with an increase in female labour force participation by nearly 60 per cent between 1983 and 1999 (Svirydzenka and Petri 2014). As such, globalization has greatly improved the lives of women, particularly those in developing countries. In addition, globalization also tends to boost women’s bargaining power by spurring change in legal anti-discrimination regimes (Pieters 2015) and women have benefited from the globalization of services (Rendall 2013). Finally, globalization in labour mobility has increasingly encouraged female migration, with evidence that women’s remittances may be higher and more resilient than those of males due to stronger linkages of women to family members but also self-insurance motives (Le Goff 2016). • While growing exports sectors have increased women’s participation in the labour force, they are facing some disadvantages compared to their male peers in this sector. In particular, exporters may be more likely to attribute characteristics such as flexibility and commitment to the job to male workers, resulting in lower wages for women (Boler et al. 2015). However, as theoretically and empirically shown in Juhn and others (2014) for the case of Mexico, a reduction in tariffs provides an incentive to productive firms to modernize their technology and export market, lowering the need for physical skills, thus increasing the wages for women in bluecollar jobs.

18.2.3 Technological Change Over the past twenty-five years, the spread of technology—especially of information and communication technologies—have expanded economic opportunities. In particular, they have shifted the type of work away from brawn and towards brain, and towards higher skilled manufacturing and services away from unskilled and routine manufacturing and agriculture. Periodic assertions over the past two centuries that technology will wipe out a large number of jobs have therefore not materialized. In contrast, the actual experience has been that automation and technological progress has done exactly the opposite—the employment rate rose during the entire period, even as more and more women moved from home to market. While there are, of course,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

cyclical fluctuations in unemployment, there is no apparent secular increase that would give credence to the fears of job losses from technological change. Autor (2015) provides an explanation for these developments by arguing that there are strong complementarities between automation and labour. But while technological change—as well as globalization discussed earlier—have had an overwhelmingly positive impact on world income and overall welfare, their impact on within-country income distribution, has varied over time. Following World War II, for example, productivity growth and growing trade co-existed with increasing equality in income distribution in advanced countries while both globalization and technological change have been associated with rising income inequality more recently (Jaumotte and Papageorgiou 2013). Technological changes have disproportionately raised the demand for capital and skilled labour over unskilled labour, by eliminating many types of routine manual jobs through automation or upgrading the skill level required to attain or keep those jobs. The higher demand for capital and skilled labour translated into disproportionately higher wages for them and a widening of the wage gap with unskilled (typically lower income) workers. In addition, financial globalization is associated with higher inequality. One explanation is that higher capital flows, including foreign direct investment, are destined for high-skill and capital-intensive sectors, which also lowers the income share of unskilled workers, exacerbating income inequality (Dabla-Norris et al. 2015; Gonzales et al. 2015b). Financial deepening is also associated with higher-income inequality, as credit is often concentrated and financial inclusion does not keep pace with deepening. This is especially pressing for women given persisting gender gaps in financial inclusion (Demirgüç-Kunt et al. 2013; Deléchat et al. 2018). On the other hand, expanding social media and information and communication technology could allow for greater access to education and information, potentially raising the share of skilled workers, especially in developing countries. Such an expansion could therefore help to alleviate income inequality. Could greater gender equality be utilized to take full advantage of the changing nature of jobs and also to mitigate rising income inequality?

18.2.3.1 Changing Nature of Jobs Autor (2015) notes that the interplay between machine and human comparative advantages amplifies the advantage of workers in problem solving skills, adaptability, and creativity. To take this argument a step further, one could argue that to the extent that automation substitutes for routine tasks, it should free up the time that women would have spent doing routine household tasks and allow them to be more involved in care work or work outside the home. Indeed, the demand for female workers in the export-oriented and ICT-enabled sectors has increased—and as women have filled these new jobs, the gender distribution of employment across sectors and across countries has changed. In addition, the nature of women’s work for given jobs has changed. Black and Spitz-Oener (2007), using data for West Germany, find that women experienced a relative increase in

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



non-routine analytic and interactive tasks commonly associated with higher skill levels, and a significant decline in routine tasks where men have seen little change. Similar changes have taken place across all countries, but female (and male) employment in the manufacturing and service sectors has grown faster in developing than developed countries, reflecting broader changes in the global distribution of production and labour. In lower-middle income countries, the shares of industry and service employment in female employment has increased by almost 8 percentage points from the mid1990s to 2020. In contrast, in high-income countries, employed women are now about 6 per cent less likely to be employed in industry than two decades ago, but more than 7 percentage points more likely to be employed in services. Going forward, both technological change and globalization will also continue to change the nature of jobs, with a large upside potential for women. Technology, especially ICT, could be the answer to increasing female economic participation in a way that is compatible with cultural and societal preferences in some countries. At the same time, improvements in ICT technology have allowed women (and men) around the world to access markets in growing numbers by lowering information barriers and reducing the transaction costs associated with market work. Because time and mobility constraints are more severe for women than men (Kochhar et al. 2017), women stand to benefit more from these developments.

18.2.3.2 Catalysing Change to Reduce Income Inequality Technological innovation is set to continue, and likely also to lead to further increasing the relative demand for higher skills, possibly exacerbating income inequality (IMF 2013), absent countervailing policies. This effect is mooted to be more pronounced in advanced countries, where the use of technology is widespread in both manufacturing and services, affecting a substantial segment of the economy. Policies to lower gender inequality could mitigate these trends. Income inequality and gender-related inequality can interact through a number of channels. First, gender wage gaps directly contribute to income inequality. Furthermore, higher gaps in labour force participation rates between men and women are likely to result in inequality of earnings between sexes, thus creating and exacerbating income inequality. Differences in economic outcomes may be a consequence of unequal opportunities and enabling conditions for men and women, and boys and girls. Gonzales et al. (2015b) find that several dimensions of gender inequality are associated strongly with income inequality across both time and countries for all income groups. However, the relevant dimensions of gender inequality vary. For advanced countries—with largely closed gender gaps in education and more equal economic opportunities for men and women—income inequality arises mainly through gender gaps in economic participation. In emerging markets and low-income countries, inequality of opportunity, in particular gender gaps in education and health, appear to pose the main obstacle to a more equal income distribution. Reducing gender inequality can therefore be a powerful way to combat income inequality in conjunction with well-designed redistributive policies. While redistribution

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &   80 70 60

R² = 0.4482

Net Gini

50 40 30 20 10 0

0

0.2

0.4

0.6

0.8

1

UN gender inequality index (re-estimated) High-income

Low-income

Middle-income

 . Income inequality vs. gender inequality Source: SWIID, United Nations, and author’s estimates.

generally has a benign effect on growth, it is only negatively related to growth in the most strongly redistributive countries (Ostry et al. 2014). More importantly, addressing deeper inequality of opportunities, such as unequal access to the labour force, health, education, and financial access, including between men and women, could have more long-lasting positive effects on the income distribution.

18.2.4 Climate Change The world’s climate is changing. Signs of a changing climate have become more prevalent, including, for example, an increase in the global temperature—as evidenced by the fact that all twelve of the warmest years on record have occurred since 1997— rising sea levels, and melting of glaciers. Climate change affects all aspects of human life, exerts significant socioeconomic cost, and is also magnifying the impact on resources and exacerbating environmental risks. Some estimates suggest that current losses could be as high as 1 per cent of global GDP, with an overwhelming majority of these losses incurred in developing countries. In addition, several factors point to mounting pressures on natural resources and the environment in the coming decades. The growth in the world’s urban population will continue while the rural population is expected to remain constant or stagnate somewhat. With population and income per capita growth slowing in advanced countries, demand for natural resources is likely to continue to be driven by EMDCs.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Export diversification (Higher Values, lower diversification)

7 6 5

R² = 0.404

4 3 2 1 0

0

0.2

0.4

0.6

0.8

1

Gender inequality index (higher values, higher gender inequality) other

low-income and developing

 . Export diversification and gender inequality, 1990–2010 Source: World Bank Development Indicators, United Nations, and IMF (2014) and IMF staff calculations.

Resource scarcity and environmental degradation are likely to have a disproportionate impact on developing economies. By 2030, close to half the world population will be living in areas of high water stress and global water demand will exceed supply by 40 per cent. Not surprisingly, given population dynamics, most of the people facing water scarcity will be living in developing countries. Can higher gender equality mitigate climate change-related vulnerabilities and foster economic diversification? Is there a gendered impact of this change?

18.2.4.1 Accelerate Economic Diversification Economic diversification, by moving employment from agriculture to other sectors, could mitigate the economic impact of weather-related shocks. The literature has now well established that diversification and structural transformation—the continued, dynamic reallocation of resources to more productive sectors and activities—are associated with higher economic growth, particularly at the early stages of development and lower volatility of growth (Papageorgiou and Spatafora 2012; IMF 2014). Structural transformation has been shown to coincide with episodes of decreases in gender inequality, in particular in the service sector. Several studies examine the relationship between women’s economic participation and structural transformation, focusing predominantly on the influence of the service sector (Rendall 2013; Ngai and Petrongolo 2014; Olivetti and Petrongolo 2014). Rendall (2013) finds that structural transformation has been important in reducing gender inequality and argues that this has happened by decreasing the labour demand for physical attributes (‘brawn’).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

Economies with lower brawn requirements offer better labour market opportunities because they allow women to take advantage of their comparative advantage in intellectual (‘brain’) attributes. Cavalcanti and Tavares (2016) link increases in female labour force participation to increases in government expenditures, leading to higher demand for services provided by the government. This in turn further encourages female labour force participation, especially when the public sector typically employs more women. But more gender equality could also help advance economic diversification, thereby mitigating economic vulnerability. Kazandjian et al. (2016) show that gender inequality decreases the variety of goods countries produce and export, in particular in lowincome and developing countries. This happens through at least two channels: First, gender gaps in opportunity, such as lower educational enrolment rates for girls than for boys, harm diversification by constraining the potential pool of human capital available in an economy. Second, gender gaps in the labour market impede the development of new ideas by decreasing the efficiency of the labour force. As a result, gender equality could be an avenue to reduce exposure of the economy to weather-related shocks, thereby mitigating the impact of climate change on the macroeconomy in particular in developing countries.

18.2.4.2 Mitigating Vulnerability to Climate Change As discussed in UNDP (2011, 2012, 2013), there is a gender aspect to climate-related challenges as agriculture, food production, and domestic energy generation are all areas dominated by women. Inequalities in access to assets and social roles may influence women’s capacity to deal with the effects of climate change. Thus, the impact on women is likely to be large and disproportionate. On the other hand, women play crucial roles in ensuring food security, such as through mobilizing communities at different stages of risk management (UNDP 2012). Better education, access to assets, finance, and to technology may allow women to play a greater role in mitigation and adaptation measures and thus to achieve multiple objectives in a mutually consistent manner. In addition, the expansion of alternative energy sources to mitigate climate change could offer opportunities for women. For example, the promotion of renewable energies that substitute for greenhouse gas emission could both create jobs for women, such as through maintenance of solar plants, but solar-powered lamps could also increase the time available to street vendors, which would benefit women (GTZ 2010).

18.3 P O

.................................................................................................................................. As more gender equal outcomes can mitigate the adverse consequences of the megatrends described above, what can policy makers do to promote more gender parity? This section gives a short overview of policies which have been previously proposed.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



18.3.1 Where do we stand? To understand the need for policy intervention, first take a look at the trends in major indicators of gender equality. Average female labour force participation across the world remains low and stagnant at around 50 per cent of the working age population, and women now represent 40 per cent of the global labour force. Female labour force participation rates—the proportion of women working or actively looking for work— have hovered around 50 per cent over the past two decades, compared with an average of about 80 per cent for men. The average rate masks substantial cross-regional differences in levels and trends: Female labour force participation rates vary from a low of 21 per cent in the Middle East and North Africa to over 63 per cent in East Asia and the Pacific and in sub-Saharan Africa (Kochhar et al. 2017). Latin America and the Caribbean experienced strong increases in female labour force participation rates of some 13 percentage points over the past two decades, while rates have been declining in South Asia, and have stayed broadly constant in Europe and Central Asia. Differences between male and female participation rates have narrowed, but the gap remains high in most regions of the world (Stotsky et al. 2016). The average gender participation gap—which is the difference between male and female labour force participation rates—has been declining since 1990, largely due to a worldwide fall in male labour force participation rates, but remains significant. The gender gap in participation varies strongly by region, with the highest gap observed in the Middle East and North Africa (51 percentage points), followed by South Asia and Central America (above 35 percentage points), and the lowest levels seen is in Organization for Economic Development and Cooperation (OECD) countries and Eastern and Middle Africa (around 12 percentage points). In addition, women dominate the informal sector, characterized by vulnerability in employment status, a low degree of protection, mostly unskilled work, and unstable earnings (ILO 2012; Campbell 2013). Furthermore, there exists a persistent gender wage gap, even accounting for education and experience.

18.3.2 Unleashing Fiscal Policy Fiscal policies can play a crucial role in increasing female labour force participation (Elborgh-Woytek et al. 2013). In particular, policy makers should create the fiscal space for priority expenditures: • Foster education. Policies to equalize enrolment rates for boys and girls would significantly boost overall education levels in low-income and developing countries. Das et al. (2015) find that female labour force participation in India would rise by 2 percentage points with an increase in spending on education of 1 per cent of GDP of Indian states. Broadly speaking, to increase girls’ participation in education and thus the human capital stock, cash transfers could be designed to

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

be conditional on sending daughters to school. Evidence suggests that education reduces fertility rates and improves the opportunities for women to remain in the formal sector. In Malawi, a cash transfer programme to current schoolgirls or recent dropouts conditional on staying in or returning to school decreased the probability of getting pregnant for recent dropouts by 30 per cent (Baird et al. 2009). In Liberia, training young women in business development and job skills increased employment rates for these women by about 50 per cent (World Bank 2012). Targeted agricultural programmes, including on the enhancement of farming techniques, would increase productivity in the sector, including for women (AfDB 2015). • Invest in infrastructure (World Bank 2011). Scaling up of infrastructure investment, such as increasing access to water, electricity, and transport reduces transaction costs and frees up women’s time to engage in market-based activities. In India, analysis using detailed household surveys has found that poor infrastructure has a dampening effect on female labour force participation, as women living in states with greater access to roads and electricity are more likely to be in the labour force (Das et al. 2015). In rural Bangladesh, upgrading and expanding the road network increased female labour supply and incomes. In rural South Africa, electrification was found to have increased women’s labour market participation by 9 per cent. Roads can enable farmers to sell their agricultural produce faster. These improvements can increase productivity and provide women who are usually the primary persons in charge of these tasks in the household with time to seek more formal employment or prolong their education. Evidence from micro-surveys in Ghana suggests halving water-fetching time increases girls’ school attendance by 2.4 per cent on average, with larger effects in rural areas (Nauges and Strand 2013). Estimates from Ethiopia suggest that female farm managers spend almost nine hours less per week on agricultural work than their male peers due to domestic work (AfDB 2015). • Improve access to health services. Adequate health care is crucial to reducing women’s obligations to time-consuming informal health care. In Bangladesh, female labour participation has benefited from the government’s Health and Community Services, and as a result the participation rate of young women almost doubled in the late 1990s (World Bank 2012b). • Implement well-designed family benefits. Better access to parental leave and high quality, affordable child care will make it easier for women to seek employment. Greater parity between paternity and maternity leave could enable women to share the burden with their partners and return to the labour market at an earlier stage. Offering flexible work arrangements and breaking down the barriers between part-time and full-time work would also help. There is also significant room for gender-equality enhancing policies on the tax side. Many fiscal policy reforms aim to boost employment of both women and men, but even measures that are gender neutral can have a disproportionally large positive effect

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



on women (World Bank 2011; Duflo 2012; Elborgh-Woytek et al. 2013). IMF (2012)—‘Fiscal Policy and Employment in Advanced and Emerging Economies’— outlines a wide range of fiscal policies to enhance the labour force participation of both women and men. • Taxation of labour income and government spending on social welfare benefits and pensions both affect labour markets similarly. They weaken the link between labour supply and income, thereby influencing the decision to participate in the labour market. Therefore, the appropriate design of benefits is important to avoid disincentives to work. • Reducing the tax burden for (predominantly female) secondary earners by replacing family taxation with individual taxation or other measures aimed at reducing marginal taxes on second earners more broadly can potentially generate large efficiency gains and improve aggregate labour market outcomes. Tax credits or benefits for low-wage earners can be used to stimulate labour force participation. More generally, governments should look at their budgets gender-responsively. Gender-responsive budgeting examines the gender impact of government expenditures, policies, and programmes and can reduce gender inequalities in education, employment, and health outcomes, among other measures. The incorporation of gender issues in Bangladesh’s national budget began in 2005 as part of the government’s efforts to promote a more inclusive society; ministries compile gender budgeting reports (World Bank 2012). Morocco has published a gender report since 2006.

18.3.3 Easing the Burden of Non-market Work In the discussion so far, women’s economic empowerment is treated as synonymous with market-remunerated work. However, non-market care work is also important and women’s role in the care economy is substantial, possibly crowding out opportunities for formal work as accounted in official GDP estimates. Women contribute substantially to raising economic welfare through large amounts of unpaid work—caring for elderly family members in addition to child-rearing and household tasks, which constrains women’s ability to participate in the labour market. On average, women spend twice as much time on household work as men and four times as much time on child care (Duflo 2012), thereby freeing up time for male household members to participate in the formal labor force. In the OECD countries, women spend about 2½ hours more than men on unpaid work (including care work) each day, regardless of the employment status of their spouses (Aguirre et al. 2012). As a result, the gender difference in total working time—the sum of paid and unpaid work, including travel time—is close to zero in many countries (OECD 2012). To ensure the burden of non-market work on women is reduced over time, the restructuring of health care systems and rethinking of social security systems in ageing societies in middle-income and advanced economies is needed. Also, as discussed

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

earlier, infrastructure investment (particularly in EMDCs) can free up women’s time and allow them more time for market-based work. Finally, gender norms should evolve and there should be no presumption that non-market care work should be performed predominantly by women. In this respect, providing and encouraging the use of paternity leave would underscore the father’s role in child rearing and reduce the burden on women.

18.3.4 Abolishing Legal Restrictions Equalizing laws for men and women has been shown to boost female labour force participation (Gonzales et al. 2015a). Gonzales et al., drawing on a large and novel panel data set of gender-related legal restrictions (World Bank 2013, 2015), find that restrictions on women’s rights to inheritance and property, as well as legal impediments to undertaking economic activities such as opening a bank account or freely pursuing a profession, are strongly associated with larger gender gaps in labour force participation. For example, Namibia equalized property rights for married women and granted women the right to sign a contract, head a household, pursue a profession, open a bank account, and initiate legal proceedings without the husband’s permission in 1996. In the decade that followed, Namibia experienced a 10 percentage point increase in its female labour force participation rate. Peru and Malawi invalidated customary law in 1993 and 1994 respectively, and both experienced significant increases in their female labour force participation rates. Equal access to productive resources for men and women would go a long way towards increasing productivity and growth in many low-income developing countries. A final move would be to making finance accessible to women. In many low-income and developing countries, the availability of microfinance has helped to reduce the gender productivity gap (Kabeer 2005), with higher credit repayment rates among women than among men.

18.4 C

.................................................................................................................................. Gender inequality has been demonstrated to impede inclusive growth across a wide range of countries. Going forward, the global economy will be confronted with ongoing megatrends that will significantly shape macroeconomic and social outcomes, such as demographic change, technological progress, globalization, and climate change. As these trends intensify, policy efforts to mitigate some of their negative consequences and to reap their potential will have to be stepped up. This chapter argues that policies which foster gender equality could have the added benefit of mitigating some of the risks posed by these megatrends. In countries facing unfavourable demographic shifts and rapid ageing, investing in women’s education and raising female labour force

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



participation could offset the drag on potential growth. The chapter finds that, for most countries, the GDP gains from doing so would more than compensate for population ageing. Technological change and globalization have benefited women as these forces have boosted demand for ‘brain’ over ‘brawn’. Going forward, these trends could continue changing the nature of jobs, creating opportunities for women.

R African Development Bank (AfDB), 2015. Economic Empowerment of African Women through Equitable Participation in Agricultural Value Chains, Abidjan: African Development Bank. Aguirre, DeAnne, Leila Hoteit, Christine Rupp, and Karim Sabbagh, 2012. Empowering the Third Billion. Women and the World of Work in 2012, New York: Booz and Company. Agénor, P.-R. and O. Canuto, 2013. ‘Gender Equality and Economic Growth in Brazil. A Long-run Analysis’. World Bank Policy Research Working Paper No. 6348. Washington, DC: World Bank. Aiyar, Shekhar and Ashoka Mody, 2011. The Demographic Dividend: Evidence from the Indian States. IMF WP/11/38. Autor, D. H., 2015. ‘Why Are There Still So Many Jobs? The History and Future of Workplace Automation’, Journal of Economic Perspectives, 29 (3), pp. 3–30. Baird, S., E. Chirwa, C. McIntosh, and B. Ozler, 2009. ‘The Short-Term Impacts of a Schooling Conditional Cash Transfer Program on the Sexual Behavior of Young Women’. Policy Research Working Paper No. 5089, Washington, DC: World Bank. Black, E. S. and A. Spitz-Oener, 2007. ‘Explaining Women’s Success: Technological Change and the Skill Content of Women’s Work’. IZA Discussion Paper No. 2803. Bloom, E., D. Canning, and J. Sevilla, 2003. ‘The Demographic Dividend. A New Perspective on the Economic Consequences of Population Change’, RAND, MR-1274. Boler, E. A., B. Javorczik, and K. H. Ulltveit-Moe, 2015. ‘Globalization: A Women’s Best Friend? Exporters and the Gender Wage Gap’. CEP Discussion Paper No. 1358. Campbell, D., 2013. ‘The Labour Market in Developing Countries’, in S. Cazes and S. Verick, eds, Perspectives on Labour Economics for Development, Geneva: ILO. Cavalcanti, T., and J. Tavares, 2016. ‘The Output Cost of Gender Discrimination: A Model Based Macroeconomic Estimate’, The Economic Journal, 126, (590). Cuberes, David and Marc Teigner, 2016. ‘Aggregate Effects of Gender Gaps in the Labor Market; A quantitative Analysis’, Journal of Human Capital, 10 (1). Dabla-Norris, E., K. Kochhar, N. Suphaphidat, F. Ricka, and E. Tsounta, 2015. ‘Causes and Consequences of Inequality: A Global Perspective’. IMF Staff Discussion Note 15/13, Washington, DC: International Monetary Fund. Das, S., S. Jain-Chandra, K. Kochhar, and N. Kumar, 2015. ‘Women Workers in India: Why So Few Among So Many?’ International Monetary Fund Working Paper No. 15/55, Washington, DC: International Monetary Fund. Deléchat, C., M. Newiak, R. Xu, F. Yang, and G. Aslan, 2018. ‘What is Driving Women’s Financial Inclusion Across Countries?’ IMF Working Paper No. 18/38, Washington, DC: International Monetary Fund. Demirgüç-Kunt, Asli, Leora F. Klapper, and Dorothe Singer, 2013. ‘Financial Inclusion and Legal Discrimination against Women: Evidence from Developing Countries’. World Bank Policy Research Working Paper No. 6416, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  -, &  

Duflo, E. 2012. ‘Women Empowerment and Economic Development’, Journal of Economic Literature, 50 (4), pp. 1051–79. Elborgh-Woytek, K., M. Newiak, K. Kochhar, S. Fabrizio, K. Kpodar, P. Wingender, B. Clements, and G. Schwartz, 2013. ‘Women, Work, and the Economy: Macroeconomic Gains from Gender Equity’. Staff Discussion Note 13/10, Washington, DC: International Monetary Fund. Fabrizio, S., R. Garcia-Verdu, C. A. Pattillo, A. Peralta-Alva, A. Presbitero, B. Shang, G. Verdier, M.-T. Camilleri, K. Washimi, L. L. Kolovich, M. Newiak, M. Cihak, I. Otker, L.-F. Zanna, and C. L. Baker, 2015. ‘From Ambition to Execution: Policies in Support of Sustainable Development Goals.‘ Staff Discussion Note 15/18, Washington, DC: International Monetary Fund. Gonzales, C., S. Jain-Chandra, K. Kochhar, and M. Newiak, 2015a. ‘Fair Play: More Equal Laws Boost Female Labor Force Participation’. IMF Staff Discussion Note 15/02, Washington, DC: International Monetary Fund. Gonzales, C., S. Jain-Chandra, K. Kochhar, M. Newiak, and T. Zeinullayev, 2015b. ‘Catalyst for Change: Empowering Women and Tackling Income Inequality’. IMF Staff Discussion Note 15/20, Washington, DC: International Monetary Fund. GTZ (Deutsche Gesellschaft für Technische Zusammenarbeit). 2010. ‘Climate Change and Gender: Economic Empowerment of WOMEN through Climate Mitigation and Adaptation?’. GTZ Working Paper. Guengant, Jean-Pierre and John F. May, 2013. ‘African Demography’, Global Journal of Emerging Market Economies, 5 (3), pp. 215–67. Hooley, J. and M. Newiak, 2016. ‘Structural Transformation and Diversification’, in A. P. Kireyev, ed., Building Integrated Economies in West Africa. Lessons in Managing Growth, Inclusiveness and Volatility, Washington, DC: International Monetary Fund. International Labour Organization, 2012. Perspectives on Labor Economics for Development, ed., Sandrine Cazes and Sher Verick. Geneva: International Labour Organization. International Monetary Fund (IMF), 2012. ‘Fiscal Policy and Employment in Advanced and Emerging Economies.’ IMF Policy Paper. Washington, DC: International Monetary Fund. International Monetary Fund (IMF), 2013. Jobs and Growth: Analytical and Operational Considerations for the Fund. IMF Policy Paper, Washington, DC: International Monetary Fund. International Monetary Fund (IMF), 2014: ‘Sustaining Long-Run Growth and Macroeconomic Stability in Low-Income Countries—The Role of Structural Transformation and Diversification—Background Notes’. Washington, DC: International Monetary Fund. International Monetary Fund (IMF), 2015. Regional Economic Outlook: Sub-Saharan Africa, April, Washington, DC: International Monetary Fund. Jaumotte, F. and C. Papageorgiou, 2013. ‘Rising Income inequality: Technology or Trade and Financial Globalization?’ IMF Economic Review, 61 (2). Juhn, C., G. Ujihelyi, and Villegas-Sanchez, 2014. ‘Men, Women, and Machines: How Trade Impacts Gender Inequality’, Journal of Development Economics, 106, pp. 179–93. Kabeer, Naila, 2005. ‘Gender Equality and Women’s Empowerment: A Critical Analysis of the Third Millennium Development Goal’, Gender & Development, 13 (1), pp. 13–24. Kazandjian, R., K. Kochhar, L. Kolovich, and M. Newiak, 2016. ‘Gender Equality and Economic Diversification’. Working Paper No. 16/140, July, Washington, DC: International Monetary Fund. Khera, P., 2016. ‘Macroeconomic Impacts of Gender Inequality and Informality in India’. IMF Working Paper No. 16/16. Washington, DC: International Monetary Fund.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

      



Kochhar, K., S. Jain-Chandra, and M. Newiak, 2017. ‘Women, Work, and Economic Growth: Leveling the Playing Field’. Washington, DC: International Monetary Fund. Le Goff, Maelan, 2016. ‘Feminization of Migration and Trends in Remittances’. IZA World of Labor 2016:220, January. Nauges, C. and J. Strand, 2013. ‘Water Hauling and Girls’ School Attendance. Some New Evidence from Ghana’. Policy Research Working Paper No. 6443, World Bank, Washington, DC. Ngai, R. and B. Petrongolo, 2014. ‘Gender Gaps and the Rise of the Service Economy’. Institute for the Study of Labor (IZA) Discussion Paper No. 8134. Organization for Economic Cooperation and Development (OECD), 2012. Closing the Gender Gap: Act Now, Paris: OECD. Olivetti, C. and B. Petrongolo, 2014. ‘Gender Gaps across Countries and Skills: Demand, Supply and the Industry Structure’, Review of Economic Dynamics, 17(4), pp. 842–59. Ostry, J., A. Berg, and C. Tsangarides, 2014. ‘Redistribution, Inequality and Growth’. International Monetary Fund, Staff Discussion Note /14/02, Washington, DC: International Monetary Fund. Papageorgiou, C., and N. Spatafora, 2012. ‘Economic Diversification in LICs: Stylized Facts and Macroeconomic Implications’. IMF Staff Discussion Note 12/13, Washington, DC: International Monetary Fund. Pieters, J., 2015. ‘Trade Liberalization and Gender Equality’. IZA World of Labor 2014:114, January. Rendall, M. 2013. ‘Structural Change in Developing Countries: Has it Decreased Gender Inequality?’ World Development, 45, pp. 1–16. Steinberg, Chad and Masato Nakane, 2012. ‘Can Women Save Japan?’ Working Paper No. 12/248, Washington, DC: International Monetary Fund. Stotsky, J., S. Shibuya, L. Kolovich, and S. Kebhaj, 2016. ‘Trends in Women’s Development and Gender Equality’. IMF Working Paper No. 16/21. Svirydzenka, K. and M. Petri, 2014. ‘Mauritus: The Drivers of Growth—Can the Past Be Extended?’ IMF Working Paper No. 14/134. Washington, DC: International Monetary Fund. UNDP, 2011. ‘Gender and Climate Change. Africa’. UNDP Policy Brief. New York: UNDP. UNDP, 2012. ‘Gender and Climate Change, and Food Security’, UNDP Policy Brief No. 4, New York: UNDP. UNDP, 2013. ‘Overview of Linkages Between Gender and Climate Change’. UNDP Policy Brief No. 1. New York: UNDP. World Bank, 2011. World Development Report 2012: Gender Equality and Development, Washington, DC: World Bank. World Bank, 2012. World Development Report 2013. Jobs, Washington, DC: World Bank. World Bank, 2013. Women, Business and the Law 2013. Removing Restrictions to Enhance Gender Equality, Washington, DC: World Bank/International Finance Corporation. World Bank, 2015. Women, Business and the Law 2013. Getting to Equal, Washington, DC: World Bank/International Finance Corporation.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  .............................................................................................................

C O U N T R Y AN D REGIONAL EXPERIENCES .............................................................................................................

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

           ’     ......................................................................................................................

´  

19.1 I

.................................................................................................................................. L America’s experience with the transformation of its production structure has gone through three entirely different historical phases. During the first one, which largely coincides with the first globalization of the late nineteenth and early twentieth centuries, economic growth was dominated by a dynamic commodity export sector, but also with the rise of modern manufacturing. Export dynamism came to end as a result of the collapse of the world economy during the Great Depression, giving way to a period in which manufacturing and modern services became the main engines of economic growth, but commodities continued to dominate the region’s export base. Rapid industrialization then came to an end in different countries between the mid- and late-1970s, and was followed by a process of premature de-industrialization accompanied in the early twenty-first century by a ‘re-primarization’ (or ‘re-commoditization’) of the export base. De-industrialization had started before the debt crisis of the 1980s but was accelerated by the market reforms adopted on a broad-based scale since the middle of that decade, and has been stronger in South America. The changes in the production structure unleashed by market reforms delivered a much poorer record in terms of economic growth than the period of rapid industrialization. During the market reform period, the commodity boom—or ‘super-cycle’ of commodity prices, as it has been called—delivered the best growth performance in 2003–13 (and, particularly, in 2003–08), but the collapse of commodity prices in recent years has generated major macroeconomic challenges and now questions the capacity of the contemporary development pattern to deliver even mediocre growth. Beyond the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

short-term challenges, the early exhaustion of the industrialization drive, together with slow growth since the 1980s indicate that the region has faced strong signs of a ‘middleincome trap’, and particularly the incapacity to develop strong endogenous technological capabilities—a feature of the region’s development that can in fact be traced back to the period of rapid industrialization. The average regional patterns outlined above have been mixed with significant diversity of national experiences in a region that is heterogeneous in terms of size and level of development of countries. Export growth and the early phases of industrial advance generated a major divergence in the levels of development of different countries. Rapid industrialization generated, in turn, diverse industrial production structures, which largely depended on the size of the economies. Finally, trends since the 1980s has led to a significant ‘north–south’ divergence, with some countries in the northern part of the region—notably Mexico and Costa Rica—largely integrating into the world economy through manufacturing. This chapter analyses the structural transformation of the Latin American economies, in particular the changing role of industrialization and export structures in the region’s development process. Following my own work, but also that of many other authors (including the editors of this volume), it takes as a point of departure the conception that capacity to constantly generate new dynamic activities, with increasing knowledge contents, is the key to rapid economic growth—a process that I have referred to in my work to as ‘dynamic efficiency’ (Ocampo 2005, 2016; see also Ocampo et al. 2009). This means, in turn, that the inability to generate new economic activities with higher knowledge content will block the development process. The chapter is divided in four sections, the first of which is this introduction. After a short review of the commodity export age, Section 19.2 looks at the period of rapid industrialization, which, following Cárdenas et al. (2000) and Bértola and Ocampo (2012), is referred to as ‘state-led industrialization’. This term captures the main features of this phase of development in a much better way than the traditional concept of ‘import-substituting industrialization’, as it involved much more than import substitution. Section 19.3 considers the process of structural change during the period of market reforms and, in particular the long de-industrialization that it generated and the more recent re-primarization of the region’s export structure. Section 19.4 presents some brief conclusions.

19.2 F  C E A  S- I

.................................................................................................................................. Commodity dependence has been an essential feature of Latin America’s integration into the world economy since colonial times (i.e. since the sixteenth century). The takeoff of modern economic development was associated with the commodity export

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



expansion that took place in the late nineteenth and early twentieth centuries, which, depending on the country, was based on tropical and non-tropical agricultural products, or on minerals and oil. It can thus be properly called the ‘commodity export age’. It involved all countries, although with differences in the timing and intensity of export expansion. By the end of the process, the Southern Cone countries (Argentina, Chile, and Uruguay) and Cuba stood ahead of the region in terms of several development indicators (per capita GDP but also urbanization and modern infrastructure development); in the largest countries (Brazil and Mexico), there were also large domestic differences between those regions that integrated more strongly into the world economy and those that did not. Export expansion was accompanied by early industrialization drives, which were supported by high levels of protection—indeed among the highest in the world at the time (Coatsworth and Williamson 2004). Domestic support for protection was no doubt associated with the fact that the rising entrepreneurial class diversified their risk by investing in both export and domestic activities. Part of the early industrialization process was related to the processing of commodities for export—for example, foundries and refining facilities in mining economies, and sugar refineries, coffee threshing, and meatpacking facilities in the relevant agricultural exporters. There were also indirect linkages associated with the consumer demand generated by rising incomes and urbanization. The magnitude of this effect depended on the size and extent of integration of domestic markets, which was determined by the development of modern infrastructure. It was also associated with the introduction of new products with high transport costs, notably beverages (sodas and beer) and cement, which were at the centre of early industrialization. Protection played a more important role in other sectors, notably in the case of textiles. By the end of the commodity export age, in 1929, manufacturing accounted for a significant share of GDP in the region’s leaders in term of development: 16 per cent in the Southern Cone, and 20 per cent in the largest country in that sub-region, Argentina. There was also an extensive network of sugar refineries in the other export leader, Cuba. In turn, the largest countries, Brazil and Mexico, were more industrialized than others in the region outside the Southern Cone, and in both cases thanks to high levels of protection (Bértola and Ocampo 2012: table 3.19). Commodity export-led growth came to an end as a result of the collapse of the world economy during the Great Depression of the 1930s and World War II.¹ This planted the seed for a new development pattern, but initially the process came about more as a result of a response to the major macroeconomic shocks that countries faced—the fall in exports and commodity prices, and the dismantling of the gold standard and associated exchange rate depreciations—rather than of any conception of a new development paradigm.

¹ See Bértola and Ocampo (2012: ch. 3) and Bulmer-Thomas (2014: chs 3–6).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

The term ‘import-substitution industrialization’ (ISI) has been the most widely used to refer to the period of rapid industrialization that took off in Latin America in the 1930s and continued after World War II until the 1970s. However, it is not a very useful term because the policies introduced during this period were not limited to import substitution but were also increasingly about the role of the state in many other spheres of economic and social development—infrastructure, financial services, education, health, housing, and social security, among others. Furthermore, dynamic domestic demand rather than import substitution was the fundamental driver of industrialization during this period and, as I will underscore below, they also included efforts to diversify the export structure and promote regional integration. The region’s most influential theory of industrialization was that developed by the Economic Commission for Latin America (ECLA; later ECLAC, when the Caribbean joined the organization) in the late 1940s and early 1950s with Raúl Prebisch at the helm.² However, industrialization was in full swing before this theory was developed. This fact was well captured in Love’s (1994: 395) statement that: ‘Industrialisation in Latin America was fact before it was policy, and policy before it was theory.’ In any way, ECLAC crafted the best theoretical justification for that strategy, along with a sense of regional identity and the promotion of regional integration processes.³ ECLAC’s views on the role of industrialization as a driver of technological change and economic growth—the main channel for the transmission of technological progress, in Prebisch’s favourite statement—and on the essential role of state intervention to promote development were in line with the views of classical development economists and even conventional wisdom in the economics profession and the realm of economic policy at the time. ECLAC’s analysis also included pessimistic assessments on the prospects of world commodity markets and terms of trade; these observations were more controversial and I will disregard them here. The policies that emerged from these views were, in any case, not proposals for Latin American countries to isolate themselves from the international economy, but rather to redefine the international division of labour in a way that would allow the region to better capture the benefits from technological progress. State-led industrialization was characterized, in any case, by a significant dependence on domestic markets, high levels of protection, and a fall in Latin America’s share in world trade, to slightly more than 4 per cent in the early 1970s, about half the level reached in 1925–29. The loss of share in world commodity markets was the main reason for this decline (Ffrench-Davis et al. 1998, and Bértola and Ocampo 2012: table 4.10), and was associated with both external and domestic factors. Among the

² In his autobiography, Furtado (1989) offers up a fascinating history of ECLAC’s early years, and Dosman (2008) provides an excellent biography of Prebisch. ³ ECLAC played a pivotal role in the creation of the Latin American Free Trade Association (LAFTA) in 1960—which would later become the Latin American Integration Association (LAIA)—as well as in the establishment of the Central American Common Market, also in 1960, and of the Andean Group in 1969 (later called Andean Community).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



external factors, the most important were the boom in Middle East oil production since the 1950s, which partly displaced earlier oil exporters, including Mexico and Venezuela, and agricultural protectionism in developed countries, which hit Argentina, Cuba, and Uruguay particularly badly. Nonetheless, commodity exports continued to play a crucial role, not only as a source of the foreign exchange that the industrial sector required to finance its imports of capital and intermediate goods, but also of government revenues in oil and mining economies, and as an engine of growth in the smaller economies. Indeed, according to Hirschman (1971), an essential feature of Latin American industrialization was the weakness of industrial interests vis-à-vis those of commodity exporters. Furthermore, most countries introduced new export promotion instruments in the late 1950s to mid1960s, to manage the balance of payments constraints generated by the collapse of commodity prices in the mid-1950s, and to benefit from the growing market for developing country manufacturing exports since the 1960s. The limited industrialization of small economies implied that their export diversification since the 1950s focused essentially on new commodities. New resource discoveries had similar effects in larger economies, notably the major oil discoveries of Mexico in the 1970s. Regional integration efforts were also put in place in the 1960s, and generated a major growth of intra-regional manufacturing trade. Various histories of ECLAC (Bielchowsky 1988; ECLAC 1998: third part; Rosenthal, 2004) have pointed out that from the 1960s on, the institution became increasingly critical of the excesses of import substitution and started to advocate a ‘mixed’ model that combined import substitution with export diversification and regional integration. This was the model—and not ISI—that was in place when the region reached its record growth in the second half of the 1960s and first half of the 1970s. The model was also ‘mixed’ in the sense that it promoted agricultural production, both for the domestic market and as part of the export diversification strategy. Although traditional agricultural exports (notably coffee and sugar) were discriminated against through taxes or unfavourable exchange rates, the model actively promoted the modernization of the agricultural sector with an elaborate set of tools of intervention. This included not only protection for sectors that competed with imports⁴ and development finance (the two major tools used to promote industrialization), but also the creation of research institutes (an area that was largely missing as a component of the industrialization strategy), marketing boards and infrastructure investment to expand the agricultural frontier. Indeed, Latin America increased (or, more properly, continued to increase) its share of world agricultural production during this phase of development.⁵ ⁴ This is a more appropriate interpretation of the information provided by Anderson and Valdés (2008) than the one that points to a widespread anti-agricultural bias of the development in place at the time. For example, Figure 1.3 in that study shows that almost all products that competed with imports were protected. ⁵ See ECLAC and FAO (1978). According to current statistical series on world agricultural production from FAO, Latin America’s share in world agricultural production (at constant prices) increased from 9.4% in 1961 to 10.5% in 1980.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

Although the process did involve more state intervention than past patterns, Latin American countries opted in fact for a less interventionist state than in other developing regions in Asia and Africa. In a sense, they went for the ‘mixed economy’ model typical of Western Europe after World War II.⁶ This was reflected in the smaller scope of state enterprises than was seen in Asia and Africa.⁷ Foreign direct investment (FDI) was welcomed for manufacturing development, and in fact the region concentrated in 1973–81 nearly 70 per cent of total FDI flows to the developing world (Ocampo and Martin 2004: Table 3.2). However, restrictions on FDI in traditional natural resource and infrastructure sectors were also put in place in several countries, which included the nationalization of several natural resource sectors (oil in Mexico in 1938, followed in later decades by tin in Bolivia, copper in Chile, and oil in Venezuela). Support for international commodity agreements was also an important feature, with the regulation of coffee markets being perhaps the most important example, starting with the first International Coffee Agreement in 1962. In turn, Venezuela became the leader behind the creation of the Organization of the Petroleum Exporting Countries (OPEC) in 1960. The Latin American industrialization process went through four different stages during the period of state-led industrialization. The first was the ‘pragmatic’ phase of import substitution, which was triggered by changes in relative prices policy responses to the external shocks of the 1930s and World War II (Diaz-Alejandro 2000). During the war, it also included the first plans to develop ‘strategic’ sectors, including for goods that were rationed for exports by developed countries during the war. The second phase, that took place between the end of World War II and the mid-1960s, can be called the ‘classical’ phase of Latin American industrialization. Industrialization now became a conscious policy strategy, which became embedded in successive development plans. However, it continued to be driven by recurrent balance-of-payments crises, some of them in the early post-war years, when the reserves accumulated during the war had been run down, but particularly after commodity prices collapsed in the mid1950s. A typical policy response to each successive crisis was to raise protection levels further, through a mix of tariffs and quantitative restrictions, multiple exchange rates and foreign-exchange rationing. Increasingly, incentives to diversify the export base (tax incentives, subsidies, tariff exemptions, or drawbacks for required intermediate goods, special credit lines, and preferential exchange rates) were also put in place. These policies were successful in diversifying the industrial and, to a lesser extent, the export structures. Industrialization was also supported by development banks, regulations on sectoral allocations of credit and interest rates, tax incentives, and

⁶ The exceptions were obviously Cuba after its 1958 revolution, and temporarily Chile during the Popular Unity Administration in the early 1970s and the Sandinista revolution in Nicaragua that began in 1978, but in the latter two cases with a more mixed economy model than the central planning that Cuba developed. ⁷ According to the data from World Bank (1995), state enterprises represented 9.7% of Latin America’s non-agricultural GDP in 1979–81 vs. 21.3% in Africa and 13.0% in Asia. In 1989–91 the shares were 9.7, 20.4, and 14.0%, respectively.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



14.0%

21.0%

13.0%

20.0%

12.0%

19.0%

11.0%

1970 prices

1990 prices

2010 prices

 . Manufacturing value added as a share of GDP, 1950–2015 Source: Author’s estimates based on ECLAC data.

Constant 2010 prices

22.0%

2015

15.0%

2010

23.0%

2005

16.0%

2000

24.0%

1995

17.0%

1990

25.0%

1985

18.0%

1980

26.0%

1975

19.0%

1970

20.0%

27.0%

1965

28.0%

1960

21.0%

1955

29.0%

1950

Constant 1970 and 1990 prices

investments by state-owned enterprises, some in new manufacturing activities but, more importantly, in infrastructure sectors. The third phase, from around the mid-1960s to the mid-1970s, can be called the ‘mature’ stage of state-led industrialization. The hallmark of this phase was the ‘mixed model’ that combined advanced import-substitution activities with an increasingly conscious export diversification strategy and regional integration efforts. Some of the larger countries (Argentina, Colombia, Chile, and Brazil) also adopted a more active foreign-exchange policy that provided for a more flexible exchange rate (crawling pegs) in order to cope with the recurrent overvaluations typical of inflation-prone economies. There were also efforts in several countries to rationalize the structure of protection. The first oil shock gave way to a fourth stage, which was characterized by a diversity of experiences. Several countries continued to follow the mixed model. Brazil, Mexico, and Venezuela launched ambitious industrial investment plans, in the first case to manage the adverse effects of the oil shock, but in the latter two to invest the increasing oil revenues. A third strategy, which was led by Chile in the mid-1970s and followed in a weaker form by the other Southern Cone countries, was to abandon the industrialization strategy and the high levels of state intervention that accompanied it. Manufacturing production grew much faster than GDP between 1929 and 1945 in all larger countries for which information is available (Brazil, Mexico, Argentina, Chile, Colombia, and Peru) and in several Central American economies (Bértola and Ocampo 2012: table 4.2). As Figure 19.1 shows, the share of manufacturing in GDP boomed after World War II, peaking in 1973–74—that is, in the transition between the third and fourth stages outlined above. Among the larger nations, increases in the industrialization coefficient between 1950 and 1974 was strongest in Brazil, Mexico, Argentina,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

and Colombia, and smaller in Chile, Peru, and Venezuela. It was also strong in Ecuador and several Central American nations, where it was mixed with the development of new commodity exports. Between 1974 and 1980, it continued to increase in Mexico and Venezuela, as well as in Ecuador and Nicaragua among the smaller countries; in Brazil, it edged down slightly, but industrial growth continued to be very dynamic. The scope of industrialization was closely linked to the size of the economies, as reflected in the composition of manufacturing value added at the end of the period of rapid industrialization. This is shown for 1974–75 in Table 19.1, in which countries are listed in order of the size of their manufacturing sectors. The sectors differentiate between traditional and non-traditional sectors, classifying among the former those that developed earlier in most countries (food and beverages, textiles, non-metallic minerals— essentially construction materials, including cement—and others) from those that were developed in the later phases (chemicals, base metals, transport, and other machinery and equipment), leaving oil refining as a category of its own (although it can also be considered as non-traditional). Brazil, Mexico, and Argentina had by then the most diversified industrial structures. In Chile and Venezuela, one particular sector played a highly significant role (basic metals and oil refining, respectively), and in Colombia and Peru traditional industries continued to represent more than half of manufacturing value added. In the smaller economies, that share reached between 60 per cent and 80 per cent. The shares of manufacturing in exports started to increase in the 1960s, led by the countries that had experienced stronger industrial development, but was a more limited process. If measured by the SITC (Standard International Trade Classification) categories 5–8, the share of manufacturing in exports increased from 8.1 per cent in 1958 to a peak of 24.2 per cent in 1973 before falling somewhat to 23.2 per cent in 1980; however, if we exclude SITC 3, to eliminate the effects of the oil export boom, the share of manufactures continued to increase from 30.7 per cent in 1973 to 37.1 per cent in 1980 of non-oil exports (Bértola and Ocampo 2012: table 4.10). By the end of the boom, primary goods and resource-based manufactures continued to concentrate 77.0 per cent of Latin America’s exports, with the rest represented basically by lowand mid-technology manufactures (see Figure 19.3(a) in Section 19.3). In contrast to the ‘Black Legend’⁸ that has surrounded the orthodox interpretations of Latin American development during state-led industrialization, GDP achieved the fastest rates of growth in the region’s history. This is shown in Figure 19.2, which estimates moving averages of decade-long growth rates (so, for example, the estimate for 1970 represents the average of 1960–70). From a lowest point during the Great Depression, growth accelerated to a rate of 5 per cent a year after World War II and peaked around 6 per cent from the mid-1960s to the mid-1970s—that is, during the hegemony of the ‘mixed model’ that combined import substitution with export promotion and regional integration. Rates of growth close to 5 per cent a year had only ⁸ This term has been mainly used to refer to the mishandling by Spain of its colonies, and particularly of the Indian populations of the Americas.

Table 19.1 Share in industrial value added A. At the end of the period of rapid industrialization (1974–75)

Brazil Mexico Argentina Venezuela Chile

Colombia

Peru

Uruguay Ecuador Bolivia Paraguay

Costa Rica

Central America

Total

30.0% 13.8% 3.3% 4.2%

22.2% 11.8% 4.8% 6.2%

17.8% 10.1% 2.9% 4.4%

27.6% 15.7% 5.6% 6.0%

25.0% 13.1% 7.1% 15.2%

32.9% 20.9% 4.0% 5.0%

42.6% 14.2% 5.3% 7.7%

43.0% 22.9% 5.4% 6.5%

50.0% 16.6% 4.3% 12.8%

49.9% 12.3% 3.7% 12.8%

48.7% 16.4% 4.6% 8.3%

19.2% 13.5% 5.0% 6.7%

51.2%

45.0%

35.1%

54.9%

60.5%

62.8%

69.9%

77.9%

83.7%

78.6%

78.0%

44.5%

Oil refineries Paper and chemical industry, excluding oil refineries Basic Metals Transport equipment Machinery and equipment and metal products Subtotal Non-traditional

3.7% 3.4% 19.6% 20.5%

5.2% 12.6%

14.0% 16.1%

3.7% 14.0%

3.0% 23.3%

9.3% 8.4%

12.9% 14.2%

2.2% 16.1%

7.8% 7.3%

4.3% 5.4%

4.4% 12.1%

5.7% 9.3%

4.7% 17.6%

8.8% 4.4% 7.4% 14.5% 21.4% 14.9%

6.0% 10.9% 14.0%

7.6% 7.4% 9.9%

30.9% 6.2% 10.0%

3.8% 4.5% 10.4%

5.9% 7.6% 8.3%

0.8% 1.4% 7.9%

1.1% 0.5% 10.0%

3.0% 0.3% 3.8%

0.3% 0.9% 5.4%

a/ a/ 4.9%

a/ a/ 7.0%

7.6% 9.2% 16.4%

57.2% 54.2%

43.5%

41.0%

61.2%

42.1%

30.2%

24.3%

27.8%

14.3%

12.0%

17.0%

16.3%

50.9%

Colombia

Peru

Brazil

Mexico Argentina Chile

Uruguay Ecuador Bolivia Paraguay Costa Rica

B. Most recent data

2013

2013

2002

2013

2013

2011

2011

2008

2010

2010

2013

Total

Food, beverages, and tobacco Textiles, apparel, leather, and footwear Non-metallic minerals Other traditional industries (wood and furniture, printing, and other manufacturing) Subtotal Traditional

23.1% 19.8% 5.4% 2.6% 3.6% 2.8% 4.0% 1.6%

37.5% 6.1% 1.7% 3.0%

30.6% 1.7% 5.9% 6.7%

28.0% 5.4% 5.1% 4.3%

34.0% 9.0% 6.8% 3.3%

44.3% 5.6% 1.6% 3.9%

41.2% 3.4% 4.7% 4.7%

49.1% 4.0% 9.1% 3.9%

55.1% 7.3% 4.7% 5.6%

48.7% 2.3% 4.7% 4.3%

24.2% 4.7% 3.6% 3.4%

36.0% 26.9%

48.3%

44.9%

42.8%

53.1%

55.4%

54.0%

66.2%

72.7%

59.9%

36.0%

Oil refineries Paper and chemical industry, excluding oil refineries Basic Metals Transport equipment Machinery and equipment and metal products Subtotal Non-traditional

6.5% 19.9% 6.8% 13.8% 16.9% 57.5%

12.3% 20.4% 6.5% 6.2% 6.4% 39.5%

15.3% 26.6% 3.6% 0.7% 8.9% 39.8%

22.5% 19.6% 5.3% 4.1% 5.7% 34.7%

13.5% 12.2% 13.8% 2.1% 5.3% 33.4%

16.3% 19.3% 1.6% 3.4% 4.0% 28.4%

13.5% 16.2% 6.3% 5.0% 4.9% 32.5%

11.2% 14.2% 6.0% 0.3% 2.2% 22.6%

0.2% 14.9% 4.3% 2.0% 5.9% 27.1%

7.2% 22.5% 2.8% 0.3% 7.3% 32.9%

10.3% 19.3% 6.9% 14.3% 13.1% 53.6%

Sources: A. Bértola and Ocampo 2012: table 4.7. B: UNIDO

14.3% 18.4% 7.1% 23.4% 10.0% 58.8%

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

Food, beverages, and tobacco 13.3% 15.3% Textiles, apparel, leather, and footwear 12.3% 15.5% Non-metallic minerals 5.5% 5.9% Other traditional industries (wood and furniture, 8.0% 5.7% printing, and other manufacturing) Subtotal Traditional 39.1% 42.4%

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´   7.0% 6.0% 5.0% 4.0% 3.0% 2.0%

0.0%

1900 1905 1910 1915 1920 1925 1930 1935 1940 1945 1950 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015

1.0%

Serie 1

Serie 2

Serie 3

 . GDP decennial growth rates (moving average annual growth rate over the decade that ends in the year indicated in the graph) Serie 1 includes Argentina, Brazil, Chile, Colombia, Cuba, Ecuador, Mexico, Peru, Uruguay, and Venezuela (the first two data points exclude Cuba and Ecuador) Serie 2 incluyes all countries, except Bolivia, Panama, Paraguay, and the Dominican Republic Serie 3 includes all countries Source: Bértola and Ocampo (2012: figure 4.5), updated with ECLAC data.

been achieved in passing during the commodity export age, and would disappear during the market reform era. Growth was also accompanied by a rapid advance in social development and, in contrast again with the ‘Black Legend’, with a high degree of macroeconomic discipline until the early 1970s (low or moderate inflation, and low fiscal external deficits). The exception to the latter rule were the Southern Cone countries and Brazil, but also an increasing number of countries after the first oil shock, which finally led to the debt crisis of the 1980s. In any case, runaway inflation was only a feature of the 1980s, and in this sense more an effect than a cause of the debt crisis (Bértola and Ocampo 2012: chs 4 and 5). Finally, the industrialization process was also accompanied by local technological development and capacity-building. This included an active learning process and an effort to adapt technologies, which gave rise to a considerable number of secondary innovations, associated with the elimination of specific bottlenecks, adjusting to smaller production scales and use of local raw materials, and redesigning products for local markets; some of these innovations also facilitated export diversification.⁹ Total factor productivity increased at a fast rate between 1960 and the mid-1970s, exceeding that of the USA, particularly in the larger economies, but then showed signs of stagnation in ⁹ See Katz (1984) and Katz and Kosacoff (2000). See also Teitel and Thoumi (1987) on the transition from import substitution to export activity in Argentina and Brazil.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



the second half of the 1970s (IDB 2010). However, the incipient innovation systems were not strong enough to create technological networks of the type that were being developed in East Asia. As a result, Latin America’s state-led industrialization was incapable of building up a solid endogenous technological base that would support the continuous diversification into technology-intensive sectors. Interestingly, agriculture rather than manufacturing was the best case in terms of development of national technological capacities in several countries.

19.3 M R  P D-

.................................................................................................................................. Whether the region could have developed more solid innovation systems over the next decades as export diversification and regional integration expanded and the protection structure was rationalized is an open question, because the development process was shocked by the macroeconomic events surrounding the debt crisis of the 1980s. The crisis was mainly the result of the boom–bust cycles in international finance associated with the recycling of petro-dollars and the shocks generated by the increase in interest rates in the USA in 1979 rather than by any structural distortions generated by state-led industrialization. In fact, it hit the economies that had liberalized in the 1970s equally or worse than those that maintained the ‘mixed model’. The crisis was characterized by recurrent external debt renegotiations, and strong balance of payments and fiscal adjustments, which led to the worst recession in the region’s history and runaway inflation in several countries, including hyper-inflation in five of them. Under the influence of changing conceptual paradigms at the world level, and the strong direct or indirect influence of the ‘structural reforms’ agenda pushed by the Bretton Woods institutions—clearly a misnomer in light of the meaning of the concept ‘structure’ in this volume and in the Latin American structuralist tradition—the region became the leader of market reforms in the developing world. The reforms that had been introduced in Chile and to a lesser extent in other Southern Cone countries in the mid-1970s became a regional wave between the mid1980s and the mid-1990s. They included widespread trade liberalization and the elimination of most restrictions on FDI, domestic and external financial liberalization, and a diverse menu of privatization experiences and opening to private investment of areas traditionally reserved for the state. As part of that agenda, most production sector policies—industrial and, to a lesser extent, agricultural policies—were essentially dismantled. Some countries kept some export diversification instruments, including the promotion of free-trade zones, encouragement of local production clusters, and horizontal (i.e. non-selective) policies, such as research funds and support for microenterprises and small and medium-sized businesses. Several development banks were maintained, but generally with more limited function. These policies were seen as

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

part of strategies to enhance ‘competitiveness’ (a term that did not carry under the new paradigm the negative connotations of ‘industrial policy’), but largely focused on supporting existing branches of activity and exports rather than diversifying the production structures, which had been an essential feature of state-led industrialization. Some exceptions to the general rule of policy interventions included protection for the automotive industry in the South American integration processes, promotion of forestry and related industries in several countries, and support for investments by individual firms with a strong potential to develop new sectors, such as the entry of INTEL in Costa Rica (Péres 2012). Some institutions, notably ECLAC and the Development Bank of Latin America (CAF), advocated for a renewed form of production sector policies appropriate for the more open economies that the region had developed. Gradually, and particularly after the crisis of the late 1990s and early 2000s, these views have been partly reflected in the greater acceptance of more active industrial policies (Devlin and Moguillanksy 2012). This includes sectorial policies aimed at exploitation of the linkages associated with new natural resources being developed. The most important have been Brazil’s policy of building new production capacity around its deep-sea oil discoveries, Uruguay’s support for its pulp and paper industries, and Bolivia’s encouragement of new manufacturing activities around its gas and mining sectors. However, there has not been a return to more ambitious industrial and, more broadly, active production sector development strategies aimed at diversifying the production structure. As we will see later in this section, support for research and development has also remained limited. The region experienced a growth recovery in the 1990s, sharply reduced inflation (with a final experience with hyper-inflation in Brazil in 1993–94), and renewed access to international capital markets. However, growth patterns since then have differed from the period of state-led industrialization in two significant ways: a much sharper business cycle and slower long-term growth. Growth was indeed interrupted for about half of a decade by the effects of the East Asia and Russian crises of 1997–98, for a shorter period by the 1994 Mexican crisis (in limited number of countries in this case) and the 2008–09 North Atlantic financial crisis,¹⁰ and, at the time of writing, is experiencing a new strong downturn in South America that has already lasted three years. This implies that external openness has made the region highly vulnerable to external shocks, both positive and negative, with (generally) procyclical macroeconomic policies tending to reinforce rather moderate externally-induced cyclical swings. Greater macroeconomic stability in terms of price and the fiscal accounts has therefore been coupled with greater growth instability. The decade 2003–13 is the only period in which the region has approached the growth performance of state-led industrialization, and particularly the first part of that

¹⁰ This is the term I will use here rather than that of global financial crisis, because although it had global effects, the financial crisis was concentrated in the USA and Western Europe.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



Table 19.2 GDP growth rates Argentina Bolivia Brazil Chile Colombia Costa Rica Cuba Dominican Rep. Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Venezuela Latin America Mexico and Central America Excluding Mexico South America Excluding Brazil

1950–80

1990–2015

3.3% 3.2% 7.0% 3.5% 5.1% 6.3% n.d. 5.8% 6.1% 4.1% 5.0% 4.3% 6.6% 4.1% 6.1% 5.4% 4.9% 2.2% 6.0% 5.5% 6.4% 5.1% 5.1% 4.0%

3.7% 4.1% 2.6% 4.9% 3.7% 4.6% 1.9% 5.3% 3.3% 3.0% 3.7% 3.7% 2.7% 3.4% 6.1% 3.1% 4.8% 3.3% 2.1% 3.1% 3.0% 4.5% 3.4% 3.9%

Source: Author’s estimates based on ECLAC data.

decade—that is, prior to the North Atlantic financial crisis (Figure 19.2).¹¹ The slowdown of growth relative to the period of state-led industrialization was a fairly generalized process, and was particularly strong for Brazil, Mexico, and Venezuela (Table 19.2). The only exceptions were the economies that had performed poorly during the previous period (the three Southern Cone economies and Bolivia), and a couple of small economies that performed well in both periods (Dominican Republic and Panama). Weak performance was also reflected in productivity trends, which have also shown poor performance after market reforms (Palma 2011). Total factor productivity experienced a strong negative trend from the early 1980s to the early 2000s (IDB 2010) and for the period 1990–2013 it showed a negative outcome if adjusted by ¹¹ Growth in 2003–08 was 5.2% a year or 3.7% per capita, not far in per capita terms from that achieved in 1967–74 (4.0% a year), the highest in the region’s history. However, per capita GDP growth comparisons over time are not as relevant as those for GDP, as Latin America has experienced over the last quarter century a ‘demographic dividend’ (falling dependency ratios) whereas it experienced during a large part of state-led industrialization a population explosion (a ‘demographic tax’). Labour force growth has in fact been similar during market reforms and state-led industrialization periods as a whole.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

human capital (ECLAC 2014: part II, ch. III). This means that the reformers’ expectation that greater openness to the rest of the world and improvements in macroeconomic management would lead to more robust productivity and economic growth has not materialized. The reasons for this poor growth and productivity performance must be found in adverse structural transformation trends. The most remarkable in this regard is the strong de-industrialization that has taken place (Figure 19.1). This process may be considered a ‘premature de-industrialization’, as it involved a reduction in the share of manufacturing in employment and GDP at much lower levels of per capita GDP than those at which this process started to take place in developed countries (Palma 2005). This trend was particularly sharp during the debt crisis of the 1980s but continued in the following quarter century of renewed growth. Whereas the phenomenon in the 1990s can be considered a direct effect of trade liberalization, the further reduction in the new century and, particularly, after the North Atlantic crisis, may be seen as a result of the Dutch disease effects generated in South America by the commodity price boom. In more conjunctural terms, macroeconomic crises have always hit manufacturing particularly hard. This is true of the 1980s, the recession that followed the East Asian crisis, and the recent downturn associated with the fall in commodity prices. De-industrialization was experienced by most countries in the region (Table 19.3) but was particularly strong in South America, the sub-region that, as we will see, experienced a renewed specialization in natural resource-based exports. The exception was the period 2003–07 (or more precisely 2002–06), particularly if we exclude Brazil, leading in fact to the fastest growth of any sub-region during the market reform era (7.6 per cent a year excluding Brazil, and 5.8 per cent a year including it). Interestingly, Mexico has been part of the de-industrialization trend since the early 2000s, indicating that the growth of manufacturing value added has been much slower than that of manufacturing exports, due both to the large import content of its exports but also the growing competition of imports with production for the domestic market. The only exceptions to the de-industrialization trade have been some Central American economies (El Salvador and Nicaragua, where it increased slightly in 1990–2015 and Honduras where it remained constant). In sectorial terms, the most important trends have been the growing share of the food and beverage industry throughout the region, the transport industries of Mexico and Brazil, and oil refining in most countries (Table 19.3). The former reflects the attraction of growing domestic demand but also growing exports of processed foodstuffs and beverages (mainly wine) by some countries. The automotive sector has benefited in Mexico from integration in the US market in the context of the North American Free Trade Agreement, and in Brazil by special protection measures in the context of MERCOSUR (Southern Common Market). Some of the smaller economies have also seen a rise of non-traditional activities, but these sectors have generally remained smaller than in the larger economies. In contrast, with the exception of the food and beverage industries, all other traditional sectors shrank in relative terms throughout the region, with only a few exceptions. Notably in this regard is the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



Table 19.3 Manufacturing value added as a share of GDP, 1990–2015 Argentina Bolivia Brazil Chile Colombia Costa Rica Cuba Dominican Rep. Ecuador El Salvador Guatemala Honduras Mexico Nicaragua Panama Paraguay Peru Uruguay Venezuela Latin America

1990

1997

2003

2007

2013

2015

17.5% 11.3% 15.7% 14.8% 16.9% 14.0% 18.3% 18.0% 13.7% 17.9% 23.5% 16.1% 17.5% 13.6% 14.2% 15.3% 16.5% 18.5% 17.5% 16.5%

16.9% 11.1% 14.9% 12.9% 13.2% 14.2% 20.9% 19.2% 13.0% 18.1% 21.4% 16.4% 18.6% 14.2% 13.5% 13.7% 16.2% 14.1% 16.7% 15.9%

15.8% 11.0% 14.0% 11.9% 13.9% 14.8% 18.6% 18.4% 13.9% 19.5% 19.8% 18.3% 17.7% 14.2% 9.3% 13.1% 15.9% 12.2% 15.0% 15.1%

15.9% 11.6% 13.9% 11.8% 14.2% 15.6% 15.7% 15.9% 13.4% 18.5% 19.1% 17.6% 17.1% 15.3% 7.8% 12.0% 16.6% 14.0% 14.0% 14.9%

15.6% 11.0% 12.0% 10.5% 11.6% 13.8% 15.6% 14.8% 12.5% 19.0% 18.4% 16.3% 16.7% 14.7% 6.4% 10.6% 15.0% 11.8% 12.1% 13.4%

14.8% 10.8% 10.8% 10.3% 11.0% 13.0% 14.9% 14.3% 12.2% 19.2% 18.1% 16.0% 17.0% 14.0% 5.7% 11.0% 13.5% 12.4% 11.7% 12.1%

Note: The data for Cuba and Venezuela shown as 2015 corresponds to the year 2014. Source: Author’s estimates based on ECLAC data.

contraction of the textile, apparel, leather, and footwear industries; a partial exception in this regard are the maquila activities in some countries in the northern part of the region. Among the traditional sectors, construction materials (non-metallic minerals) were subject to weaker contraction and tended to expand during periods of booming domestic demand. One way to understand the changes in the industrial structures that took place is through the Schumpeterian concept of ‘creative destruction’. In several cases, the destruction of firms and sectors unable to thrive in the more competitive environment may have prevailed over the creative effects of market reforms. Resources displaced from non-competitive sectors were not properly absorbed by dynamic firms and activities. So, in a sense, there was ‘destructive creation’ rather than ‘creative destruction’ (Ocampo 2016). An additional important feature, using Hirschman’s terminology, was the destruction of domestic linkages created during the period of state-led industrialization. The increasing integration of the world economy made it easier for dynamic firms to import intermediate products and equipment, which helped these firms boost their productivity, but at the cost of the destruction of existing national value chains. In turn, and to a large extent due to the instability experienced by regional integration processes (see later in this section), regional value chains did not replace the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

weakened national ones; the partial exception in this regard has been the automobile sector in the two South American integration processes (MERCOSUR and the Andean Community), where the sector received special protection. The poor aggregate productivity performance is not inconsistent with the productivity gains achieved by successful firms and sectors (agribusiness, mining, modern telecommunications, financial and transport services), which benefited from growing integration into the world economy and FDI. This is also true of the emergence of the Latin American multinational firms (‘multilatinas’). It reflects, rather, the fact that this process has been accompanied by the growth of a number of low-productivity small businesses and microenterprises, and the associated rise of labor market informality,¹² strengthening the region’s internal dualism—or structural heterogeneity, to utilize the classical ECLAC concept. What this means is the under-utilization of existing resources prevailed over the reformers’ expectations that the productivity growth in internationalized sectors would spread out into the economies. Agriculture withstood the transition to the new development model much better than manufacturing, no doubt thanks to the development of new agricultural export activities, supported in this case by stronger technological capacities in this sector in several countries. This is, notably, the case of Brazil and Argentina, and has spilled into the expansion of the agricultural sector in some other South American countries. It is also reflected in the expansion of the fruit, wine, and vegetable industries in other nations, notably in Chile, Costa Rica, and Peru. In any case, although the share of Latin America in world agricultural production continued to increase,¹³ overall growth of regional agricultural production between 1990 and 2015 was 2.5 per cent a year according to ECLAC national accounts data; this is slower than that recorded in 1950–80 at 3.7 per cent a year. This indicates that the elimination of the previous trade regime’s presumptive ‘anti-agricultural bias’ did not deliver the positive effects on agriculture that market-reform advocates had expected. So, overall, the most robust production sectors were modern services (i.e. public utilities, transport, communications, and financial and business services). In total, according to ECLAC data, the share of these dynamic services in GDP (at 2010 prices) increased from 21.2 per cent in 1990 to a peak of 26.1 per cent in 2014. In contrast to the period of state-led industrialization, where the growth of these sectors had accompanied the rise of manufacturing activities, they now kept a dynamic of their own. The subcontracting of services by manufacturing firms may have reinforced this ‘tertiarization’ of the Latin American economies and could indicate that there is a partial overestimation of the de-industrialization trend. The oil and mining sectors also

¹² According to ECLAC data, low-productivity employment (in microenterprises, domestic service, and unskilled self-employed) rose from 41.0% of employment in 1990 to a peak of 48.3% in 2002 and then fell to 43.4% in 2014, still a higher proportion than in 1990. ¹³ According to FAO data referred to above, from 10.5% in 1980 and a similar share in 1990, to 12.9% in 2013.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



grew rapidly in resource-rich countries in several periods, boosted by institutional reforms designed to open up more areas to the private sector and FDI. Trade openness was reflected in the growing share of exports in GDP: from 12.6 per cent in 1990 at 2010 prices to a 20–22 per cent range since 2003. This process had started before the debt crisis but speeded up during the crisis—exports serving as a partial mitigating mechanism—and as a result of market reforms. The region’s share in world trade also increased, thanks in particular to the rapid growth of Mexican exports, reaching about 6 per cent of world trade, still significantly below the 8 per cent reached during the export age (Bértola and Ocampo 2012: figures 1.3 and 5.8). This was accompanied by a falling share of natural resource-based exports through the 1980s and 1990s, associated with growing manufacturing exports but also to falling commodity prices. In contrast, there was a clear ‘re-primarization’ of the export structure during the super-cycle of commodity prices of the early twenty-first century. So, the share of natural-resource based exports, which had fallen from 77 per cent of total exports in the early 1980s to less than half by the end of the century, increased again to a share similar to that of 1980 by the end of the boom (Figure 19.3(a)). Export growth followed different patterns in the north and south of the region. The ‘northern’ pattern includes diversification toward manufacturing exports, most of them with a large component of imported inputs (in its most extreme form, maquila industries), and are primarily destined for the US market. This pattern is particularly strong in Mexico and Costa Rica, where it includes a growing share of mid- and hightechnology manufactures (again, with significant import components), as well as in other Central America countries, characterized in this case by a larger share of lowskilled manufactures and commodity exports (Table 19.4). The ‘southern’ pattern has changed less radically relative to old patterns, and continues to mix a significant share (b) By destination, 2013 100%

90%

90%

80%

80%

70%

70%

60%

60%

50%

50%

40%

40%

30%

30%

20%

20%

10%

10%

0%

0%

19

81

–8 2 19 85 –8 6 19 91 –9 2 19 95 –9 6 19 98 –9 20 9 01 –0 2 20 05 –0 6 20 08 –0 20 9 12 –1 3

(a) Historial evolution 100%

High-tech manufactures

Mid-tech manufactures

Low-tech manufactures

Latin United European Rest of America States Union Asia Resource-based manufactures

China

Primary Goods

 . Natural resource and technological contents of Latin American exports Source: ECLAC.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

of commodities and natural-resource-based manufactures, which have tended to increase in several countries, with a large share of manufactures in intra-regional trade. Brazil is positioned somewhere between the two groups, since it had a much more diversified export structure in 1990 (including some technology-intensive manufactures) than the other South American countries, but it is one of the countries that has experienced one of the strongest re-primarization of its export structure over the past quarter century. There is also a third specialization pattern in some smaller economies, where service exports dominate: transport and to a lesser extent financial services in the case of Panama, and tourism in the Dominican Republic and Cuba (not included in Table 19.4). Manufacturing constitutes the dominant share of exports in intra-regional trade (Figure 19.3(b)). The major weakness of intra-regional trade has been its great sensitivity to business cycles and political tensions, as reflected in the strong downswing that was experienced in the later 1990s in the two South American integration processes after an impressive boom during most of the decade. It was also visible in the collapse of Colombian exports to Venezuela in 2008–10 due to political strain between the two countries. Exports to the USA also include a sizeable share of manufacturing, which is not, however, as marked if Mexican exports are excluded.¹⁴ Exports to the European Union are much more commodity dependent, and this is particularly true of exports to Asia. In this regard, re-primarization was enhanced by growing trade with China, which became a major trading partner of Latin America in the twenty-first century, particularly after the North Atlantic financial crisis. Natural resource sectors represented over 90 per cent of sales to the Asian giant in 2013. Booming Chinese imports also supported the re-primarization process by weakening the manufacturing sectors in several (or most) countries (Gallagher 2016). Trade specialization and the nature of FDI flows have been closely interlinked. So, the ‘northern’ pattern of specialization has attracted transnational corporations that are active players in international value chains, whereas South America has attracted a larger share of FDI in natural resource sectors. Services have been important for both groups. There has been a group of Latin American firms that joined the ranks of worldclass transnational corporations. The largest ‘multilatinas’ come from Brazil and Mexico and are active players in global markets, but there is also a group of smaller firms from various countries (including several middle-sized ones) that play an important role in regional markets. The fact that Latin America’s market reforms have been relatively successful in creating new forms of integration into the world economy but have not generated rapid economic growth is one of the paradoxical effects of market reforms—and a remarkable one from the perspective of orthodox economic analysis. This fact has been subject to significant debate. A major reason, according to some authors, is that the region’s share of world markets has been increasing in sectors in which markets are not very ¹⁴ According to ECLAC data, if we exclude Mexico, primary goods and natural resource-based manufactures represented 68.3% of exports to the USA in 2014 vs. 29.9% if we include Mexico.

Table 19.4 Composition of exports of goods and services, 1990 and 2013–14 Mexico

Costa Rica

El Salvador

Guatemala

Honduras

Nicaragua

Panama

1990 2014 1990 2013 1990 2014 1990 2014 1990 2014 1990 2014 1990 2013

Dominican Rep. 1990

2014

Brazil 1990 2014

Primary goods Resource based manuf Low-tech manuf Mid-tech manuf High-tech manuf Oher goods Total goods

37.4% 8.3% 5.4% 21.3% 3.4% 0.6% 76.5%

13.4% 7.4% 8.8% 42.1% 21.1% 2.3% 95.0%

41.2% 7.5% 9.0% 4.3% 2.2% 6.3% 70.5%

15.7% 11.5% 7.2% 13.8% 16.9% 0.4% 65.7%

29.9% 5.7% 12.1% 4.9% 2.6% 0.2% 55.4%

3.1% 16.2% 44.6% 5.6% 4.7% 1.4% 75.6%

44.5% 15.9% 7.3% 4.8% 3.9% 0.1% 76.6%

29.5% 19.4% 17.7% 9.3% 2.5% 1.6% 80.1%

65.5% 9.8% 3.7% 0.9% 0.1% 0.2% 80.2%

32.5% 18.9% 7.9% 17.1% 0.4% 3.6% 80.5%

62.8% 13.5% 3.1% 1.8% 0.1% 3.7% 85.0%

29.4% 13.9% 25.8% 10.8% 0.3% 6.8% 86.9%

13.7% 5.2% 2.7% 1.0% 0.7% 0.6% 23.8%

3.3% 3.0% 6.3% 21.6% 1.9% 3.7% 14.1% 26.7% 0.5% 10.0% 15.3% 13.1% 0.1% 6.6% 11.5% 22.9% 0.1% 0.1% 2.5% 3.9% 0.1% 46.9% 9.6% 1.2% 6.1% 70.2% 59.3% 89.3%

42.5% 15.6% 5.9% 14.7% 3.7% 2.7% 85.0%

Services Transport Travel Other Total services

2.6% 16.0% 4.9% 23.5%

0.0% 3.9% 1.1% 5.0%

4.6% 13.8% 11.1% 29.5%

2.4% 14.9% 17.1% 34.3%

10.7% 10.3% 23.6% 44.6%

7.7% 11.5% 5.1% 24.4%

1.5% 7.8% 14.2% 23.4%

2.8% 11.4% 5.7% 19.9%

6.1% 4.2% 9.5% 19.8%

1.8% 11.3% 6.5% 19.5%

1.7% 3.1% 10.3% 15.0%

0.8% 7.8% 4.5% 13.1%

41.0% 12.0% 23.2% 76.2%

39.7% 0.7% 0.0% 3.8% 27.5% 25.5% 19.3% 3.9% 26.7% 3.5% 21.3% 2.9% 93.9% 29.8% 40.7% 10.7%

1.0% 2.6% 11.4% 15.0%

Primary goods Resource based manuf Low-tech manuf Mid-tech manuf High-tech manuf Oher goods Total goods

Bolivia

Chile

Colombia

Ecuador

Paraguay

Peru

Uruguay

1990 2014 1990 2014 1990 2014 1990 2014 1990 2014 1990 2014 1990 2014

1990

38.5% 24.4% 10.0% 8.8% 1.5% 0.3% 83.5%

40.8% 43.6% 81.4% 10.0% 14.4% 4.0% 19.7% 7.2% 3.7% 6.7% 6.0% 4.4% 0.5% 1.4% 0.2% 0.8% 0.9% 0.2% 78.6% 73.5% 93.8%

40.0% 14.9% 3.0% 18.9% 1.8% 4.5% 83.2%

48.4% 34.8% 2.9% 0.1% 0.0% 0.1% 86.3%

71.5% 7.9% 1.7% 0.7% 0.1% 9.7% 91.5%

52.6% 21.2% 1.9% 2.4% 0.5% 3.5% 82.2%

37.4% 41.0% 2.5% 4.4% 0.8% 1.3% 87.4%

52.5% 10.5% 10.8% 5.1% 0.4% 1.7% 80.9%

62.0% 10.4% 5.3% 6.8% 1.4% 3.1% 89.0%

74.3% 7.6% 0.9% 0.4% 0.2% 0.1% 83.5%

74.4% 10.1% 1.7% 1.9% 0.6% 3.1% 91.8%

58.9% 6.0% 3.8% 0.8% 0.0% 0.1% 69.6%

54.4% 8.6% 4.7% 2.5% 0.7% 0.3% 71.2%

37.3% 29.1% 11.8% 2.1% 0.3% 0.0% 80.6%

54.5% 23.1% 5.6% 3.1% 0.4% 0.2% 86.9%

2014

Venezuela 1990 2013 83.4% 12.6% 0.4% 1.1% 0.0% 0.1% 97.6%

Services Transport Travel Other

7.8% 3.1% 4.4% 2.0% 6.9% 5.5% 5.8% 2.8% 7.4% 1.5% 5.4% 3.7% 0.0% 3.4% 7.8% 5.1% 6.1% 5.6% 5.4% 4.5% 5.1% 2.6% 4.9% 6.2% 5.8% 5.2% 6.2% 2.7% -1.7% 6.8% 11.0% 13.9% 2.6% 8.1% 3.8% 2.1% 5.8% 4.5% 8.5% 2.0% 3.3% 1.5% 18.8% 22.4% 21.1% 2.9% 2.7% 7.5%

2.4% 0.7% 2.6% 1.0% 1.2% 0.7%

Total services

16.5% 16.8% 13.7% 8.5% 17.8% 12.6% 19.1% 11.0% 16.5% 8.2% 30.4% 28.8% 19.4% 13.1% 21.4% 26.5%

6.2% 2.4%

Source: Author’s estimates based on data from UN-COMTRADE and ECLAC data on service exports.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

Argentina

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

dynamic, in contrast to East Asia, where the opposite trend has prevailed (ECLAC 2001; Palma 2011). Hausmann (2011) has argued, in turn, that the region has specialized in sectors that offer fewer opportunities for new production activities or for making improvements in product quality–two areas of progress that are considered essential to induce rapid economic growth. In this author’s terminology, the region has tended to specialize in a portion of the ‘product space’ that offers fewer opportunities to diversify into activities that would enhance the ‘complexity’ of the production structure. More broadly, it can be argued that the essential problem has been the region’s incapacity to develop strong endogenous technological capabilities, which is essential for upgrading into sectors with high knowledge content. Thus, the region became one of the world’s most remarkable cases of what is known as the ‘middle-income trap’ (Jankowska et al. 2012; Lin and Treichel 2012; Paus 2016).¹⁵ What this means is that the sluggishness of the countries’ national innovation systems, that had been a feature of state-led industrialization, not only remained in place but was enhanced by market reforms. This is reflected in several indicators of technological development: the lower share of engineering-intensive industries, the limited levels of research and development (R&D) and, even more remarkably, the large gap in patenting in relation to both developed countries rich in natural resources and the dynamic Asian economies (Cimoli and Porcile 2011; ECLAC 2007, 2012). It is also reflected in the number of scientists or of scientific publications per million inhabitants (ECLAC 2008: table III.1).¹⁶ Table 19.5 indicates that, although R&D spending has increased since the 1990s, it did so at a much more moderate rate than in East Asia and the Pacific, and particularly in China. The current levels of R&D in Latin America (0.8 per cent of GDP) are a fraction of those of high income countries (2.5 per cent) and East Asia and the Pacific (close to 2.0 per cent) and remain below the average of the two categories of middle income countries. This is true even of Brazil, the country that records the highest levels of R&D spending in Latin America. Also, Latin America’s R&D investment comes largely from the government (over half ) vs. the pattern in leading countries in innovation, which rely more on the private sector. An important question is whether specialization in sectors where the region has a clear static comparative advantage—natural resource-based exports—are inherently unable to deliver rapid economic growth. There is a diversity of views on this question. Some authors do in fact claim that there is an inherent incapacity of natural resources to deliver rapid growth, confirming Prebisch’s claim that these sectors are not as strong as manufacturing as mechanisms for the transmission of technical progress. This is consistent with the observation that rapid economic growth in emerging and developing countries has been associated with industrialization drives (Rodrik 2014). In contrast to this view, it can be argued some developed countries have been able to ¹⁵ See an analysis of ‘middle-income traps’ by Agénor et al. (2012) and Eichengreen et al. (2013). ¹⁶ For a more extensive discussion of this subject, see ECLAC (2008: ch III and IV), ECLAC (2010: ch. III), and ECLAC (2012: ch. II).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



Table 19.5 Research and development expenditure (as % of GDP) Argentina Bolivia Brazil Chile Colombia Costa Rica Cuba Ecuador Guatemala Honduras Mexico Nicaragua Panama Peru Paraguay El Salvador Uruguay Venezuela Region Latin America & Caribbean East Asia & Pacific (excluding high income) China According to income High income Upper middle income Middle income

1996

2000

2005

2010

2014

0.42 0.33

0.44 0.29 1.00

0.38

0.61

0.30 0.30 0.35 0.07

0.11 0.39 0.48 0.05

0.15 0.43 0.55 0.13 0.04

0.52 0.16 1.16 0.33 0.20 0.48 0.61 0.40 0.04

0.40

0.45

0.54

0.26 0.07 0.33 0.09

0.04 0.32

1.00

1.24 0.38 0.20 0.56 0.41

0.38 0.11 0.07

0.25

0.15

0.07

0.07

0.27

0.21

0.37

0.06 0.07 0.34

0.08 0.33

0.46

0.56 0.73

0.62 1.19

0.77 1.70

0.82 1.96

0.57

0.90

1.32

1.73

2.05

2.18

2.34 0.68 0.64

2.25 0.89 0.85

2.43 1.20 1.13

2.46 1.57 1.36

Note: Some data refer to the year before or after that indicated in the table. Source: World Bank’s World Development Indicators.

prosper with commodity dependence.¹⁷ In recent years, Pérez (2010) has put forward the best defence of the development opportunities that natural resource sectors generate for Latin America. She argues that these sectors provide ample opportunities associated with biotechnology, nanotechnology, and environmentally-friendly products, opportunities to exploit their own value chains, and strong complementarities with Asia. Her views are confirmed at the regional level in the successful experiences with agricultural and agro-industrial exports—fruits, vegetables, and wine, but also soybean and corn exports—that include important technological content in the production and handling of the associated products. In contrast, her (correct) evaluation is ¹⁷ One interesting (now classical) analysis, is the comparative history of Scandinavian and Latin American economic history in the essays collected in Blomström and Meller (1991).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

that Latin America is too far behind in other technology-intensive sectors and is no longer a low-wage region, and so faces strong barriers to competing in both hightechnology sectors associated with ICT and in low-skilled manufactures. It should be added that the cyclical effects of commodity dependence may generate negative long-term effects. The basic reason is that commodity price cycles, reinforced by pro-cyclical capital flows, may generate an equally pro-cyclical performance of macroeconomic variables that have a strong effect on non-commodity tradable sectors—both exporting as well as import-competing—particularly strong real exchange rates fluctuations—real appreciation during commodity booms, depreciation during crises. This has negative cyclical effects on non-commodity sectors during commodity booms, and more generally increases the risks that these sectors face by making the profitability of investing in them highly volatile. A final paradox of the effects of market reforms has been the low correlation between GDP growth and export growth during market reforms (see, in particular, Bértola and Ocampo 2012: figure 5.11). Notably, in this regard, the rapid growth of Mexico’s manufacturing exports has delivered one of the poorest growth records since 1990, indeed quite similar to that of Brazil, a country that has experienced one of the strongest de-industrialization processes (Table 19.2). This appears to demonstrate that the destructive effects of liberalization—that is, the destruction of sectors and pre-existing national value chains—had a strong impact in the region’s two largest economies that prevailed over the positive effects of the new export activities that these countries developed—manufacturing exports in Mexico and natural-resource exports in Brazil. With similar or in some cases weaker export growth records, Chile, Costa Rica, the Dominican Republic, Panama, and Peru have delivered the fastest rates of growth in the region in 1990–2015 (over 4.5 per cent a year—see Table 19.2). This means that the development of new export sectors in these countries, which included a mix of high-technology manufactures in Costa Rica, agricultural activities with higher technological content and mining in Chile and Peru, and service exports in the Dominican Republic and Panama, had a net positive effect in terms of the creation of production capacities. In Chile, the most important among the faster growing countries in terms of prior industrial development, the destructive effects of liberalization were also felt earlier, in the 1970s.

19.4 C

.................................................................................................................................. Despite the ‘Black Legend’ that has surrounded it, state-led industrialization was a period of success for Latin America in terms of structural transformation and economic growth. In contrast, and in open contradiction of orthodox expectations, market reforms have delivered slow economic growth. This is associated with the destruction of previous production capacities, and particularly a fairly generalized de-industrialization process,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



that prevailed over the positive effects of the development of new export activities. Beyond the effects of reforms, Latin America has faced a strong ‘middle-income trap’, the conditions for which were sown in the last stages of state-led industrialization, when the region failed to make the transition to the development of strong innovation systems and higher levels of research and development. The challenges that the region faces have mounted with the recent collapse of the super-cycle of commodity prices, which had delivered the only period since market reforms during which performance approached the records achieved during the period of state-led industrialization. Renewing economic growth will require a return to active production sector strategies, breaking policy patterns that became entrenched during the period of market reforms. The essential focus of the new strategies must be the increased knowledge content of the new economic activities. This should include re-industrialization but also modern services and the exploitation of the opportunities of technological upgrading natural resources. This effort should go hand in hand with consolidation of the advances made in the coverage of the education system, although at the same time significantly improving its quality, and with significant efforts to reduce the lags in infrastructure development, particularly the transport infrastructure—two issues that go beyond the scope of this chapter. Given the sluggish growth of international trade which has become an essential feature of the world economy since the North Atlantic financial crisis, it is essential not only to boost competitiveness and improve the quality of the export basket, but also to strike a balance between the domestic and external markets. There are three options, which can be mixed in diverse ways according to national conditions. The first is to use the opportunities offered by domestic markets and, particularly, by the rise of the middle class. However, a focus on domestic markets will only work for Brazil and, to a much lesser extent, for some middle-sized countries. A second, broader-based opportunity is the revitalization of regional integration processes, which are characterized by a strong manufacturing trade structure. This requires, however, overcoming the political constraints that have been weakening the South American integration processes in recent years. The third strategy is diversifying exports in two different ways: by improving their knowledge content, and by expanding exports to the fast-growing Asian economies, particularly China. These two elements are complementary, since one of the main challenges is that of diversifying the region’s exports to the Asian giant.

A The literature on the issues covered here is massive; for this reason, references included are highly selective. This chapter draws from my joint book with Luis Bértola on the economic history of Latin America (Bértola and Ocampo 2012). I am grateful to Cristina Gutierrez for research assistance.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

R Agénor, Pierre-Richard, Otaviano Canuto, and Michael Jelenic, 2012. ‘Avoiding Middle-Income Growth Traps’, Economic Premises, No. 98, November, Washington, DC: World Bank. Anderson, Kym and Alberto Valdés, 2008. ‘Introduction and Summary’, in Kym Anderson and Alberto Valdés, eds, Distortions to Agricultural Incentives in Latin America, Washington, DC: World Bank, pp. 1–58. Bértola, Luis and José Antonio Ocampo, 2012. The Economic Development of Latin America since Independence, Oxford: Oxford University Press. Bielchowsky, Ricardo, 1988. ‘Cincuenta años de pensamiento de la CEPAL: Una reseña’, in Cincuenta años de pensamiento de la CEPAL, Vol. 1, Santiago de Chile: Fondo de Cultura Económica and ECLAC, pp. 9–62. Blomström, Magnus and Patricio Meller (eds), 1991. Diverging Paths: Comparing a Century of Scandinavian and Latin American Economic Development, Washington, DC: InterAmerican Development Bank and John Hopkins University Press. Bulmer-Thomas, Victor, 2014. The Economic History of Latin America since Independence, 3rd edn, Cambridge: Cambridge University Press. Cárdenas, Enrique, José Antonio Ocampo, and Rosemary Thorp (eds), 2000. Industrialization and the State in Latin America: The Postwar Years, Volume 3: An Economic History of Twentieth-Century Latin America, Houndmills: Palgrave, in association with St Antony’s College, Oxford. Cimoli, Mario and Gabriel Porcile, 2011. ‘Learning, Technological Capabilities and Structural Dynamics’, in José Antonio Ocampo and Jaime Ros, eds, The Oxford Handbook of Latin American Economics, New York: Oxford University Press, ch. 22. Coatsworth, John H. and Jeffrey G. Williamson, 2004. ‘Always Protectionist? Latin American Tariffs from Independence to Great Depression’, Journal of Latin American Studies, 36(2), pp. 205–32. Devlin, Robert and Graciela Moguillanksy, 2012. ‘What’s New in the New Industrial Policy in Latin America’, Policy Research Working Paper No. 6191, September, Washington, DC: World Bank. Díaz-Alejandro, Carlos F., 2000. ‘Latin America in the 1930s’, in Rosemary Thorp, ed., Latin America in the 1930s: The Role of the Periphery in the World Crisis’, Volume 2: An Economic History of Twentieth-Century Latin America, Houndmills: Palgrave in association with St Antony’s College, Oxford, ch. 2. Dosman, Edgar J., 2008. The Life and Times of Raul Prebisch, 1901–1986, Montreal: McGillQueens’ University Press. ECLAC (Economic Commission for Latin America and the Caribbean), 1998. Economic Survey of Latin America and the Caribbean 1997–1998, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean), 2001. A Decade of Light and Shadow: Latin America and the Caribbean in the 1990s, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean), 2007. Progreso técnico y cambio estructural en América Latina, Santiago: ECLAC and IDRC. ECLAC (Economic Commission for Latin America and the Caribbean), 2008. The Productive Transformation 20 Years After: Old Problems, New Opportunities, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean), 2010. Time for Equality: Closing Gaps, Opening Trails, Santiago: ECLAC.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’   



ECLAC (Economic Commission for Latin America and the Caribbean), 2012. Structural Change for Equality: An Integrated Approach to Development, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean), 2014. Economic Survey of Latin America and the Caribbean, Santiago: ECLAC. ECLAC (Economic Commission for Latin America and the Caribbean), and FAO (Food and Agricultural Organization), 1978. 25 años en la agricultura de América Latina: Rasgos principales 1950–1975, Cuadernos de la CEPAL, No. 21, Santiago: ECLAC. Eichengreen, Barry, Donghyun Park, and Kwanho Shin, 2013. ‘Growth Slowdowns Redux: New Evidence on the Middle-Income Trap’, National Bureau of Economic Research, Working Paper No. 18673, January. Ffrench-Davis, Ricardo, Oscar Muñoz, and Gabriel Palma, 1998. ‘The Latin American Economies, 1959–1990’, in Leslie Bethell, ed., The Cambridge History of Latin America, Latin America: Economy and Society Since 1930, Vol. 6, Cambridge: Cambridge University Press, ch. 4. Furtado, Celso, 1989. La fantasía organizada, Bogotá: Tercer Mundo. Gallagher, Kevin P., 2016. The China Triangle: Latin America’s China Boom and the Fate of the Washington Consensus, New York: Oxford University Press. Hausmann, Ricardo, 2011. ‘Structural Transformation and Economic Growth in Latin America’, in José Antonio Ocampo and Jaime Ros, eds, The Oxford Handbook of Latin American Economics, Oxford: Oxford University Press, ch. 21. Hirschman, Albert O., 1971. ‘The Political Economy of Import-Substituting Industrialization in Latin America’, in Albert O. Hirschman ed., A Bias for Hope: Essays on Development and Latin America, New Haven: Yale University Press, ch. 3. IDB (Inter-American Development Bank), 2010. The Age of Productivity: Transforming Economies from the Bottom Up, Washington, DC: Inter-American Development Bank. Jankowska, Anna, Arne Nagengast, and José Ramón Pere, 2012. ‘The Product Space and the Middle-Income Trap: Comparing Asian and Latin American Experiences’, OECD Development Centre, Working Paper No. 311, April. Katz, Jorge, 1984. ‘Domestic Technological Innovations and Comparative Advantage: Further Reflections on a Comparative Case-study Program’, Journal of Development Economics, 16 (1–2), pp. 13–37. Katz, Jorge and Bernanrdo Kosacoff, 2000. ‘Technological Learning, Institution Building and the Microeconomics of Import Substitutions’, in Enrique Cárdenas, José Antonio Ocampo, and Rosemary Thorp, eds, Industrialization and the State in Latin America: The Postwar Years, Volume 3: An Economic History of Twentieth-Century Latin America, Houndmills: Palgrave, in association with St Antony’s College, Oxford, ch. 2. Lin, Justin Yifu and Volker Treichel, 2012. ‘Learning from China’s Rise to Escape the MiddleIncome Trap: A New Structural Economics Approach to Latin America’, Policy Research Working Paper No. 6165, August, Washington, DC: World Bank. Love, Joseph L., 1994. ‘Economic Ideas and Ideologies in Latin America Since 1930’, in Leslie Bethell, ed., The Cambridge History of Latin America, since 1930. Economy, Society and Politics, Cambridge: Cambridge University Press, vol. 6, ch. 7. Ocampo, José Antonio, 2005. ‘The Quest for Dynamic Efficiency: Structural Dynamics and Economic Growth in Developing Countries’, in José Antonio Ocampo, ed., Beyond Reforms: Structural Dynamics and Macroeconomic Vulnerability, Palo Alto: Stanford University Press, World Bank and ECLAC, ch. 1.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



´  

Ocampo, José Antonio, 2016. ‘Dynamic Efficiency: Structural Dynamics and Economic Growth in Developing Countries’, in Akbar Noman and Joseph E. Stiglitz, eds, Efficiency, Finance, and Varieties of Industrial Policy, New York: Columbia University Press, ch. 3. Ocampo, José Antonio, and Juan Martin (eds), 2004. América Latina y el Caribe en la era global, Bogotá: ECLAC and Alfaomega. Ocampo, José Antonio, Codrina Rada, and Lance Taylor, 2009. Growth and Policy in Developing Countries: A Structuralist Approach, New York: Columbia University Press. Palma, Gabriel, 2005. ‘Four Sources of “De-Industrialization” and a New Concept of the “Dutch-Disease” ’, in José Antonio Ocampo, ed., Beyond Reforms: Structural Dynamics and Macroeconomic Vulnerability, Palo Alto: Stanford University Press, World Bank and ECLAC, ch. 3. Palma, Gabriel, 2011. ‘Why Has Productivity Growth Stagnated in Latin America since the Neo-Liberal Reforms?’, in José Antonio Ocampo and Jaime Ros, eds, The Oxford Handbook of Latin American Economics, Oxford: Oxford University Press, ch. 23. Paus, Eva, 2016. ‘Latin America’s Middle Income Trap: Where the Washington Consensus Meets Globalization’, Paper presented at the conference on the ‘Political Economy of the Middle Income Trap’, International Development Institute, King’s College, London, 24–25 February. Péres, Wilson, 2012. ‘Industrial Policies in Latin America’, in Adam Szirmai, Wim Naudé, and Ludovico Alcorta, eds, Pathways to Industrialization in the Twenty-First Century: New Challenges and Emerging Paradigms, New York: Oxford University Press, ch. 8. Pérez, Carlota, 2010. ‘Technological Dynamism and Social Inclusion in Latin America: A Resource-Based Production Development Strategy’, CEPAL Review, No. 100 (April), pp. 121–41. Rodrik, Dani, 2014. ‘The Past, Present and Future of Economic Growth’, in Franklin Allen et al., Toward a Better Global Economy, Oxford: Oxford University Press, ch. 2. Rosenthal, Gert, 2004. ‘ECLAC: A Commitment to a Latin American Way toward Development’, in Yves Berthelot, ed., Unity and Diversity in Development Ideas: Perspectives from the UN Regional Commissions, Bloomington, IN: Indiana University Press (United Nations Intellectual History Project), ch. 4. Teitel, Simón and Francisco E. Thoumi, 1987. ‘De la sustitución de importaciones a las exportaciones manufactureras de la Argentina y el Brasil’, Desarrollo Económico, 27 (105), pp. 29–60. World Bank, 1995. Bureaucrats in Business: The Economics and Politics of Government Ownership, Washington, DC: World Bank, Policy Research Report No. 4.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

    ’        An Exception and the Rule ......................................................................................................................

 

20.1 I

.................................................................................................................................. T process of industrialization and development is associated with a structural transformation of economies, which is a multidimensional phenomenon. In a longterm perspective, its most important dimension is structural change in the composition of output and employment over time. This is reflected in changing shares of the primary sector, secondary sector, and tertiary sector in national income and total employment for an economy. The primary sector is made up of agriculture, livestock, forestry, and fishing, although it is often described simply as agriculture. The secondary sector is made up of manufacturing, mining, construction, and utilities (electricity, gas, water), although it is often described simply as industry. The tertiary sector is made up of a diverse range of services, as distinct from goods, and is often described simply as services. In the literature on the subject, the words ‘structural transformation’ and ‘structural change’ are frequently used interchangeably. Strictly speaking, however, the former is an outcome while the latter is a process. The object of this chapter is to analyse the process of structural change in India since 1950 and discuss the path to structural transformation which has turned out to be different from that of most countries. The chapter is in four parts. In Section 20.2, it develops an analytical framework to examine the relationship between economic growth and structural change, in theory and history, to explore directions of causation. Section 20.3 outlines the broad contours of structural change in the composition of output and employment in

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

independent India, situated in historical perspective, in terms of changes in the relative importance of the primary, secondary, and tertiary sectors. In Section 20.4, the chapter considers the patterns of change through a disaggregated analysis to focus on agriculture, manufacturing, construction, and services which are divided further into two categories, in terms of their changing shares of output and employment. Section 20.5 discusses India’s path to structural transformation, over six decades during which two phases are discernible, to highlight the process of services-led growth starting around 1980, and reflects on the future to stress the importance of manufacturing and industrialization.

20.2 S F  A C

.................................................................................................................................. In the literature on structural transformation, analytical constructs are derived from stylized facts rather than economic theorizing or analytical abstractions. Thus, patterns of structural change observed in the process of development provide the foundations. In the earlier stages, when income levels are low, the share of the agricultural sector in both output and employment is overwhelmingly large. At the next stage, as industrialization proceeds, the share of the manufacturing sector in output and employment rises, while that of the agricultural sector falls. At an advanced state of development, after industrialization, the share of the manufacturing sector in both output and employment diminishes, while that of the services sector rises. This is the classical pattern, observed by Fisher (1935), Clark (1940), and Kuznets (1966), from the experience of countries that industrialized during the second half of the nineteenth century and the first half of the twentieth century. It is, then, possible to make an analytical distinction between three stages in the structural transformation of economies that succeeded at industrialization and development in the past (Nayyar 1994). In the first stage, there is absorption of surplus labour from the agricultural sector into manufacturing at existing levels of real wages and productivity (Lewis 1954). This is associated with a decline in the share of agriculture and a rise in the share of manufacturing in both employment and output, somewhat more in the former. The process can be described as labour absorption at the extensive margin. In the second stage, there is a transfer of labour from low productivity to higher productivity occupations in manufacturing, while, at the same time, there is an increase in the average productivity of labour in both sectors, so that real wages rise in both. There is a further increase in the share of manufacturing and a further decrease in the share of agriculture, more pronounced in output than in employment. The process can be described as labour use at the intensive margin. In the third stage, the share of the agricultural sector continues to decline, even as the share of the manufacturing sector is maintained while that of the services sector rises but, after a stage, when labour is no longer available from agriculture or from domestic

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



personal services, the share of the services sector increases at the expense of the manufacturing sector, more in employment than in output. This outcome in industrialized countries has also been described as de-industrialization (Rowthorn and Wells 1987). The focus of such conventional thinking is on economic growth in which structural change is an associated outcome. It is postulated that the income elasticity of demand for industrial goods is higher than that for agricultural goods, while the income elasticity of demand for services is even higher than that for industrial goods (Fisher 1935). The expansion of markets creates new demands so that new production activities follow. And growing economies almost always follow the sequence of moving, in terms of relative importance, from the primary sector to the secondary sector and then to the tertiary sector. In this characterization, productivity growth in the services sector is slower than in the manufacturing sector, because the scope for attaining it through capital accumulation, scale economies, or technical progress is much less (Baumol 1967). This provides an explanation for why the share of the services sector in total employment increases further (Rowthorn and Wells 1987). At the same time, the increase in its share of total output is attributable mostly to an increase in the relative price of services (Baumol 1967). It is important to recognize that patterns of structural change are not simply an associated outcome of economic growth. Indeed, the causation might also run in the opposite direction. In fact, the unconventional, heterodox, perspective stresses that structural change drives economic growth (Schumpeter 1942; Hirschman 1958; Chenery 1960; Ocampo et al. 2009; Nayyar, 2013). This proposition was also implicit in the Lewis (1954) model, where the transfer of surplus labour from the agricultural sector to the manufacturing sector, at a subsistence plus wage, increases the profits of capitalists, reinvestment of which is a source of capital accumulation and economic growth. The Kaldor (1966) model went much further in developing this causation to suggest that the manufacturing sector is the engine of growth in economies. This was set out in terms of three laws. First, there is a positive relationship between the growth of manufacturing output and the growth of GDP, which is explained partly by the absorption of surplus labour from the agricultural sector into the manufacturing sector. Second, the growth of manufacturing output leads to the growth of productivity in manufacturing, which is attributable to static and dynamic scale economies, where the former depend on plant size or output levels at any point in time, while the latter derive from learning-by-doing that is a function of cumulative past output (Arrow 1962) or cumulative production experience (Kaldor 1962) over time. Third, the growth of manufacturing output is associated with an overall increase in productivity in the economy driven by spillover effects elsewhere. In the relationship between economic growth and structural change, it is clear that the causation could run in both directions. But that is not all. The experience of late-comers to industrialization since 1950 suggests that the pattern and sequence of structural change might not be uniform across economies and could follow different paths. It would seem that during the second half of the twentieth century and the first decade of the twenty-first century, most developing countries have moved from the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

first stage, in which agriculture dominates, to the third stage, in which services dominate, without necessarily going through the second stage in which manufacturing dominates. There are some exceptions to this pattern but not many. To begin with, of course, the share of the industrial sector in both output and employment does increase almost everywhere, perhaps less than in the classical Clark–Kuznets worldview, but then stabilizes or even declines. In contrast, the share of the services sector begins to rise along with the fall in the share of the agricultural sector but, after a point, its share in both output and employment rises at the expense of the industrial sector. The former is not surprising, as the services sector is a source of labour absorption and employment creation in developing countries, but the latter is unexpected because it is widely believed that manufacturing is the engine of productivity growth. In this context, it is important to note that the world has changed, particularly in the services sector, from the time that stylized facts about structural change were formulated by the pioneers (Nayyar 2012). Given the massive increase in the size of firms, it is more profitable to procure services, such as those in the spheres of law, accounts, transport, or finance, from specialist providers rather than produce them within the firm (Coase 1937). Indeed, telecommunications, financial, software, or business services are now organized in a manner that strongly resembles the manufacturing sector, for scale economies or technical progress are easily incorporated to increase efficiency in providing these services. The revolution in transport, communication, and information technologies has meant that hitherto non-traded services now enter into crossborder transactions in international trade (Nayyar 1988; Nayyar 2012). In this changed world, the services sector could also drive economic growth in terms of Kaldor’s first two laws, by raising growth in GDP and in manufacturing productivity, with some possibility of spillover effects in the economy as a whole implicit in the third law (Nayyar 2012). In examining structural transformation, it is worth considering available evidence on the nature of causation. In a study based on data for a cross-section of fifty-seven developing countries and transition economies, grouped into twelve regions, during the period 1970–2006, Ocampo et al. (2009), analyse the relationship between structural change and economic growth. The annual growth rate of GDP per capita is juxtaposed with changes in the shares of agriculture and industry in total GDP. The scatter plots for the period show a negatively sloped regression line for decreases in the agricultural output share and a positively sloped regression line for increases in the industrial output share for the entire sample of twelve country-groups. However, the relationship between falling agricultural shares or rising industrial shares and economic growth is clear only for four country-groups in Asia (East Asia, Southeast Asia, China, and South Asia) that registered sustained growth. In contrast, the other eight country-groups that experienced slow growth (semi-industrialized countries mostly from Latin America but including South Africa and Turkey, Central America and the Caribbean, Middle East and North Africa, and Eastern Europe) or stagnant growth (smaller Andean countries, sub-Saharan Africa, other Africa, and former USSR), reveal a random scatter. Similarly, the fast growth regions also had

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



rapidly rising service sector shares, but there was no apparent relationship for the lagging growth regions. It would seem that structural change and economic growth are necessary but not sufficient to drive each other. The direction of causation does run in both directions but is strong and positive only in countries where there are virtuous circles of cumulative causation that reflect success in development, while it is weak or absent elsewhere. Quite apart from the direction of causation, the history of development experience does establish the relationship between economic growth and structural change. In the colonial era, during the first half of the twentieth century, for Asia, Africa, and Latin America, taken together, there was almost no change in the composition of output and employment. The near absence of structural change was associated with slow rates of growth and negligible industrialization in these continents over five decades (Nayyar 2013). Development experience in the post-colonial era, 1950–2010, provides a sharp contrast. But both economic growth and structural change in Asia was much faster in Asia than in Africa or Latin America (Nayyar 2013). Structural change in Asia was a driver of economic growth, because it moved labour from low-productivity to highproductivity sectors, but structural change in Latin America and Africa was not conducive to economic growth, because it did not (McMillan and Rodrik 2011). Generalizations from cross-country studies serve an important purpose but cannot suffice, because unfolding reality at the country-level is neither as uniform nor as simple as the stylized facts. There are some late-comers to development in Asia such as Korea and Taiwan, followed by China, that have followed the more classical pattern of structural change where the decline of agriculture is juxtaposed with the rise of manufacturing followed by an increase in the relative importance of services. However, other late-comers to development in Asia, such as India, have followed the nontraditional pattern of structural change, where the decline of agriculture has been juxtaposed with some increase in manufacturing but a much greater increase in services. Most countries in Asia conform to this pattern. So do most countries in Africa or Latin America and the Caribbean. It is, therefore, worth analysing the Indian experience to consider whether such a pattern represents a different path to structural transformation.

20.3 C  S C  I

.................................................................................................................................. It is worthwhile to begin with a historical perspective, so that it is possible to draw a comparison between the colonial era and independent India. In 1900–01, the share of the primary sector (agriculture, livestock, forestry, and fishing) in GDP at current prices was 63.3 per cent, while that of the secondary sector (mining, manufacturing, mining, small-scale and cottage industries) was 11.2 per cent, and that of the tertiary sector (railways and communications, government services,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

commerce and transport, professions, house property, and domestic services) was 25.5 per cent; and, in 1946–47, these shares were 53.3 per cent, 15.2 per cent, and 31.5 per cent respectively (Sivasubramonian 2000). Agriculture was dominant in the primary sector (52.2 per cent and 42.1 per cent, respectively), while small-scale and cottage industries were the largest constituent of the secondary sector (9 per cent and 7.4 per cent respectively); manufacturing was smaller to begin with, but its share did increase from 1.8 per cent to 7.3 per cent. It was the two World Wars that provided an impetus to industrialization, as the relative importance of manufacturing rose partly during 1914–18 but mostly during 1939–45. It would seem that there was some structural change in the composition of output. In sharp contrast, there was almost no change in the composition of employment. In 1901, the share of the primary sector in the total workforce was 75 per cent, while that of the secondary sector was 10 per cent and that of the tertiary sector was 15 per cent; and, in 1951, these shares were 76 per cent, 10 per cent, and 14 per cent respectively. In fact, the corresponding shares were almost the same in 1911, 1921, 1931, and 1941 Sivasubramonian, 2000). It would seem that, during the last half century of colonial rule in India, structural change in the economy was confined to some change in the composition of output with no change in the composition of employment. And the pace of economic growth in that era was indeed slow. There are alternative estimates of economic growth during the period 1900/01–46/47. The Sivasubramanian (2000) estimates suggest that GDP growth was 1 per cent per annum while growth in GDP per capita growth was 0.2 per cent per annum, while the Maddison (1985) estimates suggest that these growth rates were 0.8 per cent per annum and 0.04 per cent per annum respectively. National income would have doubled in 70 years with growth at 1 per cent per annum and in 87.5 years at 0.8 per cent per annum (Nayyar 2006). Clearly, modest structural change, such as it was, did nothing to drive economic growth. The latter half of the twentieth century provides a sharp contrast in both. The changes in the composition of output and employment in independent India, over the six decades since 1950, are outlined in Figure 20.1, based on time series statistics for GDP and the available data for selected years on employment. It shows that the share of agriculture in total output witnessed a substantial decline (from more than one-half to less than one-fifth), while the share of industry increased (from oneseventh to one-fourth) and the share of services rose more (from about one-third to almost three-fifths). The share of agriculture in total employment diminished (from three-fourths to one-half ), while that of industry increased (from one-tenth to one-fourth) and that of services rose less (from about one-seventh to more than onefourth). It would seem that the share of agriculture fell far more in output than in employment whereas, in proportionate terms, the share of industry increased more in employment than in output and the share of services increased more in output than in employment. The underlying factors are analysed in Section 20.4. Figure 20.1 also reveals that these trends were uneven over time. It is possible to discern three phases. Between 1950 and 1970, the shares of the three sectors

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



90.0 80.0 70.0

Per cent

60.0 50.0 40.0 30.0 20.0 10.0 19 5 19 0–5 52 1 19 –5 5 19 4– 3 5 55 19 6–5 56 7 19 –5 6 9 19 0–6 62 1 19 –6 6 3 19 4–6 6 5 19 6–6 68 7 19 –6 7 9 19 0–7 72 1 19 –7 7 3 19 4–7 76 5 19 –7 7 7 19 8–7 8 9 19 0–8 8 1 19 2–8 84 3 19 –8 8 5 19 6–8 8 7 19 8–8 9 9 19 0–9 9 1 19 2–9 9 3 19 4–9 96 5 19 –9 9 7 20 8–9 0 9 20 0–0 02 1 20 –0 0 3 20 4–0 0 5 20 6–0 0 7 20 8–0 10 9 20 –1 12 1 –1 3

0.0

Agriculture, forestry, and fishing (GDP) Industry (GDP) Services (GDP)

Agriculture, forestry, and fishing (employment) Industry (employment) Services (employment)

 . Changes in the composition of output and employment in India, 1950–51 to 2013–14 (per cent) Source: CSO, National Accounts Statistics and NSSO, Surveys on Employment.

changed slowly. Between 1970 and 1990, the pace of change was significantly faster. Between 1990 and 2010, the changes in shares of the three sectors gathered further momentum. It is also clear that the changes in output shares were, in general, more pronounced than the changes in employment shares throughout the six decades. There were, however, differences between sectors. The falling share of the agricultural sector, as well as the rising share of the services sector, in output and employment both gathered momentum from the early 1990s. However, the share of the industrial sector in output peaked in the early 1990s and fluctuated thereafter around the same level until 2010 and declined thereafter. In sharp contrast, the share of the industrial sector in employment continued to increase steadily from the early 1990s until the early 2010s, if anything at a faster pace. This might seem paradoxical. It deserves an explanation, which is provided in Section 20.4. It is important to recognize that there were asymmetries between changes in output shares and employment shares for each of the three sectors. There was a sharp drop in the share of the agricultural sector, but this decline was far greater in its share of output when compared to its share of employment. Of course, its employment share was always much higher than its output share. There was a significant increase in the share of the industrial sector, but this increase was more pronounced in its share of employment as compared with its share in output. There was a substantial rise in the share of the services sector, but this rise was far greater in its share of output as

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

compared with its share of employment. Yet, in both the industrial sector and the services sector, the output share was consistently higher than the employment share. However, the gap between the two shares, which was always small for the industrial sector and large for the services sector, narrowed over time in the industrial sector and widened over time in the services sector. The outcomes of these patterns of structural change in India over the six-plus decades are worth highlighting. Between 1950–51 and 2013–14, the share of agriculture and allied activities in GDP dropped by 34 percentage points; the share of industry increased by 11 percentage points; and the share of services rose by 23 percentage points (Table 20.1). Between 1951 and 2011–12, the share of agriculture and allied activities in total employment dropped by 26 percentage points, the share of industry rose by 14 percentage points, and the share of services increased by 12 percentage points (Table 20.2). This overall picture suggests that structural change in the economy of independent India has been significant, even if it does not conform to the classical pattern in terms of sequence or the relative importance of industry and services. But this is somewhat deceptive, since these aggregates conceal as much as they reveal. It is, therefore, essential to disaggregate the primary, secondary, and tertiary sectors, into their constituent sub-sectors, for a more meaningful analysis of changes in the composition of output and employment over time. The primary sector is made up of agriculture, forestry, and fishing. The secondary sector, also termed industry, is made up of manufacturing, construction, mining, and utilities. The tertiary sector, also termed services, is made up of a wide range of diverse services that are grouped into a dozen categories at two-digit level in statistical classifications. In the primary sector, agriculture is the dominant sub-sector, so much so that its changing shares in output and employment over time are mirrored in the shares for the sector as a whole. In the secondary sector, industry, manufacturing, and construction are the dominant sub-sectors, which account for its changing shares in output and employment over time, while the shares of the other two sub-sectors, mining and utilities, are small proportions. Thus, the analysis that follows seeks to focus on agriculture in the former, and manufacturing and construction in the latter. The constituents of the services sector, however, are many more and wide ranging. It is possible to explore the heterogeneity of the services sector in terms of analytical distinctions, production characteristics, or employment attributes (Nayyar 2012). There are analytical distinctions that can be made between organized and unorganized economic activity, public and private provision, or intermediate and final demand. There are production characteristics such as capital-intensity, skill-intensity and labour-intensity, or technological-levels and scale-economies. There are employment attributes such as barriers-to-entry for job-seekers implicit in educational-status or skill-levels. Each of these matter, but their incorporation would make the construction of any satisfactory taxonomy much too complex for the present exercise. Obviously, the tertiary sector is much too diverse for such a simple disaggregation. It is made up of eleven sub-sectors each of which also represent an aggregation of

Table 20.1 Structural changes in the composition of output in India: 1950–51 to 2013–14 (in percentages) Sectors

1960–61

1970–71

1977–78

1980–81

1990–91

2000–01

2010–11

2011–12

2012–13

2013–14

1. Primary sector of which agriculture

51.7

42.5

42.1

37.1

35.5

29.1

23.1

18.2

17.9

17.5

18.2

45.4

37.4

37.0

32.1

30.3

24.7

19.5

15.8

15.5

15.1

2. Secondary sector of which manufacturing construction

14.1

19.3

20.5

23.2

24.4

26.6

26.1

27.2

27.2

26.2

24.8

10.5 2.7

13.7 4.0

13.8 4.7

15.4 4.9

16.2 4.8

16.2 5.6

15.4 6.0

14.8 7.9

14.7 8.2

14.1 8.1

12.9 7.8

3. Tertiary sector of which Group A sub-sectors Group B sub-sectors

34.2

38.2

37.4

39.5

40.1

44.3

50.8

54.6

54.9

56.3

57.0

14.8 19.4

16.6 22.6

16.6 21.3

18.7 21.0

19.6 20.5

20.8 23.6

23.1 28.0

25.9 28.7

26.1 28.8

26.4 29.9

TOTAL (1+2+3)

100

100

100

100.0

100

100

100

100

100

100

100

Notes: The percentages have been calculated from data on GDP at factor cost in current prices The tertiary sector is divided into two groups of services. Group A is made up of the following services sub-sectors: 1. Wholesale and Retail Trade; 2. Hotels and Restaurants; 3. Transport and Storage; 4. Other Social, Community and Personal Services. Group B is made up of the following services subsectors: 5. Communication Services; 6. Financial Services; 7. Real Estate and Renting Services; 8. Business services; 9. Public Administration and Defence; 10. Health Services; and 11. Education Services Source: CSO National Accounts Statistics

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

1950–51

Sectors

1951

1961

1972–73

1977–78

1983

1987–88

1993–94

1999–2000

2004–05

2011–12

1. Primary sector of which agriculture

74.7

76.2

73.9

70.8

68.7

64.0

64.6

61.7

58.5

48.9

64.3

60.6

57.7

54.5

56.0

52.3

45.7

2. Secondary sector of which manufacturing construction

10.1

12.5

13.8

16.9

14.6

15.8

18.1

24.3

9.9 1.8

10.6 2.3

12.0 3.8

10.4 3.1

10.7 4.3

11.7 5.6

12.6 10.6

3. Tertiary sector of which Group A sub-sectors Group B sub-sectors

15.2

16.3

17.6

19.1

20.7

22.5

23.4

26.8

11.9 4.9

12.1 5.5

12.7 6.4

14.6 6.1

15.9 6.6

16.5 6.9

18.1 8.7

TOTAL (1+2+3)

100

100

100

100

100

100

100

100

10.7

13.1

100

11.3

14.8

100

Notes: The figures for 1951 and 1961 are from the Census of India and not strictly comparable with figures for later years. The National Sample Survey in 1983 was for the calendar year, while in other years it was for the financial year. The disaggregated data for sub-sectors are available starting 1977–78. The tertiary sector is divided into two groups of services. Group A is made up of the following services sub-sectors: 1. Wholesale and Retail Trade; 2. Hotels and Restaurants; 3. Transport and Storage; 4. Other Social, Community and Personal Services. Group B is made up of the following services sub-sectors: 5. Communication Services; 6. Financial Services; 7. Real Estate and Renting Services; 8. Business services; 9. Public Administration and Defence; 10. Health Services; 11. Education Services. Source: National Sample Survey Organisation, Surveys on Employment.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

Table 20.2 Structural changes in the composition of employment in India: 1950–51 to 2011–12 (in percentages)

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



what could be divided further. Given the heterogeneity, for the purpose of analysis, this chapter classifies the entire gamut of services into two categories. The first—Group A—is made up of: (i) wholesale and retail trade; (ii) hotels and restaurants; (iii) transport and storage; and (iv) other social, community, and personal services, The second—Group B—is made up of: (v) communication services; (vi) financial services; (vii) real estate and renting services; (viii) business services; (ix) public administration and defence; (x) health services; and (xi) education services. The logic of the distinction between these two groups needs to be made explicit. The constituents of Group A are essentially labour-intensive services. These sub-sectors are generally characterized by low barriers-to-entry for job-seekers and low technologicallevels for provision. The constituents of Group B are essentially capital-intensive, human-capital intensive, or skill-intensive services. These sub-sectors are generally characterized by high barriers-to-entry for job-seekers and high technological levels for provision. Of course, such a generalization of characteristics is not without its limitations, because there is a dualism within each of these sub-sectors, particularly those in Group A. For example, wholesale trade ranges from large firms to small entrepreneurs, while retail trade is even more dichotomized with brand chain-stores or supermarkets at one end and corner or roadside shops at the other. Hotels and restaurants range from the five-star variety at one end to cheap or roadside places for beds or food. Similarly, transport services range from civil aviation with jet aircraft, through the railways, to trucks for goods or a range of more primitive vehicles for passengers or goods that ply roads. Storage services range from sophisticated warehousing facilities to stand-alone spaces run by individuals. The dualism might not be so sharp or widespread in Group B sub-sectors, but it does exist in some. Real estate services are provided both by large firms and single individuals. Education services are mostly in universities, colleges, or schools but there are also teaching shops and private tutors. Health services are provided not only by large hospitals, established clinics, primary health centres, or private practices, but also by doctors or individuals from different systems of medicine. Notwithstanding such dualism, the distinction between Group A and Group B services does serve an analytical purpose.

20.4 P  C: T U S

.................................................................................................................................. It is possible to understand the story underlying the observed patterns of structural change in India on the basis of a disaggregated analysis which recognizes sub-sectors within sectors: agriculture in agriculture, forestry and fishing; manufacturing and construction in industry; and Group A and Group B sub-sectors in services. Table 20.1 presents evidence on structural changes in the composition of output for

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

selected benchmark years during the period from 1950–51 to 2013–14, while Figure 20.2 outlines these trends based on time-series data. Table 20.2 presents evidence on structural changes in the composition of employment during the period from 1951 to 2011–12 for benchmark years based on available employment statistics, although disaggregated data for the specified sub-sectors are available only from 1977 to 1978, so that Figure 20.3 outlines these trends during the period from 1977–78 to 2011–12. The discussion that follows draws upon these tables and figures to examine, in turn, the changing shares of the primary, secondary, and tertiary sectors, disaggregated by sub-sectors, in output and employment. The primary sector is constituted by agriculture, forestry, and fishing, among which agriculture is overwhelmingly important. It accounted for more than 85 per cent of output and more than 90 per cent of employment in the primary sector through the six-plus decades. However, between 1950–51 and 2012–13, its share of India’s GDP dropped from 45 per cent to 15 per cent and the decline was continuous (Table 20.1). But its share of total employment in the economy was much higher and dropped far less, from 64 per cent in 1977–78 to 46 per cent in 2011–12 (Table 20.2); this share would have been in the range of about 68 per cent in 1951. Such rising disproportionalities between output and employment shares are striking. It would seem that this pattern of structural change was rather asymmetrical, because labour absorption outside the agricultural sector was far slower than the contraction in its share of national income. The secondary sector—industry—comprises manufacturing, construction, mining, and utilities. Manufacturing and construction, taken together, accounted for around 85 per cent of output and more than 90 per cent of employment in industry through the period under review (Table 20.1). There were, however, significant differences in the relative importance of manufacturing and construction in the economy and this changed over time. The share of manufacturing in GDP rose from 10.5 per cent in 1950–51 to its peak level of 17.3 per cent in 1979–80, and fluctuated in the range of 16 per cent until 1995–96 when it was 17.3 per cent once again, to hover in the range 15–16 per cent until 2007–08, but declined slowly thereafter to 12.9 per cent in 2013–14 (Table 20.1 and Figure 20.2). On the other hand, its share of total employment in the economy, which would have been around 9 per cent in 1951, increased from 10 per cent in 1977–78 to 12 per cent in 1987–88, dipped to 10–11 per cent until the end of the 1990s, to recover thereafter so that it was 12.6 per cent in 2011–12 (Table 20.2 and Figure 20.3). It seems that the employment share of manufacturing contracted less than its output share. The two shares were about the same circa 1950. Subsequently, from the mid-1970s until the mid-1990s, the share of manufacturing in output progressively exceeded its share in employment but this asymmetry diminished significantly during the 2000s. The share of construction in GDP increased steadily, without interruption, from less than 3 per cent in 1950–51 to about 8 per cent in 2010–11 and remained around that level until 2013–14 (Table 20.1). In sharp contrast, its share of total employment in the

0.0

0.0 1950–51 1952–53 1954–55 1956–57 1958–59 1960–61 1962–63 1964–65 1966–67 1968–69 1970–71 1972–73 1974–75 1976–77 1978–79 1980–81 1982–83 1984–85 1986–87 1988–89 1990–91 1992–93 1994–95 1996–97 1998–99 2000–01 2002–03 2004–05 2006–07 2008–09 2010–11 2012–13

Per cent 1950–51 1952–53 1954–55 1956–57 1958–59 1960–61 1962–63 1964–65 1966–67 1968–69 1970–71 1972–73 1974–75 1976–77 1978–79 1980–81 1982–83 1984–85 1986–87 1988–89 1990–91 1992–93 1994–95 1996–97 1998–99 2000–01 2002–03 2004–05 2006–07 2008–09 2010–11 2012–13

0.0

1950–51 1952–53 1954–55 1956–57 1958–59 1960–61 1962–63 1964–65 1966–67 1968–69 1970–71 1972–73 1974–75 1976–77 1978–79 1980–81 1982–83 1984–85 1986–87 1988–89 1990–91 1992–93 1994–95 1996–97 1998–99 2000–01 2002–03 2004–05 2006–07 2008–09 2010–11 2012–13

Per cent Per cent

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     

Agriculture, forestry, & fishing

60.0

Industry

60.0

Services Manufacturing

Group A sub-sectors



60.0 (a) Primary sector

50.0

40.0

30.0

20.0

10.0

Agriculture

(b) Secondary sector

50.0

40.0

30.0

20.0

10.0

Construction

(c) Tertiary sector

50.0

40.0

30.0

20.0

10.0

Group B sub-sectors

 . Share of the primary, secondary, and tertiary sectors in GDP, dis-aggregated by sub-sectors in India, 1950–51 to 2013–14 (per cent)

Source: CSO National Accounts Statistics.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  (a) Primary sector 70 60

Per cent

50 40 30 20 10 0

1977–78

1983

1987–88

1993–94

1999–00

Agriculture, forestry, & fishing

2004–05

2011–12

Agriculture

(b) Secondary sector 70 60 Per cent

50 40 30 20 10 0

1977–78

1983

1987–88 Industry

1993–94

1999–00

Manufacturing

2004–05

2011–12

Construction

(c) Tertiary sector 70 60 Per cent

50 40 30 20 10 0

1977–78

1983 Services

1987–88

1993–94

Group A sub-sectors

1999–00

2004–05

2011–12

Group B sub-sectors

 . Share of the primary, secondary, and tertiary sectors in employment, dis-aggregated by sub-sectors in India, 1950–51 to 2011–12 (per cent) Source: Table 20.2.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



economy rose rapidly from 1.8 per cent in 1977–78 to 10.6 per cent in 2011–12 (Table 20.2); this share would have been less than 1 per cent in 1951. It seems that the employment share of construction expanded far more than its output share between 2004–05 and 2011–12. In fact, in 1977–78, the share of construction in GDP was 3 percentage points higher than in employment and this difference remained positive in the range of 2 percentage points until 2004–05 but, in 2011–12, its employment share was more than 2 percentage points higher than it output share. The rise in the share of construction in employment during the 2000s, which coincided with a stagnation followed by a decline in the share of manufacturing in output, explain why the employment share of industry (secondary sector) continued to increase steadily from the early 1990s to the early 2010s, even though its output share stagnated from the early 1990s until 2010 and diminished thereafter. The share of the tertiary sector—services—in GDP increased from 34 per cent in 1951 to 56 per cent in 2012–13. Over this period, the share of Group A sub-sectors rose from 15 per cent to 26 per cent, while the share of Group B sub-sectors rose from 19 per cent to 30 per cent (Table 20.1). Thus, their contribution to the increased significance of the services sector was roughly equal. It is, however, worth noting that there was little change in the first two decades, so that the increase in shares began after early 1970s. Although the GDP share of Group B sub-sectors was somewhat higher than that of Group A sub-sectors throughout, the upward trajectories in both moved closely together from the mid-1970s (Figure 20.2). The share of services in total employment increased from 15 per cent in 1951 to 27 per cent in 2011–12. But there was little change in this share until the early 1970s. Between 1977–78 and 2011–12, the share of Group A sub-sectors in employment rose from 12 per cent to 18 per cent, while that of Group B sub-sectors rose from 5 per cent to 9 per cent. Obviously, both sets of sub-sectors contributed to the increasing employment share of services to the economy, but the contribution of Group A was far greater than that of Group B. Indeed, the employment share of Group A services was roughly double that of Group B services throughout (Table 20.2 and Figure 20.3). This is hardly surprising. First, Group A services are much more labour-intensive in provision. Second, Group A services have far fewer barriers-to-entry for job-seekers, particularly in the informal sector segments of these sub-sectors. The shares of both Group A and Group B service sub-sectors in GDP were higher than their respective shares is total employment. However, the asymmetry was much less in Group A than in Group B. Between 1977–78 and 2011–12, on average, the output share exceeded the employment share by 7.5 per cent for Group A sub-sectors and by 18.5 per cent for Group B sub-sectors. However, over the same period, this gap was in the range 6–8 per cent for Group A but rose from 16 per cent to 21 per cent for Group B (Figure 20.2 and Table 20.2). This disproportionality, particularly the latter, is indeed striking. The asymmetries in shares of output and employment are the highest for agriculture in the primary sector, followed by Group B services, while Group A services in the tertiary sector come next. Initially, such asymmetry between output and employment

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

shares in manufacturing was close to what it is in Group A services sub-sectors but is now much less. This asymmetry is lowest in construction, where the employment share now exceeds the output share. These asymmetries are attributable to differences in levels of, and growth in, output per worker between sectors. Table 20.3 presents evidence on trends in GDP per worker, disaggregated by sub-sectors, at constant prices from 1977–78 to 2011–12. It also sets out average annual growth rates in GDP per worker and GDP at constant prices, disaggregated by sub-sectors, for the periods 1977–78 to 1993–94 and 1993–94 to

Table 20.3 Levels of, and growth in, GDP per worker in India, disaggregated by sub-sectors: 1977–78 to 2011–12 A. GDP per worker at constant prices in 2004–05 Rupees Sectors/subsectors

1977–78

1983

1987–88

1993–94

1999–2000

2004–05

2011–12

Primary sector Agriculture Secondary sector Manufacturing construction Tertiary sector Services A Services B

15794 16355 50632

17054 17841 51372

15682 14626 53384

20253 21521 66773

24329 23720 86130

24298 23618 89946

37378 34867 119228

38433 120760 58701 44276 94661

42732 89717 67196 49949 105632

42132 59940 75273 55276 115935

58300 88683 89770 58508 165312

80394 86616 126394 81851 236388

84715 89396 147176 98832 262681

143305 82872 237622 153326 412759

Total

27684

30931

33620

41537

57290

64922

110941

B. Growth in GDP per worker and GDP: per cent per annum

Sectors/sub-sectors

Growth in GDP per worker

Growth in GDP in constant 2004–05 prices

1977/ 78–1993/94

1950/ 51–1977/78

1977/ 78–1993/94

1993/ 94–2011/12

1993/ 94–2011/12

Primary sector Agriculture Secondary sector Manufacturing construction Tertiary sector Services A Services B

1.57 1.73 1.74 2.64 1.91 2.69 1.76 3.55

3.46 2.72 3.27 5.12 0.38 5.56 5.50 5.21

2.34 2.21 5.59 5.38 5.48 4.50 4.18 4.81

3.31 3.08 5.03 5.07 4.17 6.31 5.15 7.38

3.07 2.92 7.56 7.30 8.66 8.66 8.38 8.86

Total

2.57

5.61

3.60

4.84

7.04

Note: The growth rates in GDP per worker are calculated as point-to-point compound growth rates. The average annual rates of GDP growth have been calculated by fitting a semi-log regression equation LnY a + bt and estimating the values of b. Source: CSO National Accounts Statistics and NSSO Surveys on Employment.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



2011–12. In addition, GDP growth rates are provided for the period 1950–51 to 1977–78 to facilitate comparison. There was an increase in output per worker in every sector over time. If 1977–78 is treated as the base year when output per worker was 100, in 2011–12, its level was 212 in agriculture, 373 in manufacturing, 346 in services A, and 436 in services B. However, differences in levels of output per worker, which were significant to start with, widened over time. If output per worker was 100 in the economy as a whole in 1977–78, by 2011–12, its relative level decreased from 59 to 31 in agriculture, from 139 to 129 in manufacturing, and from 160 to 138 in services A, but increased from 342 to 372 in services B. This comparison is not made for construction because of data inconsistencies; output were worker was highest in 1977–78 and lowest in 1987–88 possibly because estimates of employment in construction, based on surveys, were too low in 1977–78 and too high in 1987–88 compared with other years. In fact, output per worker in construction witnessed little change in the other five years so that growth was mildly negative. It is no surprise that changes in the share of these sub-sectors in total output is attributable to differences in their respective GDP growth rates. The evidence in Table 20.3 provides confirmation. Differences in growth rates of GDP and GDP per worker in each of these sub-sectors explain the asymmetries between their shares in output and employment. From 1977–78 to 1993–94, growth in output per worker as a percentage of growth in GDP was 56 per cent in agriculture, 52 per cent in manufacturing, 34 per cent in services A, and 48 per cent in services B. From 1993–94 to 2011–12, this percentage was 94 per cent in agriculture, 70 per cent in manufacturing, 65 per cent in services A, and 59 per cent in services B. Obviously, increases in output per worker accounted for a much larger proportion of GDP growth during the later period when compared with the earlier period, so that employment creation—hence the absorption of surplus labour from agriculture—slowed down sharply. Ironically enough, in agriculture, growth in output per worker accounted for much of the growth in output in the later period. It would seem that the reported share of agriculture in total employment remained high simply as a residual employer of the last resort for many engaged in rural non-farm activities. The process of structural change in India was slow until around 1970 but gathered momentum thereafter. However, given the available evidence on employment, the two discernible periods are: 1950–51 to 1977–78 and 1977–78 to 2011–12. For the earlier period, employment statistics, disaggregated by sub-sectors, are not available. Thus, it is only possible to consider structural change in terms of the three major sectors. Between 1950–51 and 1977–78, the share of the primary sector in GDP fell by 13.2 percentage points, while the share of industry rose by 7.7 percentage points and the share of services rose by 5.5 percentage points. Over the same period, the share of the primary sector in employment decreased by just 3.9 percentage points, while the share of industry increased by 2.8 percentage points and the share of services increased by 1.1 percentage points. Structural change in the economy was much greater in the later period. Between 1977–78 and 2011–12, as a proportion of GDP, the share of the agricultural sector fell

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

by 22.1 percentage points and the share of manufacturing declined by 0.7 percentage points, while the share of construction increased by 3.3 percentage points, the share of Group A services rose by 7.4 percentage points and the share of Group B services rose by 8.8 percentage points. The decrease in the share of forestry & fishing and an increase in the share of mining & utilities accounted for the difference. Over the same period, as a proportion of total employment, the share of the agricultural sector fell by 18.6 percentage points, while the share of manufacturing increased by 2.7 percentage points, the share of construction jumped by 8.8 percentage points, the share of Group A services rose by 6.2 percentage points, and the share of Group B services increased by 3.6 percentage points. The residual difference was attributable to a drop in the share of forestry & fishing. The essentials contours of the underlying story are clear. During the first phase, until the late 1970s, structural change was slow and modest, which was more visible in output than in employment. This pattern of change was also conventional. The primary sector, which declined in relative importance, was replaced largely by the secondary sector, led by manufacturing, while the services sector accounted for the residual. The story was altogether different during the second phase from the late 1970s to the early 2010s. Structural change was much faster, more significant, and rather unconventional. The rapid decline in the share of agriculture, more pronounced in output than in employment, continued. The share of manufacturing in output, which peaked in 1979–80, fluctuated around that level until 2007–08, but declined slowly thereafter, while its share in employment continued to fluctuate around the level attained in the late 1980s. The reduced share of the agricultural sector in output was captured mostly by the services sector and in small part by construction. However, almost half of the labour displaced from the agricultural sector was absorbed in construction while one-third was absorbed by labour-intensive services with low barriers to entry for job seekers. Therefore, less than one-fifth of the labour that moved out of agriculture was absorbed in manufacturing and in human-capitalintensive or technology-intensive services.

20.5 D P  T

.................................................................................................................................. There are two distinct phases of economic growth in India since independence. In the first phase, from 1950–51 to 1979–80, growth in GDP and GDP per capita was 3.5 per cent and 1.4 per cent per annum respectively. This provided a sharp contrast with the near-stagnation in the colonial era. Indeed, if we consider the twentieth century in its entirety, the turning point in economic performance, or the structural break in economic growth, is 1951–52 (Nayyar 2006). This growth performance was about the same as in most countries. It was neither as good as East Asia nor as

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



bad as Africa. It was average (De Long 2003). During the second phase, 1980–81 to 2010–11, growth in GDP and GDP per capita was 6.6 per cent per annum and 4.9 per cent per annum respectively, which was much faster than elsewhere in the developing world, except for China. If we consider the second half of the twentieth century, the turning point in economic growth is 1980–81. It is not 1991, when economic liberalization began (Rodrik and Subramanian 2005; Nayyar 2006). However, economic growth slowed down starting in 2011–12, although it was still much higher than elsewhere except for China. This was attributable in part to the Great Recession in the world economy—in the aftermath of the global financial crisis in late 2008—which persists, and in part due to domestic factors and policies. The two phases of structural change in India broadly coincided with the two phases of economic growth. During 1950–80, structural change was significant in sharp contrast with the colonial era, even if the pace was modest. The pattern of change was conventional but the causation ran in both directions. Economic growth drove structural change, as the income elasticity of demand for industrial goods, and also services, was higher than that for agricultural goods. Structural change drove economic growth, as surplus labour from agriculture was absorbed into industry, particularly manufacturing, and also services, at higher levels of productivity. During 1980–2010, much like economic growth, the pace of structural change also gathered momentum. Once again the causation ran in both directions. Economic growth drove structural change, because the income elasticity of demand for services turned out to be even higher than that for industrial goods so that the contribution of manufacturing was far less. Structural change drove economic growth, as surplus labour from agriculture was absorbed into Group A services, and into construction, at much higher levels of productivity than in agriculture (Table 20.3). But that was not all. The rapid rise in the share of Group B services in GDP, even if the increase in their employment share was much less, was attributable to the striking increase in productivity as output per worker in this sub-sector more than quadrupled between 1977–78 and 2011–12 (Table 20.3). It is clear that, during 1980–2010, economic growth in India was led by the services sector (Banga 2005; Dasgupta and Singh 2005; Rakshit 2007; Eichengreen and Gupta 2011). This was so even during 2011–15, when economic growth slowed down. In this, India was very different from other countries in Asia that witnessed rapid economic growth led by the manufacturing sector. But India was also an outlier in two other respects. First, the share of industry, as well as manufacturing, in GDP peaked at much lower levels, at 27 per cent and 17 per cent respectively, compared with Korea, Taiwan, China, Indonesia, Malaysia, and Thailand, at 40+ per cent and 30+ per cent respectively (ADB 2013). Second, the share of services in employment was much lower than that in GDP, compared with Korea, Taiwan, China, Indonesia, Malaysia, and Thailand, where the corresponding shares were much closer to each other (Ghose 2016). What were the consequences of differences in the pace of economic growth and the process of structural change, particularly during the second phase, between sectors for the economy at a macro level?

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

The rising disproportionality between the share of agriculture in output and in employment, similar for forestry and fishing, means that GDP per capita in the primary sector—predominantly agriculture—has been less than one-tenth of GDP per capita in the non-agricultural sector since 1990–91. Employment creation possibilities in the agricultural sector have also declined sharply, as the employment elasticity of output growth dropped from 0.26 during 1993/94–2004/05 to 0.42 during 2004/05–2011/12 (IHD 2014). Thus, the disproportionately large share of employment in the primary sector, mostly agriculture, is a deceptive residual number. In this, a significant proportion might work in agriculture during peak seasons but derive much of their incomes from non-agricultural rural employment. Such underemployment does not provide adequate, let alone sustainable, livelihoods. Rural poverty in India persists at high levels. Indeed, in 2011–12, among people below the poverty line, a large number (around 250 million) and a large proportion (almost three-quarters) lived in rural India. The share of the secondary sector—industry—in GDP and in employment, during 1950–80, rose largely because of the manufacturing sector. The pace of industrialization was impressive in comparison with the past and with most developing countries. However, during 1980–2010, and even more so thereafter, the share of industry in GDP was maintained by utilities and construction rather than manufacturing, while its share in employment rose because of construction. In fact, the story of industrialization since 1991 is somewhat dismal, as compared with India’s own past performance and the more recent performance of other countries. The stagnation and decline in the share of manufacturing in GDP (Figure 20.2), juxtaposed with India’s declining share of manufacturing value-added in, and manufactured exports from, the developing world (Nayyar 2013), might reflect the beginnings of de-industrialization. In the first phase, 1950–80, the share of the services sector in GDP registered an increase that is largely attributable to Group A services, although its employment share changed little. But the transformation was striking during 1980–2010. This was driven by output growth rates that were significantly higher than GDP growth for the economy. The rapid growth in the services sector was not notional. It was real. The increase in the relative prices of services, vis-à-vis agriculture and industry, was negligible, while the outsourcing of services from industrial firms was not significant enough to support the idea that services growth was just disguised manufacturing activity (Nayyar 2012). What, then, were the underlying factors? On the demand side, growth in the services sector was driven by private final consumption expenditure, but government final consumption was not a significant factor (Rakshit 2007; Nayyar 2012). A growth accounting exercise suggests that, since 1991, private final demand accounted for about one half of the growth in service-sector output, while the other half was divided between outsourcing by the industrial sector and exports, with the latter taking an increasing share over time (Eichengreen and Gupta 2011). A systematic econometric analysis, based on National Sample Survey data for household consumer expenditure in 1993–94 and 2004–05, shows that aggregate consumption expenditure on education, health, entertainment, personal, communication, and transport services absorbed a larger share of household budgets as

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



expenditures (incomes) increased (Nayyar 2012). In other words, the income elasticity of demand for these consumer services was high. This might appear surprising in a country such as India where income levels are low. The explanation might lie in increasing income inequalities together with household preferences for this range of consumer services (Rakshit 2007). On the supply side, starting in the early 1990s, deregulation in sectors such as communications, finance, and business, along with privatization in sectors such as education and health, also provided a stimulus to growth in these producer services and consumer services (Nayyar 2012). The story of rapid economic growth in India during the period since 1980 conforms to the Lewis (1954) model with a slight twist. Rural–urban migration led to the absorption of unskilled labour drawn from low productivity occupations in agriculture to higher productivity occupations in the urban informal services sector. Output per worker in Group A services was higher than in agriculture to begin with in 1977–78, but this gap between the former and the latter widened rapidly thereafter. The much higher productivity growth even in unskilled labour-intensive services in the informal sector was attributable to the far better infrastructure in urban India, as compared with rural India, and this positive externality made an enormous difference (Nayyar 2012). Economic growth was driven in exactly the same manner by the absorption of surplus labour from agriculture into construction, a sub-sector of industry, where levels of output per worker were much higher than in agriculture even though, unlike services, there was little productivity growth in construction. Manufacturing was simply not part of this story. Is such services-led growth sustainable in India? There has been some slowdown in growth since 2011–12. Yet, the process remains service-led so far. In the longer term, however, there are limits that need to be recognized. In the services sector as a whole, employment growth has been much lower than output growth. What is more, such employment creation has been concentrated in Group A services, particularly the informal sector, using unskilled labour with low barriers-to-entry for job seekers. In fact, much of the employment expansion is in Group A services where the quality of employment is low, while there is little employment expansion in Group B services where the quality of employment is high (Nayyar 2012). Therefore, the possibilities of moving unskilled labour from low productivity occupations in the former to higher productivity occupations in the latter, or even from the informal sector to the formal sector within the former, are very limited. In other words, the services sector has led growth through labour absorption at the extensive margin, but might not be able to sustain this growth through labour absorption at the intensive margin. In this situation, the importance of manufacturing cannot be stressed enough. Economic development is not only about economic growth. It is also about the capabilities of an economy to organize and transform its productive activities. This is simply not possible without industrialization (Nayyar 2013). The economic history of the now developed high-income countries provides confirmation (Chang 2002). Indeed, the experience of success stories in development since 1950 also shows

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

that no country has achieved even middle-income status without industrialization (McMillan and Rodrik 2011; Nayyar 2013; ADB 2013). Manufacturing also shows faster growth in lower-income countries, so that it narrows the productivity gap between industrialized and developing countries more rapidly, suggesting an unconditional convergence to the frontier (Rodrik 2013). It must, of course, be recognized that the world economy has changed. The Great Recession, which followed in the aftermath of the global financial crisis, persists. Growth in international trade has slowed down more than in output. There could be a return to protectionism, particularly in industrialized countries. Technical progress— robotics or artificial intelligence—that replaces labour in production processes, is likely to constrain labour-intensive manufactured exports from developing countries. Even so, the rationale for manufacturing-led growth in India is strong. Indeed, industrialization is an imperative. First, it is the path to employment creation. Most new entrants to the labour force, an estimated 8 million every year, are low-skilled workers; the employment of these workers is the only means of mobilizing India’s most abundant resource—people—for development. And the employment elasticity of output growth in the organized manufacturing sector is higher than in the organized services sector. Between 1999–2000 and 2011–12, it was 0.65 in the former and 0.43 in the latter (Ghose 2016). Rapid manufacturing growth will also drive employment growth in other sectors. Second, it is a potential source of economic growth, not only for labour absorption at the extensive margin but also for labour use at the intensive margin, with a potential for moving workers from lower to higher productivity employment in manufacturing. Moreover, given the low shares of manufacturing in GDP and employment, juxtaposed with the enormous size of the domestic market, the potential for manufacturing growth is large. The renewed focus and the suggested emphasis on manufacturing, for sustaining the pace of economic growth in India, is not meant to be at the expense of the services sector which has led growth since 1980. Indeed, it is essential to build on this past success which has been almost unique to India. In this process, manufacturing and services are valuable complements rather than substitutes. There are potential synergies to be exploited. The much needed investment in physical infrastructure and the spread of education in society, would alleviate constraints on growth to help realize the untapped potential in both manufacturing and services. Rapid output growth in manufacturing and services would provide an impetus to employment creation in both sectors on the demand side, just as it would enhance productivity increases in both sectors through the exploitation of static and dynamic scale economies on the supply side. This strategy would be the equivalent of ‘walking on two legs’. In this quest, the agricultural sector, characterized by near stagnation, cannot be forgotten since rural India is home to half its population. Output growth and productivity increases in agriculture would create incomes that could enlarge the size of the domestic market for both manufacturing and services. The moral of the story is that, while economic growth might be led by a sector, the process of structural transformation cannot be completed if there is a sector, or sectors, that are laggards.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 ’     



R ADB, Asian Development Bank, 2013. ‘Asia’s Economic Transformation: Where to, How and How Far?’ in Key Indicators for Asia and the Pacific 2013, Manila: Asian Development Bank. Arrow, Kenneth J., 1962. ‘The Economic Implications of Learning by Doing’, Review of Economic Studies, 29, pp. 155–73. Banga, R., 2005. ‘Critical Issues in India’s Service-Led Growth’, ICRIER Working Paper Series No. 171, New Delhi: International Council for Research on International Economic Relations. Baumol, W. J., 1967. ‘Macroeconomics of Unbalanced Growth: The Anatomy of Urban Crisis’, American Economic Review, 57, pp. 415–26. Chang, Ha-Joon, 2002. Kicking Away the Ladder: Development Strategy in Historical Perspective, London: Anthem Press. Chenery, Hollis B., 1960. ‘Patterns of Industrial Growth’, American Economic Review, 50, pp. 624–54. Clark, Colin, 1940. The Conditions of Economic Progress, London: Macmillan. Coase, R. H., 1937. ‘The Nature of the Firm’, Economica, 4, pp. 386–405. Dasgupta, S. and Ajit Singh, 2005. ‘Will Services be the New Engine of Indian Economic Growth?’, Development and Change, 36, pp. 1035–57. De Long, J. B., 2003. ‘India since Independence: An Analytic Growth Narrative’, in Dani Rodrik, ed., In Search of Prosperity: Analytic Narratives on Economic Growth, Princeton, NJ: Princeton University Press. Eichengreen, B. and P. Gupta, 2011. ‘The Service Sector in India’s Road to Economic Growth’, NBER Working Paper No. 16757, Cambridge, MA: National Bureau of Economic Research. Fisher, A. G. B., 1935. The Clash of Progress and Security, London: Macmillan. Ghose, Ajit K., 2016. India Employment Report 2016: Challenges and Imperative of ManufacturingLed Growth, Institute for Human Development, New Delhi: Oxford University Press. Hirschman, Albert O., 1958. The Strategy of Economic Development, New Haven, CT: Yale University Press. IHD (Institute for Human Development), 2014. India Labour and Employment Report, New Delhi: Academic Press. Kaldor, Nicholas, 1962. ‘Comment on Economic Implications of Learning by Doing’, Review of Economic Studies, 29, pp. 246–50. Kaldor, Nicholas, 1966. Causes of Slow Rate of Growth in the United Kingdom, Cambridge: Cambridge University Press. Kuznets, Simon, 1966. Modern Economic Growth: Rate, Structure and Spread, New Haven, CT: Yale University Press. Lewis, W. Arthur, 1954. ‘Economic Development with Unlimited Supplies of Labour’, The Manchester School, 22, pp. 139–91. Maddison, Angus, 1985. ‘Alternative Estimates of the Real Product of India: 1900–46’, Indian Economic and Social History Review, 22 (2), pp. 201–10. McMillan, Margaret and Dani Rodrik, 2011. ‘Globalization, Structural Change and Productivity Growth’ in M. Bacchetta and M. Jansen, eds, Making Globalization Socially Sustainable, Geneva: ILO-WTO. Nayyar, Deepak, 1988. ‘Political Economy of International Trade in Services’, Cambridge Journal of Economics, 12, pp. 279–98. Nayyar, Deepak, 1994. ‘International Labour Movements, Trade Flows and Migration Transitions’, Asia and Pacific Migration Journal, 3, pp. 31–48.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Nayyar, Deepak, 2006. ‘India’s Unfinished Journey: Transforming Growth into Development, Modern Asian Studies, 40, pp. 797–832. Nayyar, Deepak, 2013. Catch Up: Developing Countries in the World Economy, Oxford: Oxford University Press. Nayyar, Gaurav, 2012. The Service Sector in India’s Economic Development, New York: Cambridge University Press. Ocampo, Jose Antonio, Codrina Rada, and Lance Taylor, 2009. Economic Structure, Policy and Growth in Developing Countries, New York: Columbia University Press. Rakshit, Mihir, 2007. ‘Services-Led Growth: The Indian Experience’, Money and Finance, ICRA Bulletin, February, pp. 91–126. Rodrik, D., 2013. ‘Unconditional Convergence in Manufacturing’, Quarterly Journal of Economics, 128, pp. 165–204. Rodrik, D. and A. Subramanian, 2005. ‘From Hindu Growth to Productivity Surge: The Mystery of the Indian Growth Transition’, IMF Staff Papers No. 52, pp. 193–228. Rowthorn, Robert E. and John R. Wells, 1987. De-Industrialization and Foreign Trade, Cambridge: Cambridge University Press. Schumpeter, Joseph A., 1942. ‘The Creative Response in Economic History’, Journal of Economic History, 7, pp. 149–59. Sivasubramanion, S., 2000. The National Income of India during the Twentieth Century, New Delhi: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

   , – ......................................................................................................................

 

S transformation refers to the reallocation of resources from sectors of lower to those of higher productivity. Egypt needs this reallocation in order to attain the GDP growth rates that would enable it to meet the economic and social challenges confronting the country. Prominent among these challenges are providing a level of income by which Egyptians can obtain the goods and services that enable them to lead lives that they value and, in view of Egypt’s history, safeguarding the country from external economic and political pressures. Since in a modern economy most incomes are earned through employment, this effectively means that the economy must grow at a rate that would create the required number of decent jobs while keeping the external deficit within manageable bounds. Many factors drive the structural transformation of an economy. These include the economic strategy adopted by the country—such as the relative roles assigned to the public and private sectors—the fiscal, monetary, trade, and other incentives that encourage or hold back the transforming impulses; the institutional structure, especially the supremacy of the rule of law and the working of the judicial system, the functioning of the bureaucratic apparatus, the effectiveness of the education and technical training system; the efficiency of financial markets; and others. Underpinning these are issues of political economy, that is, the interaction of politics and economics, which determine which of the factors driving structural change are supported by the regime, and with how much vigour. Egypt’s economic transformation since the Free Officers’ revolution of July 1952 has been relatively slow. Egypt requires a transformation not only in the structure of the GDP, but also in its labour market, the external sector, the public finances, and major institutions, if it is to deal successfully with the challenges it faces. The underlying reason for the slow transformation was a political-economy framework that did not prioritize economic and employment growth.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

21.1 E’ U E P

.................................................................................................................................. Egypt’s basic economic problem is defined by the country’s geographic and demographic structures. Although the country comprises nearly 386,000 square miles, only a narrow strip of about 15,000 square miles in the Nile Valley and the Delta is cultivable. Of the total population, estimated in mid-2015 at 85 million, 98 per cent is crammed into this slender band, giving it a density of more than 5,500 per square mile. The population has increased rapidly from 19 million in 1947 to about four and a half times that number in 2015. Even on the (optimistic) assumption of a significant slowdown in the growth rate, the medium variant of the United Nations population projections estimates Egypt’s population in 2050 at 122 million. However, the population contains a large ‘echo’ generation below the age of 10 that will enter the labour market in the near future. In 2012 there were about 8 million Egyptians aged 25–29, but also more than 11 million below the age of 5 years. This cohort will in due course enter the labour market, creating another very large youth ‘bulge’ and another substantial demand for jobs. Thus the economy will not only have to create jobs for large numbers, but it will also have to keep doing so for an extended period. On the other hand, between 1947 and 2015 the country’s cultivated area increased by barely 25 per cent. Intensive cultivation to some extent compensated for the deterioration in the man–land ratio, but even so the cropped area (the cultivated area times the cropping intensity) between 1947 and 2015 only increased from 9.1 million feddans to about 14 million feddans.¹ A feddan in 2015 was thus expected to support 6.3 persons, compared with 2.1 in 1947. The theme of a large and rapidly growing population confronted by a relative fixity of arable land resonates as a leitmotif in Egypt’s economic experience since the 1950s. The Egyptian economy requires structural changes because, as Little (1967: 258) notes, ‘The fact that Nile development can never again keep pace with population is at the root of Egypt’s economic problem.’ This conflict underlies the country’s major economic problems, and its implications permeate issues arising in virtually all the sectors. Thus, to consider only two examples, the fixity of the usable land means that agriculture cannot be a major provider of employment for the expanding labour force; therefore, Egypt’s development strategy must aim at transforming the economy in the direction of manufacturing and services. It also means that Egypt cannot feed itself from its own agriculture and must rely heavily on imports; indeed, nearly 60 per cent of Egypt’s wheat requirements are imported. But imports have to be paid for in foreign exchange; this means that the economy must sustain a high level of export competitiveness in order to obtain the increasing requirements of foreign exchange; as Richards (1993: 244) points out, ‘food security’ for Egypt essentially means ‘foreign exchange security’. ¹ A feddan = 1.038 acres = 0.42 hectares.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



This chapter is divided into two parts. Section 21.2 describes and analyses the transformations that have taken place in major areas of the economy; Section 21.3 discusses the critical political-economic policies that chiefly accounted for the slowness of the transformation.

21.2 S T, 1965–2015

.................................................................................................................................. Since a principal reason for wanting to transform the structure of its economy is to accelerate its growth, it would be useful to briefly review the performance of Egypt’s macroeconomic indicators.

21.2.1 GDP and its Components 21.2.1.1 Economic Growth As is the case with many developing countries, Egypt’s economic data are continually being revised and improved. The databases of the World Bank and the IMF enable one to make reasonably consistent judgements from the mid-1960s, but the story for earlier years is more reliable for general trends and rates of change than for absolute levels. It is also advisable to cross check key elements of the data with estimates from different sources. With this caveat in mind, it appears that between 1947 and 1952 the GDP in real terms grew at about 5 per cent a year as the economy recovered from the effects of World War II (Hansen and Nashashibi 1975: 11–15 and table 1-1). Growth slowed between 1952 and 1955 to about 2 per cent a year. After the Suez Canal war of 1956, the government began a push for economic development, supporting it by expansionary fiscal and monetary policies; this accelerated the growth rate to about 6 per cent. This story is also in line with Mead (1967: tables I-A-6 and I-A-8) who estimates the average growth of the GNP between 1945 in 1963 in constant 1954 prices at about 4.3 per cent a year.² Thus, broadly speaking, the Egyptian economy appears to have expanded at an average rate of between 4 and 4.5 per cent a year in real terms between 1945 and 1965. Hansen and Marzouk (1965: 3, table 1.1), on the basis of some generally plausible assumptions, attempt computations for earlier years; they estimate that real GNP per capita in 1914 was about the same as in 1952 roughly LE 45 in 1954 prices. Thus it appears that the average Egyptian was no better off in 1952 than in 1914. Using the IMF/World Bank database, GDP in constant 2005 prices increased from $13 billion in 1965 to more than $136 billion in 2015. The average growth rate

² A detailed account of the 1939–62 period appears in Hansen and Mead (1965).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

of 4.7 per cent a year concealed substantial year-to-year fluctuations—the coefficient of variation for the five decades as a whole was 57.9 per cent.

21.2.1.2 Population and Per Capita Income Estimates of Egypt’s population go back to pharaonic times—the Roman historian Titus Flavius Josephus provided a figure of 7.5 million (excluding Alexandria) for the first century .³ The first modern census took place in 1897, and censuses were conducted every 10 years until 1947. Hostilities with the UK, France, and Israel postponed the 1957 census until 1960, and a sample enumeration was performed in 1966 in place of the scheduled 1967 census. From 1976 onwards, censuses have been carried out as planned. From 1960 to 2015, Egypt’s population is estimated to have increased from 26 million to 85 million, a growth rate of about 2.3 per cent a year. It is estimated that between 1947 and 1952 GDP per capita increased by about 3 per cent a year. From 1950 to 1956, per capita income probably fell slightly, while from 1957 to 1964 its average growth was between 3.0 and 3.5 per cent a year.⁴ The IMF/World Bank database shows GDP per head in constant 2005 prices increasing from $406 in 1965 to about $1,600 in 2015, an average growth rate of 2.6 per cent a year. This series shows greater fluctuation than that of the GDP. Indeed, in some years—such as 1966, 1967, 1973, and 1991—real per capita income fell. For the period as a whole, the coefficient of variation was 97.6 per cent, much higher than that for the GDP growth series.

21.2.1.3 Structural Changes The differential growth of sectors over the 50-year period changed the structure of the GDP. The share of agriculture dropped from 29 per cent of the total to 15 per cent, while that of industry increased from 27 per cent to 39 per cent. A large part of this increase represented the emergence and rapid expansion of the petroleum sector. The services sector showed fewer fluctuations, veering between 45 per cent in 1965, increasing up to 52 per cent in some years, but by 2015 returning to around its 1965 share (see Figure 21.1). The story of industrial growth must be interpreted with some care. The classification of ‘industry’ includes the mining sub-sector, including petroleum. Separating the manufacturing component from total industrial value-added shows a progressive decline in the contribution of the manufacturing sub-sector, so that its share in the GDP in 2015 was less than in 1965. The share of petroleum had increased very substantially. The disappointing performance of the manufacturing sub-sector was a problem for policy makers, because it is perhaps the most important sector for generating decent jobs.

21.2.1.4 Investment, Savings, and Total Factor Productivity Economic growth is driven principally by a combination of investment and productivity improvements. The modest performance of these elements is the immediate

³ Bowman and Rogan (1999: 6).

⁴ Hansen and Nashashibi (1975: 14).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



Structure of GDP (%)

60 50 40 30 20 10 0

1965

2015 Agriculture

Industry

Services

 . Structure of GDP, 1965 and 2015 (per cent) Sources: World Bank, World Development Indicators database.

reason why the structure of Egypt’s economy changed only slowly. The deeper reasons lie in the interaction of political and economic factors—for example, the ascendancy of groups that favoured consumption and imports over those favouring saving and exports—which impeded policy reforms that would bring about the required institutional and other changes. These matters are explored in Waterbury (1983), Hinnebusch (1988), el-Mikawy and Handoussa (2002), Farah (2009), and in more detail in Ikram (2006, 2018). Hansen and Nashashibi (1975: 14–15) estimate that between 1947 and 1957 gross fixed investment accounted for only 12–13 per cent of GDP. It increased to about 19 per cent from 1957 to 1964, but in the face of rising inflation the authorities opted to restrict demand by cutting back investment. The rebuilding of the armed forces and other requirements after the June 1967 war with Israel further reduced the resources for investment. At the end of the 1960s, the share of investment in GDP was about the same as it had been in 1947.⁵ For roughly two-thirds of the period 1965–2015, investment accounted for less than 20 per cent of GDP; see Figure 21.2. This rate was insufficient to maintain the GDP growth rate necessary to generate sufficient jobs for the increasing labour force, substantial numbers of which therefore remained unemployed, underemployed, or pushed into precarious jobs in the informal economy (Radwan 1998; Krafft and Assaad 2013; World Bank 2014a).To put matters in perspective, for three or more decades East Asian countries such as South Korea, Taiwan, Malaysia, Singapore, and Hong Kong maintained investment rates of about 35 per cent of GDP, while that for China frequently exceeded 40 per cent. The insufficient level of Egypt’s investment rate was not compensated for by improvements in total factor productivity (TFP), that is, the efficiency with which factors of production are combined in use. Several studies indicate that TFP growth played only a minor role in propelling Egypt’s GDP, accounting for perhaps around ⁵ See also Bruton (1983).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

40 35 30 25 20 15 10 5

19

65 19 67 19 69 19 71 19 73 19 75 19 77 19 79 19 81 19 83 19 85 19 87 19 89 19 91 19 93 19 95 19 97 19 99 20 01 20 03 20 05 20 07 20 09 20 11 20 13 20 15

0 Gross capital formation (% of GDP)

Gross domestic savings (% of GDP)

 . Investment and savings, 1965–2015 (percentage of GDP) Sources: World Bank, World Development Indicators database.

only 10 per cent of GDP growth between 1965 and 2015; see Mohammed (2001), Boopen et al. (2009), The Conference Board (2015), IMF (2015), World Bank (2015). Examinations of earlier periods confirm the same result; thus, for example, Maddison (1970, with weights for labour and capital corrected by more recent studies) would have productivity growth accounting for about 11 per cent of the growth of GDP between 1950 and 1965. Most of Egypt’s GDP growth came from employing additional labour and capital; the result more of ‘perspiration than inspiration’, as Krugman (1997) put it in a similar context. To again put matters in perspective, consider that for the high-performing East Asian countries (Hong Kong, Singapore, Indonesia, Malaysia, and Thailand, Korea, and Taiwan) as a group, the average growth of real GDP between 1960 and 2003 was 6.5 per cent a year; of this, capital contributed about 48 per cent, labour 25 per cent, and TFP growth 27 per cent.⁶ A discussion of the principal conceptual issues relating to TFP measurement and of Egyptian data problems will be found in Ikram (2006: 101–16). The experience of China, especially since the late 1970s, is especially impressive. The World Bank (1997) estimated that between 1978 and 1995, roughly 30–58 per cent of China’s GDP growth was accounted for by TFP growth. Yueh (2013) attributed around 45 per cent of the country’s growth for three decades from 1979 to capital accumulation, and about 30 per cent to TFP growth. Within the TFP, growth transfers of labour and capital from public sector enterprises to the private sector explained about 8 per cent of total GDP growth, while approximately 20 per cent of growth was explained by institutional improvements. This highlights not only the importance of TFP to total GDP growth, but also the crucial contribution that institutional improvements make to TFP growth.⁷ Key issues pertaining to institutions that have a special bearing on ⁶ World Bank (1993b: 60–70); Kim and Hong (1997: 183, table 8–5); Thorbecke and Wan (1999: 3–20); Stiglitz and Yusuf (2001: 16, tables 1.3 and 1.4). ⁷ Yueh (2013).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



Egypt’s economic development—such as the bureaucracy, the commercial judicial system, the tax system, the education and training system—are discussed in Ikram (2006: 280–315). An important structural change in Egypt’s investment pattern was in the respective roles of public and private investment. With the sequestration of British and French assets after the Suez Canal war of 1956, and especially with the large-scale nationalizations of productive and financial entities in 1961, the public sector’s role metamorphosed from supporting the private sector to dominating the economy. In 1952, the public sector accounted for an estimated 13 per cent of GDP and 28 per cent of gross capital formation; by 1962, while still accounting for only 18 per cent of GDP, the sector undertook nearly 74 per cent of gross capital formation.⁸ The government’s share in economic activity continued to rise, and Hansen (1975: 203) estimated that in 1973 about 90 per cent of investment and between 63 and 70 per cent of the total availability of resources was accounted for by the public sector.⁹ See also Bruton (1983). The pendulum began to swing back from the early 1990s. Under the Economic Reform and Structural Adjustment Program (ERSAP) agreed to with the IMF and the World Bank in 1991, Egypt began to tilt the balance towards the private sector. The effect on investment was slow. The share of public investment in the total dropped sharply; however, the share of private investment increased only gradually—from 1991 to 2011 (the end of the Mubarak era) it did not exceed 15 per cent of GDP, and total investment continued to hover around 20 per cent of GDP. Private investment exceeded public only from 2006, but the steadily falling share of public investment ensured that total investment even by 2015 had not reached its pre-1991 share of GDP. Egypt’s inability to mobilize sufficient domestic savings remained a perennial problem, contributing both to the insufficiency of domestic investment and to the necessity of borrowing externally. Between 1965 and 2015, the investment rate averaged 20.2 per cent, the domestic savings rate only 13.5 per cent. The actual picture might be a little less bleak than appears at first sight, because with substantial inflows of remittances from expatriate Egyptians, the country’s national savings are higher than its domestic savings. However, a large gap remained between investment and total savings. The shortfall was financed through foreign aid and commercial borrowing from abroad. The borrowing and the amount of foreign aid not provided as a grant added to Egypt’s external indebtedness and preempted an increasing amount of the country’s foreign exchange earnings servicing the debt. The ratio of debt servicing to total foreign exchange earnings reached its zenith in 1986, when paying all the debt servicing and clearing the accumulated arrears would have consumed 114 per cent of the country’s

⁸ Mansfield (1965: 136); O’Brien (1966: 100, 107); and Mead (1967: 272–3). See also Wahba (1994: 73) who puts the private sector’s share of GDP in 1961 at 76 per cent. O’Brien (1966: 107–8) considers the published figures to understate the share of government in GDP, but even after correcting for this concludes that at least two-thirds of GDP took place outside the government’s contribution. ⁹ Availability of resources = GDP (market prices) + Indirect Taxes + Imports. Resource use = Consumption + Investment + Exports.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

foreign exchange earnings. Egypt would have to use up all its foreign exchange earnings and even borrow additional amounts simply in order to meet its debt obligations; it would have to borrow still more to purchase essential imports. Figure 21.2 shows investment and savings as a percentage of GDP from 1965 to 2015 and highlights the gap between them. Figure 21.2 and the discussion of TFP’s contribution encapsulate the essence of Egypt’s continuing economic problems: The investment rate was insufficient to expand the economy at a rate that would fully employ the labour force, while domestic savings fell short even of even the anaemic investment rate. The persistent gap between savings and investment compelled Egypt to turn to foreign savings, which meant a continual piling up of external debt. Behind the savings–investment performance lie deeper institutional issues, such as matters of governance, the structure of incentives, the working of the bureaucracy and the commercial judicial system, and weaknesses in the education and training system; for a discussion of these issues see Ikram (2006: 280–315).

21.2.2 Structure of Employment In a modern economy, employment is the most important means of earning income; questions concerning the structure of the labour market thus constitute perhaps the most important issues facing policy makers. Unfortunately, any long-term study of unemployment in Egypt is bedevilled with problems arising from inconsistencies of definition, coverage, measurement, and interpretation, both within and between the different data series.¹⁰ Thus no single source can serve as a definitive basis for long-term analysis. This explains Fergany’s (1995) remark that monitoring changes in unemployment in Egypt can border on detective work. Recent data, especially from the mid-1990s, are much better and thus analysis based on them can be more securely grounded.

21.2.2.1 Growth and Structural Change in Employment The Egyptian labour force is estimated to have increased from 7.8 million in 1960 to about 28 million in 2015, an annual average growth rate of about 2.4 per cent. As with GDP, the principal change was the drop in the share of agriculture; employment in this sector fell from 54 per cent of total employment in 1960 to 28 per cent in 2015. Between these years, the share of employment in industry increased from 10 per cent to 25 per cent and that in the services sector from 36 per cent to 47 per cent (see Figure 21.3).

¹⁰ Principally, the Population Census, the Labor Market Sample Surveys, and the Egypt Labor Market Panel Surveys, plus special studies by analysts such as Samir Radwan, Ragui Assaad, Nader Fergany, and others.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



60 50 40 30 20 10 0

1960

1980 Agriculture

Industry

2015 Services

 . Structure of employment, 1960–2015 (percentage of total employment) Sources: 1960, World Bank (1978: volume 6, statistical appendix, table 1.11); 1980, 2015, World Bank, World Development Indicators database.

A brief review of the progress of employment in Egypt since 1960 would highlight the following points. Unemployment for most of the period between 1960 and 1975 remained less than 4.5 per cent of the labour force; indeed, until 1966 it remained below 1.5 per cent. This felicitous outcome was a result of the growth spurt engineered by the importsubstitution strategy adopted by the revolutionary regime, and the government’s policy of guaranteed employment to all graduates of universities and technical institutions. The launch of President Sadat’s infitah (‘open-door’ policy) in 1974 produced a massive inflow of external resources that boosted economic growth to an average of over 8 per cent per year. However, policy distortions resulted in growth occurring to a large extent in non-tradable sectors, namely construction and financial services, which together accounted for less than 30 per cent of employment growth between 1976 and 1986. The main policy distortions were an overvalued exchange rate and domestic interest rates that were very low or negative in real terms, creating a large bias in favour of capital-intensive methods of production. This bias was reinforced by huge subsidies on the use of energy. By artificially depressing the price of capital relative to that of labour, the subsidies acted as an indirect tax on labour and thus encouraged entrepreneurs to invest in capital-intensive rather than in labour-intensive activities. Radwan (2002) aptly terms the years 1975–85 the ‘decade of jobless growth’. The resource inflow enabled the government to step up its employment programme, and employment in the civil service (i.e. government service excluding public sector enterprises) grew faster than in any other sector in the economy. Assaad (1995) highlights two mutually reinforcing structural effects that emerged as a result of the government taking on the role of primary employer in the economy. First, by 1986 the government sector employed almost 75 per cent of all graduates and 60 per cent of employed females, that is, the labour market became segmented along educational and gender lines. Second, government sector pay strongly influenced wages in the private sector. Because of the large weight of the government sector in the economy, the pay it offered to graduate labour set a minimum expected level in other sectors of the economy, raising its relative cost.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

The economic boom, and much of the external resource inflow, came to an end with the collapse of oil prices in 1982. The price of Egypt’s petroleum exports plummeted from an average of $33.60 per barrel in 1981 to $14.29 per barrel in 1987.¹¹ Between these years, the share of oil-generated receipts in government revenues dropped from more than 20 per cent to 9 per cent. The government did not formally abandon its guaranteed employment programme, but strung out the process so that graduates had to wait for several years before receiving an appointment. Between 1987 and 1993, economic growth declined, and unemployment touched 10–11 per cent in certain years, and increasingly was comprised of educated young people in their twenties entering the labour force for the first time. The informal sector became an important source of employment in the 1990s (accounting for over 20 per cent of employment), but it was not sufficient to absorb the numbers entering the labour force. Economic growth resumed more briskly after 1993, but was hit by two major blows in 1997. The first was the financial crisis of East Asia that impacted on Egypt’s trade and investment inflows. The second setback was an attack in November by fundamentalist groups on foreign tourists in the major tourist site of Luxor. The economy recovered somewhat towards the end of the century, but the growth rate still remained insufficient to create the number of jobs required for the expanding labour force. The outcomes of Egypt’s labour market during this period highlight the importance that must be paid to the composition of growth, and not only to its rate. A number of studies—for example, Karshenas (1994), Radwan (1997), Fergany (1999), Assaad (2000), Galal (2002)—pointed out that the primary cause of unemployment in Egypt was a structural deficiency in the demand for labour that resulted principally from the capital-intensive character of economic growth following the fortuitous expansion in the inflow of external resources after 1975. This change was reflected, for example, in estimates of the employment elasticity of output growth in the manufacturing sector. The elasticity declined from 0.43 in 1975 to 0.28 in 1992, and was generally lower than in some of Egypt’s principal competitors—for example, Malaysia (0.84), Indonesia (0.66), Thailand (0.72), or Jordan (1.04). These high employment elasticities showed that the export-oriented industrialization pursued by countries, especially in East Asia, embodied a high degree of labour intensity. Some estimates for Egypt, for example, Fergany (1998, 1999) put the economy-wide constant prices employment elasticity at only 0.054 between 1990 and 1995, and 0.12 in manufacturing. Fegany calculated that over this period, LE 1 million of output growth produced only eight job opportunities. World Bank (2014b: 3–4) pointed out that in in Egypt for the period 1996–2012 the correlation between growth in output and in employment was weaker than one would expect. The Bank attributed this to policies that artificially lowered the cost of capital; and to the long period of government-guaranteed employment, which provided jobs regardless of how the economy was performing. The exchange rate and trade policy bias in favour of capital-intensity (discussed in Section 21.2.3) was reinforced by monetary policies that for most of the period 1960–90

¹¹ Weighted according to the prices received and the shares contributed by each oilfield in exports.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



held nominal interest rates to below 5 per cent, even in the face of high inflation. The result was that real interest rates remained low, and for many years, for example, 1975 to 1991, they were negative—World Bank (1993a); Fry (1995); Mohieldin (1995, 1998). The low and negative real interest rates reduced the cost of capital and provided another inducement to prefer capital-intensive technologies over those that were more labour-intensive. Fiscal policies further increased the incentive for capital-intensive methods of production. Energy was heavily subsidized, with energy subsidies reaching over 7 per cent of GDP in 2014, which, as the World Bank (2015: xi) noted, amounted to more than the combined spending on health, education, and public investment.¹² All in all, therefore, government policies made less than optimal use of Egypt’s most abundant resource, namely, the large and increasing labour force, and tilted the economy’s structure in a direction not consistent with comparative advantage.

21.2.2.2 Trends in Structural Transformation of the Labour Market The World Bank (2014b: 25) highlighted three long-term trends that from the 1990s were transforming the structure of the Egyptian labour market. First, informal private sector employment increased from 30.7 per cent of the labour force in 1998 to 40.0 per cent in 2012. Second, employment in the public sector fell from 34.0 to 27.1 per cent. Third, formal private sector employment remained more or less constant, between 13.0 and 13.5 per cent of the labour force. Accompanying the structural changes was a decline in labour productivity (measured as output per worker). Labour productivity grew at an annual average rate of 2.4 per cent between 1990 and 2011. The overall figure conceals an average growth rate of 2.7 per cent during 1990–2002, which declined to 2.2 per cent during 2003–11. These changes were brought about by different forces. In the first period, the chief impetus to productivity growth was provided by labour moving out of low productivity agriculture into the higher productivity services sector (employment in industry was not noticeably affected). In the second period, sectoral shifts were smaller; productivity growth came mainly from improvements in certain sub-sectors of industry, especially mining, but not from a reallocation of labour to higher productivity sectors.¹³ The most crucial development in the structure of Egypt’s labour market since the 1990s was the increasing share of employment in informal jobs.¹⁴ Growth slowdowns since the mid-1990s were principally marked not by increasing the rate of unemployment but by decreasing the quality of employment, that is, a shift towards informal jobs. Data from the Egyptian Labor Market Panel Survey (ELMPS) show that informal employment increased from 53 per cent of the labour force in 1998 to 61 per cent in 2012. ¹² Energy subsidies had a further detrimental effect in that they were regressive, with the richest 20 per cent of Egyptian households capturing 60 per cent of all energy subsidies. ¹³ World Bank (2014b: 22). ¹⁴ Informal workers are defined by World Bank (2014b: 15–16) as those who lack both social insurance and a written contract, and who are not in the farm sector or the formal sector. Formal sector workers, following the standard ILO definition, are those who work for the government or for public enterprises, or any working arrangement that provides either social insurance or a formal-written work contract.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

The structural reason is that the Egyptian labour market is dominated by micro enterprises, which employ fewer than ten workers, account for nearly 95 per cent of all establishments, tend to be part of the informal economy, and are unable to take advantage of economies of scale. They also tend to cluster in low-productivity subsectors: for example, about 35 per cent of employment was in retail trade, in which the average firm size amounted to only two employees (the owner and one other). Despite their individually small size, micro establishments accounted for more than 70 per cent of total employment in 2012. At the other end of the scale, firms that employed more than 100 employees accounted for only 16 per cent, while squeezed in between were the small and medium-sized establishments that employed between 10 and 100 workers and accounted for about 12 per cent of all jobs. A serious problem for employment in the Egyptian economy is that the performance of these predominantly small firms differs from that typically found among small firms in other countries. Numerous studies covering different countries and over many years—for example, Mansfield (1962), Hall (1987), Hart and Oulton (1996), Davidsson et al. (2006), Ayyagari et al. (2011), Hsieh and Klenow (2012), Haltiwanger et al. (2013)—show that younger and smaller firms have higher employment growth rates than older and larger firms. Moreover, the studies show that firms grow over time, and firm age is an important factor in the ability to create jobs. However, the World Bank (2014b: 108–9, and figure 5.6) demonstrated that establishments in Egypt showed little relation between firm age and job growth and that, on average, older establishments employed hardly any more workers than did younger establishments. There was a conspicuous absence of the so-called ‘gazelle’ firms, that is, typically, the 5–10 per cent of firms that in developed countries rapidly bound ahead and deliver 50–80 per cent of aggregate employment creation, and are the major transformers of the economic structure. Adding to Egypt’s problem was that the share of employment in large establishments declined from 23 to 16 per cent between 1996 and 2006. Most of the large establishments had not grown ‘organically’ as a result of their competitive dynamism. As the World Bank (2014b: 106) notes—of the 206 biggest establishments (those with more than 1,000 employees), 131 were founded before 1976 in the period of state-led industrialization before Sadat’s infitah (the ‘open-door’ policy) began to reverse the roles of the public and private sectors. The biggest firms therefore did not grow because of their higher efficiency and productivity, but were ‘born large’ in the Nasser period and its immediate aftermath.

21.2.3 Structural Transformation of the External Sector For virtually the entire period 1965–2015 Egypt’s external sector was characterized by severe structural weaknesses, and showed a tendency to transform only slowly. The weaknesses included a narrow export base dominated by hydrocarbon exports, which

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



accounted for approximately one-third of export earnings, and a commodity composition, the earnings from which grew at a markedly slower rate than world trade. Merchandise imports generally remained two to three times earnings from merchandise exports, and the deficit on the balance of visible trade was met by inflows from invisibles, such as worker remittances, Suez Canal dues, tourism, and foreign aid. The ‘invisible’ items accounted for almost two-thirds of total foreign exchange receipts; thus over the period as a whole, Egypt remained principally an exporter of services. The country’s economic policies impacted on the external sector in such a manner that growth in the GDP triggered an earlier and sharper increase in imports than in foreign exchange earnings. The external deficit had to be controlled, because it was limited by the country’s access to external financing, which fell well short of demand. Managing the deficit by restricting imports—the strategy favoured by the authorities— left the capital stock underutilized and labour under- or unemployed. The slow structural transformation of the balance of payments, therefore, acted as the ultimate constraint on Egypt’s growth during the period. The principal factor accounting for this outcome was a persistent anti-export bias in the policy framework, created by exchange rate and trade policies that accentuated the fall in the country’s productivity and export competitiveness. The continuing decline in Egypt’s share of world trade tells the story. In 1950, for every $100 of world exports, Egypt accounted for about one dollar; by 1965 the share had fallen to 37 cents, by 1973 to 20 cents, and by 2015 had dropped further to 14 cents. A demand decomposition of growth between 1965 and 2005 showed that nearly 85 per cent of the growth in GDP resulted from increases in domestic demand, and only about 15 per cent from foreign demand. An indication that many of Egypt’s exports were not in line with the country’s comparative advantage is provided by estimates of the index of revealed comparative advantage.¹⁵ Calculations of the Revealed Comparative Advantage index by the World Bank for 1970, 1980, and 1992, and by Egypt’s Ministry of Trade for 2005 showed that Egypt’s comparative advantage lay mainly in sectors related to food, textiles and clothing products, and in some manufactures. The country’s comparative advantage was lowest in capital-intensive industries, and yet it was these industries that received the most encouragement. Changes in the structure of Egypt’s exports occurred very slowly. A ‘constant market share analysis’ of Egypt’s exports at the three-digit level for the decade 1990–2000 showed that Egypt’s export growth (which averaged less than 2 per cent a year) was unable to benefit from the buoyancy in international trade (which was growing at 6 per cent), because the country’s export basket was insufficiently weighted by the commodities that were growing fastest in world trade; the details are reported in Ikram (2006: 129–31). Further confirmation of the substantial mismatch between Egypt’s export ¹⁵ If xij is the value of country i’s exports of commodity j and Xtj is the country’s total exports, its revealed comparative advantage index is: RCAij = (xij/Xtj)/(Xiw/Xtw) where the w subscripts refer to world totals.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

basket and the dynamic elements in world trade was provided by a TradeCAN (Trade Competitiveness Analysis of Nations) analysis for the 1990s. This showed that only 27 per cent of Egypt’s exports consisted of ‘rising stars’, that is, products whose shares were growing both in world imports and also in Egypt’s exports.¹⁶ The basic reason behind the lacklustre performance of exports was an incentive framework—driven principally by the exchange rate and import protection measures—that created a strong anti-export bias. First, the nominal exchange rate was fixed for long periods, and with inflation in Egypt generally higher than in her competitors, the real effective exchange rate kept appreciating; its trajectory was determined fortuitously by the difference between inflation rates in Egypt and in competing countries. The authorities were loath to devalue the nominal exchange rate to compensate for the appreciation of the real effective rate, because (i) it would have increased the prices of, and consequently the subsidies for, essential consumer items (subsidies could amount to 20 per cent of budgetary expenditures); and (ii) it would have required mobilizing larger amounts of domestic resources to service external debt. Second, there was a conviction that Egypt’s exports would not respond adequately to a devaluation, a condition sometimes described as ‘elasticity pessimism’. Proponents of this view believed, in effect, that the Marshall–Lerner–Robinson condition for a devaluation to improve the trade balance was not met in Egypt.¹⁷ Justification for such an attitude was provided by econometric estimates of changes in the demand for Egypt’s exports in response to exchange rate changes that showed that Egypt’s exports were inelastic with respect to price. Both theory and experience have not been kind to this view. Orcutt (1950) had shown that the results obtained in most estimations of trade elasticities had to be viewed with reservation, because the data and the methods employed (single equation estimation by Ordinary Least Squares) tended to bias the estimates downwards. Orcutt’s paper has been subjected to a large amount of discussion and empirical testing, and his main finding holds, namely, that regressing total exports on the average price will in general not give a fair estimate of the elasticity of demand.¹⁸ In particular, the downward bias created by not using a simultaneous equation framework is likely to remain significant, and short-term elasticities will tend to be lower

¹⁶ Government of Egypt/World Bank (2000); Mohammed (2001). ¹⁷ Formally, the condition states that if trade is initially balanced and the elasticities of supply of exports and imports are infinite, devaluation will improve the trade balance if the absolute sum of the foreign elasticity of demand for exports and the home elasticity of demand for imports (measured in the same currency) exceeds unity. That is, devaluation will improve the current account of the balance of payments if: |Em| + |Ex| > 1 where Em is the price elasticity of demand for imports, and Ex is the price elasticity of demand for exports. The condition is named after Marshall (1924: appendix J); Robinson (1937: 138–46); Lerner (1944: 377–9). ¹⁸ For example, Neisser and Modigliani (1953: 226–30); Harberger (1957: 5089); Haberler (1958: 3–9); Klein et al. (1961: 126–36); Prais (1962: 260–70); and more recently Winters (1991: 276–8), Thirlwall and Gibson (1992: 166–8), and Joshi and Little (1994: 284–91).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



than their long-term estimates. The relevant econometric issues are summarized in Ikram (2006: 135–9). More methodologically-robust studies, such as Pearson (1997), Nathan Associates (1999) examined the performance of Egypt’s international trade in 1980–98 and their estimates of the elasticities were relatively high. Nathan Associates estimated the tradeweighted import demand elasticities for 1997 at 1.1 in the short run and 2.5 in the long. The corresponding export elasticities were also higher: for all products and markets, the 1997 trade-weighted average elasticity was 0.89 in the short run and 0.81 in the long run. Thus, the absolute sum of the export and import elasticities was much larger than unity, and the Marshall–Lerner–Robinson condition was comfortably met. The experience of high-exporting countries similarly pours cold water on the notion that fast export growth can go hand-in-hand with an appreciating real exchange rate. All the East Asian high exporting countries (China, Korea, Taiwan, Malaysia, Singapore, Hong Kong, Thailand, and Vietnam) kept their exchange rates competitive. But Egypt’s real effective exchange rate appreciated for most of the quarter century from 1980. Moreover, Egypt’s own experience when it did devalue, for example in 2004, showed a rapid boost in exports. Changes in the export structure were also not encouraged by the degree of protection given to imports, which made producing for the domestic market substantially more profitable than doing it for the external. From 1960 to 2004 Egyptian tariffs remain generally higher than those of its neighbours. World Bank studies estimated tradeweighted tariffs in Egypt at 48 per cent in 1976 and 45 per cent in 1981; even in 1996 the rate averaged 31 per cent, significantly exceeding the world average of 8.2 per cent and that of developing countries (21.4 per cent). A major reduction and simplification of the tariff structure had to wait until 2004.¹⁹ Throughout the period, Egypt also employed a variety of non-tariff measures, generally dressed up as quality control or health protection measures. ‘In practice, however,’ commented the World Bank (1998: 32), ‘quality control has become [another] means to protect local industry.’ Nathan Associates (1998: 5) estimated that in the 1990s, ‘quality control’ measures increased costs by 5–90 per cent to the producers and importers on whom they impacted, diminished exports by at least 9–12 per cent and GDP by more than 1 per cent. Such measures thus further increased the attractiveness of the domestic over the international market, and slowed down the transformation of the export structure.

21.2.4 The Public Finances Structural changes in Egypt’s public finances were also slow. From 1960 until the mid1970s the public finances were dominated by two themes. First, Egypt had been in a

¹⁹ World Bank (1978) and Ikram (1980) provide a discussion of the earlier years.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

state of war since 1948; this meant continuing heavy expenditures (not always fully recorded in the budget) on defence. Second, the move towards planning and ‘Arab Socialism’ with the initiation of the Ten Year Comprehensive Plan in 1960 required heavy expenditures on public investment, especially in import-substituting manufacturing. Another aim of the ‘Arab Socialism’ strategy was to achieve greater equity in consumption. In order to attain this, many commodities (as many as eighteen in several years) considered to be ‘essential’ were subsidized.²⁰ Subsidies became an important item in the budget from 1973, increasing from less than 2 per cent of GDP in 1971 to about 5 per cent in 1973 and to 10 per cent in 1978. Over the 50-year period from 1965 to 2015, subsidies accounted for 12 per cent of total expenditure in the budget; in certain years they could be much higher, for example, during 1975–81 they consumed an average of nearly 22 per cent of budgetary expenditures. Fiscal reforms from 2004 succeeded in moderating the subsidy bill, and by 2015 it claimed 9 per cent of expenditures. Between 1960 and 2015, the budget was continually in deficit, with the gap veering between 22.6 per cent of GDP in 1975 and 1.6 per cent in 1995. Over the period as a whole, the deficit averaged 11.5 per cent of GDP (in 2015 it was about 12 per cent). The deficit persisted because of a basic structural weakness—expenditures responded to domestic inflation while revenues responded chiefly to exogenous influences. Expenditures on consumer subsidies, public sector salaries and pensions, government purchases, and interest on domestic debt rose in line with domestic inflation and the expansion of government employment; since the base kept growing, expenditures increased pari passu. Revenues, however, marched to a different drummer. The exogenous influences were transmitted through the growing dependence of the budget on ‘economic rents’; that is, payments in excess of those required to keep a factor of production in use. The definition is sometimes used in a more elastic sense; the World Bank (1983: 4) for example, calls them ‘exogenous’ resources ‘in the sense of their having very little to do with the productivity of Egypt’s domestic labor force’. The charge that Egypt had become a ‘rentier’ state has been levelled by several writers and institutions, and is virtually a staple of an analyst’s Egyptian toolbox—for example, World Bank (1974, 1983); IMF (1976); Roy (1980); Abdel-Fadil (1983); Waterbury (1983); Richards (1984, 1991); el-Beblawi (1987); Springborg (1989); Hansen (1991); Sadowski (1991); Ikram (2006, 2018); Marcou (2008); Farah (2009); Soliman (2011); and Kandil (2012). The main exogenous resources are foreign aid, Suez Canal tolls, remittances from expatriate Egyptian workers, a part of the payments received for petroleum exports, and a part of the receipts from tourism. The common characteristic of the ‘economic rents’ garnered by Egypt is that their burden fell on foreigners and not on Egyptians. Such measures were politically attractive because they enabled Egyptian regimes to raise resources without alienating ²⁰ The definition of ‘essential’ could be a little idiosyncratic: even cigarettes and halawa (a dessert) were subsidized.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



their own constituency. The concomitant structural weakness is that the generation of these resources was largely outside the control of Egyptian policy makers. Suez revenues are determined by traffic through the Canal, and thus depend on the health of international trade. Tourism is a luxury item and expenditures on it respond to economic conditions in the countries from which tourists originate and to their view of the security situation in Egypt and the region, and how these compare with alternative destinations. Revenue from petroleum taxes is contingent on oil prices that are determined by cartels and producers outside Egypt, while the amount of external aid is decided in foreign capitals. Ex-patriate worker remittances depend upon the health and attitudes of the host economy. There have also been periods when, for political reasons, the host country has expelled Egyptian workers. Largely as a result of the disjunction between the structural influences impacting on them, budgetary revenues and expenditures responded differently to changes in the variables on which they were levied or expended. Ahmed (1984: 41) and Ikram (2006: 165–7, 173–4) showed that almost all the expenditure items were more buoyant with respect to the underlying variables than were those on the revenue side. The regularly higher buoyancy values for expenditure tend to confirm that Egypt’s budgetary problems were largely structural, arising chiefly from ‘consistently earning in Centigrade and spending in Fahrenheit’, as one observer put it.

21.3 T P E  S R

.................................................................................................................................. The second part of this essay is a brief political-economy analysis of why the authorities were reluctant to push policies that would promote more rapid transformation of the economy.²¹ At any given time there is a certain distribution of winners and losers in an economy. Policy reforms that would substantially alter the economy’s structure are also likely to drastically change this distribution. What makes policy reform particularly difficult in developing countries is that the pre-reform group of economic winners has generally achieved that position by attaining political strength through wealth or by forming coalitions with centres of power, such as the monarchy, the military, the bureaucracy, or the religious establishment. In order to bring about major reforms of the economic structure, the resistance of such interest groups has to be overcome; see Olson (1965, and especially 1982). All regimes, whether democratic or authoritarian, seek to prolong their term in office. However, the stakes are much higher in an authoritarian environment—regimechange is likely to result not simply in a loss of office, but quite possibly of liberty ²¹ This section draws particularly heavily on Ikram (2018).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

and even of the life and limb of the autocrat, his family, and close associates. An authoritarian regime—such as has ruled Egypt from 1952 to 2011—is therefore likely to go to considerable lengths to hold down popular dissent and to create coalitions that would buttress its position; see Alesina (1992). A detailed discussion of the political economy of reform in Egypt is provided by Ikram (2018). The political-economy behaviour of Egyptian regimes has been the subject of considerable study: a partial sampling would be Baker (1978), Cooper (1979), Roy (1980), Waterbury (1983, 1985), McDermott (1988), Hinnebusch (1988), Springborg (1989), Hansen (1991), Wahba (1994), Ikram (2006, 2018), Marcou (2008), Soliman (2011), and Kandil (2012). These studies offer different emphases and nuances, but the principal underlying theme could be paraphrased as follows. From 1952 until 2011 Egypt was ruled by authoritarian regimes, who were exceptionally vulnerable because they lacked the legitimacy of a free, democratic election. These governments sought to preempt popular dissent by forging an implicit compact: the regime would ease the economic burden on the people through a combination of consumer and other subsidies, and by minimizing the resources extracted from them by way of taxes; in return, the people would be politically quiescent. With the population growing rapidly and its consumption and other expectations steadily increasing (between 1965 and 2015 real public consumption increased at an annual average rate of 4.4 per cent, almost twice the population growth rate), this strategy compelled rulers to rely heavily on external economic rents and exogenous resources, even though it heightened Egypt’s vulnerability to foreign pressures.

Moreover, from 1975, regimes built coalitions with large businesses by providing them with special advantages; creating an environment of ‘crony capitalism’. The World Bank (2014a) and Diwan et al. (2015) provide quantitative details of the firm-specific privileges (such as import protection, energy subsidies, tax concessions, favoured availability of bank credit, preferential permission to purchase government-developed land, reduction of regulatory burdens, etc.) that were made available to the group of politically-connected business elites.²² As a result of these advantages, the differential in profits between the politically connected firms and the unconnected was more than two times larger between 2005 and 2010, but disappeared after the fall of the Mubarak regime in February 2011. The higher profitability of politically connected firms could be entirely explained by their access to non-tariff measures²³ and energy subsidies; without these advantages they were no more profitable than unconnected firms. The distortions in the economic structure provided politically-connected businesses with very large economic rents, which would disappear if the structure were reformed; thus the business elites had

²² A useful (more qualitative) discussion is Loewe (2013). ²³ Egypt’s import tariffs were reduced at the end of the 1990s and protection against imports was increasingly provided through the use of non-tariff measures.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



every incentive to exercise their financial and political influence to impede the transformation of the economic structure. Given the forces supporting the status quo, how can structural transformation be brought about? What would it take to upset the existing equilibrium and to change the dynamics between coalitions? Drazen (2000) provides a survey of several models that have been proposed as answers. The best explanations emphasize the part played by crises. In such circumstances, trying to maintain the status quo would impose unacceptable political and economic costs on the pre-crisis beneficiaries; for empirical discussions of relevance to developing countries see, for example, Nelson (1990); Przeworski and Limongi (1993); Bruno and Easterly (1996); Lora (1998); Drazen and Easterly (1999); Haggard (2000); Lora and Olivera (2004). Crises have played an important role in the structural transformation of Egypt’s economy. Thus, the 1952 Egyptian revolution by the Free Officers had its genesis in the outcome of the 1948 war with Israel; the 1956 sequestrations of British and French assets resulted from the Suez Canal War; the 1961 large-scale nationalizations followed the breakup of the United Arab Republic with the secession of Syria from the union; the abandonment of ‘Arab socialism’ and the move towards the ‘open-door policy’ (the infitah) in 1974 had its roots in consequences of the Arab-Israeli war of 1973. Of course, a direct threat to Egypt was not always necessary—the major reforms in the 1990s followed Egypt’s decision to participate in the war with Iraq. The war was fought mainly by Western troops; Egypt’s role was to provide the conflict with an international fig leaf, namely, that the war was not simply a matter of the West versus Arabs. The participation rewarded Egypt with a programme of generous debt write-offs and financial assistance that enabled the country to emerge from under a heavy burden of external debt. The massive reduction of the debt burden and the provision of a much larger volume of external assistance under this Economic Reform and Structural Adjustment Program (ERSAP) provided Egypt with the resources to cushion the impact of reform policies; detailed discussions of this program are in Subramanian (1997) and Ikram (2006, 2018).

R Abdel-Fadil, M., 1983. Speculations on the Question of the Egyptian Economy, Cairo: Dar al-Mustaqbal al-Arabi. Ahmed, S., 1984. ‘Public Finance in Egypt.’ Staff Working Paper No. 639. Washington, DC: World Bank. Alesina, A., 1992. ‘Political Models of Macroeconomic Policy and Fiscal Reform.’ Policy Research Working Paper No. 970. Washington, DC: World Bank. Assaad, R., 1995. The Effects of Public Sector Hiring and Compensation Policies on the Egyptian Labor Market, Cairo: Economic Research Forum. Assaad, R., 2000. The Transformation of the Egyptian Labor Market: 1988–1998, Cairo: Economic Research Forum.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Ayyagari, M., A. Demirguc-Kunt, and V. Maksimovic, 2011. ‘Small vs. Young Firms across the World: Contribution to Employment, Job Creation, and Growth’. Policy Research Working Paper No. 5631. Washington, DC: World Bank. Baker, R. W., 1978. Egypt’s Uncertain Revolution under Nasser and Sadat, Cambridge, MA: Harvard University Press. Boopen, S., R. Sawkut, and S. Ramessur, 2009. ‘Using Growth Accounting to Explain Sources of Growth: The Case of COMESA’, International Journal of Business Researchi, 9 (3). Bowman, A. K. and E. Rogan (eds), 1999. Agriculture in Egypt: from Pharaonic to Modern Times, Oxford: Oxford University Press for the British Academy. Bruno, M. and W. Easterly, 1996. ‘Inflation’s Children: Tales of Crises That Beget Reforms’. American Economic Review Papers and Proceedings, 86, pp. 213–17. Bruton, H. J., 1983. ‘Egypt’s Development in the Seventies’, Economic Development and Cultural Change, 31 (July), pp. 679–704. Conference Board, 2015. Total Economy Database. Available at: http://www.conference-board. org/retrievefile.cfm?filename=The-Conference-Board-2015-Productivity-Brief-SummaryTables-1999-2015.pdf&type=subsite accessed 27 March 2016. Cooper, M., 1979. ‘Egyptian State Capitalism in Crisis: Economic Policies and Political Interests, 1967–1971’, International Journal of Middle East Studies, 10 (4), pp. 481–516. Davidsson, P., F. Delmar, and J. Wiklund, 2006. Entrepreneurship and the Growth of Firms, Cheltenham: Edward Elgar. Diwan, I., P. Keefer, and M. Schiffbauer, 2015. ‘Pyramid Capitalism: Political Connections, Regulation, and Firm Productivity in Egypt’. Policy Research Working Paper No. 7354. (Macroeconomics and Fiscal Management Global Practice Group). Washington, DC: World Bank. Drazen, A., 2000. Political Economy in Macroeconomics, Princeton, NJ: Princeton University Press. Drazen, A. and W. Easterly, 1999. ‘Do Crises Induce Reform? Simple Empirical Tests of Conventional Wisdom’. Working Paper. el-Beblawi, H., 1987. ‘The Rentier State in the Arab World’, Arab Studies Quarterly, 9 (4). el-Beblawi, H., 2008. ‘Economic Growth in Egypt: Impediments and Constraints (1974–2004)’. Commission on Growth and Development, Working Paper No. 14, Washington, DC: World Bank. el-Mikawy, N. and H. Handoussa (eds), 2002. Institutional Reform and Economic Development in Egypt, Cairo: Economic Research Forum. Farah, N. R., 2009. Egypt’s Political Economy, Cairo: American University in Cairo Press. Fergany, N., 1995. Recent Trends in Participation in Economic Activity and Open Unemployment in Egypt, Cairo: Al Mishkat. Fergany, N., 1998. Dynamics of Employment Creation and Destruction: Egypt, 1990–95, Cairo: Al Mishkat. Fergany, N., 1999. An Assessment of the Unemployment Situation in Egypt, Cairo: Al Mishkat. Fry, M. J., 1995. Money, Interest, and Banking in Economic Development, 2nd edn, Baltimore MD: Johns Hopkins University Press. Galal, A., 2002. ‘Employment and Unemployment in Egypt’. Policy Viewpoint 11 (June), Cairo: Egyptian Center for Economic Studies. Government of Egypt/World Bank, 2000. ‘Plan of Action for Export Promotion’. Report of the Joint Task Force of the Government of Egypt and the World Bank, Cairo: World Bank. Haberler, G., 1958. ‘Introduction to “Problems in International Economics” ’, Review of Economics and Statistics, 40 (supplement), pp. 3–9.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



Haggard, S., 2000. ‘Interests, Institutions, and Policy Reform’. In A. Krueger, ed., Economic Policy Reform:The Second Stage. Chicago: University of Chicago Press, pp. 21–57. Hall, B. H., 1987. ‘The Relationship between Firm Size and Firm Growth in the US Manufacturing Sector’, Journal of Industrial Economics, 35, pp. 583–600. Haltiwanger, J., R. S. Jarmin, and J. Miranda, 2013. ‘Who creates Jobs? Small versus Large versus Young’, Review of Economics and Statistics, 95 (2), pp. 347–61. Hansen, B., 1975. ‘Arab Socialism in Egypt’, World Development, 3 (4), pp. 201–11. Hansen, B., 1991. Egypt and Turkey: The Political Economy of Poverty, Equity, and Growth, New York: Oxford University Press. Hansen, B. and G. Marzouk, 1965. Development and Economic Policy in the UAR (Egypt), Amsterdam: North-Holland. Hansen, B. and D. Mead, 1965. ‘The National Income of the UAR (Egypt), 1939–62’, in S. Goldberg and P. Deane, eds, Studies in Short-Term National Accounts and Long-Term Economic Growth, Income and Wealth Series 11, London: Bowes and Bowes. Hansen, B. and K. Nashashibi, 1975. Foreign Trade Regimes and Economic Development: Egypt, New York: National Bureau of Economic Research. Harberger, A., 1957. ‘Some Evidence on the International Price Mechanism’, Journal of Political Economy, 65 (6), pp. 506–21. Hart, P. E. and N. Oulton, 1996. ‘The Growth and Size of Firms’, Economic Journal, 106 (3), pp. 1242–52. Hinnebusch, R. A., 1988. Egyptian Politics under Sadat, updated edition, Boulder, CO: Lynne Rienner. Hsieh, Chang-Tai, and P. J. Klenow, 2012. ‘The Life Cycle of Plants in India and Mexico’, Working Paper No. 18133. Cambridge, MA: National Bureau of Economic Research. Ikram, K., 1980. Egypt: Economic Management in a Period of Transition, Baltimore: Johns Hopkins University Press. Ikram, K., 2006. The Egyptian Economy, 1952–2000: Performance, Policies and Issues, London: Routledge. Ikram, K., 2018. The Political Economy of Reforms in Egypt: Issues and Policymaking since 1952, Cairo: American University in Cairo Press. IMF (International Monetary Fund), 1976. ‘Arab Republic of Egypt: Recent Economic Developments’, Washington, DC: IMF. IMF (International Monetary Fund), 2015. ‘Arab Republic of Egypt: 2014 Article IV Consultation—Staff Report’. Joshi, V. and I. M. D. Little, 1994. India: Macroeconomics and Political Economy, 1964–1991, Washington, DC: World Bank. Kandil, H., 2012. Soldiers, Spies, and Statesmen: Egypt’s Road to Revolt, London: Verso. Karshenas, M., 1994. ‘Structural Adjustment and Employment in the Middle East and North Africa’. Working Paper, Cairo: Economic Research Forum. Kim, K.-S. and S.-D. Hong, 1997. Accounting for Rapid Growth in Korea, 1963–95, Seoul: Korea Development Institute. Klein, L. R., R. J. Ball, A. Hazlewood, and P. Vandome, 1961. An Econometric Model of the United Kingdom, Oxford: Basil Blackwell. Krafft, C. and R. Assaad, 2013. ‘The Structure and Evolution of Employment in Egypt: 1998–2012’. Working Paper No. 805, Cairo: Economic Research Forum. Krugman, P., 1997. ‘What Ever Happened to the Asian Miracle?’ Fortune, 13 (4), pp. 26–7. Lerner, A. P., 1944. The Economics of Control, London: Macmillan.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Little, T., 1967. Modern Egypt, New York: Praeger. Loewe, M., 2013. ‘Industrial Policy in Egypt 2004–2011’. Discussion Paper No. 13/2013, Bonn: Deutsches Institut für Entwicklungspolitik. Lora, E., 1998. ‘What Makes Reforms Likely? Timing and Sequencing of Structural Reforms in Latin America’. Working Paper, Office of the Chief Economist. Washington, DC: InterAmerican Development Bank. Lora, E. and M. Olivera, 2004. ‘What Makes Reforms Likely: Political Economy Determinants of Reforms in Latin America’, Journal of Applied Economics, VII (1), pp. 99–135. Maddison, A., 1970. Economic Progress and Policy in Developing Countries, London: Allen and Unwin. Mansfield, E., 1962. ‘Entry, Gibrat’s Law, Innovation, and the Growth of Firms’, American Economic Review, 52 (5), pp. 1023–51. Mansfield, P., 1965. Nasser’s Egypt, London: Penguin Books. Marcou, J., 2008. L’Égypte contemporaine, Paris: Le Cavalier Bleu. Marshall, A., 1924. Money, Credit and Commerce, London: Macmillan. Mead, D. C., 1967. Growth and Structural Change in the Egyptian Economy, Homewood, IL: Richard D. Irwin. McDermott, A., 1988. Egypt from Nasser to Mubarak: A Flawed Revolution, London: Croom Helm. Mohammed, N., 2001. ‘Sources of Economic Growth in Egypt: Past Experience, Experience of Other Countries & the Way Ahead’. Paper prepared for the Cairo University conference on Growth Strategies for Late Comers, with Special Reference to the Arab Region, Cairo, 12–14 May. Mohieldin, M., 1995. ‘Courses, Measures, and Impact of State Intervention in the Financial Sector: The Egyptian Experience’. Paper presented at a conference on The Changing Role of the State in Economic Development and Growth, Rabat, Morocco, January. Mohieldin, M., 1998. ‘Financial development in Egypt’. Paper prepared for the World Bank Resident Mission in Cairo. Nathan Associates, 1998. Enhancing Egypt’s Exports, Cairo: USAID. Nathan Associates, 1999. The Elasticities Approach to Egypt’s Balance of Payments and Equilibrium Exchange Rate, Cairo: USAID. Neisser, H. and F. Modigliani 1953. National Income and International Trade, Urbana, IL: University of Illinois Press. Nelson, J. M. (ed.), 1990. Economic Crisis and Policy Choice, Princeton, NJ: Princeton University Press. O’Brien, P. K., 1966. The Revolution in Egypt’s Economic System, Oxford: Oxford University Press. Olson, M., 1965. The Logic of Collective Action, Cambridge, MA: Harvard University Press. Olson, M., 1982. The Rise and Decline of Nations, New Haven, CT: Yale University Press. Orcutt, G. H., 1950. ‘Measurement of Price Elasticities in International Trade’, Review of Economics and Statistics, 32 (2), pp. 117–32. Pearson, D. W., 1997. ‘Trade Prospects for Egypt’, London: Institute for Middle East Studies. Prais, S. J., 1962. ‘Econometric Research in International Trade: A Review’, Kyklos, 15 (3), pp. 560–79. Przeworski, A., and F. Limongi, 1993. ‘Political Regimes and Economic Growth’, Journal of Economic Perspectives, 7, pp. 51–70.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

   , –



Radwan, S., 1997. Job Creation and Poverty Alleviation in Egypt: Strategy and Programs, Geneva: ILO. Radwan, S., 1998. ‘Towards Full Employment: Egypt into the 21st Century’. Egyptian Center for Economic Studies (ECES), Distinguished Lecture Series 10, Cairo: ECES. Radwan, S., 2002. ‘Employment and Unemployment in Egypt: Conventional Problems, Unconventional Remedies’. Egyptian Center for Economic Studies (ECES). Working Paper No. 70, Cairo: ECES. Richards. A., 1984. ‘Ten Years of Infitah: Class, Rent, and Policy Stasis in Egypt’, Journal of Development Studies, 20 (4), pp. 323–38. Richards, A., 1991. ‘The Political Economy of Dilatory Reform: Egypt in the 1980s’, World Development, 19 (12), pp. 1721–30. Richards, A., 1993. ‘Food, Jobs, and Water: Participation and Governance for a Sustainable Agriculture in Egypt’, in M. A. Faris and M. H. Khan, eds, Sustainable Agriculture in Egypt, Boulder, CO: Lynne Rienner. Robinson, J., 1937. Essays in the Theory of Employment, Oxford: Basil Blackwell. Roy, D. A., 1980. ‘The Egyptian Economy: Conversation on Social Contract, Economic Development, and Policy Alternatives’. Mimeograph. Sadowski, Y. M., 1991. Political Vegetables, Washington, DC: Brookings Institution. Soliman, S., 2011. The Autumn of Dictatorship, Palo Alto, CA: Stanford University Press. Springborg, R., 1989. Mubarak’s Egypt: Fragmentation of the Political.Order, Boulder, CO: Westview Press. Stiglitz, J. E. and S. Yusuf (eds), 2001. Rethinking the East Asia Miracle, New York: Oxford University Press for the World Bank. Subramanian, A., 1997. ‘The Egyptians Stabilization Experience’. Working Paper No. 18, Egyptian Center for Economic Studies (ECES), Cairo: ECES. Thirlwall, A. P. and H. Gibson, 1992. Balance of Payments Theory and the United Kingdom Experience, 4th edn, London: Macmillan. Thorbecke, E. and H. Wan (eds), 1999. Taiwan’s Development Experience: Lessons on role of Government and Market, Boston, MA: Kluwer Academic. Wahba, M. M., 1994. The Role of the State in the Egyptian Economy, 1945–1981, Reading: Ithaca Press. Waterbury, J., 1983. The Egypt of Nasser and Sadat: The Political Economy of Two Regimes, Princeton, NJ: Princeton University Press. Waterbury, J., 1985. ‘The “Soft State” and the Open Door: Egypt’s Experience with Economic Liberalization, 1974–1984’, Comparative Politics, 18 (1), pp. 65–83. Winters, A., 1991. International Economics, 4th edn, London: HarperCollins. World Bank, 1974. Arab Republic of Egypt: Economic Situation and Prospects, Washington, DC: World Bank. World Bank, 1978. Arab Republic of Egypt: Economic Management in a Period of Transition, 6 vols, Washington, DC: World Bank. World Bank, 1983. Egypt: Issues of Trade Strategy and Investment Planning, Washington, DC: World Bank. World Bank, 1993a. Egypt: Financial Policies for Adjustment and Growth, 3 vols, Washington, DC: World Bank. World Bank, 1993b. The East Asian Miracle, Washington, DC: World Bank. World Bank, 1997. Arab Republic of Egypt: Egypt—Issues in Sustaining Economic Growth, 4 vols, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

World Bank, 1998. Egypt in the Global Economy, Washington, DC: World Bank. World Bank, 2014a. Jobs or Privilege? Washington, DC: World Bank. World Bank, 2014b. More Jobs, Better Jobs: A Priority for Egypt, Washington, DC: World Bank. World Bank, 2015. Egypt: Promoting Poverty Reduction and Shared Prosperity. Middle East and North Africa Region, Washington, DC: World Bank. Yueh, L. Y., 2013. China’s Growth: The Making of an Economic Superpower, Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

        ......................................................................................................................

       The Real Story Beneath ......................................................................................................................

   ,  ,    

22.1 I

.................................................................................................................................. V N is an economy in transition. Central planning is gradually giving way to a more market-based economic system. The country officially embarked on this transformation in 1986 and significant development has taken place since the early 1990s. After thirty years of change, Viet Nam has achieved remarkable results in both economic growth and poverty reduction. Average annual growth rates recorded their highest levels during the first decade of reform and poverty reduction was also at its most significant during the 1990s. The average annual growth rate of this period was about 7.4 per cent and the poverty headcount ratio fell from 58.2 per cent in 1992 to 25 per cent in 2000. The second decade of reform from 2000 to 2010 saw more radical moves towards market economic principles and an opening up in terms of international trade, including membership of the WTO in 2007. Growth rates continued to be high and a further reduction in the poverty headcount ratio to about 12 per cent by 2012 was recorded. Recently, economic growth has tapered off, and the average annual growth rate was less than 6.0 per cent during 2012–15. Several studies¹ suggest that Viet Nam needs to change its economic growth model in order to develop sustainably into the future. ¹ See for example Ohno (2009); Tho (2013); and World Bank and Ministry of Planning and Investment (2016).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

This chapter aims to investigate structural change in the Vietnamese economy during the intensive reform period from 2000 to 2012. While twelve years is a relatively short period, it nevertheless represents a significant stage in Viet Nam’s modern socioeconomic history. This is an epoch in which Viet Nam experienced comprehensive structural transformation after a decade of high growth, characterized by new economic management principles and the lifting of the USA’s embargo. Specific policy moves included, first, in 2000 a major change in Viet Nam’s business environment with the introduction of an enterprise law, which eased the entrance of private firms into business. As a result, the number of firms increased markedly. Within twelve years, the number of new firms increased by more than eight times to 346,777. Second, during the first decade of the new millennium, Viet Nam took a series of bold steps in terms of international economic integration. It started with the Viet Nam–USA free trade agreement (FTA) in 2000, which had a profound impact on the economy. Viet Nam also became a member of the WTO in 2007 and committed to various FTAs. By 2009, better economic performance helped move the country from low-income to lowermidde-income status when GNI per capita increased to US$1,100 from US$980 in 2008. Against this background, we provide in this chapter a detailed analysis of structural changes in Viet Nam during 2000–12. We make use of the rich Social Accounting Matrix (SAM) data available,² and rely on analytical tools that range from input– output multipliers to decomposition of SAM multipliers over time and structural change decomposition. Going beneath existing descriptions of structural change in Viet Nam, these methods help to provide a deeper investigation into the ongoing transformation process over the last thirty years. Together with the work by Tarp et al. (2003) and McCaig and Pavcnik (2013) this helps in identifying policy implications for the next period of Viet Nam’s development. In sum, we offer insights into key features of a successful case of structural transformation in a low income and open economy from which much can be learnt for other countries in transition. The results confirm that Viet Nam has indeed experienced significant structural transformation in the economy over the last several decades. The economy has industrialized rapidly and is now much less dependent on natural resources. The analysis also shows that Viet Nam has managed to exploit abundant labour and comparative advantages in agricultural products and has benefited from international economic integration. The expansion of external demand, mainly through institutional changes associated with trade liberalization, has been a major driver of output growth.³ The country has, moreover, managed to take advantage of being a ‘latecomer’ in upgrading its production capacity through imports. Importantly, structural transformation has been broad-based, resulting in rapid poverty reduction.⁴

² This includes two detailed SAMs for Viet Nam for the years 2000 and 2012. See Jensen et al. (2004) on the former; and note that the latter was generated as part of the background analytical work for this chapter, see CIEM and UNU-WIDER (2016). Hoai et al. (2016) can be consulted for further detail and technical material underlying this chapter. ³ See also Abbott et al. (2009). ⁴ See Arndt et al. (2012) for further background.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Our analysis also suggests that it is time for Viet Nam to make concerted efforts aimed at transforming the economic growth model toward a more sustainable one which does not rely so much on expanding capital and labour. Technological upgrading has played a relatively modest role so far, and the advantages of low labour costs are running out of steam. The benefits of external demand extension can further strengthen internal economic integration and technological upgrading, but productivity dynamism is needed to sustain future growth. We conclude that Viet Nam needs to apply a more proactive and aggressive approach in terms of these two aspects to avoid being caught in a ‘middle income trap’ as suggested by Tho (2013). Section 22.2 briefly presents a few selected head-line ratios from the two SAMs used, while Sections 22.3–22.5 contain our analytical results. They include changes in industry interactions (backward and forward linkages), decomposition of SAM multipliers, and decomposition of structural change (GDP and employment). We conclude in Section 22.6 with a summary and policy recommendations.

22.2 S H- R

.................................................................................................................................. This section presents an analysis of key economy-wide ratios of Viet Nam, extracted from the 2000 and 2012 SAMs. The basic macroeconomic measures shown in Table 22.1

Table 22.1 Economy-wide economic ratios, comparing 2000 and 2012 (%) Value added/gross value of production W&S/gross value of production GOS/gross value of production GOS/value added W&S/value added Imports/total supply Domestic supply/total supply Intermediate sales/demand Household demand/demand Government expenditure/demand Investment/demand Exports/demand

2000

2012

41.4 22.8 18.6 45.0 55.0 19.4 80.6 40.1 21.1 3.5 10.1 18.5

32.7 21.5 11.2 34.3 65.7 19.8 80.2 48.3 16.5 1.6 7.2 21.3

Note: W&S is wages and salaries, GOS is gross operating surplus, value added is measured at factor costs, total supply excludes margins and taxes and demand excludes redistributed margins and change in stocks. Source: Authors’ calculations based on a 2000 and 2012 SAM; see Jensen et al. (2004), CIEM (2014), Hoai et al. (2016), and CIEM and UNU-WIDER (2016).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

reflect the overall process of structural transformation during 2000–12. Some key features include: • Viet Nam’s economy has experienced high growth during the past decade, mainly driven by the extension of production inputs. Moreover, while the economy grew at 6.8 per cent annually in the 2000s, value added as a share of gross value of production fell significantly from 41.4 per cent to 32.7 per cent. This decrease can mainly be attributed to the decline in the rewards of the production factor capital. The share of operating surplus has significantly dropped from 18.6 per cent in 2000 to 11.2 per cent in 2012 of gross value of production while the labour share remained at around 22 per cent during the period. Over the last decade, Viet Nam has benefited from low wage labour and rapid extension of capital relative to GDP. However, these advantages seem to be less obvious in the future as labour is expected to become relatively more expensive. Already, the share of wage and salaries of value added rapidly increased from 55.0 per cent in 2000 to 65.7 per cent in 2012 while the share of capital dropped from 45 per cent in 2000 to 34.3 per cent. The World Bank and Ministry of Planning and Investment (2016) found a similar trend which may imply that the allocation of capital has been suboptimal, thereby depressing returns. This could be a consequence of inefficiency due to poor performance of public investment and state-owned enterprises (SOEs), perhaps in labour intensive industries, at the cost of the return to capital; this has led to calls for stronger reform in public investment and SOEs to move towards higher levels of technology so as to sustain growth in the future. • Trade liberalization has played an important role in the transformation process. During 2000–12, Viet Nam signed two major free trade agreements: the Viet Nam– USA bilateral FTA and WTO membership. As a result, the share of exports in total demand increased from 18.5 per cent in 2000 to 21.3 per cent. Although imports have been used to extend domestic supply and increased rapidly with the openingup process, it is notable that its share in total supply remained at around 19 per cent. This suggests that, at least at the macro level, domestic production was not impacted negatively in spite of the rapid move to integrate into the global economy. This may partly be due to the fact that the share of gross domestic capital formation to GDP has remained a relatively high share of GDP (34 per cent), it is one of the main drivers of the high growth. This was helped by Viet Nam being quite active in attracting foreign direct investment (FDI). The FDI sector contributed about 14 to 16 per cent (in terms of ownership) to GDP from 2000 to 2012. Another factor is that over the period, the establishment of the oil extraction industry and petroleum refining resulted in a significant reduction in the fuel import bill. • Total demand consists of intermediate sales, domestic final demand and exports. There has been a shift towards intermediate sales and exports, away from household, government, and investment demand. This suggests that the outward orientation has been accompanied by a higher level of industrial integration. The activities which are the main beneficiaries of this macro-level change will be discussed in Section 22.3.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



In Table 22.2 we show selected activity level ratios derived from the two SAMs available to us. Continuing the trend from the 1990s, the 2000s have seen a declining share of the primary sectors, alongside a gradual expansion of medium and high technological industries and service sectors. As such, Table 22.2 confirms that major shifts have indeed taken place in the structure of the economy. This includes the decline of agriculture as a contributor to GDP, down to 16 per cent from 27 per cent, and of mining down, from 10.5 per cent to 7.3 per cent. Labour-intensive industries (textiles, clothing, and leather) improved relatively, with the share of this sector increasing from 3 per cent to 7.5 per cent between 2000 and 2012. Food processing and beverage industries do not show improvement but other, more capital intensive industries (metal products and electrical machineries)

Table 22.2 Activity real growth and nominal shares in GDP (at factor costs), comparing 2000 and 2012 Share in GDP (%)

Share in employment (%)

Share of exports (%)

Share of imports (%)

Total

2000

2012

2000

2012

2000

2012

2000

2012

Agriculture Mining Food & beverage Text, cloth, & leath Oth manuf Petrol ref & oils Chemicals Plast & rubb prods N-met min prods Metal prods Gen machinery El mach & appl Transprt equip Utilities Construction Trade Acc, rest, & touris Transport Comm srv & publ Fin & bus serv Health Government Education Other services

26.7 10.5 5.5 3.0 1.8 0.1 1.5 1.0 1.7 0.8 0.3 0.8 1.5 3.3 5.6 10.4 3.4 2.1 2.0 6.0 1.5 3.0 3.7 3.9

16.0 7.3 5.3 7.5 3.0 0.8 1.5 1.2 1.6 2.5 0.8 3.0 1.1 3.0 6.2 11.0 3.1 4.1 1.5 9.5 1.3 2.9 4.0 1.6

64.4 0.8 1.5 3.6 2.0 0.01 0.2 0.2 0.6 0.7 0.1 0.1 0.2 0.3 3.1 8.3 2.7 1.9 1.3 0.7 1.0 2.1 2.0 2.1

47.4 0.6 2.1 5.0 3.0 0.02 0.3 0.3 0.9 1.0 0.1 0.4 0.5 0.5 6.4 12.3 4.2 2.9 0.6 1.8 0.9 3.1 3.4 2.3

11.1 21.3 15.4 22.3 5.6 0.6 0.6 0.4 0.4 1.0 3.4 3.3 0.6 0.0 0.0 0.2 5.7 2.9 1.0 2.3 0.3 0.0 0.2 1.4

6.4 10.1 17.8 20.1 6.4 0.1 1.6 3.5 2.2 4.5 1.2 17.2 1.1 0.0 0.0 0.0 4.1 2.1 0.1 0.8 0.2 0.0 0.3 0.2

1.5 0.9 3.4 13.1 7.5 11.6 11.0 4.9 1.8 7.7 10.1 8.5 7.8 0.2 0.0 0.0 1.9 1.7 0.2 3.2 0.2 0.0 2.0 0.8

8.3 3.1 5.6 8.4 5.6 10.8 7.8 5.5 0.7 13.7 7.3 14.8 1.8 0.1 0.0 0.0 1.0 0.7 0.3 2.9 0.4 0.0 0.9 0.2

Note: For calculating GDP growth, it has been deflated using deflators. The shares are calculated in current prices at factor costs. Source: See Table 22.1.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

as well as services (transportation and financial and business services), saw their GDP shares increase. The shift in the employment share of the primary sector mirrors those of sectoral GDP shares. However, services, in particular trade and construction, absorbed relatively more labour than manufacturing. Employment in the service sector increased from about 25 per cent in 2000 to 38 per cent in 2012 while that of manufacturing increased from 10 per cent to 14 per cent. Exports have also shifted away from mining and agriculture, in particular towards light manufacturing and other sectors such as metal products and machinery and transport equipment as well as a number of services. While significant increases can be noted for metal products, electrical machinery, and transport equipment (the last from a low base), the general machinery share has dropped. It is notable that while the share of agricultural exports decreased, its share of imports increased rapidly from 1.5 per cent to 8.3 per cent. This may reflect changing household demand patterns towards food products that are not yet fully supported by local agricultural supplies, such as dairy. The reverse applies to transport equipment and chemicals, the latter perhaps due to the establishment of the petroleum industry. On the other hand, electrical machinery made a jump in its share of exports and its share of imports also increased significantly. This reflects that foreign direct investment has poured into this industry over the period, mainly in the product assembly stage, to take advantage of low labour costs.

22.3 B  F L M

.................................................................................................................................. This section investigates in more detail the ongoing changes in internal interactions among the industries of the economy. We explore both backward and forward linkage multipliers as described and explained in more detail in Hoai et al. (2016). These linkage concepts focus on interactions amongst activities, ignoring household income and expenditure loops. In Figure 22.1 the backward linkage multipliers are ranked for 2000 and then the difference between 2000 and 2012 is added. It can be seen that, on average, the economy has become much more connected in terms of backward linkages, by about 20 per cent. The activities with the highest increases in their connectivity are: petrol refinery, and oils; communication services and publishing; agriculture; transport equipment; transport services; food and beverages; and chemicals. At the end of the period, some of the highest backward linkage multipliers are recorded for petroleum—due to its recent establishment—but also transport equipment and communication services as well as food and beverages. The latter can be explained by firms in this industry taking advantage of agriculture’s progress. During the 2000s,

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



2.5 2.0 1.5 1.0 0.5 0.0

fo o te d & xt , c be n- lot ver m h, ag et & e m le a in th ot pro h ds m m et an co al p uf el nst rod tra ma ruc s n ch tio ac spo & a n c r p pl , re t eq pl as st, ui t & & pm ru tou bb ri pr s ge o n av ds m er ac ag h e ch ine r e go mi y ve ca r n ls m en fin & tra t bu de co a s s m gr er m ic v sr ulu v& r e p m ubl pe tr inin tro an g l r spo ef r t ed & o u c i ls at io ot he he n r s alt er h vi ut ces ili t ie s

–0.5

Backward multiplier 2000

Backward multiplier change

 . Backward linkage multipliers for 2000 and the change between 2000 and 2012 Source: see Table 22.1.

Viet Nam was ranked second in the world market for exports of rice, coffee, and cashew nuts, that is, processed foods which contribute to higher integration and multipliers. Viet Nam also managed to keep protecting its domestic market for transport equipment by maintaining high import tariffs. Although these tariffs are set to reduce in future under WTO and other FTA arrangements, the policy attracted many foreign firms, in particular from Japan, investing in the transport equipment industry. We can also see that agriculture has become more connected to the rest of the economy over time in terms of backward linkages. This might be related to switching to ‘home grown’ inputs from chemicals and indirectly, petroleum refining and oil exploitation, which were previously imported. Also, foreign direct and domestic investment increased in the sectors that provided inputs to agriculture such as animal feeds. Both plastic and rubber product producers and chemicals show improved backward linkages. Textile, footwear, and leather do not seem to have made much gain in this regard, but hold on to their relatively high level. Figure 22.2 shows that, on average, the forward linkage multiplier has not increased as much as the backward linkage multiplier. Plastic products, metal products, and non-metallic mineral products producers had the highest forward linkages in 2000. By 2012 this had changed, with petroleum and chemicals now taking the lead on the back of the establishment of a petroleum refinery industry. The economy also appears to have become more energy intensive, as one would expect, with the forward linkages of utilities increasing. Other main increases in forward linkages are recorded for mining (linked to crude oil exploration) and agriculture. The latter ties in with shifts in demand towards more processed foods. On the other hand, Hoai et al. (2016) had noted a shift away

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,     3.0 2.5 2.0 1.5 1.0 0.5 0.0 –0.5

pl as t&

ru b n- m b pr m eta od et l s pe mi pro tro n p ds l r ro ef ds & ch oi e m ls ica co o util ls m th iti m m es el srv anu m & f a p fin ch ub t ra & & l ns b app po us l rt se e r ag qui v ric pm te u xt , c a ltur lo ve e ge th, rag n & e m le ac at hi h fo od m ner & in y be in v g ac c, tr era re an ge st, sp & or t t ed our uc is at io ot he he n r s alt er h vi c go tr es ve ad co r n e ns m t r u en ct t io n

–1.0

Forward multiplier 2000

Forward multiplier change

 . Forward linkage multipliers for 2000 and the change between 2000 and 2012 Source: see Table 22.1.

from intermediate sales for transport equipment, communication services, electrical machinery, and non-metallic minerals products producers to final demand categories such as household expenditure, investment demand and exports. This may explain that their forward linkage multipliers have declined.

22.4 D  SAM M

.................................................................................................................................. A decomposition of SAM multipliers has been undertaken previously for Viet Nam by Roland-Holst and Tarp (2003). We follow a similar approach, albeit at a different level of detail, one that is common across the two years for which SAMs at our disposal, allowing for an intertemporal comparison. SAM multiplier decomposition focuses on interactions (in terms of monetary flow) within and amongst activities/commodities, factors of production, and institutions. In order to do so effectively, we expand the SAMs that were used for the linkage multiplier analysis in Section 22.3 with a disaggregation of the production factor labour and of households into a rural and an urban category.⁵ ⁵ Further disaggregation into common categories for 2000 and 2012 is either not possible or not convenient. For details, see Hoai et al. (2016).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



We employ an additive SAM multiplier decomposition method to investigate changes in the structure of the economy between 2000 and 2012.⁶ The main components of the decomposition are blocks covering: • twenty-four activities and twenty-four commodities; • incomes generated by three factors (urban and rural labour and capital); • factor incomes distributed to institutions (urban and rural households and enterprises); • household expenditures; and • transfers to other institutions. The decomposition identifies: • ‘own’ multiplier effects in and amongst activities/commodities and amongst institutions; • ‘spill-over’ effects, for example from activities on households via the factor payments and the other way around; and • ‘feedback’ effects which may occur when activities pay factors, followed by the distribution of factor incomes to households and their spending on commodities result in higher production of activities. Among the three sets of effects, our results reveal that the ‘own’ multiplier effect is greater than the two others (Figures 22.3 and 22.4). The ‘own’ multiplier effect is very similar to the backward linkage multipliers discussed in Section 22.3 and their analysis is therefore not repeated here. The spill-over of production activities on urban and rural household income are shown in Figure 22.3. It is notable that rural households have, on average, benefited more from production expansion. The average spill-over multiplier for rural households was about 0.42 versus the one for urban which was 0.33. This finding is in line with the results of a study done by Arndt et al. (2012) using a 2003 Viet Nam SAM. This also helps to explain the success of Viet Nam in reducing poverty, whereas a majority of the poor remain in rural areas. It should, however, be noted that the average spill-over multiplier tends to decrease over time for both rural and urban households, indicating that reducing poverty further is becoming more difficult. Figure 22.3 also demonstrates that spill-overs mainly emanate from services, with public sector services in particular becoming more dominant. Services are in general characterized by relatively high labour intensity while new minimum wage legislation that was enacted during the period of observations also contributed in this regard. Textiles, clothing, and leather is the only manufacturing activity that has improved its impact on urban household income over the period of observation. In general, the growth

⁶ See Hoai et al. (2016) for more detail.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,     Urban

–0.2 –0.1 other services education fin & bus serv government health comm srv & publ acc, rest, & touris utilities mining trade transport average n-met min prods food & beverage plast & rubb prods chemicals oth manuf text, cloth, & leath agriculture construction el mach & appl gen machinery tranport equipm petrol ref & oils metal prods

0.0

0.1

0.2

Rural 0.3

0.4

0.5

0.6

–0.4 agriculture education government health food & beverage mining other services comm srv & publ acc, rest, & touris utilities average oth manuf trade fin & bus serv plast & rubb prods chemicals n-met min prods construction transport tranport equipm gen machinery petrol ref & oils text, cloth, & leath metal prods el mach & appl

–0.2

0.0

0.2

0.4

0.6

2000 Spillover on HH including urban

2000 Spillover on HH including rural

2000–12 Spillover on HH including urban

2000–12 Spillover on HH including rural

0.8

 . Spill-over effects from activities on household income for 2000 and the change between 2000 and 2012 Source: see Table 22.1.

of three industries, textiles, clothing, and leather; construction; and trade, created the largest increases in spill-overs to household income in both rural and urban areas. Trade’s improved impact on rural households can be explained by the geographic spread of these activities. Agriculture and food processing remain important contributors to rural household income with a less important role for urban households. Figure 22.4 shows household feedback effects on activities. The results suggest that rural household incomes have a relatively higher spill-over impact on activities as they tend to buy more locally (directly and indirectly). However, urban households were catching up over the period studied. Overall, there is a modest increase for urban households but rural households offer less benefit to production activities in 2012 relative to 2000. Agriculture; food and beverages; textiles, clothing, and leather; and trade services remain the main beneficiary activities of the spill-over effects from both household groups. But for all the activities mentioned, spill-over has become less intense. As expected, urban household incomes have a broader spread of spill-over effects and this is intensifying more than for rural households. The main activities that benefit from household incomes more over the period are financial and business services; transport; communication services; electrical machinery (which here includes appliances); and petroleum. The last is evidence of this product now being available locally, more so than in 2000. In general, the higher economic integration (multipliers) noted in Section 22.3 helps to indirectly enhance feedback effects.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

       Urban –0.1 agriculture food & beverage trade text, cloth, & leath acc, rest, & touris fin & bus serv average other services utilities oth manuf transport equipm education chemicals transport comm srv & publ metal prods health plast & rubb prods el mach & appl n-met min prods mining gen machinery petrol ref & oils government construction

0.0

0.1

0.2



Rural 0.3

0.4

0.5

–0.3 –0.2 –0.1 0.0 0.1 agriculture food & beverage trade text, cloth, & leath average other services fin & bus serv chemicals oth manuf acc, rest, & touris transport equipm utilities education transport health metal prods plast & rubb prods n-met min prods comm srv & publ mining el mach & appl gen machinery petrol ref & oils government construction

0.2 0.3

0.4 0.5 0.6

2000 Spillover of HH including urban

2000 Spillover of HH including rural

2000–12 Spillover of HH including urban

2000–12 Spillover of HH including rural

0.7

 . Feedback effects from households on activities for 2000 and the change between 2000 and 2012 Source: Table 22.1.

22.5 D  S C

.................................................................................................................................. In addition to analysing and comparing ratios and multipliers, a more direct way of investigating structural transformation and change is to break the total change in output down into contributions by various components. The most obvious way is to disaggregate total change in output by change in final demand and a change in technology. The latter is interpreted here as more inter-industry interaction amongst activities. To undertake a decomposition of structural change with the two SAMs it is, however, necessary that they are valued at the same prices. We selected to inflate the 2000 SAM to 2012 prices, using a number of published deflators, including the GDP Deflator, PPI, Import Price Index, Export Price Index and the CPI, as discussed in Hoai et al. (2016).

22.5.1 Decomposition of Change in GDP (Value Added) While standard decomposition analysis is initially conducted in terms of gross output (see Hoai et al. 2016), policy makers may be more interested in changes in value added.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

We therefore extend the decomposition of the total change in the gross value of production of the changes in technology and final demand into a three-way decomposition. In doing so, the total change in value added can be attributed to activities adding more (or less) value per unit of gross output, technology effects, and final demand effects, as shown in the following equation: 1 Δv ¼ Δˆv ðL2000 f2000 þ L2012 f2012 Þ 2|fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Value Added Intensity Effect

1 þ ½ˆv 2000 ΔLf2012 þ vˆ 2012 ΔLf2000  2 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl ffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Technology Effect

1 þ Δf ðˆv 2000 L2000 þ vˆ 2012 L2012 Þ 2 |fflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl{zfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflfflffl} Final Demand Effect

in which v is a vector of value added/output ratios and ^v is a diagonal matrix with the elements of v on the main diagonal. L stands for the Leontief inverse, capturing technology. f is a n  1 column vector of final demand aggregated across the relevant components (household expenditure, government expenditure, investment demand, and exports). The results are shown in Table 22.3. The most striking conclusion at the economy-wide level is that final demand is driving the increase in GDP with negative contributions from the technology effect and the value added intensity effect. So, while GDP more than doubled in real terms from 2000 to 2012, the contribution by activities themselves adding value is negative while the interaction amongst local industries also had an adverse impact. A repeat of this decomposition for wage earnings and gross operating surplus separately (but not shown here) suggests that the negative value added intensity effect is mainly due to the latter. This is consistent with the lower economy-wide share of gross operating surplus in GDP when comparing 2000 to 2012 in the two SAMs as discussed in Section 22.2 above. The impact of adding value has been negative across a wide range of activities in particular for agriculture, chemicals, transport equipment, accommodation, transport, health, and communications services. It is possible that these activities have faced increased competition and as a result, margins have been squeezed.

22.5.2 Decomposition of Change in Employment It is also possible to decompose the change in employment between 2000 and 2012. In order to do so, we replace the value added/output ratios v and ^v in Section 22.5.1 with employment/output ratios e and ^e respectively. The results are shown in Table 22.4. Unlike with value added in Table 22.3, we now report changes relative to the initial levels of employment. The first observation to make here is that total employment has increased by 36.8 per cent over the period as a whole. Growth in employment is entirely driven by changes in final demand, as can be seen in the last entry of the first row

Table 22.3 Decomposition of change in value added between 2000 and 2012 due to changes in value added intensity, changes in technology, and changes in final demand Total

Source: see Table 22.1.

va coef eff

tech eff

fd eff

ini va level

ch in va

va coef eff

tech eff

fd eff

1,496,559

640,943

203,518

2,341,020

1,447,411

100.0

42.8

13.6

156.4

% of Δva

% of Δva

% of Δva

% of Δva

100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

136.5 227.0 71.6 23.4 15.3 126.7 67.0 58.6 20.4 13.8 58.3 36.0 240.5 24.5 9.3 23.8 58.8 30.5 123.0 4.2 98.0 53.5 18.8 162.4

27.1 161.5 64.4 3.9 9.6 193.0 88.3 32.8 20.2 50.6 84.8 24.5 109.1 11.7 10.8 103.6 82.3 37.0 36.0 23.4 180.1 90.1 31.2 452.9

209.4 288.5 107.1 80.5 105.7 33.7 78.6 125.8 99.8 63.1 73.5 111.6 231.5 112.8 79.8 179.8 241.1 93.6 259.0 127.6 378.1 136.6 112.4 515.3

149,956 92,476 87,636 185,185 70,172 24,422 30,721 26,287 31,087 65,542 20,509 80,542 18,016 52,220 106,127 162,595 41,391 93,256 23,794 201,400 12,467 41,003 71,228 6,521

204,721 209,946 62,715 43,295 10,720 30,935 20,572 15,405 6,345 9,021 11,953 29,021 43,333 12,789 9,893 38,688 24,339 28,465 29,263 8,476 12,218 21,949 13,368 10,589

40,626 149,330 56,481 7,231 6,711 47,123 27,133 8,610 6,287 33,189 17,393 19,697 19,649 6,092 11,495 168,409 34,084 34,470 8,571 47,130 22,452 36,945 22,215 29,533

314,051 266,800 93,870 149,121 74,181 8,234 24,160 33,082 31,029 41,374 15,069 89,865 41,700 58,918 84,739 292,315 99,814 87,252 61,628 257,005 47,137 55,999 80,075 33,601

315,620 306,438 67,259 33,465 18,900 592 14,618 9,809 18,029 8,014 3,538 8,221 14,961 36,666 74,356 156,447 51,827 27,530 20,640 93,993 24,641 43,256 43,544 55,048

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Agriculture Mining Food & beverage Text, cloth, & leath Oth manuf Petrol ref & oils Chemicals Plast & rubb prods N-met min prods Metal prods Gen machinery El mach & appl Transprt equipm Utilities Construction Trade Acc, rest, & touris Transport Comm srv & publ Fin & bus serv Health Government Education Other services

ch in va

Total

Agriculture mining Food & beverage Text, cloth, & leath Oth manuf Petrol ref & oils Chemicals Plast & rubb prods Nmet min prods Metal prods Gen machinery El mach & appl Transprt equipm Utilities Construction Trade Acc, rest & touris Transport Comm srv & publ Fin & bus serv Health Government Education other services Source: see Table 22.1.

ch in emp

emp inp coeff eff

tech eff

fd eff

ini emp level

13,822

32,466

185

46,473

37,600

157 15 534 1,237 779 5 63 94 234 258 30 152 217 133 2,092 3,184 1,123 769 189 676 89 784 1,031 385

23,740 143 641 2,078 1,660 319 196 135 44 1,365 270 399 97 46 453 720 256 1,743 1,119 50 276 429 100 309

2,948 154 454 241 241 318 152 51 77 935 194 123 85 17 190 3,360 687 821 171 135 347 685 366 491

20,949 282 721 3,555 2,197 6 107 177 355 688 106 428 229 163 1,449 5,825 2,067 1,691 1,101 761 711 1,040 1,297 566

24,200 300 557 1,351 770 4 88 61 238 270 44 55 62 104 1,179 3,130 1,014 729 473 263 394 798 737 780

ch in emp

emp inp coeff eff

tech eff

fd eff

36.8

86.3

0.5

123.6

% of ini emp

% of ini emp

% of ini emp

% of ini emp

0.6 4.9 95.9 91.5 101.1 121.7 71.2 154.4 98.4 95.5 67.5 276.4 352.2 128.0 177.5 101.8 110.7 105.5 40.0 257.3 22.6 98.2 139.9 49.3

98.1 47.6 115.0 153.8 215.6 7,243.3 222.9 222.7 18.3 505.9 613.6 724.8 157.9 44.5 38.4 23.0 25.3 239.0 236.8 19.1 70.0 53.7 13.5 39.6

12.2 51.4 81.4 17.8 31.3 7,218.2 172.1 84.9 32.5 346.4 440.0 223.9 138.4 16.3 16.1 107.4 67.8 112.6 36.1 51.3 88.1 85.8 49.7 62.9

86.6 94.1 129.4 263.1 285.4 146.7 122.0 292.2 149.3 255.0 241.0 777.2 371.7 156.2 122.9 186.1 203.8 231.9 232.9 289.5 180.6 130.3 176.1 72.6

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Table 22.4 Decomposition of change in employment between 2000 and 2012 due to changes in employment intensity, changes in technology, and changes in final demand

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



(123.6 per cent). The labour productivity effect can be seen to have reduced the total demand for labour over this period by 86.3 per cent with a slight decrease in the use of labour due to technology (in terms of inter-industry interaction) shifting toward industries that require more capital. At the detailed level there are only a few exceptions to the negative economy-wide employment input effect. These are construction and a few service sectors. These industries can be characterized as relatively high labour intensive with some having relatively limited scope to increase labour productivity. The technology effect has been beneficial to demand for labour for most manufacturing activities, except for textiles, clothing, and footwear and non-metallic minerals. Overall, this suggests that the increased integration of the Viet Nam economy that was noted Section 22.3 has contributed to higher labour absorption in most manufacturing. The large increase in the technology effect of petroleum refinery activity is due to its establishment during the period of observation (2009), but this is in stark contrast to its employment intensity effect.

22.6 S  P I

.................................................................................................................................. We have employed several SAM based models to explore structural transformation nd change in Viet Nam with a focus on the period 2000–12. We built on earlier work to provide a longer term perspective, being cognizant of the fact that the Vietnamese economy is a fast moving entity. A series of significant changes have taken place in recent years. Our first finding is that the economy of Viet Nam is rapidly moving away from its dependence on natural resources, mining, agriculture, and food processing towards more intricately involved production activities. This is not to say that agriculture and food related industries have not grown as a whole. Together, they have grown at about 2.5 per cent per annum in real terms. It is, however, clear that other industries are rapidly taking over and shifting the Vietnamese economy in a more diversified direction. While this shift is spearheaded by textiles, clothing, leather, and footwear, it is also clear that the next generation of more capital intensive activities seems to lie in the metal products and general and electrical machinery. This confirms a continuous transformation of the economy during the 2000s as compared to the 1990s, see for example McCaig and Pavcnik (2013). Surprisingly, the transport equipment industry is not yet prominent in the newly emerging patterns of production, although transport services are, as well as financial and business services. With the decline of agriculture, further urbanization is boosting construction and non-metallic minerals industries. In this regard, Viet Nam has pursued a transformation process that is very much in the expected direction, see Tarp et al. (2003) and Tarp (2017).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

Given the patterns of industry growth, it is also no surprise that our next level of analysis, which employed various types of multipliers, showed a more integrated economy at the end of the period of observation. By making a distinction between backward and forward linkages, we found that this phase of development is characterized by a bias towards backward linkages, with further room for the development of forward linkages. This suggests that there should now be opportunities to develop the extent to which industries sell intermediate inputs locally. Further analysis of trade data may be required to determine whether this can be linked to current regional crossborder value chain patterns in which Viet Nam appears to remain in its position of supplying intermediate level goods. How to add more value to those supplies is another area in need of further investigation. We furthermore found that Viet Nam could benefit more from international integration if the linkages of some export-oriented products improved such as textiles, clothing, and footwear and electrical machinery and appliances. In particular, special attention should be paid to developing the food processing industry. It has the highest backward linkage in the economy, while its share in GDP has slightly reduced. Based on our SAM decomposition analysis, it emerges that urbanization itself may be contributing to the relative (but not absolute) decline of agriculture. Demand patterns are shifting away from food towards higher value goods, as suggested by the changing spill-over effects on Viet Nam’s production activities. The higher benefits received by rural households from the growth process have contributed to the success of Viet Nam in poverty reduction over the past decade. This channel of influence may, however, become less important in the times ahead and more effort by means of specific policies will therefore be needed to further reduce poverty. The lowest hanging fruits have been picked, so to speak. Our decomposition of structural change took the analysis a step further and confirmed that changes in GDP as well and employment were mainly driven by final demand as foreseen in a previous study by Tarp et al. (2002). The change in technology (interpreted here as the change in industry interaction) made only a minor contribution. This implies that integrating into the international economy has been a key driver in Viet Nam’s growth over the period 2000–12. It would appear that to sustain growth in the future, new policy measures must be pursued. The study by the World Bank and Ministry of Planning and Investment of Viet Nam (2016) suggests six groups of measures, ranging from private sector development to innovation and institutional development, each of which can be considered as potentially contributing to strengthening further domestic integration. Where technology effects did make a substantial contribution it included activities such as food and beverages; petroleum; chemicals; metal products; general machinery; transport equipment and transport, although for all these activities final demand remains the main source. Thus, for a number of important manufacturing sub-sectors, inter-industry interaction has contributed positively to the change in gross output between 2000 and 2012, confirming the higher backward linkage multipliers for these activities.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



The decomposition of change in GDP highlighted the rather counter-intuitive observation that the contribution by the component that captures value added per unit of output actually made a negative contribution to GDP growth for most activities. Nevertheless, this echoed the initial observation that the share of value added in gross value of production had declined over the period of observation, while the share of wages and salaries in value added itself had increased. Both shifts occurred at the cost of the role of the return to capital. It reflects the rapid expansion of the capital stock before and during the period of observation, implying that the returns to capital may have declined substantially. The same trend has been found in the study by the World Bank and Ministry of Planning and Investment of Viet Nam (2016). Possibly, diminishing returns have been at work, likely combined with inefficiency due to poor performance of public investment and SOEs. This calls for stronger reform of public investment and SOEs in Viet Nam in order to sustain growth in the future. Tho (2013) argues that without radical reform in this respect, Viet Nam will fall into a middle income trap. We also showed that while GDP has increased in real terms over the period of observation, the contribution by activities themselves adding value, is negative while the interaction amongst local industries has also had an adverse impact. The negative impact of adding value occurred for a wide range of activities, in particular for agriculture; chemicals; transport equipment; accommodation; and transport, health, and communications services. This possibly also shows that these activities have faced increased competition and as a result, margins have been squeezed, which suggests the need to transform further to sustain growth in the future. The lack of technological upgrading was also highlighted by the decomposition of structural change in employment. While growth in employment is entirely driven by changes in final demand, the labour productivity effect was shown to have reduced the total demand for labour with a slight decrease in the use of labour due to interindustry interaction, shifting demand towards industries that require less labour inputs. Notable and encouraging exceptions to this general observation are found in a number of newly emerging manufacturing industries. In order to upgrade the technology, improvement in human capital is a necessary condition. We conclude by noting that the Viet Nam case of structural transformation provides illuminating insights into how success can be pursued under globalization. Three key points stand out: • Structural transformation is a continuous process of movement from low to higher levels of development. Viet Nam managed this process well during the 1990s and 2000s. The economic structure was gradually transformed from being dependent on agriculture and mining towards labour intensive and subsequently more capital intensive manufacturing and services. • The benefits of extending external demand helps most of the domestic economy if internal economic integration is being developed. Viet Nam managed quite well in this regard. However, areas calling for improvement include the internal linkages relating to industries with comparative advantage such as textiles and clothing, and food processing.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



   ,  ,    

• Developing labour-intensive industries during the transformation process, especially at an early stage of transformation, can play a decisive role in achieving broad-based structural transformation. Finally, the various decompositions of structural change carried out in this chapter suggest that it is time for Viet Nam to start considering changing its economic growth model, introducing technological upgrading more deliberately. The advantage of low labour costs is starting to taper off and more and targeted technology upgrading is required. While considerable success has been achieved over the past quarter century, Viet Nam needs to start taking a more proactive and aggressive approach in the area of technology.

R Abbott, P., J. Bentzen, and F. Tarp, 2009. ‘Trade and Development: Lessons from Vietnam’s Part Trade Agreements’, World Development, 37 (2), pp. 341–53. Arndt, C., A. Garcia, F. Tarp, and J. Thurlow, 2012. ‘Poverty Reduction and Economic Structure: Comparative Path Analysis for Mozambique and Vietnam’, Review of Income and Wealth, 58 (4), December. CIEM, 2014. A Viet Nam Social Accounting Matrix for the Year 2011, Hanoi: CIEM. CIEM and UNU-WIDER, 2016. A 2012 Social Accounting Matrix (SAM) for Viet Nam, Hanoi: CIEM and Helsinki: UNU-WIDER. Hoai, D. Thi Thu, F. Tarp, D. van Seventer, and H. Cong Hoa, 2016. ‘Growth and Structural Transformation in Viet Nam during the 2000s’. 2016/108, Helsinki: UNU-WIDER. Jensen, H.T., J. Rand, and F. Tarp, 2004. A New Viet Nam Social Accounting Matrix for the Year 2000, Hanoi: Science and Technics Publishing House. McCaig, B. and N. Pavcnik, 2013. ‘Moving out of Agriculture: Structural Change in Vietnam’. NBER Working Paper No. 19616, Cambridge, MA: NBER. Ohno, K., 2009. ‘Avoiding the Middle Income Trap: Renovating Industrial Policy Formulation in Vietnam’, ASEAN Economic Bulletin, 26 (1). Roland-Holst, D. and F. Tarp, 2003. ‘Globalization, Economic Reform, and Structural Price Transmission: SAM Decomposition Techniques with an Empirical Application to Vietnam’. MPRA Paper No. 29367, Munich: University Library of Munich. Tarp, F., 2017. Growth, Structural Transformation, and Rural Change in Viet Nam, Oxford: Oxford University Press. Tarp, F., D. Roland-Holst, and J. Rand, 2002. ‘Trade and Income Growth in Vietnam: Estimates from a New Social Accounting Matrix’. MPRA Paper No. 29395, Munich: University Library of Munich. Tarp, F., D. Roland-Holst, and J. Rand, 2003. ‘Economic Structure and Development in an Emergent Asian Economy: Evidence from a Social Accounting Matrix for Vietnam’. Journal of Asian Economics, 13 (January), pp. 847–71. Tho, T. V., 2013. ‘The Middle-Income Trap: Issues for Members of the Association of Southeast Asian Nations’. ADBI Working Paper No. 421, Tokyo: Asian Development Bank. World Bank and Ministry of Planning and Investment of Viet Nam, 2016. Viet Nam 2035: Toward Prosperity, Creativity, Equity, and Democracy: Overview, Washington, DC: World Bank and Ministry of Planning and Investment of Viet Nam.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

           The Chinese Experience ......................................................................................................................

 

M transition, like economic development, is a long-term process that is intertwined with structural change. In economic development, restructuring is driven primarily by the processes of urbanization and industrialization; as a result, many of the structural regularities can be summarized by linking them to GDP per capita. The logic of market transition is different. By definition, market transition involves giving planners less control over the output mix, and allowing domestic and international markets to play a determinative role. In all command economies, ‘planner sovereignty’ involved a focus on strategic heavy industries, including steel and machinery. As a result, all transition economies in their initial stages had to downsize over-developed heavy industrial sectors, and expand labour-intensive, consumer-oriented light manufacturing. However, since different economies entered the market transition process at very different levels of GDP per capita, the associated structural changes played out in very different ways. This chapter examines China’s experience with structural change during transition. It looks at structural change during three distinct periods of economic reform and growth.¹ In each case, the pattern of structural change is shown to be related to the choices policy makers made with respect to reform and market transition. Each period yields specific ‘lessons’ about the relationship between policy and structural change. The final section pulls these lessons together and suggests some generalizations that apply to China’s experience as a whole. ¹ These periods roughly correspond to the periods of ‘reform strategy’ laid out in Naughton (2007). Specific dates have been adapted in order to utilize the successive Industrial and Economic Censuses, which provide detailed structural data for 1980, 1985, 1995, 2004, 2008, and 2013. In order to correspond with the detailed data source, the first period has been adjusted to cover 1980–95, and the second to cover 1995 through 2013. Annual data releases omit the important very small-scale sector (and do so in inconsistent ways). Therefore, systematic structural analysis must depend on the periodic census.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

In the first period, 1978 through 1995, the predominant form of structural change in industry was the movement towards a more labour-intensive output basket. This structure was more in line with China’s underlying comparative advantage, but in traditional developmental terms involved a step ‘backward’ from the over-developed heavy industries China had built during the command economy period. During the second period, from 1995 through to about 2010 (industrial census data from 2008 and 2013 are deployed), China resumed a broad-based industrial advance, which displayed elements of a classical pattern of middle-income structural change, moving ‘upstream’ toward more capital and skill-intensive industries. Finally, since 2010, China has begun to move toward a service economy. The share of industry in GDP is still extremely high, and movement has been gradual, but it is clear that a new era of structural change has begun. In each of these periods, a clear relationship exists between policies adopted to foster market transition and specific structural change outcomes. However, the specific structural changes were rarely successfully targeted by policy makers: instead, they developed as unanticipated outcomes of other market-oriented policies.

23.1 R   L- G P: 1980–1995

.................................................................................................................................. During the period 1978 through to 1995, there was an unusually close correspondence between the strategy of economic reform and the corresponding structural change. This correspondence has been noted in the literature, by authors approaching the issues from different standpoints. Naughton (1995) described the reform strategy of ‘growing out of the plan’, which froze the existing state-owned command economy and allowed new entries driven by market profitability. Clearly, this approach had structural implications, since profit rates were systematically higher and entry barriers were systematically lower in labour-intensive sectors. Lin et al. (1996) described the phenomenon directly from the standpoint of structural change. They showed that the inappropriately capital-intensive industrialization strategy of the command economy period was, with reform, yielding to a more balanced growth strategy. The shift toward more labour-intensive production was in line with China’s factor endowments at that time, and the economy derived important productivity and growth advantages.

23.1.1 Structural Change in Industry, 1980–1995 It should be emphasized how closely integrated these approaches were in practice. Figure 23.1 shows that structural change during the 1980 to 1995 period was identical with change in the ownership and size composition of industry. Panel

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



A of Figure 23.1 shows that in 1980, Chinese industry consisted mostly of stateowned firms operating in relatively capital-intensive sectors. Moreover, the truly small-scale sector was relatively unimportant. Indeed, since Chinese ‘large-scale’ firms at this time were certainly not large scale by international standards, it would be more accurate to say that China’s industrial landscape was dominated by Panel A: 1980 Composition of industrial output Small scale, village & below (Foreigninvested) Small SOEs & city, township collectives

Large SOEs

Monopoly sectors

Scale-intensive

Ordinary manufacturing

Light, labourintensive manufacturing

Ordinary manufacturing

Light, labourintensive manufacturing

Panel B: 1995 Composition of industrial output Small scale: village, co-op, & private Foreigninvested

Collectives

SOEs

Monopoly sectors

Scaleintensive

 . Change in industrial composition, 1980–95

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

medium-size single plant facilities, nearly all of which were publicly owned. Of course, foreign-invested firms played no role in the industrial economy. By 1995, as Panel B of Figure 23.1 shows, this picture had changed completely. The entry of small-scale firms into primarily labour-intensive sectors (the upper right corner of Panel B) had transformed the industrial sector. The sectoral composition of the state sector had changed very little, but new-entrant firms had a completely different sectoral composition. Entry was the means through which structural change was realized. Indeed, Figure 23.1 is constructed by using the share of output in each industrial sector that was produced by state firms in 1995. Monopoly sectors were more than 75 per cent produced by SOEs: these include only resource extraction, utilities, and tobacco. Scale-intensive sectors include those industrial sectors for which SOEs accounted for between 40 per cent and 62 per cent of output. They include metals, chemicals, and machinery, plus pharmaceuticals and synthetic fibres, as well as coal mining, food processing, and beverages. Coal mining was an exception to the general rule that the state should own mineral resources: in practice, small-scale mining was allowed to grow rapidly during the 1980s. Food processing and beverages were sectors in which state control over procurements gave state firms a competitive advantage well into the 1990s. ‘Labour-intensive manufacturing’ consists of all sectors in which the state share was under 26 per cent. It includes all the standard labour-intensive sectors (garments, toys, etc.) but also stone mining and lime kilns and building materials production, which, under Chinese conditions at that time, were often produced in labour-intensive processes using limited capital. ‘Ordinary manufacturing’ consists of those sectors in which the state share was between 26 per cent and 36 per cent of total output, including textiles, instruments, electric machinery, and electronics. This observation means that China was in fact deriving three simultaneous types of productivity advantage during the 1980s and early 1990s from the entry process: (i) Private ownership—easily exercised in the relatively small-scale firms that were driving change—gave improved incentives to manager-owners. (ii) The entry of smallscale firms allowed specialization in specific activities better suited to stand-alone firms than to all-encompassing factories. (iii) Sectoral change permitted much better utilization of China’s factor endowment. In the event, it was the ‘township and village enterprise’ (TVE) phenomenon that attracted the most attention. Indeed, TVEs as a group allowed the Chinese economy to benefit from all three of these productivityenhancing effects. Through the 1980s, an increasingly large share of TVEs were in fact private firms. Even the village-owned firms benefited from close, long-term, face-to-face interactions that approximated the benefits of private ownership. TVEs, especially in the Lower Yangtze region, quickly took over a range of sub-contracting businesses with larger urban enterprises, benefiting from specialization and lower costs. Most fundamentally, TVEs faced factor costs that closely approximated China’s underlying factor endowments—cheap labour and expensive capital—and developed in ways that were consistent with those costs and opportunities.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



23.1.2 The Policy Approach of the 1980s At the time, policy-makers certainly did not envision such a clean process of structural change. Rather, as is well known, policy making at this time was one of ‘crossing the river by groping for stepping stones’, as Chen Yun famously put it. Rather, policy makers were struggling to find a viable reform strategy that would stabilize the economy and promote growth. However, two aspects of the policy challenges China faced were particularly relevant to the question of structural transformation. First, policy makers were deeply concerned about employment. In the 1980s, China was just beginning its ‘demographic dividend’. While celebrated recently in the literature as a crucial element of high-growth periods, the demographic dividend presents an obvious policy challenge. During the 1980s, China’s overall labour force was increasing at 3 per cent annually, and to accommodate some rural-to-urban migration, urban employment had to expand even more rapidly. At the very beginning of the reform process, 1978–80, this chronic problem was made acute by the return to the city of millions of young adults who had been sent to the countryside during the Cultural Revolution. The employment challenge was immense. It can be said that whatever industry produced, it had to absorb labour.² Second, policy makers were very aware of the slow growth of consumption and living standards during the previous two decades. China’s reform era arguably began with the desire to ‘give agriculture a chance to catch its breath’, a policy adopted at the crucial Third Plenum of December 1978. More concretely, this meant reducing the burden of compulsory procurement on farm households, cutting back government investment, and giving households a chance to improve their consumption. The aggregate investment rate was reduced intentionally from 38 per cent of GDP in 1978 to just under 32 per cent in 1982–83 in order to release resources for consumption. Obviously, this government-led shift in the structure of demand facilitated the shift toward a ‘lighter’ industrial structure. At the same time, the lower capital intensity of the expanding labour-intensive industrial sectors meant that rapid growth could continue despite a lower investment rate. The opportunity to engage in international trade complemented these two fundamental drivers of policy change. As China re-engaged with world markets, exports began their steady climb from under 5 per cent of GDP in 1978 to over 20 per cent by 1995. Entry by foreign-invested firms, particularly those from Hong Kong and Taiwan, played a key role in this process. As Panel B of Figure 23.1 shows, the initial entry of foreign firms was spread across a range of sectors, but many of these were involved in exports. By 1996, foreign-invested firms accounted for 40 per cent of China’s total exports, and a much larger share of incremental exports. Moreover, by opening up an export processing regime, China permitted exporters to import raw materials and ² Of course, this was also the period in which China adopted its draconian birth control policies, driven by fears about employment pressure and over-population.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

components duty-free (as long as they were assembled into export goods). Again, by 1996, exports made by processing imported materials and components accounted for 56 per cent of all exports. Since exports were far more labour-intensive than the imported commodities used to produce them, this system facilitated the shift toward a lighter industrial structure.

23.1.3 Discussion It is evident that China’s shift away from a heavy industrial structure, and toward a lighter, more labour-intensive structure, was the result of a broad re-orientation of development strategy. The objectives of employment creation and increased consumption share took over from forced draft industrialization as the primary concerns of policy makers. The ability to access the global market made the shift much more feasible. Yet at the early stage of economic transition, the reorientation of policy was only possible to the extent that a successful reform strategy could be devised. Thus, while structural change at first sight appears to be an offshoot of a particular reform strategy, it is more accurate to say that structural change was limited by the pace of feasible reforms. Reform of the existing institutional set-up would necessarily have been timeconsuming, since both the state ownership system and the inherited price system were difficult to change. Entry by small-scale domestic firms, and by foreign-invested firms, allowed the re-orientation of development strategy to take place quickly. This in turn allowed productivity gains to be reaped quickly and growth of living standards to accelerate. What looked at the time like a transition strategy, should perhaps in retrospect be attributed to the stage of development. Other transitional economies, in the Soviet Union or Eastern Europe, did not have a latent comparative advantage in labourintensive production the way China did. The labour force in each of these economies was growing slowly and had already begun ageing. China and Vietnam, by contrast, with comparatively good human capital and very low wages, could re-orient development strategy to take advantage of their competitive advantages.

23.2 I U, 1995–2013

..................................................................................................................................

23.2.1 Structural Change in Industry, 1995–2013 A very different picture of industrial structural change emerged after 1995. Instead of shifting ‘backwards’ to labour-intensive sectors, industry grew broadly and shifted into more sophisticated capital- and skill-intensive sectors. Growth was rapid across the board. Labour-intensive sectors did not shrink much in relative terms (and grew rapidly in absolute terms). Rather, natural resource sectors, and especially the oil

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



Table 23.1 Composition of industrial output 1995

2008

2013

Food and beverages Labor-intensive manufacturing Materials-intensive industry Capital and technology intensive Electronics All manufacturing Resource extraction Utilities

12.9% 22.1% 28.8% 14.3% 1.1% 79.1% 15.7% 5.2%

9.4% 20.0% 27.7% 22.4% 7.1% 86.5% 6.9% 6.6%

9.7% 19.1% 28.1% 23.6% 7.2% 87.7% 6.5% 5.8%

N.B. ‘Early stage’ (food & labour-intensive) Sophisticated manufacturing (Capital/tech intensive & electronics)

35.0% 15.3%

29.3% 29.4%

28.8% 30.9%

extraction and refining sectors shrank in relative terms (while still expanding, but very modestly, in absolute terms). Table 23.1 shows the change in structure in the industrial sector during this period (in 2013 constant prices). What emerges is a classic pattern of structural change amid rapid growth (real output grew at 13 per cent per year). Structural change reflected a middle-income country undergoing a healthy process of upgrading. ‘Early stage’ industries such as food products and labour-intensive manufactures lost share, as did resource extraction. Petroleum extraction and refining experienced the biggest declines in relative share: even then, oil extraction has continued to inch upward (at 2 per cent annually) and refining also grew (at 9 per cent annually). Skill-intensive and capitalintensive sectors grew especially rapidly and gained share. Electronic hardware was the fastest-growing sector, followed by transport machinery (including automobiles). The China electronics industry has many distinctive features, in addition to rapid growth. First, it is a relatively labour-intensive sector, with capital per worker 25 per cent less than the industrial average. Second, it is the industrial sector with the largest share of foreign-invested firms. In 2013, 69 per cent of electronics output was produced by foreign invested firms. Of course, these two features are closely related, and both are the result of the rapid integration of China into cross-border electronics production networks. The architects of electronics global production networks moved many of the labour-intensive stages of the production process to China. By 2013, there were more than 10 million workers in China’s electronics hardware industry, over 7 per cent of total industrial workers. Because of its importance and distinctive features, generalizations about structural change depend on how we treat the electronics industry. On one hand, electronics hardware can be classified with other capital- and technology-intensive industries as a ‘second stage’ industry, requiring complementary skills in technology and management, and developed infrastructure. In that case,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

‘sophisticated’ industry increased its share remarkably between 1995 and 2008, almost doubling from 15.3 per cent to 29.4 per cent. This increase in share equals the decline in share of early industries and natural resources. China’s industry has become much more sophisticated. On the other hand, if electronics hardware is classified as a labourintensive industry, then the share of broad labour-intensive industries (food products, labour-intensive manufactures and electronics) does not change at all between 1995 and 2008, amounting to 36 per cent of total output in both years. In other words, labour-intensive manufacturing, broadly defined, grew just as fast as total industry during this period, but the composition of labour-intensive manufacturing shifted dramatically in the direction of electronics. In fact, both these classification schemes provide useful insights into structural change in China in this period. Growth has encompassed an exceptionally broad range of industries, and there is very little evidence of China losing competitiveness in labour-intensive industries or growing out of the labour-intensive phase of industrialization. While Chinese industry certainly became more sophisticated, there is little evidence of changing comparative advantages from a factor endowment perspective. Labour-intensive sectors boomed, particularly when electronics is considered. Note that this is true through 2013 as well. Indeed, structural change slowed slightly after 2008, and the annual shifts in composition from 2008 to 2013 were considerably smaller than those from 1995 to 2008.

23.2.2 The Policy Approach of the 1990s and 2000s Economic reforms accelerated after 1995, and policy makers adopted dramatic programmes to restructure industry, the financial system, and foreign trade and investment. In the wake of this period of massive domestic change, China entered the World Trade Organization (WTO) in 2001, and a series of WTO-related commitments drove further structural change in the years to 2008. Once again, reform-related policy decisions were crucial drivers of structural change. However, in contrast to the earlier period, ease of entry for new firms was no longer the predominant factor shaping industrial restructuring. Instead, successful reforms had an immediate impact in opening up new sources of demand. Indeed, the proliferation of the sources of demand was perhaps the most unanticipated feature of the 1998 to 2008 period. Reforms unlocked two qualitatively different sources of demand: external markets in the wake of WTO entry; and demand for housing and furnishings. Both of these created a knock-on demand for upstream industrial products, which the industrial sector was now entirely capable of providing. The WTO story is well known. Freed to export and—crucially—to purchase complementary imports, Chinese firms entered an extraordinary range of export markets. Exports soared from 18 per cent of GDP in 1998–99 to a peak of 35.5 per cent of GDP in 2006. Moreover, Chinese exports quickly took on an extremely diversified profile (Rodrik 2006; Schott 2008). At the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



same time, Chinese exports remained predominantly labour-intensive, particularly when electronics exports are classified as labour-intensive, following the discussion in Subsection 23.2.1. Reform of the housing market was also a crucial driver, one less often acknowledged than the impact of WTO membership. A bold programme of housing privatization in 1998 converted most Chinese urban dwellers into homeowners. With the housing construction market (though not land markets) already largely private, the way was cleared for a housing construction boom that has lasted twenty years and surprised most observers in its magnitude and longevity. Evidently, derived demand from the epochal housing boom led to the expansion of materials-intensive industries such as for steel and building materials. These changes on the demand side would not have had the impact they did without complementary changes on the supply side. In this, we see again the critical importance of the reforms put in place during the Zhu Rongji era, and especially in 1996 to 2002. The 1990s reforms created a far more mixed industrial ownership structure. With many larger state firms now restructured or privatized, and many experienced engineers and managers from the state sector now launched into business, the latent skills and sophistication of China’s industry could now finally be brought fully into play. The supply elasticity of traditional industries, including the materials-intensive industries, was high, and China’s large machinery industry began a broad process of upgrading and import substitution, even as it adapted quickly to the opportunity of providing light consumer-oriented machinery. External opening played an important role on the supply side as well. The case of electronics was discussed earlier; and a similar dynamic is evident in the second-mostrapidly growing industrial sector in this period, transportation machinery. Transport machinery was also the sector with the second-largest share of output produced by foreign-invested firms (39 per cent, and slightly higher, 45 per cent in autos, in 2013). Foreign investment played a crucial role in the upgrading of China’s industry into more technologically demanding sectors (whether capital-intensive or not). It should perhaps go without saying, but the expansion of materials-intensive industry would also have been impossible without the opening of external sources of raw materials. China’s demand reshaped global markets for iron ore and copper during the first decade of the twenty-first century; equally, cheap, high-quality iron ore and copper enabled China’s materials-intensive industries and housing boom during that decade.

23.2.3 Discussion In the most straight-forward and obvious way, China’s industrial boom of the early twenty-first century was driven by its economic reform policy, reflecting the longlasting effect of the reforms put in place at the end of the twentieth century (including WTO membership). Aggregate structural change was not particularly evident because reforms ended up unlocking demand in so many different areas, including areas that

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

had not been well anticipated by economists or policy makers. Since wages were still low and labour was abundant, there were no particular pressures in the form of changing relative factor prices to drive restructuring. The result was a remarkably diverse industrial boom in which, except for oil, virtually all sectors participated. Of course, this environment of explosive growth of an entire network of industries also provided a fertile environment for new kinds of entrepreneurial activity and interindustrial specialization. It follows directly from these observations that this great industrial boom was certainly not the product of sectoral targeting (one of the possible definitions of industrial policy). Indeed, during this period, sectoral targeting played an insignificant role; new technology-oriented industrial policies only began to emerge gradually after 2006 and certainly had no impact until after 2009 (Chen and Naughton 2016). This may have been an advantage. Chinese proponents of industrial policy at the time nursed unrealistic visions of building a domestic-ownership centred automobile industry and consolidating the steel industry. Luckily, those visions were laid out in vague programmatic documents without enforcement mechanisms; their feebleness meant that they had little impact. Had they been aggressively implemented, they would certainly have obstructed the rapid industrialization China experienced. Policy was essential, but it was reform policy directed at making markets work, rather than sectoral policy designed to privilege certain industries. While there was still a relationship between entry barriers and ownership composition, the simple relationship that had existed in the 1980s no longer applied. Conditions of entry had expanded enormously compared to those fifteen years earlier. Financial institutions existed and were largely profit oriented; accumulated wealth was now seeking investment outlets; and restructured and privatized SOEs became important producers and market actors. Nevertheless, entry was not completely free either, but was driven more by formal and informal regulatory barriers. Informal regulation prevented the emergence of strong, diversified private steel makers, for example; while formal regulation maintained the state’s monopoly of the oil sector. It is not completely accidental that the worst performing sectors (petroleum extraction and refining) were also those in which entry barriers remained impermeable, even if natural resource endowment played the most important role. Finally, trends across industrial sectors in the relationship between real output growth and prices display the distinctive and healthy pattern we expect to see in a vigorous growth process driven by demand development and productivity growth. Specifically, we expect to see that industrial sectors growing most rapidly show the lowest rates of sectoral price inflation. In this period, the correlation between the 35sector industrial growth rate and the implied sectoral deflator was a dramatic .61. This is a normal result in rapidly growing economies, the result of the interaction between supply- and demand-side factors. Resource constraints and slow productivity growth push up prices and constrain demand; conversely robust demand growth stimulates greater effort at productivity improvement. The current price structure of output changes less than the fixed price structure of output. Overall, in this period,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



we see an exceptionally broad-based process of growth, driven by the interaction of expanding demand and supply, unleashed by economic reforms.

23.3 T E  L- G   S   S- E  2010

.................................................................................................................................. In the wake of the explosive growth of the first decade of the twenty-first century, China began to undergo profound changes in its underlying economic conditions. Beginning in about 2004, wages for unskilled workers began to increase rapidly, overtaking growth in skilled wages for the subsequent decade. After 2010, aggregate growth began to slow, dropping from the 10 per cent average annual rate between 1978 and 2010 to the 6.7 per cent recorded in 2016. By most indications, the era of highspeed growth was over, and China was entering a period of medium–high growth in the 5–7 per cent range. These changes merit book-length treatment, and can obviously only be addressed in a very partial and incomplete sense here. Still, it is possible to ask what the preliminary evidence is in terms of economic restructuring. We have already seen that the speed of industrial restructuring slowed down after 2008, comparing relatively early multi-year periods.³ However, it is possible to expand our field of vision and look at the relationship between industry and services.

23.3.1 Services in the Economy A substantial literature, dating back to Kuznets and Chenery, describes the succession of large-scale sectors in the development process. Recent work has re-invigorated the structural approach and given it new and stronger micro-foundations (e.g. Herrendorf et al. 2013; Ju et al. 2015). The fundamental empirical generalization from the earlier work has withstood the test of time: there is a regular progression of growth that leads to an increasing share of industry up through about $10,000 per capita GDP (in 1990 purchasing power parities), after which the industrial share begins to decline and growth is increasingly service-sector led (Matsuyama 2008; Herrendorf et al. 2014). China is near this threshold. Moreover, slowing growth and the emergence of excess

³ Information from the Economic Censuses, conducted at approximately five-year intervals, is essential to analyse industrial structure. If we date the end of the high-speed growth era to 2010, the comparison between the 2008 and 2013 censuses provides only a glimpse of the earliest years of the transition to a new economic structure. We must wait for the Economic Census anticipated in 2018 to look more deeply at industrial structural change.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



  60% Tertiary: Services

50% 40%

Secondary: Industry + Construction 30% 20%

2016

2015

2013

2012

2011

2010

2009

2008

2007

0%

2014

Primary: Agriculture

10%

 . Composition of GDP: current prices

capacity in heavy industrial sectors has signalled that China is beginning the era of service-led growth. Policy in China has embraced this transition. Policy makers have acknowledged the ‘new normal’, which includes both slower overall GDP growth and a shift to new growth drivers, including prominently the shift to services. Moreover, initial data seem to confirm that this shift is taking place. Figure 23.2 shows that current price GDP has indeed begun a historical shift in the direction of services, which have increased share significantly since 2011 and accounted for 51.6 per cent of GDP in 2016. The official media have given high visibility to this shift to a service-led economy. However, other data indicate that the shift is far less robust than would initially seem. Figure 23.3 shows that once the data are converted into constant prices, that increase in service share becomes much smaller and less persistent. The service sector share was basically unchanged from 2008 through 2014, and then began a modest increase in 2015 and 2016. In other words, the bulk of the increase of the service has been caused by changes in relative prices, rather than by differential real growth rates. This warrants further discussion. Differential price trends are indeed quite striking. Figure 23.4 shows that the implicit deflators for the secondary and tertiary sectors grew in tandem from 2000 through 2008, and then diverged. Service sector prices continued to climb, at an average annual rate of 4.7 per cent, while industrial prices peaked in 2011 and then began to decline (increasing at only a 2.2 per cent average annual rate over the entire period). As of 2016, the service sector deflator was 46.5 per cent above the industrial deflator on a 2000 base. These price trends reflect changes in the real economy. On the demand side, steadily growing service sector prices reflect the shift in demand toward services, and the relatively weak demand for industrial products, particularly heavy industrial products.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



60% Tertiary: Services

50% 40%

Secondary: Industry + Construction 30% 20% Primary: Agriculture

2016

2015

2014

2013

2012

2011

2010

2008

2007

0%

2009

10%

 . Composition of GDP: production side (2015 constant prices)

250

200 Tertiary 150 Secondary 100 Index 2000 = 100

2016

2015

2014

2013

2012

2011

2010

2009

2008

2007

2006

2005

2004

2003

2002

2001

0

2000

50

 . Implicit deflators: secondary and tertiary sectors

Indeed, the trend of falling industrial prices since 2011 is entwined with global commodities price trends, which have generally fallen during that time. Chinese demand has in turn been a major driver of global commodities prices, most obviously for iron ore, copper, and, to a lesser extent, oil. Thus, these differential price trends are themselves a reflection of structural change and part of a process that draws capital, labour, and human capital into the service sector: price trends are part and parcel of the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

structural transformation of the economy. However, they are only an initial phase of the structural transformation. This point warrants further discussion. It might be supposed that sectoral deflators are influenced by trends in wage rates. Moreover, since many services are labour-intensive, perhaps the increasing relative price of services simply reflects the pass-through of increasing wage costs. After all, this is a period of rapidly increasing unskilled wages: wages of rural–urban workers (nongmingong) grew 13.8 per cent annually between 2009 and 2015, and wages in a sample of private urban firms (predominantly small scale) grew at almost exactly the same pace (NBS 2016a, 2016b). It might be reasonable to expect that prices of labourintensive services were pushed up by wage costs. In fact, this explanation is not supported by the data. First, as Figure 23.5 shows, the aggregate service sector is in fact very diverse, and only certain sub-sectors are labour-intensive. Indeed, Figure 23.5 shows that the service sector can be usefully broken down into three sub-sectors based on capital and human capital-intensity. Traditional labour-intensive services are in the southwest quadrant, and include retail, lodging, and household services (as discussed below, manufacturing and construction are also in this quadrant), In the northeast quadrant is a cluster of capital- and human-capital-intensive service sectors, including most producer (business) services. Finally, in the southeast we see a cluster of humancapital-intensive service sectors, including education, research, culture, and health care. The location of manufacturing in Figure 23.5 shows that Chinese manufacturing is in

Sector intensity of capital and human capital

7

Finance

Real Estate

6

Capital/labour ratio

5 Business Services

4

Utilities

3 2 Transport

1 Construction

0

0

10

20

Information Infrastructure Services Services

Mining Manufacturing Retail & Wholesale Lodging Household Services

30

40

50

Cultural Services

Research & Tech. Services Health

60

70

Share with secondary education

 . Sector intensity of capital and human capital

80

Education

90

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



aggregate still fairly labour-intensive.⁴ Capital per worker in manufacturing is far below that of transport, infrastructure services, and business services. Correspondingly, human capital per worker (here proxied by the share of workers with a secondary school or higher education) in manufacturing is below that in most service sectors. Finally, the implicit deflators calculated for service sectors do not correspond with their labour intensity: prices have increased most rapidly in business services, such as finance and real estate, and least rapidly in transport and retailing. Increasing wages cannot explain trends in relative prices between industry and services.

23.3.2 The Policy Environment for a Service-led Economy China launched a new round of reforms at the November 2013 ‘Third Plenum’. These reforms were to include a ‘decisive role’ for market forces and increased openness. However, implementation has been uneven. A key index of reform progress could be the openness of the service sector, now acknowledged as one of the key drivers of growth going forward. China’s service sector may lag industry in the marketization process. First, state ownership is more prominent in capital- and human-capitalintensive service sectors than it is in manufacturing. Table 23.2 shows the role of state ownership in the labour force of different categories of non-agricultural workers. Using the same categories shown in Figure 23.5, labour-intensive services have similar levels of government ownership to mining and manufacturing, below 10 per cent. (Labourintensive services, not surprisingly, account for the majority of service-sector labour, but much less than half of service-sector capital.) However, government ownership is much more prominent in human-capital-intensive services, where governmentcontrolled entities employ 62 per cent of the labour force; and more prominent in

Table 23.2 Non-agricultural workforce Number of workers

Government own. share

(million)

(per cent)

Mining and manuacturing Labour-intensive services Capital & human capital-intensive services Human capital-intensive services

144 184 42 69

9.9 8.1 34.4 61.8

Total

439

18.4

Economic census, 2013

⁴ To be sure, this observation is partially dependent on the fact that the Economic Census of 2013, the source of the data in Figure 23.5, includes the substantial small-scale and household sector in manufacturing and other sectors. There is tremendous heterogeneity within industry, and indeed within industrial subsectors.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

capital- and human-capital-intensive services, where it accounts for 34 per cent of all workers.⁵ While each sub-sector is different, this high degree of government ownership strongly suggests that regulatory barriers inhibit the entry of private firms into these sectors. Indeed, detailed studies of key sectors such as banking, education, and culture indicates that entry barriers are in fact formidable. Support for such a judgement is also provided by the OECD Investment Restrictiveness Index which finds China’s service sector to be far more protected from foreign investment than the manufacturing sector, and well above developing country averages (OECD 2017). Immediately after the 2013 Third Plenum, there were hopes that the service sector would benefit promptly from a new round of liberalization policies. The Shanghai Free Trade Zone was designated a national ‘pilot’ zone to confirm its role as a test bed for new measures opening the service sector to international participation. However, policies in the Shanghai Zone have not advanced as rapidly as hoped, and those policies that have been adopted have generally not diffused to the rest of the country.

23.3.3 Discussion China clearly stands on the threshold of a substantial shift towards a service economy. However, the analysis in Subsections 23.3.1 and 23.3.2 suggest that such a shift has barely begun. The structure of demand has clearly swung in the direction of services, which is the first step in the transition to a new phase of growth. This can be seen in the more rapid growth in services prices, movement that is not explained by the rapid growth in unskilled wages. However, movement in the underlying structure of output has been much slower. Most of the structural change disappears when prices are properly accounted for. It is worth emphasizing that this pattern of price change is the opposite of what we observed in the case of industrial restructuring during the period 1995–2008. In that industrial restructuring, prices increased most rapidly in the sectors where output was growing most slowly, meaning that real structural change was more rapid than structural change measured in current prices. In the shift to a service economy, prices are growing more rapidly in the more rapidly growing service sectors, meaning that real structural change was less rapid than that measured in current prices. This is not merely a statistical phenomenon. We would normally expect that in a healthy process of restructuring, prices would drop most rapidly in rapidly growing sectors, because booming demand is met by productivity improvement on the supply side. The demand and supply side are working together to facilitate structural change. So far, this is not happening with the Chinese shift to a service economy. This means that we are still in the early phases of structural transformation. Demand has shifted towards the service sector, but relative trends in productivity improvement and cost reduction have so far not emerged to facilitate this shift. That means there is still an ⁵ These calculations include workers in government and in state-controlled public-service undertakings (shiye danwei), as well as enterprises.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



enormous potential—and a long way to go—for China to transform itself into a service economy. Why has the structural transformation so far been slow? It is impossible to say for sure on the basis of preliminary and fragmentary evidence, but we have a primary suspect. So far, reform policy has not sufficiently targeted an open and competitive service sector. Formal and informal regulatory barriers still reserve most of the capital and human-capital-intensive service sectors for state-controlled entities. China still awaits a further acceleration of service sector development when reform policy catches up with existing shifts in demand.

23.4 C

.................................................................................................................................. The preceding analysis has shown the close correspondence between China’s reform strategy and the process of structural change. These patterns break down into three distinct periods, during which different reform policies correspond to clear and distinct structural outcomes. This close correspondence might also be due to the ‘meta-strategy’ of Chinese economic reform. China’s distinctive approach to market transition is often called ‘gradualist’ and is contrasted to the ‘Big Bang’ approach adopted in Poland and, to a certain extent, Russia. However, ‘gradualism’ does not usefully characterize the key elements of China’s reform strategy; in this context it might be more appropriate to characterize Chinese transition strategy as ‘growth oriented’. That is, the essence of China’s strategy has been that marketization and liberalization were harnessed to growth potential (rather than to coherence of reform design). In other words, marketization was always a form of restructuring, privileging certain sectors that were deemed likely to respond to powerful incentives and lower entry barriers. Inevitably, policy makers targeted sectors where the (revealed) growth potential was high, in an ‘opportunistic’ search for growth. They adopted incremental policies that had structural implications. This short chapter has illustrated some of the structural features of different periods of reform policy making. While reform policy has clear structural implications, these are different from the structural regularities of the development process. The interaction between reform policy and development policy depends on stage of development, and is different for each economy. In this sense, the China experience reflects China’s own unique conditions and reveals the extraordinary potential of another period of structural change.

R Chen Ling and Barry Naughton, 2016. ‘An Institutionalized Policy-making Mechanism: China’s Return to Techno-Industrial Policy’, Research Policy, 45, pp. 2138–52. Herrendorf, Berthold, Richard Rogerson, and Akos Valentinyi, 2013. ‘Two Perspectives on Preferences and Structural Transformation’, American Economic Review, 103(7), pp. 2752–89.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Herrendorf, Berthold, Richard Rogerson, and Akos Valentinyi, 2014. ‘Growth and Structural Transformation’, Handbook of Economic Growth, 2, pp. 855–941. Ju, Jiandong, Justin Yifu Lin, and Yong Wang, 2015. ‘Endowment Structures, Industrial Dynamics, and Economic Growth’, Journal of Monetary Economics, 76, pp. 244–63. Lin, Justin Yifu, 2012. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: The World Bank. Lin, Justin Yifu, Fang Cai, and Zhou Li, 1996. The China Miracle: Development Strategy and Economic Reform, Hong Kong: Chinese University Press. Matsuyama, Kiminori, 2008. ‘Structural Change’, in L. Blume and S. Durlauf, eds, The New Palgrave Dictionary of Economics, 2nd edn, Basingstoke: Palgrave Macmillan. Naughton, Barry, 1995. Growing Out of the Plan: Chinese Economic Reform, 1978–1993, New York: Cambridge University Press. Naughton, Barry, 2007. The Chinese Economy: Transitions and Growth, Cambridge, MA: MIT Press, 2007. (Chinese translation 2011.) NBS (National Bureau of Statistics), 2016a. ‘Annual Survey of Rural Non-Agricultural Workers [nongmingong]’. Accessed at: http://nbs.gov.cn NBS (National Bureau of Statistics) 2016b. ‘Wages in Private and Non-Private Employers’. Accessed at: http://nbs.gov.cn OECD, 2017. ‘Investment Restrictiveness Index’. Accessed at: http://www.oecd.org/invest ment/fdiindex.htm Rodrik, Dani, 2006. ‘What’s so Special about China’s Exports?’ China & World Economy 14, pp. 1–19. Schott, Peter K., 2008. ‘The Relative Sophistication of Chinese Exports’, Economic Policy 23, pp. 5–49. State Council First Economic Census Leadership Small Group Office 中国经济普查年间 2004。 北京：中国统计出版社，2006. Four volumes. State Council Second Economic Census Leadership Small Group Office, 中国经济普查年间 2008。 北京：中国统计出版社，2010. Five volumes. State Council Third Economic Census Leadership Small Group Office 中国经济普查年间 2013。 北京：中国统计出版社，2015. Five volumes.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

                                                        ......................................................................................................................

 

24.1 I

.................................................................................................................................. S K (officially, the Republic of Korea) has achieved rapid economic growth for several decades, since the early 1960s. In 1996, South Korea finally joined the OECD and the ranks of high-income economies. Its economic growth is noteworthy, because its initial conditions were quite similar to many African countries, in that South Korea underwent several decades of colonial rule, several years of civil war, and a period of hunger and food shortage in the 1950s, and reliance on US food aid. Its experience was worse in terms of resource endowments, with all the minerals located in North Korea. Furthermore, although it launched a series of five-year economic plans, beginning in the early 1960s with a new political leadership (ex-military, President Park), Korea was once in the same situation as other developing countries, in terms of facing continual external imbalances with persistent trade deficits, until the late 1980s (Lee and Mathews 2010; Lee 2016: ch. 1). While the initial emphasis of Korea’s industrial policy was the promotion of labourintensive sectors for earning dollars by exporting in the 1960s and 1970s, the government put a new emphasis on technological development, mostly since the 1980s, with some preparation in the 1970s. The preparation for such a policy shift was started with the establishment of government research institutes (e.g. Korea Institute of Science and Technology (KIST) in the 1970s to conduct problem-solving R&D for private firms and to transfer the R&D outcomes to them. Beginning in the mid-1980s, a decisive policy shift occurred, when the government encouraged private, in-house R&D by allowing

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

tax exemptions for R&D expenses, and even initiating public–private joint R&D to break into higher-end segments and sectors involving bigger and riskier projects (Lee and Kim 2009; Lee 2013b). This policy initiative succeeded in building the competitive and high-end manufacturing sector, which was an important factor that led to a trade surplus in 1986, for the first time in modern Korean history. Since then, Korea has been able to overcome the persistent trap of external imbalances or stop–go cycles of crisis and reforms. It was Amsden (1989) that attributed such successful economic catch-up to industrial policy by the government, getting prices wrong and creating rents for targeted sectors. Industrial policy in Korea, under the leadership of the Economic Planning Board (EPB), has more or less followed the practices of Japan, which is well documented in the influential work of Johnson (1982), who attributed the Japanese miracle to the role of one super ministry called MITI (Ministry of International Trade and Investment) in Japan. One of the first definitions of industrial policy was in Johnson (1982), who defined it as policies that aim to improve the structure of a domestic industry in order to enhance a country’s international competitiveness. While Japan and Korea have made remarkable success in catch-up development, owing to industrial policy, some other countries followed the free market principle of the so-called Washington Consensus and focused on macroeconomic stabilization and trade and financial liberalization. While the latter group also experienced some economic growth, it tended to be short-lived or of the stop–go cycle type, because those following the Washington Consensus principles failed to bring up capabilities of private sectors (Lee and Mathews 2010). While Rodrik (1996) noted the importance of sequential or gradual adoption of ten policies of the Washington Consensus in East Asia, different from the simultaneous adoption in Latin America. He missed the fact that East Asia had further built up and upgraded capabilities, since the mid-1980s, before moving to more marketization (the next five policies in the Washington Consensus) (Lee and Mathews 2010). When we see catch-up growth as the process of capacity building, what we have in mind is the capacity of private corporations. The capacity of latecomer economies to grow capable private companies is the most important and fundamental criterion to determine the success or failure of economic development or growth. The corporations may initially be state-owned firms (e.g. the Pohang Iron and Steel Company (POSCO) in Korea), when the risks for private capital are too high. The idea, however, is to move them towards private ownership (i.e. eventually make them ‘public’ through an initial public offering (IPO)), after they build up certain levels of capabilities or competitiveness. Thus, this chapter considers the essence of industrial policy to be building the capabilities of private firms to sustain long-term economic growth, rather than picking winners or providing protection for some firms or sectors (Lee 2013a). Among various aspects of capacities, the emphasis should be on technological capabilities, because without these, sustained growth going beyond the middle-income trap is impossible (Lee 2013c). In this era of open market competition, private

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



companies cannot sustain growth if they continue to rely upon cheap products; they need to be able to move up the value-chain to higher-value added goods, based on continued upgrading and improvement and technological innovation. Furthermore, another important feature of the Korean model is that these private companies have been ‘locally owned’ companies, including locally controlled joint ventures (JVs), not foreign controlled subsidiaries of the multinational corporations (MNCs). MNCs are always shopping around the world, seeking cheaper wages and bigger markets. Therefore, they cannot be relied upon to generate sustained growth in specific localities or countries, although they can serve as useful channels for knowledge transfer and learning. In what follows, we discuss the role of the government or industrial policy in this process of capability building, with a focus on the financing aspect of the policy implementations. This chapter can be regarded as a sequel to Lee (2013a) and Lee (2015). The former has a more theoretical focus, discussing the three types of failures— market, system, and capability—failures as a justification for government activism, whereas the latter discusses the different tools of industrial policy at different stages of development. In Section 24.2, we first elaborate the nature of financial control by the government, which has been one of the enabling conditions for industrial policy since the 1960s (its take-off). We also explain the roles and evolution of key developmental banks, such as the Korea Development Bank, Ex-Im Bank, and Industrial Bank for small and medium-sized enterprises (SMEs). Section 24.3 elaborates the three episodes of industrial policy and financial arrangement in these cases, such as the case of the establishment of POSCO, targeted development of bottleneck technologies for SMEs, and leapfrogging into digital TV since the mid-1990s. Then, Section 24.4 concludes the chapter with a discussion of the implications for African economies.

24.2 T F S  I P  K

..................................................................................................................................

24.2.1 Financial Control and Industrial Policy The serious scarcity of capital in the 1960s and 1970s in Korea forced firms to depend heavily on credit for raising finance beyond retained earnings. In the absence of effective capital markets, the state used its control over the banking system to channel domestic and foreign savings to selected industries or firms (Lee and Lee 2016: ch. 2). The new regime that took power in 1961 nationalized the commercial banks, and thus, the banks were owned by the government until 1980, when they were privatized. Although many banks have been privatized, the Korean government still maintains

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

effective control over the banking institutions through its personnel policies. In Korea, the government exercised almost direct control over private sectors through their control of credit. For an effective state activism or industrial policy, state ability for financial control was critical. One often does not notice the critical difference between the state’s financial control through credit allocation and other control instruments, such as tariffs, import quotas, tax incentives, and entry or trade licences. First, financial control implied more discretionary control. With credit allocation, the state can control not only the financial ability of firms, but can also impose a firm’s compliance in other matters. Second, a qualitative difference was that the state’s financial control was not based on its political authority, which was the case for other instruments that are supported by legislation or regulations; rather the state’s financial control was based on the state’s economic power, which was associated with its ownership of banks. Third, whereas most other controls, except licensing, were aimed at specific industries or sectors, and thus, affect firms only indirectly, financial control was directly aimed at individual firms. In this regard, a simple but fundamental fact should be noted: the state’s financial leverage over firms carried the power of control because firms had a strong motivation to better their performances and because firms believed credit supply to be critical. In Korea, the firms’ motivation for success derived from private ownership and an expectation that firms would be the beneficiaries of their own good performance. Thus, even if big business firms were under so-called soft budget constraints due to their special connections with state agencies, that did not necessarily lead to weak motivational efficiency, as it did in socialist firms, but can, in fact, lead to exactly the opposite behaviour, that is, excessive risk taking.¹ Korea had a huge saving gap in the 1960s, with domestic savings at 9 per cent of GDP and gross investment 15 per cent of GDP, and thus, had to rely on foreign borrowing to fill the gap. That is why exports were so important and the critical binding constraint for growth for an economy at lower and middle income stages. Despite the low income and thus low domestic saving, Korea had maintained a higher investment rate, and one of the reasons for this was the low-interest rates, suppressed by the government. So, Korea was basically under a condition of financial repression, but it may be considered as ‘financial restraint’, in the words of Hellmann et al. (1997), in that the real interest rates had been at least positive. Despite this suppressed interest ratio, the domestic savings ratio in Korea had continued to increase, owing to the growth of income associated with strong investment over the decades (Cho 1997); the domestic savings rates had increased from 9 per cent in the early 1960s to about 30 per cent in the mid-1980s.

¹ Park (1990) mentioned risk taking in the form of excessive and duplicative investment in the heavy industry drive in Korea in the late 1970s.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



In the Korean experience, the banking sector had always been supposed to ‘serve’ the real sectors by providing a stable supply of the so-called ‘growth money’ at affordable rates, whereas the manufacturing or production sectors had always been given priority. Of course, such a practice had been possible because Korea established several development banks, such as Korea Development Bank, Ex-Im Bank, and Industrial Bank and also, because most of the commercial banks were under government ownership or control until they were privatized in the mid-1980s. With a very small margin between the lending and deposit interest rates, the profitability of banking sectors was very low, which boosted the profitability of the manufacturing sector, so that private investment flowed into manufacturing rather than into financial businesses. Furthermore, manufacturing sectors were often earning rents, owing to entry control by the government in adjusting the ‘optimal number’ of firms in each sector, in consideration of the market size, so that the admitted firms were, in effect, guaranteed minimum levels of profits (rents), which can be a source of investment funds for the next period. Making the rate of return in certain industrial sectors higher than interest rates can be another means to direct industrial policy, especially in a situation facing high interest rates. In Korea, this tradition of implementing entry control in many sectors had been regarded as a type of industrial policy modelled on Japanese practices (Johnson 1982). The practice had two meanings. The first was to sort out the ‘good’ and ‘bad’ producers, and the second was to allow stable profits for the selected producers, so that they were assured long-term profits, and may be encouraged to invest more in fixed capital for business expansion. This practice also had the effect of having the return rates higher than interest rates, which was also good for boosting private investment. Simply put, the idea was that, for instance, five firms with profits in a sector are better than ten firms with no profits. Such a practice of entry control had been one of the typical tools of industrial policy in the past in Japan and they were copied in Korea.

24.2.2 The Roles and Evolution of Several Developmental Banks 24.2.2.1 Korea Development Bank (KDB) The Korea Development Bank (KDB) has been the main vehicle for policy loans, or so-called development financing in Korea, with its value of assets of 269.7 trillion won (US$ 232.5 billion) in 2016. The bank was established in 1954. The main function of Korea Development Bank was to provide funds for industry, especially for manufacturing, agriculture, and mining. In the 1950s, the bank’s main source of funds was foreign aid from the USA and, using these funds, KDB invested in basic industries such as the fertilizer and cement industries and recovery of a power plant destroyed during the Korean War (Korean Economy Compilation Committee 2010). Private firms needed endorsements from the finance minister to obtain KDB loans when

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

the source of funds was not aid money. Thus, the overall size of policy loans from KDB was relatively small. Until 1960, their policy loans were less than 16 billion won (US$250 million). In 1961, the new government changed the development strategy from importsubstitution industrialization to export-led industrialization. To do that, they made the supply of policy loans the main ‘duty’ of the financial sector, with KDB as the pillar bank in this regard. Laws concerning the KDB were revised four times in the 1960s, and the bank’s registered capital increased from 40 million won (US$0.32 million) in 1961 to 150 billion won (US$520 million) in 1969, with the legal right to borrow money from foreign countries (KDB 2014). Using these funds, policy loans from KDB increased by about twelve times from 20.3 billion won (US$162.7 million) in 1961 to 239.13 billion won (US$608.6 million) in 1972 (Son 2013). Most of loans were used for production facilities, and 55.9 per cent of the funds, out of the total loans made in Korea for production facilities, were provided by KDB (KDB 2014). KDB also provides guarantees for loans when Korean firms borrow from foreign financial institutions. The amount of the guarantees increased from 18.1 billion won (US$139.2 million) in 1963 to 600.3 billion won (US$1.73 billion) in 1971 (KDB 2014). The Korean government started fostering heavy and chemical industries in 1973. These industries have certain characteristics, which require a large amount of investment and a long time horizon to be profitable. To supply such investment funds for heavy and chemical industrialization, the KDB acquired loans from foreign financial institutions, such as the Asian Development Bank (ADB) and the International Bank for Reconstruction and Development (IBRD), and issued foreign currency bonds in the international capital market. Funds from foreign financial institutions and foreign currency bonds increased rapidly from 4.79 billion won (US$12.2 million) in 1972 to 478.7 billion won (US$989 million) in 1979 (KDB 2014). As a result, about half of KDB funds came from foreign countries in 1979. The remaining funds of the KDB mainly come from the ‘National Investment Fund’, which was raised by issuing bonds sold to households and private banks. The government forced every commercial bank to buy the bonds of the National Investment Fund, using as much as 20 per cent of the annual increase in their savings deposits (Nam 2009). Using these funds, policy loans from the KDB increased very rapidly, by about ten times, from 318.47 billion won (US$799.5 million) in 1973 to 3.12 trillion won (US$6.4 billion) in 1980. These funds were mainly used for heavy and chemical industries, such as shipbuilding, steel, machinery, chemical, automobile, and electronics industries. In the 1980s, the focus of industrial policy changed from a sector selective industrial policy to bottleneck technology development (Shin and Lee 2012). In accordance with the change, since 1981, the law concerning the KDB has specified that KDB could provide funds for R&D in the emerging industries. Also established, in 1984, was the ‘Korean Technology Financing Corporation’, to match increasing venture capital demand. Due to these changes, the rate of increase of KDB policy loans slowed down in the 1980s, which increased from 3.12 trillion won (US$6.4 billion) in 1980 to 10.59 trillion won (US$15.8 billion) in 1989. In terms of the source of funds, the share of

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



loans from foreign countries and foreign currency bonds decreased as the Korean economy grew. Instead, the share of domestic bonds and deposits increased from 13.2 per cent in 1980 to 80.4 per cent in 1989 (KDB 2014).

24.2.2.2 Export–Import Bank of Korea The Export–Import Bank of Korea was established in 1976, to provide long-term policy finance to exporting firms, active in the export of capital goods. Given that the capital goods sectors were one of the least developed sectors (Lee and Kim 2016), strong export financing was needed to offset some of the competitiveness disadvantages facing the Korean firms. The new strategy of the Korean government, since the mid-1970s, to promote heavy and chemical industrialization also targeted exports of capital goods. To do that, long-term export financing was needed at that time, because international markets for capital goods were basically buyers’ markets, and many foreign buyers required deferred payment conditions for sellers. Furthermore, it usually takes a long time to make capital goods, so it is very difficult for firms in latecomer countries to export without long-term financial support, if foreign buyers demand deferred payment conditions (Export–Import Bank of Korea 1996). To support domestic exporting firms facing the deferred payment condition from the foreign buyers, the Export–Import Bank of Korea provided long-term policy loans, with repayment periods as long as ten years. The annual interest rate was 7 per cent, which was, relatively, a very low level. After the establishment of the Export–Import Bank of Korea, the amount of export financing increased from 134.2 billion won (US$277 million) in 1977 to 444.3 billion won (US$918 million) in 1979, and to 774.7 billion won (US$890 million) in 1985. The share of exports, supported by loans from the Export–Import Bank of Korea, among total exports also increased from 1.5 per cent in 1977 to 4.8 per cent in 1984 (Export–Import Bank of Korea 1996). From 1976 to 1985, 76.5 per cent of their export financing went to the shipbuilding industry. It supported rapid export growth in the shipbuilding industry for which the annual average export growth rate was 20.5 per cent during this period. Remaining funds went to other heavy and chemical industries. In the late 1980s, exports of shipbuilding and plants decreased due to a change in the international conditions, so that the amount of export financing decreased to 314.4 billion won (US$356 million) in 1986. In response to this decrease, the Export–Import Bank of Korea expanded the list of target industries to include electronics and electrical instruments. From the late 1980s, the Korean economy posted a trade surplus for the first time, and thus, some of the government regulations against outbound foreign investment by domestic firms were relaxed. In accordance with the easing of the regulations, the Export–Import Bank of Korea provided policy loans to Korean firms that invested in foreign countries. Facing rising wage rates in Korea, firms in the light industries, such as textiles, tried to move their production facilities to developing countries which had cheaper labour costs. Therefore, the Export–Import Bank of Korea provided them with financial services. As a result, loans for international investment increased

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

from 5.17 million dollars in 1987 to 574 million dollars in 1995 (Export–Import Bank of Korea 1996).

24.2.2.3 Industrial Bank of Korea (IBK) for the SMEs The Industrial Bank of Korea (IBK) was established by the Korean government in 1961. Its main function was to provide loans for SMEs. The law on the IBK specified that the share of the SMEs in its total loans should be at least 90 per cent (IBK 2011). In addition to firms in manufacturing, mining, and transportation, since 1973, firms in the construction, commerce, and service sectors could also be regarded as client SMEs for IBK. Nevertheless, the main focus was manufacturing SMEs with respect to IBK’s contribution to an export-led industrialization strategy. One difference between IBK and either KDB or the Export–Import Bank of Korea was that the majority of IBK’s funds came from deposits by households and firms, and the share from international borrowings or from the National Investment Fund was small. However, compared to KDB and the Export–Import Bank of Korea, the sizes of their policy loans were relatively small. The amount of a policy loan from IBK was about one-third of that from KDB in the 1960s and 1970s. To support export-led industrialization, IBK increased their policy loans very rapidly from 2.1 billion won (US$16.8 million) in 1961, to 52.7 billion won (US$170 million) in 1970, and to 645 billion won (US$1.3 billion) in 1979 (Son 2013). The share of SME loans from IBK, among the total amount of SME loans, was 21.7 per cent in 1970. IBK also provided SMEs with consulting services or technology guidance (with the UN Development Programme (UNDP)) to improve the competitiveness of SMEs. In the early take-off period in Korea, the main focus of development strategy was on a selected number of big businesses, which were the leading exporters during the period. In particular, heavy and chemical industrialization, from the mid-1970s, targeted big businesses that could meet the requirement of a certain size of fixed capital investment. With this policy background, the SMEs were not the main focus of industrial policy. However, the new regime, which came into power in 1980 after the death of President Park, introduced some changes to industrial policy, such as a shift from sector-specific targeting to technology-specific targets. Another change, since the mid 1980s, has been to allocate more resources for SMEs in technologyintensive businesses. Since 1981, IBK has provided policy loans for SMEs that make various intermediate goods, such as diverse parts, industrial materials, and tools, and sell to big firms. Since 1986, IBK has provided long-term policy loans to these SMEs. In 1989, IBK began providing policy loans for small firms whose potential for future growth was good but their number of employees was less than 50. These small firms usually had difficulty in getting loans from commercial banks, which required some value of collateral. Thus, IBK provided policy loans to these firms without requiring much collateral. Due to this kinds of financial support, policy loans from IBK increased from 970.5 billion won (US$1.6 billion) in 1980 to 6.69 trillion won (US$9.45 billion) in 1990.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



24.3 S  I P  F

..................................................................................................................................

24.3.1 Industrial Policy to Develop a Strategic Sector: The Case of Pohang Steel The growth and development of the steel industry in Korea has been represented by a state-owned company, POSCO.² Typically, state activism is justified when there is a certain level of positive externalities, such as that of market failure prevailing in terms of the gap between private and social returns. POSCO’s case fits into this category for state intervention. Steel is an input in diverse sectors of production. Given the high degree of the scale economy and a limited size of the domestic market, throughout the history of Korea, steel goods were certain to be underproduced if left with private firms, and private monopoly would charge much higher prices. Reliance on imported steel alone would lead to no benefits from backward and forward linkages. Under these conditions, entry by establishing a state-owned enterprise (SOE) seemed to be the rational choice in the context of the Korean economy in the past. During the reconstruction period after the Korean War (1950–53), the rising domestic demand for steel products led to a need for the construction of an integrated steelworks.³ At the time, most Korean steelmakers used scrap iron, rather than pig iron, as raw material. With scrap metal running out, the need for a stable supply of pig iron increased. In addition, Korean steel firms in those days were small and specialized in only one segment of the whole process of steel production. This inefficient separation underlined the advantage of having an integrated steel mill. In the absence of private capitalists able to take on a heavily capital-intensive, integrated steel project, a government initiative was inevitable. However, the Korean government’s six attempts for eleven years between 1958 and 1968 all foundered. The main reason for the failure lay in financing the projects. Opposing the Korean government’s plan to build an integrated steel mill, the World Bank and the US Agency for International Development (USAID) expressed concerns about Korea’s ability to repay foreign loans and questioned the need for a large-capacity steel mill in a small developing economy (D’Costa 1999: 64; Song 2002: 57). Rather, the World Bank and USAID suggested developing steel-consuming industries first, for example machinery, automobile, and shipbuilding (Song 2002: 57). The Korean government rejected their opinion and insisted that steel-consuming industries were not a prerequisite for the successful development of the steel industry, and that the steel industry should grow first for the effective development of steel-consuming sectors. Former President Park Chung-hee took the initiative and gave top priority to the steel project in the second

² This subsection relies on Lee (2015).

³ This paragraph is based on Lee (2015).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

five-year economic development plan (1967–71). The steel project was one of the three key projects of the plan. The others were the Ulsan petrochemical complex and the Gyungbu Expressway (Song 2002: 42–3). The Korean government created the state-owned steel firm POSCO in 1968. The government held 56.2 per cent of the company’s shares, and the remaining 43.8 per cent were held by the state-run Korea Tungsten Co. Two years later, the company commenced construction of the initial phase of the nation’s first integrated steelworks in Pohang. The long-lasting principal problem of financing was overcome by ‘ingenious’ methods (D’Costa 1999: 63–4). Through agreements with the Japanese government in 1969, the Korean government allocated part of the war reparation funds from the Japanese to the Pohang project. A total of $73.7 million from the war reparation funds for three years was assigned to the first phase. Another loan worth $50 million was provided by Japan’s Export–Import Bank. Japanese sources accounted for approximately 60 per cent of the capital needs of the first phase (Song 2002: 76). The rest was covered by local capital. Table 24.1 presents the sources of financing by phase. Direct investment from the government accounted for 11.3 per cent of the project’s total costs. The government’s intervention and assistance enabled POSCO to access domestic and foreign sources, accounting for approximately 66 per cent. Domestic sources of finance were state-run and private bank loans with very low interest rates that were actually negative in reality. To mobilize resources from abroad, the government negotiated with foreign lenders on behalf of its national producer and guaranteed POSCO’s loan payments. Evident from Table 24.1 is the increasing share of POSCO’s own funds from 0 per cent (Phase I) to

Table 24.1 Financing for the Pohang project (US$ million)a Period

Govt. capitalb

Domestic fundsb

Own fundsb

Foreign capitalb

Total costsb

I

1970–1973

II

1973–1976

III

1976–1978

IV-1

1979–1981

IV-2

1981–1983

111 (33.2) 19 (3.2) 225 (16.2) 121 (7.8) 0 (0.0)

26 (7.7) 39 (6.5) 101 (7.3) 336 (21.7) 47 (13.3)

0 (0.0) 157 (26.6) 293 (21.1) 327 (21.1) 189 (53.4)

197 (59.1) 376 (63.6) 768 (55.4) 768 (49.5) 118 (33.3)

334 (100.0) 591 (100.0) 1,387 (100.0) 1,552 (100.0) 354 (100.0)

476 (11.3)

549 (13.0)

966 (22.9)

2,227 (52.8)

4,218 (100.0)

Phase

Total

Note: a Original units are in Korean won. The won–dollar exchange rates used in the conversion are calculated by averaging the daily exchange rates for each phase: 361.00, 448.89, 484.00, 555.36, and 729.31 won/dollar for phases I, II, III, IV-1, and IV-2, respectively. b Percentage of total costs in parentheses. Source: Song (2002, 118).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



53.4 per cent (Phase IV-2), whereas that from foreign capital declined from 59.1 per cent to 33.3 percent over the same period. These changes indicate that POSCO’s ability to generate internal funds was gradually enhanced while the government nurtured the industry through various instruments, which is addressed in the next Subsection 24.3.2. In 1970, the steel mill’s first-stage construction commenced in Pohang. By 1983, its production capacity had expanded four times. Additional integrated steelworks were constructed in Gwangyang in the mid-1980s. As a result, Korean steel production increased sharply. By 1993, the only Korean integrated steel firm broke the 30 million tonne mark, which placed Korea in sixth place in global crude steel production. During the period 1973–93, the compound annual growth rate (CAGR) of Korea’s crude steel output was 21.2 per cent, whereas that of the world was 0.7 per cent. In 1998 and 1999, POSCO became the world’s biggest steel producer, surpassing the former top producer Nippon Steel (Lee and Ki 2017). Currently, POSCO has two integrated steelworks in Pohang and Gwangyang, and it produces approximately two-thirds of Korea’s total steel output. Notably, this successful development was made possible by the combination of government activism and the SOE’s aggressive technological learning and capability building. In its early stage, POSCO simply purchased and used stabilized or standard technologies and facilities. At the time, overseas training was the primary source of learning. In the 1980s, as POSCO increasingly threatened rival companies in the global export market, access to a foreign knowledge base became more difficult than before. Thus, POSCO established its own R&D system, which was composed of three parties: industry (POSCO), a university (Pohang University of Science and Technology (POSTECH)), and an institute (Research Institute of Science and Technology (RIST)). The in-house R&D system facilitated the company’s stage-skipping catch-up, as it adopted the most up-to-date technologies and facilities in the second steel mill project. The building of POSCO’s technological capabilities can be considered a path-following catch-up at the initial stage and a stage-skipping catch-up at the later stage, according to the classification of the three types of catch-up proposed by Lee and Lim (2001). As a matured industry, technological uncertainty was low in steel production. Furthermore, the Koreans’ entry and expansion at a late stage took advantage of the window of opportunity associated with the lowered price of factory equipment and facilities during global recessions, namely, the first and second oil shocks (Lee and Ki 2017). Market uncertainty decreased through government and private efforts to develop automobiles, shipbuilding, and other steel-consuming industries. Finally, after its stable establishment in terms of international competitiveness, this SOE was completely privatized in 2000. The Steel Industry Promotion Law was announced on 1 January 1970, three months before construction of the first phase of the Pohang plant. The law, which was valid for ten years, empowered the government to grant POSCO various financial and administrative support for (i) access to long-term and low-cost foreign capital; (ii) the purchase of equipment and raw materials; (iii) the construction of port facilities, water and electricity systems, roads, and railroads; (iv) research and technical training;

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

(v) reduced prices on electricity, gas, and water; and (vi) discounts for rail transport and port dues (D’Costa 1999: 65; Lee and Ki 2017). At the same time, the law made changes in the Regulation Law on Tax Reduction and Exemption and in the Tariff Law. POSCO was exempted from corporate tax and received an 80 per cent tariff cut on the import of equipment (Nam 1979: 78).⁴ After an extension of another twenty years, the Steel Industry Promotion Law was discontinued in 1986 (D’Costa 1999: 65). Construction of the first phase for a production capacity of 1.03 million tonnes was completed between 1970 and 1973. By 1983, four expansions had been carried out, increasing the total capacity of the Pohang Mill to 9.6 million tonnes (D’Costa 1999: 65). Empowered by the law, the government was able to provide a large fund for the Pohang project in various forms. The government pumped $476 million into the project. Additionally, in the form of infrastructure support, tax and tariff cuts, and discounts for public utility charges, the government invested approximately $840 million (Song 2002: 118–19). When passed, the Steel Industry Promotion Law was criticized as benefiting only POSCO. To be eligible for the previously mentioned government support: (i) a steel firm should have an integrated steel mill with more than one million tonnes annual capacity; and (ii) the government should hold over 50 per cent stake in that company. POSCO was the only firm to meet those criteria. As a way to establish the steel industry, Park’s administration concentrated all available resources on the single, state-owned POSCO and its integrated steel mill rather than creating an environment for private firms to grow in a market mechanism and with free competition. The absence of a capitalist class for a capital-intensive steel project enables us to argue that such direct intervention was inevitable and justifiable at the time. Since 1973, POSCO received a further boost through a substantial change in the economic growth policy of the Park administration. The Heavy and Chemical Industrialization (HCI) Program (1973–79), designed to shift the Korean economy away from the low value-added light industries, selected six heavy and chemical industries for intensive nurture: steel, petrochemicals, automobiles, machine tools, shipbuilding, and electronics (D’Costa 1999: 65). This programme accelerated POSCO’s growth in two ways. First, the government strengthened its support for the steel industry, mainly through low-interest financing and tax cuts. Second, and more importantly, the HCI drive made the government realize the necessity for the expansion of the Pohang plant and, furthermore, the construction of an additional integrated steel plant. The selected sectors from the programme were mostly steel intensive; thus, a significant increase in steel demand was expected from these industries. As a result, following the announcement of the HCI strategy, the Pohang plant was expanded four times from 1973 to 1983. Construction of the second steel plant began in 1985 against the backdrop of the thriving heavy industry (Song 2002: 99, 159–60).

⁴ Steel firms with an integrated mill with more than 100 thousand tonnes annual capacity were eligible for this tariff cut.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



24.3.2 Industrial Policy to Develop Bottleneck Technologies for the SMEs The Industrial Base Technology Development Projects (IBTDPs) for the period of 1987–1991 (which was later renamed as ‘Industrial Technology Development Projects’) symbolizes the shift to a functional, promotion-type industrial policy from the earlier industrial policy of sector-promotion (Korea Industrial Technology Evaluation Institute 2007). This shift was initiated by the abolishment in July 1986 of the Industry Promotion Law, which targeted seven sectors, and by the recognition of a new law, the Industrial Development Law, in the same year. This law established the legal basis for the implementation of the firm survey on their demand for specific industrial technologies, and for implementing various projects to develop ‘industrial base technologies’ (see Table 24.2). The IBTDPs were intended and implemented to develop the so called bottleneck technologies that can be commonly applicable to a large number of SMEs, preferably in the form of tripartite joint R&D by the private–academic–public labs. Also, the ministry in charge changed from the Ministry of Science and Technology to the Ministry of Trade and Industry for this IBTDP. As can be seen in Table 24.2, about half of the funding for each project was from the government budget. One of the noteworthy features of the IBTDPs was trying a bottom-up approach, compared to the previous top-down approach, to identify key bottleneck technologies by conducting large-scale surveys to firms (see Tables 24.3 and 24.4). From 1987 to 1991, five rounds of surveys were conducted, with a spent budget of 1,885 million won, which led to the identification of 1,329 needed technologies. Out of these, 934 technologies were funded for development, with a success rate of 84.4 per cent. In this scheme (Table 24.3), the technologies were classified into several categories, such as those to be funded by these projects and to be developed domestically, and those that could be imported rather than developed domestically. Table 24.4 shows the diverse financing options for the different identified technologies. Only those classified

Table 24.2 Industrial Base Technology Development Projects (IBTDPs), 1987–95 Unit: 100 mil. won

No. of projects (new) (continuing) Investment amount (government budget) (private sector funds)

1987–1990

1991

1992

1993

1994

1995

Total

945 617 328 2,152 1,026 1,126

551 333 218 1,369 712 657

480 179 301 1,395 727 668

420 258 162 1,706 887 819

422 173 249 2,644 1,414 1,230

464 243 221 3,701 1,908 1,793

3,282 1,803 1,479 12,967 6,674 6,293

Source: Korea Industrial Technology Evaluation Institute (2007: 45, Table 3–8).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Table 24.3 Outcomes of the survey to identify the ‘needed’ industrial technologies 1986 1987 1988 1989 1990 1991 Number of technological areas (number of the units in charge of the survey) Number of experts involved the surveys Number of the participating firms Budget for the surveys (million won)

A Total number of technologies identified for projects

185 ( ) 981 733 247

225 7 852 724 241

102 200 200 (9) ( ) (27) 492 1,205 1,416 535 1,107 5,994 240 251 701

No. of technologies identified as needed to be developed

581 562

564

417

638

947

No. of technologies identified as needing further guidance and assistance

118 168

117

56

105

217

No. of technologies identified to be imported

837 202

202

46

75

165

1,536 932

883

519

Total

219 ( ) 818 585 205

818 1,329

Source: Korea Industrial Technology Evaluation Institute (2007: 12, Table II-3).

Table 24.4 Implementation plan of the technology development projects identified by the demand surveys Classification Characteristics of the technologies

Support plan

Group I

• Technologies badly needed in the • To be funded by this IBTDPs and/or production sites of the firms other policy loans • Basic (generic) technologies identified as common bottlenecks • Technologies with high commercialization • Possibilities and ones that are soon expected to be developed by existing firms

Group II

• Long-term, large-scale projects • Technologies that require more and broader, basic researches to be successfully developed

Group III

• Technologies easily developed with direct grant to the involved firms

• To be funded by the Targeted (focused) R&D Projects administered by the MOST

• To be provided loans for the required expenses for R&D: • Long-term, low-interest rate loans (Industrial Development Funds) • General policy loans (recommending loans from technology development funds by the Korea Development Bank or Industrial Bank of Korea)

Source: Korea Industrial Technology Evaluation Institute (2007: 12, Table II-4).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



in Group I (for instance, those identified as badly needed technologies in many production sites or regarded as common bottleneck technologies) were intended to be funded by this IBTDP. In comparison, those in Group III, such as technologies easily developed with direct grants to the firms involved, are to be developed by direct grants to the specific firms from banks such as the KDP or the IBK.

24.3.3 Industrial Policy for Leapfrogging: Digital TV by Public–Private Joint R&D In line with the tradition of Neo-Schumpeterian economics, there has been proposed a thesis of leapfrogging by Perez and Soete (1988).⁵ This idea of leapfrogging emphasizes the importance of utilizing emerging technological opportunities in the process of catching up. Perez and Soete (1988) focused on how a catching-up country, not being bound by costly investment in capital goods and the infrastructure of the old paradigm, can leapfrog into a new technological paradigm ahead of advanced countries. Seen from this view, the emergence of digital technology, since the 1990s, has also offered an opportunity for latecomers to try leapfrogging. Actually, in the mid 1990s, Korean companies emerged as world leaders in several innovative digital products (Lee et al. 2005). Korea was the first country in the world to develop the CDMA-based (Code Division Multiple Access) digital mobile telecommunication. Also, it was via an LG product that the UK enjoyed its first digitally broadcast TV programmes, and via Samsung products that Americans watched the historic launch of the space shuttle Discovery. Samsung and LG command numerous world firsts in terms of technologies and licences in related fields of digital technology. Since the late 1990s, Samsung and LG have enjoyed top market shares in digital TVs, both in the UK and in the USA. At the time of writing, the absolute majority of TV exports by Korea is of digital TVs, which have replaced analogue TVs. This signifies the shift from analogue to digital goods as Korea’s main export item. Here, we will consider the story of the emergence and growth of the digital TV industry in Korea and, thereby, examine the role of industrial policy in this episode of leapfrogging into digital TV by Korean firms. The period of analysis is from the early 1990s to 2002–03, and here I rely heavily on Lee et al. (2005), which is a detailed case study. Initial actions toward high definition television, HDTV, by the Korean government and firms were heavily influenced by the Japanese lead in analogue HDTV. A Japanese research group came to Korea during the 1988 Seoul Olympic games and staged a promotional tour of their achievements in the hope that the Koreans would follow their way, as in the past. Recognizing that HDTV would be a next-generation, hot consumer item with immense technological and market potential, the Korean government first established the Committee for Co-Development of HDTV in 1989. This committee ⁵ This subsection relies on Lee et al. (2005).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

had representatives from three ministries (Ministry of Commerce, Industry and Energy; Ministry of Information and Communication; and Ministry of Science and Technology) and seventeen institutions comprising private firms, government research institutes (GRIs), and universities. The Korean government wanted to promote HDTV as one of the most important export items for the next generation, for the twenty-first century. The government initiated a grand research consortium for HDTV. It was led by the Video Industrial R&D Association of Korea, the Korea Electronics Technology Institute (KETI), and the Korea Institute of Industrial Technology (KITECH), joined by Samsung, LG, Hyundai, Daewoo Electronics, and other private firms. The Video Industrial R&D Association of Korea assumed the supervision of the progress of all the research projects. It evaluated technical aspects of the projects, coordinated opinions among firms involved in the R&D consortium, and collected research proposals and details on the progress of each of the research projects from the firms. Administrative work for the whole research project was carried out initially by the Korea Institute of Industrial Technology (KITECH) and later, by Korea Electronics Technology Institute (KETI), a spin-off institute from KITECH. The administrative work included preparing reports for the progress of the research project and for reporting details of R&D expenditures and administrative work for technical licensing fees. In addition, KITECH and ETRI carried out both the coordination of smaller consortiums and R&D in two specific fields of the whole project. The research project’s first task was to interpret and absorb the foreign knowledge and eventually to develop HDTV sets. The total budget for the five years, 1990–94, was 100 billion Korean won (roughly US$100 million), with the government and the private sector to each pay half of the total. Right after the start of the Korean project, General Instruments (GI), a leading American firm in digital TV technology, staged a historic demonstration of the possibility of digital TV in 1990. The head of the research team at GI was a Korean American, named Dr Woo-Hyun Paik, who later joined LG Electronics, in 1998, as the CTO (Chief Technology Officer). Now, with the Korean research project for HDTV decisively underway, in spring 1991, digital HDTV targeted US markets, leaving behind Japanese- or European-led analogue HDTV. The problem was that the US standard had not yet been determined. In this regard, one interesting strategy by the Korean team was the decision to develop several alternative standards simultaneously, with different private companies in charge of different standards. At that time, four leading standards were identified in the USA. Thus, Samsung was chosen or assigned to develop the standard set by GI and the MIT coalition; LG that of the Zenith and AT&T coalition; Daewoo that of RCA; and Hyundai the standard set by Farouja. The public–private coalition encouraged private firms to stick to these risky R&D activities by channelling R&D funds and forming a network of researchers from firms, universities, and GRIs. In the project, there was a clear division of labour among the participating units. The whole project was divided into digital signalling (satellite and terrestrial), display (CRT, LCD, PDP), and ASIC chips (application-specific integrated circuits chips, encoding, decoding, demultiplexer, display processor). Each unit, GRI,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



or private firm, was assigned to different tasks with some intentional overlaps among them; namely, two units took the same task to avoid the monopoly of research outcomes. This government-led consortium had the effect of providing private companies with the legitimacy of the project; and without this, the companies admitted, their project would have stopped because they could not have just kept pouring money into a project with uncertain cash outcomes. Furthermore, the consortium provided the firm’s R&D team with the opportunity to meet and collaborate with university and other public sector researchers. The R&D staffs, during a subsequent interview, acknowledged that what had been particularly helpful was the interaction with university professors—especially those who had just returned from the USA with PhD degrees in digital technology-related fields.

24.4 C R  I  A

..................................................................................................................................

24.4.1 The Korean Experience of Financing Industrial Development For an effective industrial policy, the state’s ability to control financial resources in a national economy is often critical. Financial control implies more discretionary control, such that the state, with its power in credit allocation, can control not only the financial ability of firms, but can also assure the firm’s compliance in other matters, such as industrial policy implementation. In the Korean experience, the banking sector was intended to ‘serve’ the real sector by providing a stable supply of so-called ‘growth money’ at affordable rates, whereas the manufacturing or production sectors always had been given priority. Of course, this practice was possible because Korea established several development banks, such as the Korea Development Bank, Ex-Im Bank, and the Industrial Bank. Also, most of the commercial banks were government-owned until the mid-1980s and continued to be influenced by the government even after privatization. Additionally, manufacturing sectors often earn rents, due to entry control exerted by the government in adjusting the ‘optimal number of the firms’ in each sector, considering the market size so that admitted firms may be guaranteed a minimum level of sorts for profits (rents) that can be invested in the next period. Thus, making the rate of return in certain industrial sectors higher than interest rates is one of the tools for industrial policy, especially when relief from high interest rates is needed. Diverse cases of industrial policy and financing may have some policy implications for economies in Africa, which are trying to build their industrial bases. Tools of policy and financing can be different, depending upon the nature of the sectors and projects. For a project like physical infrastructure, or those with strong externality, the practices

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

of POSCO in the Korean steel industry may be applicable. More direct intervention, in the form of SOEs, can be justified. Building oil or gold refineries in Africa can be accomplished by using these kinds of SOEs, which can be privatized later, as in the case of POSCO. Korean Air, the top airline in Korea, was also an SOE. For targeted development of certain technologies in Africa, especially for medium-sized enterprises (MEs), the bottom-up approach taken in the IBTDPs, and executed in Korea’s economic past, can have useful implications, in terms of how to identify ‘needed technologies’ by conducting firm surveys and arranging for diverse financing tools. Finally, in an effort to break into newly emerging sectors or businesses, the public–private joint R&D or foreign–domestic joint R&D practised in Korea’s past can be a useful device of industrial policy for the necessary sharing of knowledge, funds, and risks.

24.4.2 External Imbalances and Industrial Policy for Export Manufacturing in Africa It is not surprising that many countries in Africa at low-income stages have had trade deficits for many years. That is basically due to weak export capabilities, compared with an ever-strong demand for imported goods in African economies. Korea also went through three decades of trade deficits, until it recorded its first trade surplus in 1986; since then, it has maintained a trade surplus (Lee 2013b). Korea, in the early 1960s, had a 1 to 9 ratio of exports to imports, which is much worse than a typical country in Africa. Thus, Korea had a huge savings gap with domestic savings at only 9 per cent of GDP and gross investment at 15 per cent of GDP, thus relying on foreign borrowing to fill the gap. This illustrates why exports are so important and are the critical binding constraints for growth for an economy at lower or middle income stages. Given that getting out of a trade deficit may take several decades, a country at a lower income stage may find it necessary to take transitory measures to manage the balance of payments. In looking for specific policy tools, the past experience of Korea could be useful. In the 1960s and 1970s, Korea maintained tight centralized control on foreign exchanges within the economy, with all export earnings (foreign currencies) first put under the control of the government (Bank of Korea), and then allocated for ‘justifiable uses’, like payment for imports of capital goods (Amsden 1989). One of the reasons for the tight control of foreign exchange under the closed capital market in the early period had to do with the fact that export promotion and free capital mobility cannot work together; export promotion often involves undervaluation of currencies (or typical economic conditions in emerging economies tend to involve frequent depreciation), which is a signal or incentive for people to take their money abroad (or put their money in foreign currency-dominated bank accounts). In these practices, imports of ‘non-necessaries’ such as luxury consumer goods, tended to be discouraged by high tariffs, diverse non-tariff barriers, or social campaigns, and it was difficult to get permission to use dollars. For instance, even imports of foreign fruits (e.g. bananas) was discouraged by high tariffs or non-tariff barriers.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



In general, tariffs tended to be low for capital goods while very high for consumer goods, which Korea aimed to promote for exportable goods—which was termed asymmetric protection in Shin and Lee (2012). Such protection was found to have significant impacts, not on TFP (total factor productivity) changes, but on the volume and market shares of Korean export products. These practices also meant that there was tight control of capital outflow (capital flight); for instance, ordinary people could not have their bank accounts in foreign countries, and foreign banks were not allowed to open business in Korea until the late 1980s. Despite low income and, thus, low domestic savings, Korea maintained a higher investment rate, and one of the reasons for this was the low interest rates, with rate hikes suppressed by the government. Despite this suppressed interest ratio, the domestic savings ratio in Korea continued to increase, owing to the growth of income associated with strong investment over the decades. This experience may have some implications for African countries, including Uganda where interest rates are currently very high, over 24 per cent, in spite of inflation rates not being that high, whereas very low interest rates are applied to savings deposited in banks. This situation is very bad for private investment and reflects the asymmetric power and dominance of the lender over borrower, and also the dominance of the banking sector over the real sector. If both sides have equal power, interest rates for savings should also be high. In other words, financial markets appear to be oligopolistic and unbalanced in the power of supply and demand, and can be said to be a state of market failure—which may justify some form of government intervention. In other words, the banking sector is earning extra rents associated with oligopoly. This is quite the opposite of the desirable state of the productive sector enjoying rents, as in Korea’s past, where the banking sector had always been tasked to ‘serve’ the real sector by providing a stable supply of so-called ‘growth money’ at affordable rates, and the manufacturing or production sector had always been given priority.

24.4.3 Dilemma and Prospects of the Resource-based Development in Africa In situations in many African countries, such as Uganda, despite competitive exchange rates (undervaluation or deprecation), exports tend not to respond. This situation is not particularly surprising, because competitive exchange rates would only work in an economy with a strong manufacturing basis. Relatedly, Ramanayake and Lee (2017) find a negative effect of undervaluation on growth in mineral-exporting groups, and positive (no significant) effects of undervaluation in manufacture-exporting groups. This finding is consistent with the fact that if currency is more undervalued in countries that depend greatly on natural resource exports, then less income is earned in terms of dollars, because natural resource exports are often insensitive (inelastic) to exchange rates. Thus, there is an important contrast between manufacture- versus

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

mineral-exporting countries, such that depreciation often tends to exert countercyclical effects of recovering exports and growth in economies with a strong manufacturing base (or non-negative effects on average), which is not the case in mineral-exporting economies. These mineral-exporting economies face the growth-impeding and procyclical effects of undervaluation during times of weak performance of an economy with a typical balance-of-payment crisis. The growth-impeding and procyclical effects of undervaluation underscore the difficulties facing economic growth in mineralexporting economies and, thus, the dilemma of the so called resource-based development model. In other words, the nature of the curse is not only a symptom associated with the Dutch disease, but it also means being stuck in the resource-based sector with little chance of entering manufacturing, due to the countercyclical effects of the low valuation of currencies. Therefore, while entry into, and promotion of, manufacturing sectors would be a desirable long-term development goal for typical countries in Africa, the condition of already-free capital mobility and already-privatized banking sectors indicates that the role of the government in promoting manufacturing would have limited impacts, except in a few countries such as Ethiopia. Low valuation of currency would lead to capital flight and less domestic savings available for investment, and control of interest rates for boosting investment in industrial sectors is not especially feasible under the private (or foreign) dominance of commercial banking. The situation of Kenya, which recently tried a form of interest ceiling, indicates the dilemma. If domestic effort to promote exports is limited, FDI is, of course, an option but attracting FDI in the manufacturing sector has not been easy in many African countries. If this is the case, a more radical or innovative idea, for instance, for a country like Uganda, might be leapfrogging into IT service or ‘Smart Agriculture’, bypassing the stage of manufacturing. A case of leapfrogging has been taking place in India, which bypassed manufacturing to leapfrog into IT service as the engine of growth (Lee 2013c: 178–205). There is also a growing recognition that agriculture is no longer a traditional industry but a ‘high-tech’ sector, now called the sixth industry, as a combination of the primary, secondary, and tertiary industries. It is combined with IT or digital technologies, as it braces for the benefits of new innovations, recently associated with the so called Fourth Industrial Revolution. An example would be the Netherlands, which is leading ‘Smart Farming and Dairy’. In 2015, its export value in agriculture was the second largest in the world, or 438 billion euro, with a share of 20 per cent in the total exports of the country. Agriculture may be a more attractive sector to attract FDI than manufacturing in some African economies, like Uganda, in terms of its comparative advantages. Of course, the agro-food industry and processing segment of the primary sector industry can also be a good option for industrial development. In this regard, a good example is the case of a brand of coffee company, called ‘Good African Coffee’, established by an entrepreneur from Uganda named Rugasira (2013), which is already successful in the global market with its brands and sales network in Europe and North America. This case is important because this company does not export crude or unprocessed coffee, but high-valued, processed, and branded coffee.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



R Amsden, Alice, 1989. Asia’s Next Giant: South Korea and Late Industrialization, Oxford: Oxford University Press. Cho, Yoon Je, 1997. ‘Government Intervention, Rent Distribution and Economic Development in Korea’, in M. Aoki, H.-K. Kim, and M. Okuna-Fujiwara, eds, The Role of Government in East Asian Economic Development, Oxford: Oxford University Press, pp. 208–32. D’Costa, Anthony P., 1999. The Global Restructuring of the Steel Industry: Innovations, Institutions, and Industrial Change, London and New York: Routledge. Export–Import Bank of Korea, 1996. 20 Years of History of Export–Import Bank of Korea (in Korean), Seoul: Export–Import Bank of Korea. Hellmann, T., K. Murdock, and J. Stiglitz, 1997. ‘Financial Restraint: Towards a New Paradigm’, in M. Aoki, H.-K. Kim, and M. Okuna-Fujiwara, eds, The Role of Government in East Asian Economic Development, Oxford: Oxford University Press, 163–207. IBK (Industrial Bank of Korea), 2011. 50 Years of History of Industrial Bank of Korea (in Korean), Seoul: Industrial Bank of Korea. Johnson, Chalmers, 1982. MITl and the Japanese Miracle: The Growth of Industrial Policy, 1925–1975, Stanford, CA: Stanford University Press. KDB (Korea Development Bank), 2014. 60 Years of History of Korea Development Bank (in Korean), Seoul: Industrial Bank of Korea. Korea Industrial Technology Evaluation Institute, 2007. Evolution of Industrial Technology Policy: With a Focus on the 20 Years of Industrial Technology Development Projects (1987–2006) (in Korean), Seoul: Korea Industrial Technology Evaluation Institute. Korean Economy Compilation Committee, 2010. The Korean Economy: Six Decades of Growth and Development, Vol. 1 (in Korean), Seoul: Korean Economy Compilation Committee. Lee, Keun, 2013a. ‘How Can Korea Be a Role Model for Catch-up Development? A Capability-based View’, in Augustin K. Fosu, eds, Achieving Development Success: Strategies and Lessons from the Developing World, Oxford: Oxford University Press, pp. 25–49. Lee, Keun, 2013b. Schumpeterian Analysis of Economic Catch-up: Knowledge, Path-creation, and the Middle Income Trap, Cambridge: Cambridge University Press. Lee, Keun, 2015. ‘Capability Building and Industrial Diversification’, in Jesus Felipe, ed., Development and Modern Industrial Policy in Practice: Issues and Country Experience, Cheltenham: Edward Elgar Publishing, pp. 70–93. Lee, Keun, 2016. Economic Catch-up and Technological Leapfrogging: Path to Development and Macroeconomic Stability in Korea, Cheltenham: Edward Elgar Publishing. Lee, K. and J. H. Ki, 2017. ‘Rise of Latecomers and Catch-up Cycles in the World Steel Industry.’ Research Policy, 46 (2), pp. 365–75. Lee, K. and B. Y. Kim, 2009. ‘Both Institutions and Policies Matter but Differently for Different Income Groups of Countries: Determinants of Long-run Economic Growth Revisited’, World Development, 37 (3), pp. 533–49. Lee, K. with Y. Kim, 2016. ‘Technological Catch-up in the Capital Goods Sector’, in Keun Lee, ed., Economic Catch-up and Technological Leapfrogging: Path to Development and Macroeconomic Stability in Korea, Cheltenham: Edward Elgar Publishing, pp. 231–49. Lee, K. with H. Y. Lee, 2016. ‘Historical Origins and Initial Conditions for Economic CatchUp’, in Keun Lee, ed., Economic Catch-up and Technological Leapfrogging: Path to Development and Macroeconomic Stability in Korea, Cheltenham: Edward Elgar Publishing, pp. 15–29.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Lee, K. and C. Lim, 2001. ‘Technological Regimes, Catching-up and Leapfrogging: Findings from the Korean Industries’, Research Policy, 30 (3), pp. 459–83. Lee, K. and J. A. Mathews, 2010. ‘From Washington Consensus to BeST consensus for World Development’, Asian-Pacific Economic Literature, 24 (1), pp. 86–103. Lee, K., C. Lim, and W. Song, 2005. ‘Emerging Digital Technology as a Window of Opportunity and Technological Leapfrogging: Catch-up in Digital TV by the Korean Firms’, International Journal of Technology Management, 29 (1–2), pp. 40–63. Nam, Duck-Woo, 2009. On the Road to Economic Development (in Korean). Seoul: Samsung Economic Research Institute. Nam, J., 1979. Characteristics and the Structure of Supply and Demand of the Steel Industry (in Korean), Seoul: Korea Development Institute. Park, Y. C., 1990. ‘Development Lessons from Asia: The Role of Government in South Korea and Taiwan’, American Economic Review, 80 (2), pp. 118–21. Perez, C. and L. Soete, 1988. ‘Catching-up in Technology: Entry Barriers and Windows of Opportunity’, in Giovanni Dosi, Christopher Freeman, Richard Nelson, Gerald Silverberg, and Luc Soete, eds, Technical Change and Economic Theory, London: Pinter Publishers, pp. 458–79. Ramanayake, Sanika and Keun Lee, 2017. ‘Differential Effects of Currency Undervaluation on Economic Growth in Mineral- vs. Manufacture-Exporting Countries: Revealing the Source of the Vicious Procyclicality in the Resource-cursed South’, in Jorge Niosi, ed., Innovation Policy, Systems and Management, Cambridge: Cambridge University Press. (Paper presented at the 2016 Conference of the International Schumpeter Society, Montreal, Canada.) Rodrik, D., 1996. ‘Understanding Economic Policy Reforms’, Journal of Economic Literature, 34 (1), pp. 9–41. Rugasira, A., 2013. A Good African Story: How a Small Company Built a Global Coffee Brand, London: Bodley Head. Shin, Hochul and Keun Lee, 2012. ‘Asymmetric Protection Leading Not to Productivity but Export Share Changes: The Case of Korean Industries, 1967–1993’, Economics of Transition, 20 (4), pp. 745–85. Son, Sang-Ho, 2013. Assessment, Analysis and Prospect of Korea’s Policy Loan (in korean), Seoul: Korea Institute of Finance. Song, Sungsoo, 2002. ‘The Historical Development of Technological Capabilities in Korean Steel Industry: Posco from the 1960s to the 1990s’. PhD Thesis (Program in History and Philosophy of Science). Seoul: Seoul National University.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

        ......................................................................................................................

                 Structural Transformation and Industrial Policy ......................................................................................................................

- 

25.1 I

.................................................................................................................................. T should be no doubt that Taiwan has been a star performer economically in the post-war period. According to Maddison (2010),¹ Taiwan’s per capita income in 1950 was only $916, measured in 1990 international dollars (PPP), which was far lower than that of not only Latin America but also the Philippines, and less than one tenth of that of the USA. After more than half a century, by 2008, in Maddison’s estimates, Taiwan’s per capita income had increased 22.8-fold. It hence ranked along with South Korea at the top of developing countries in terms of the rate of growth in this post-war period.² Therefore, by 2014, this exceptional growth record has made Taiwan reach the rank of sixteenth in the world in terms of per capita PPP GDP, surpassing many of the OECD countries.³ Looking back historically, the speed and magnitude of Taiwan’s post-war development indeed looks impressive. When the Japanese colonists were forced to withdraw from Taiwan due to Japan’s defeat in 1945, Taiwan was still a typical colonial economy relying upon exporting sugar and rice to a protected Japanese market, which vanished after Japan’s defeat. Over 90 per cent of Taiwan’s exports were primary goods in 1939. The industrialization after 1937 was to support Japanese military activities in the south ¹ This is from Maddison (2010). The units are 1990 International Geary-Khamis dollars. ² This calculated from Maddison (2010), using each country’s per capita GDP of 2008 divided by that of 1950. The ranking excludes the top-ranked Equatorial Guinea, which is an oil-rich country with less than one million people. ³ According to IMF World Economic Outlook (April 2015), Taiwan’s PPP per capita GDP was $47,899, about 85 percent of the USA’s ($56,421). http://www.imf.org/external/pubs/ft/weo/2015/01/ weodata/weoselgr.aspx

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

Table 25.1 Major economic indicators I, 1951–2014 Average annual growth rates of Year 1951–60 1961–70 1971–80 1981–90 1991–2000 2001–14 1951–2014

Real GDP

Population

GNP per capita*

Gross capital formation**

Industrial production***

Exports

CPI

8.1 9.7 9.8 7.6 6.3 3.6 7.3

3.6 3.1 2.0 1.4 0.9 0.4 1.8

4.5 6.8 7.7 6.4 5.1 3.3 5.5

14.1 15.4 13.9 7.3 7.5 0.3 9.1

11.9 16.5 13.8 6.2 5.1 4.3 9.4

22.1 26.0 29.5 10.0 10.0 5.5 16.3

9.8 3.4 11.1 3.1 2.6 1.1 4.8

Note: *only till 2013. **Figure before 1969 was deflated by indexes with 1986 as the base; those afterwards were by indexes with 1996 as the base. ***Figures for 1995 and before exclude quarrying. Source: (1) DGBAS, http://www.dgbas.gov.tw/mp.asp?mp=1 http://61.60.106.109/task/sdb http:// www.dgbas.gov.tw/ct.asp?xItem=9522&ctNode=2857 (2) Taiwan Statistical Data Book, various years; (3) DGBAS, http://www.dgbas.gov.tw/ct.asp?xItem=9522&ctNode=2857

Pacific, and the plants were all owned and managed by Japanese.⁴ Indigenous industrial development would be a post-war phenomenon. Luckily, in 1945, Taiwan’s traditional exports were able to swiftly turn to the Chinese market in place of the Japanese one. However, just four years later, the export market disappeared again after the Nationalist regime was defeated by the Communists and had to retreat to Taiwan at the end of 1949. Taiwan had to find other outlets for its exports. Fortunately, Taiwan was able to embark on a path of sustained industrialization in the early post-war period, so that it managed to gradually reduce its dependence on primary exports. The share of rice and sugar in Taiwan’s exports declined from 74 per cent in 1952, to 47 in 1960, 22 in 1965, and a mere 3.2 per cent in 1970,⁵ clearly demonstrating the fruits of industrialization in the first twenty post-war years. Taiwan has been able to grow from a low income to a high income economy in the postwar decades because it managed to sustain its quick pace of development throughout this period. Its GDP and per capita GNP grew at an average annual rate of 9.2 and 5.8 per cent respectively in the first thirty post-war years, 1951–80, and 6.3 and 4.9 per cent from 1981 to 2014 (see Table 25.1). It faced various challenges along the way, but managed to adapt to the new environment and transform itself at every turn. That is, it has successfully undergone several rounds of structural transformation. For example, recently, after successful ⁴ For discussion of the Japanese colonial period, see Ho (1978); Gold (1981); Cumings (1984); and Myers and Peattie (1984). ⁵ See Chu (2017).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



upgrading, its high-tech industry has been the major manufacturer of information technology (IT) and communication products in the world in the last two decades.⁶ This chapter will examine how Taiwan managed to transform its economy in the post-war era. The discussion will follow a chronological order, from importsubstitution industrialization in the 1950s, to export promotion and secondary import substitution in the 1960s and 1970s, to entry into the high tech sector from the 1980s, and to liberalization and globalization in the 1990s. It will be shown that at every turn the transformation was successful because it was facilitated by suitable industrial policies, which were adaptive to the changing situation.

25.2 I S   1950

.................................................................................................................................. At the end of World War II, when the Nationalists came to Taiwan in 1945, they took over an economy badly affected by the war, with latent inflationary pressure. The Japanese export market disappeared overnight, and was replaced by the Chinese market. During 1945 to 1949, although the Nationalist government managed partly to recover local industrial production, especially utilities, Taiwan’s economy nonetheless was affected by the ensuing civil war and hyperinflation on the mainland. Preparing for the central government’s eventual retreat to Taiwan at the end of the year, Taiwan’s Provincial government, under Chen Cheng, undertook several important policy measures in 1949. Chen implemented initial land reform in June 1949, fixing the rent at 37.5 per cent of output, so as to stabilize the rural areas. He also carried out monetary reform, issuing a new currency, the New Taiwan Dollar, using gold reserves shipped from the mainland as support. Martial law and other strict political control measures were put in place as well. In the short term, the monetary reform was relatively effective, so that annual inflation rate dropped from 1360 per cent in the first half of 1949 to 66 per cent in the second half after the reform.⁷ In the first few years after the Nationalist government retreated to Taiwan in December 1949, it took further measures to successfully stabilize the economic and political order. The odds seemed formidable, including political uncertainty, a huge influx of immigrants from the mainland, persistent fiscal and balance of payment deficits, and chronic shortages of foreign exchange and material resources. However, probably due to lessons learned from political and economic governance failures on the mainland, the Nationalist government reformed itself, and began to manage its economic affairs prudently. Moreover, other than prudent fiscal management, the Nationalist government needed to have tremendous amounts of material resources to establish economic stability. For that, it relied first on gold reserves to shore up the value of the currency, ⁶ This will be discussed in Section 25.5.

⁷ This section draws upon Chu (2017).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

and then on US aid, which arrived after the Korean War broke out in 1950 and continued until 1965. The US military and economic aid helped to stabilize the Nationalist rule both economically and politically. During this period, economic aid amounted to approximately US$ 1.5 billion in total, which was almost equal to that of the total balance of payment deficit.⁸ As mentioned above, to consolidate Taiwan as their last bastion, the Nationalists implemented rent reduction immediately in 1949, fixing the rent at 37.5 per cent of output. In 1953, they pushed further and implemented the Land-to-the-Tiller programme. The compensation given to landlords consisted partly of shares in the four industrial state-owned enterprises (SOEs). Consequently, the local elite could no longer rely upon rural rents for a living and had to engage in modern business. Moreover, there were two other factors which helped to ensure the lasting success of land reform. One was that the government paid careful attention to the agricultural sector, assuring the farmers adequate supply of the necessary inputs for production. The second factor was that the newly emerged industrial sector provided opportunities for those elite who had to leave the rural sector. The rise in agricultural productivity and output helped to secure ample food supply for the enlarged population and to keep the wage level low for the industrial sector. It also contributed to an improvement in income distribution. Moreover, making industrialization possible required that the government be able to extract surplus from a very productive agricultural sector. Because the agricultural sector had made significant gains in productivity, and land reform had redistributed income in favour of tenants, meant that the sector was better able to bear the heavy fiscal burden. Furthermore, the former tenants’ newly acquired land had the potential to be turned into capitalized assets in the modernization process. In a way, the success of the land reform started a virtuous cycle. Taiwan’s post-war industrialization has no doubt been very much a state-led development. In that early period, the Nationalist government had almost all the essential policy tools at its disposal, especially the right to allocate US aid and foreign exchange to finance industrial projects. And the government indeed used these tools to single-mindedly promote economic growth.⁹ Right from the outset, industrialization was the clear objective. In the 1950s, besides trying to restore the economy to its pre-war level, the government promoted importsubstitution industrialization due to severe foreign exchange constraints. During 1951–53, it push-started the cotton textile industry, the main target industry, by bearing most of the risks and responsibilities itself. A small number of other industries, including utilities, fertilizers, and some consumer essentials, were also targeted and enjoyed prioritized allocation of resources. It should be noted that the Nationalists did not expand the SOE sectors and instead promoted private enterprises. Although most of the non-agricultural US aid went to ⁸ Chao (1985: 8). ⁹ For discussions of major policies, see Ho (1978); Amsden (1979); Gold (1981); Wade (1990); and Chu (2017).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



support the large SOEs, especially the utilities and transportation, a significant part of it was used to promote new manufacturing industries. Most of the projects went into private hands, including the current top conglomerate, Formosa Plastic Company. In these industrial projects the government played the role of entrepreneur, drafting the investment plans from scratch and handling them all the way up to handover to the would-be private industrialists. The share of private enterprises in manufacturing output, therefore, increased from 41 per cent in 1952 to 70 per cent in 1966.¹⁰ Despite retreating to Taiwan, Chiang Kai-shek intended to ‘recover’ the mainland eventually. Paradoxically this provided the strong political will necessary for supporting the post-war developmental projects. The economic bureaucracy, also a legacy from the mainland period, enjoyed great autonomy in promoting industrialization under authoritarian rule. As a result, the economy quickly recovered and stabilized in the early 1950s, and began to embark on the route to rapid industrialization, which has continued to the present time. Section 25.3 will discuss the subsequent change in policy. That is, the policy regime switched from import substitution to export promotion in 1958, and rapid export growth of manufactured goods followed.

25.3 P R   S  E- G  1958

.................................................................................................................................. Regarding ways to lessen the foreign exchange constraint, as well as import control and import substitution industrialization, promoting exports, if feasible, could be a more effective way.¹¹ In addition, the scale of the domestic market was obviously too small to allow industry-level scale economies to be realized and thereby sustain growth. For example, the cotton textile industry reached self-sufficiency within just two years and began to accumulate excess capacity. However, the foreign exchange regime was designed to facilitate import substitution, and had over-valued exchange rates and established a complicated set of multiple exchange rates. The government then had to design schemes, that is, get the prices ‘wrong’,¹² as a way to encourage firms to export. With the benefit of hindsight, the switch to an export-promotion policy regime seemed a logical next step for a government eager to find ways to sustain growth and push industrialization. However, due to the fear of unforeseen risks and resistance from vested interests, reform took place only after a prolonged round of heated debate ¹⁰ CEPD, Taiwan Statistical Data Book, various years. ¹¹ This section draws upon Lin (1973: chs 4–6) and Chu (2017). ¹² Amsden (1989) coined the term, ‘getting the prices wrong’, in her seminal work on South Korea’s post-war economic development. The term means that the latecomer state has to provide subsidies to the disadvantaged latecomer firms so as to alter the prevailing market prices in order to induce the latecomer firms to embark on the learning process.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

among the economic bureaucrats and the ruling elite. Nonetheless, the Foreign Exchange Reform did take place and was successfully implemented in two steps in 1958, converting the multiple exchange rates into a unitary rate, devaluing the currency significantly, and adopting various export promotion programmes. Furthermore, to promote overall economic development, the government enacted a 19-Point Program for Economic and Financial Reform and an important Statue for Encouraging Investment in 1960. The Statue remained in effect until 1990 when it was replaced by the Statue for Promoting Industrial Upgrading. It put in place the framework needed to reduce investment barriers and to provide tax favours to investors. The policy regime switch was not as drastic as it seemed, however, because the extent of trade liberalization was rather limited, and the domestic market continued to be largely protected. The new policy has more to do with subsidizing exports than trade liberalization. Nonetheless, exports indeed started to grow very rapidly under the new incentive structure. Taiwan’s cotton textile products began to be subject to import restraints in the USA market as early as 1962. The textile industry, including the cotton and man-made-fibre sub-sectors, continued to be the leading sector at the outset of the export-led growth. This occurred long before apparel exports began to take off in the late 1960s, showing the beneficial effects of import substitution (see Table 25.2).¹³

Table 25.2 Major economic indicators II, 1952–2014 GDP by industry (%) Gross fixed Real GDP per capital formation Exports as a Trade balance % of GDP (US$million) Agriculture Industry Services Year capita (US$) as a % of GDP 1952 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2014

213 164 229 393 978 2,385 3,290 8,124 12,918 14,704 16,051 18,503 22,648

11.3 16.6 17.0 21.7 31.3 30.7 19.5 23.1 25.7 24.4 21.9 20.7 21.0

8.0 11.5 19.4 30.4 39.9 52.6 52.5 44.5 46.3 52.2 61.0 71.5 68.1

71 133 106 43 643 78 10,678 12,639 9,330 11,218 15,817 23,364 39,670

32.2 28.5 23.6 15.5 12.7 7.7 5.7 4.0 3.3 2.0 1.7 1.6 1.9

19.7 26.9 30.2 36.8 39.9 45.7 44.6 38.9 33.1 30.5 31.3 31.1 34.1

48.1 44.6 46.2 47.7 47.4 46.6 49.7 57.0 63.5 67.5 67.1 67.2 64.0

Sources: (1) DGBAS, http://www.dgbas.gov.tw/mp.asp?mp=1 http://61.60.106.109/task/sdb http:// www.dgbas.gov.tw/ct.asp?xItem=9522&ctNode=2857 (2) Taiwan Statistical Data Book, various years.

¹³ Chu (2008).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



25.4 S I S

.................................................................................................................................. Changing the policy from import substitution to export promotion, however, did not imply that the government intended to stop practising industrial policy to promote industrialization. On the contrary, the government used a secondary import substitution programme to establish upstream production to supply inputs to the exporting downstream industries. Nonetheless, protection schemes came with strings attached; that is, there were time limits and contingent price and quality conditions.¹⁴ Actually, the promotion of upstream industry was part of the plan, even when the light industries began to grow in the 1950s. A plant to manufacture man-made fibre was established with government help in the mid-1950s. The automobile industry made a start in 1956. Plans for the steel and petrochemical industries began to be discussed in the 1950s as well. Due to difficulties in obtaining technology and capital, the first naphtha cracking plant did not begin operation until 1968, and the first integrated steel mill began construction only in the early 1970s. Both were undertaken by SOEs, socializing investment risks deemed unbearable by the private sector at the time. All these were part of the secondary import-substitution policy, which was to promote industrial deepening. In the 1970s, the level of US support, which had been crucial for the survival of the Nationalist government on Taiwan, began to lessen. US–PRC relations started to thaw, although the USA and PRC did not establish diplomatic relations until January 1979. This created a legitimacy crisis for the Nationalist regime. In addition, the first oil crisis in 1973 brought about an economic crisis around the same time. In response, the government enacted the Ten Construction Projects from 1974 to 1979, most of which were in the plan anyway. The projects included six major infrastructure projects, one nuclear power plant, and three industrial projects: the integrated steel mill, and expansion of the petrochemical plants and shipyards. These helped to stimulate the economy in the short term, to build up infrastructure, and to sustain and deepen industrialization in the long term.

25.5 E  H T

.................................................................................................................................. As soon as the plan for heavy industry was more or less in place in the early 1970s, the government began to plan for the next growth industry, that is, electronics. Adopting a different policy approach this time, the government set up the National Science Council and public research laboratories, such as the Industrial Technology Research ¹⁴ Chu (2001). This is similar to the Korean case as described in Amsden (1989).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

Institute (ITRI), in the early 1970s, and started the first IC project in 1976. Later these consecutive IC projects were spun off from ITRI, and the spin-offs, mainly United Microelectronics Corporation in 1980 and Taiwan Semiconductor Manufacturing Company in 1987, now comprise the main part of Taiwan’s integrated circuit (IC) industry. This policy environment had also supported the spectacular growth of Taiwan’s IT industry. A policy network was in place to develop locally produced key components after local production of mature IT products became possible. Thus, by now, the majority of the world’s IT products are made by Taiwanese firms. Taiwan’s industrial prowess operates mostly behind the scenes, because its leading firms are mainly subcontractors for firms in advanced countries. In recent years, due to successful industrial upgrading, Taiwan has become one of the world’s largest producers of IT products, semiconductors, liquid-crystal-display units, and man-made fibres. Taiwan’s information and communication technology (ICT) products continue to occupy a substantial share of the world market in 2014, for example notebook computers (85 per cent), motherboards (85 per cent), tablets (39 per cent), servers (86 per cent),¹⁵ IC foundry (71 per cent) and IC design (22 per cent).¹⁶ Amsden and Chu (2003) have studied how Taiwan managed to upgrade and enter into high-tech in recent years. The entry strategy of Taiwanese firms into the high-tech industry has been one of playing second mover or doing subcontracting. Lacking frontier technology, the firms enter when the product becomes mature, and earn profits based upon efficient and low-cost manufacturing and timely delivery. They have to absorb the technology and expand production quickly. These firms have mostly relied upon locally trained engineers, as well as some returnees from abroad. Support from the education system, accumulated manufacturing experiences, and local production networks provided the necessary conditions for the emergence of these firms. On the other hand, the government’s industrial policy helped to set up the right environment and the crucial institutions, and assisted the advancement of the industry along the way. As a result, the main players in Taiwan’s IT industry are the large nationally owned firms, not foreign capital, and they have been able to capture a large share of global IT production, as shown above. Domestically, the share of the IT and electronics sector in total manufacturing value added rose from around 13 per cent in 1990 to 46 per cent in 2014 (Table 25.3).¹⁷ In the meantime, an increasing portion of offshore production shifted to mainland China, reaching 92 per cent in 2014.¹⁸ Moving production to China allowed Taiwanese firms to have access to an almost unlimited supply of cheap and efficient labour, hence allowing them to greatly expand their scale of operations.

¹⁵ Market Intelligence and Consulting Institute (MIC), 2015, Information Industry Yearbook 2015, p. 15. ¹⁶ ITRI, 2015, Semiconductor Industry Yearbook 2015, 6–17, 6–30. ¹⁷ MOEAa (2015). ¹⁸ MIC (2015: 16).

Table 25.3 Distribution of manufacturing value-added, 1971–2014 Unit: %

sum

1975

1980

1985

1990

2000

2010

2014

12.83 23.23 4.32 3.29 8.75 5.35 9.42 3.97 6.2 1.06 4.2 11.86 4.3 1.22

11.94 23.86 2.88 2.6 9.2 4.91 8.71 3.7 5.31 1.11 3.05 16.81 4.71 1.21

7.85 18.74 2.65 3.50 8.87 7.99 8.37 4.33 5.15 5.67 3.96 13.25 6.55 3.03

8.44 17.49 4.31 3.54 8.76 5.32 9.80 3.97 6.71 5.24 3.73 13.80 5.75 3.14

6.79 12.55 2.61 3.58 9.43 4.80 9.36 3.78 7.64 5.52 4.75 18.58 7.10 3.51

7.08 6.02 1.04 2.16 12.81 2.29 4.63 2.76 13.57 8.18 5.80 26.85 4.68 2.14

4.23 2.08 0.31 1.55 12.67 2.29 2.57 1.66 10.15 4.37 4.99 48.19 3.58 1.36

4.18 2.20 0.35 1.67 13.69 2.11 2.61 1.29 9.33 5.01 4.90 46.39 4.06 2.21

5.75 17.21 3.28 1.13 4.06 3.06 4.79 1.21 7.37 7.12 1.60 14.99 0.38 0.92

100

100

99.91

100

100

100.01

100

100

0.01

Source: MOEAa, Yearbook of Industrial Production Statistics by Taiwan Area, R.O.C., various years.

Changes 1971–2000

Changes 2000–14 2.9 3.8 0.7 0.5 0.9 0.2 2.0 1.5 4.2 3.2 0.9 19.5 0.6 0.1 0.01

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

Food, beverages, and tobacco Textile, apparel, and leather Wood and furniture Paper and printing Chemicals and products Petroleum refining Rubber and plastics Non-metallic minerals Basic metal Metal products Machinery Electrical and electronics machinery Transport equipment Miscellaneous

1971

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

Most of the successful second-movers in Taiwan, however, have not pursued R&D-intensive and own-brand strategies to catch up. Second-movers expand by relying upon accumulated organizational capabilities based on subcontracting manufacturing, hence implying path dependence in development.¹⁹ Thus, the strategy of choice for most has been upgrading subcontracting, cross-industry subcontracting, and then own-brand manufacturing, in that order. Among the structural factors affecting a firm’s strategic choice, industrial policy has been crucial. South Korea has produced some successful global brands, supported by the state’s national champion policy and long-term commitment to the Chaebŏl. China has also adopted a highly ambitious national champion strategy. The fact that the government in Taiwan has never adopted a national champion strategy helps to partly explain the evolutionary path of Taiwan’s second-movers, and attests to the importance of industrial policy. Taiwan’s second-movers did move along the upgrading path stated above, entering other areas, especially communications and video products, and related parts and components. Therefore, despite the trend of moving production offshore in certain segments, total employment in the electronics sector did not decline over the last two decades.

25.6 L  D A

.................................................................................................................................. In sum, until 1986, except for the high-tech industry, which relies on a different set of policies, the overall industrial policy was export promotion but accompanied by secondary import substitution and protection of the domestic market. Most of the banks were publicly owned; the government could therefore direct credits to support its industrial policy. Nonetheless, the Nationalist government was much more restrained in this regard when compared with that of the South Korean case, probably due to historical experiences. The government had also successfully managed to maintain macroeconomic stability throughout the years, by keeping the budget mostly in balance and the inflation rate low. That is, before 1986, Taiwan’s economy had been operating under a stable export-promotion regime, in which foreign exchange was under control, the exchange rate was kept stable and undervalued, and the domestic market was protected. However, past success of the aforementioned export-led growth model created circumstances which made the next round of transformation inevitable. Indeed, starting from 1986, Taiwan’s economy began its great transformation. Although the government still tried to guide the process with developmental intentions, it was basically passive in making the necessary adjustments after its hands were forced. ¹⁹ Chu (2009).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



That is, this time, it failed to adopt a forward-looking policy to guide unavoidable transformation. While external trade was more or less in balance before 1980, trade surplus and exchange reserves had begun to accumulate rapidly since 1980. The imbalance between the progressive export sector and the protected domestic sector left the increasingly wealthy domestic consumers ever more dissatisfied. A sustained trade imbalance between the USA and East Asia eventually led to the signing of the Plaza Accord in 1985, which forced the New Taiwan dollar to appreciate starting from 1986. Its currency value had risen 40 per cent against the US dollar by 1989. Under pressure from the USA, the government lessened foreign exchange controls and began to reduce tariff rates, remove non-tariff trade barriers, and phase out the tariff rebate programme. In the meantime, substantial asset bubbles began to appear in the local stock and housing markets. The wage level began to rise significantly, and the share of industry in GDP started to decline. Meanwhile, the government also began to liberalize the internal economic environment. After Chiang Ching-Kuo lifted martial law in mid-1987, the government began to open up (to both foreign and local firms) various domestic markets, in which the number of operating licences had previously been limited and more or less frozen since the early post-war period. Among the newly liberalized markets, the most important ones were the modern services, such as banking, telecommunications, transportation, and mass retailing. Significantly, at the same time, the government began to improve the cross-Strait relationship by allowing citizens to visit relatives on the mainland for the first time since 1949. Privatization of state-owned enterprises began two years later. Thus, democratization, liberalization, and globalization went hand in hand within a short period of time. It should be stressed that this was a process of managed liberalization, even though the extent of its success can be debated. In hindsight, the government probably should have implemented the reform earlier in a more forward-looking way. However, it proved to be a difficult transformation, changing from a developmental state model, in which growth was given priority, to one in which political, social, and economic goals had to be renegotiated and realigned.

25.7 G

.................................................................................................................................. The pace of globalization has been swift in Taiwan since the late 1980s. The flow of inward and outward foreign direct investment (FDI) has increased significantly (see Table 25.4). Inward FDI now mostly flows into the modern service sectors, as entry restrictions continue to lessen. By the time Taiwan formally entered the WTO in 2002, the domestic market had already gradually become quite open. Outward FDI mostly took place from the late 1980s on. The first wave was when labour-intensive production moved offshore, initally to the ASEAN countries and later to China. In the last few years, the high-tech industry has also begun to move

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

mass production lines to China. For example, the ratio of overseas production of Taiwan’s IT industry increased from 43 per cent in 1998, to 78 per cent in 2003, and to 92 per cent in 2014.²⁰ High-tech firms are now under intense pressure to upgrade their operations again.

Table 25.4 Globalization, 1952–2015 Inward FDI

Outward FDI-Total

Outward FDI- to China

US $bn

Number of cases

US $bn

Number of cases

1952–90 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

5.772 389 411 323 389 413 500 683 1,140 1,089 1,410 1,178 1,142 1,078 1,149 1,131 1,846 2,267 1,845 1,711 2,042 2,283 2,738 3,206 3,577 3,789

13.3 1.8 1.5 1.2 1.6 2.9 2.5 4.3 3.7 4.2 7.6 5.1 3.3 3.6 4.0 4.2 14.0 15.4 8.2 4.8 3.8 5.0 5.6 4.9 5.8 4.8

873 365 300 326 324 339 470 759 896 774 1,391 1,387 925 714 658 521 478 464 387 251 247 306 321 373 493 462

3.1 1.7 0.9 1.7 1.6 1.4 2.2 2.9 3.3 3.3 5.1 4.4 3.4 4.0 3.4 2.4 4.3 6.5 4.5 3.0 2.8 3.7 8.1 5.2 7.3 10.7

237 264 9,329 934 490 383 8,725 1,284 488 840 1,186 5,440 10,105 2,004 1,297 1,090 996 643 590 914 887 636 554 497 427

0.2 0.2 3.2 1.0 1.1 1.2 4.3 2.0 1.3 2.6 2.8 6.7 7.7 6.9 6.0 7.6 10.0 10.7 7.1 14.6 14.4 12.8 9.2 10.3 11.0

10.5 27.8 190.7 59.5 80.5 56.8 149.8 61.7 38.3 51.4 63.4 199.5 194.0 205.2 245.4 177.1 154.1 239.4 237.6 517.7 388.9 158.0 175.6 140.9 102.1

0.7 0.9 0.3 1.0 2.2 3.2 0.5 1.6 2.6 3.1 2.3 2.2 2.0 3.5 4.6 7.0 10.0 16.6 12.1 16.0 16.2 20.1 16.6 20.7 25.7

Sum 1991–2000 2001–10 2001–15

6,747 15,389 30,982

31.3 66.3 92.3

5,944 6,032 7,987

23.9 38.6 73.7

22,974 24,265 27,266

17.1 80.2 137.8

Average 72.7 223.3 213.3

1.6 7.6 11.7

Year

Note: FDI = foreign direct investment. Source: Adapted from MOEAb (various years).

²⁰ MIC (various years).

US $bn

Share in total Average value outward per case FDI(%) (US $mn)

Number of cases

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



The destination of Taiwan’s outward FDI has become increasingly concentrated on Mainland China, making up over 70 per cent of the total outflow (Table 25.4). The situation in export trade is similar.²¹ In sum, although Taiwan’s economy has become increasingly globalized, the external relationship has been dominated by the crossstrait relation. As economic liberalization began in a big way after 1986, the major part of labourintensive production moved overseas. Two kinds of exports took its place. One was upstream input, the growth of which originally took place via secondary import substitution. For example, fabrics and man-made fibres became major exports after the downstream apparel production moved out. The other major exports came from the newly emerged high-tech industry, which became the major force in exports. The share of heavy industrial goods in total exports increased from 46 per cent in 1990 to 81.3 per cent in 2014, and the share of goods with high technology content rose from 27 per cent in 1990 to 52 per cent in 2014.²² The changing product mix of Taiwan’s exports from the late 1980s until the present shows that the share of capital-intensive products, technology-intensive products, and high-tech products in Taiwan’s exports have all increased in this period, indicating that broadbased industrial upgrading has indeed taken place in this round of industrial transformation.

25.8 S: M O   D M

.................................................................................................................................. As discussed in Section 25.6, liberalization of the domestic market, especially the service sector, began in earnest in the late 1980s.²³ This liberalization has both internal and external components: externally opening up to foreign investors and internally allowing new domestic entrants in previously restricted markets. This is particularly important in the modern service sectors, such as finance, communications, transportation, and mass retailing. Regulation of foreign entries more or less followed the original developmental state model, that is, fostering local firms while opening up gradually to foreign competition. For example, when opening up telecommunications services, a former government monopoly, local entrants were encouraged to find foreign partners, but only as minority shareholders. The ceiling on foreign shareholding was later lifted, as their technology and managerial assistance became non-essential. The government policy of

²¹ CEPD, Taiwan Statistical Data Book, 2015, pp. 225–6. In 2014, the share of exports heading to Hong Kong and China amounted to 39.8 per cent, while the share heading to the USA was 11.1 per cent. ²² CEPD, Taiwan Statistical Data Book, 2015, pp. 216–18. ²³ This section draws from Chu and Hung (2002) and Amsden and Chu (2003: ch. 4).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

a managed opening of modern services permitted local firms to capture a second-mover advantage.²⁴ It is generally believed that, in terms of global competitiveness, Taiwan’s services sector is inferior to its manufacturing sector; nonetheless, maybe due to this policy of managed opening, the level of foreign shares in services in Taiwan is still low compared to that in many other developing countries. For the government to manage local entries, however, was much more complicated, because it involved domestic politics and occurred during the democratization process. In general, most of the existing large-scale business groups tried to participate in some of the newly opened markets. The government demonstrated a tendency to try to accommodate as many qualified applicants as the situation allowed. For example, in 1991, regulating the entry of new banks, the government finally approved fifteen applications, three times the originally planned five new banks. Other hotly pursued opportunities included entry into non-bank financial services, cable TV, electric power plants, mobile and fixed-line telecommunication services, high-speed rail, media, and so forth. One byproduct of this round of internal liberalization is that the share of the business groups in economic activities has greatly increased because the groups have been the main participants in the newly liberalized sectors.

25.9 I S T

.................................................................................................................................. Several industry trends are observable in Taiwan. Its exports mainly consisted of labour-intensive products in the earlier post-war period and technology- and capitalintensive products in the later period. Taiwan’s exports came mainly from small and medium-sized enterprises (SMEs) in the earlier period and from large-scale firms in the later period. During both periods, subcontracting has been the dominant business model. At present, the leading industrial enterprises in Taiwan are high-tech subcontractors and medium-tech upstream input producers. Very few large-scale firms have their own brands, and Taiwan’s few global brands are mostly owned by non-major firms.²⁵ Part of the reason why Taiwan has been able to maintain healthy growth during the last sixty years is that its industry has been very adaptive. In other words, Taiwan has been able to find new growth industries as its comparative advantage has shifted. As a result, it is interesting to note that, in terms of the size of employment, the manufacturing sector has not lost much ground during the last two decades, with the total number of employees around 2.4 million. This is evidence that Taiwan still relies very much on its industrial sector to maintain growth and global competitiveness.

²⁴ Amsden and Chu (2003: 151).

²⁵ Chu (2009).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



The industrial structure continued to change as Taiwan’s comparative advantage shifted due to its own success. For example, in the early post-war period, the first leading sector was the textile and apparel industry. At the height of its influence, it contributed over 23 per cent of manufacturing value added in the early 1970s. As shown in Table 25.3, since then its share has continued to decline, reaching a mere 2 per cent in 2014. The last two columns of Table 25.3 display the changes in each sector’s share in total manufacturing value added during 1971–2000 and 2001–14. In the former period, the declines in textiles and plastics were offset by the increases in chemicals, basic metals, and electronics. This shows that the more traditional industries gradually moved their operations overseas, especially to China. In the latter period, however, almost all sectors except electronics experienced decreases in their shares of total manufacturing value added. The share of the electronics sector rose steeply from 26.85 per cent in 2000 to 46.39 per cent in 2014, and the share of the top three industries now accounts for 69.4 per cent.

25.10 S T

.................................................................................................................................. Industrial production has been the driving force right from the beginning of the postwar period. The share of agricultural production in GDP decreased from 38 per cent in 1953 to below 10 per cent after 1978, while the share of manufacturing in GDP increased from 12.9 per cent in 1952 to 29 per cent in 1970, reached a peak at 39.4 per cent in 1986, and then began to decline to 29 per cent by 2014. The share of services remained stable, around 48 per cent, in the first thirty post-war years, and then steadily rose to 64 per cent by 2014 (Table 25.1). The shifting pattern of GDP composition among the primary, secondary, and tertiary sectors in Taiwan closely resembles that of the more advanced countries, of course, indicating the steady advancement of Taiwan’s economy. The changes in the employment composition run pretty much parallel to those of the GDP. Comparing the initial conditions of the developing countries at the beginning of the post-war period, Chang (2005) found that the conditions really varied among the Asian, African, and Latin American countries, and that the only condition in which the East Asians were clearly superior to all others was in social infrastructure, that is, the literacy rate and life expectancy. In this regard, Taiwan had a favourable starting position and has improved on that continuously since then. The relevant indicators show good progress. Table 25.5 shows that life expectancy for females and males increased from 60.3 and 57.4 in 1952 to 83.2 and 76.7 in 2014 respectively. The literacy rate increased from 63.6 per cent in 1952 to 98.5 per cent in 2014. Taiwan’s income distribution had been relatively equal until recently, if compared with most other developing countries. In terms of household income, the ratio of the top fifth to the bottom fifth was 5.25 in 1964, which fell to 4.17 in 1980, but rose to 5.55 in 2000, and increased sharply to 6.08 in 2013 (see Table 25.6).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

Table 25.5 Social indicators, 1952–2014 Year 1952 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2014

Population Labour force Life expectancy (4) (1,000 participation Unemployment Male Female persons) (1) rate (%) (2) rate (3) 8,128 10,792 12,628 14,676 16,150 17,805 19,258 20,353 21,304 22,216 22,690 23,162 23,434

66.5 62.4 58.2 57.4 58.2 58.3 59.5 59.2 58.7 57.7 57.8 58.1 58.5

4.4 4.0 3.3 1.7 2.4 1.2 2.9 1.7 1.8 3.0 4.1 5.2 4.0

57.4 62.3 65.1 66.7 68.3 69.6 70.8 71.3 71.9 73.8 74.5 76.1 76.7

Infant Literacy mortality (5) rate (6)

60.3 66.4 69.7 71.6 73.4 74.6 75.8 76.8 77.7 79.6 80.8 82.6 83.2

44.7 35.0 24.1 16.9 12.6 9.8 6.8 5.3 6.4 5.9 5.0 4.2 3.6

63.6 78.7 81.2 87.6 87.1 89.7 91.5 93.2 94.4 95.6 97.3 98.0 98.5

Sources: (1) DGBAS, http://eng.stat.gov.tw/mp.asp?mp=5 and www.dgbas.gov.tw/mp.asp (2) CEPD, Taiwan Statistical Data Book, Taipei, various years; (3) DGBAS, http://eng.stat.gov.tw/mp.asp?mp=5 and www.dgbas.gov.tw/ct.asp?xItem=9522&ctNode=2857 (4) Ministry of Interior, http://sowf.moi. gov.tw/stat/year/list.htm (5) Ministry of Health and Welfare, http://www.mohw.gov.tw/cht/DOS/ Statistic.aspx?f_list_no=312&fod_list_no=5481 (6) Ministry of Education, https://stats.moe.gov.tw/

Table 25.6 Changes in income distribution, 1964–2014 Ratio of income share of top 20% to that of bottom 20% Year

Gini coefficient

Without government transfers

With government transfers

1964 1970 1980 1990 2000 2005 2010 2014

0.321 0.294 0.278 0.312 0.326 0.340 0.342 0.336

4.31 5.53 6.57 7.45 7.72 7.40

4.17 5.18 5.55 6.04 6.19 6.05

Source: Adapted from DGBAS, Survey of Household Income and Expenditure, various years, http://statdb.dgbas.gov.tw/pxweb/Dialog/statfile9L.asp

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



Some related policies and aspects of economic development helped to bring about these beneficial results in the earlier period. Land reform helped to build the basis for social and income equality. Labour-intensive export production provided ample jobs for the young leaving the agricultural sector. Industrial districts had also been quite dispersed and spread jobs widely, until recently when the electronics boom concentrated new high-tech jobs in the north of the island. Implementation of industrial policy relied mostly on SOEs, not on large private firms such as the chaebŏl in South Korea, hence leaving room for the SMEs. But, in the later period, globalization and industrial concentration in electronics led to increased geographical concentration of employment in the northern part of the island, and a higher skill premium for the better educated, thus a worsening of income distribution. Education policy emphasized mass education in the early period, vocational education in the later period, and higher education only recently. In the earlier period, the education policy had served economic development well, by supplying a large number of good quality unskilled and semi-skilled workers, and ample supply of competent engineers. The higher education system has been greatly expanded since the 1990s. The results of educational reforms and the plan of further liberalization have been the subjects of heated debate in recent years. Heath policy focused on a public health programme in the early period, emphasizing control of the spread of communicable diseases and implementation of a birth control programme. The health programme has been expanding along with economic progress throughout the years. A universal health care plan was put into effect in 1998, as the major social welfare programme offered in Taiwan to this day.²⁶ Even though Taiwan’s per capita income has approached the advanced country level, the size of the social welfare programme remains relatively small. The share of government expenditure was only 16.4 per cent of GDP in 2014.²⁷ Besides a universal medical insurance programme, there are a limited number of social welfare programmes to provide the necessary social safety net in a changing society, in which the traditional family has come to provide fewer and fewer supportive functions. Probably due to the effects of globalization and democratization, the tax burden, which had been relatively light in the earlier post-war period, has been decreasing since the 1990s. The ratio of tax revenue to GDP has declined from 19.2 per cent in 1980 to 12.3 per cent in 2014. The budget had been more or less in balance before the late 1980s. Since then, the fiscal deficit has been increasing rapidly; its share in total government expenditures has averaged more than 30 per cent since the 1990s and shows few signs of improving.²⁸

²⁶ Chen (2011). ²⁷ CEPD, 2015 Taiwan Statistical Data Book, 179. ²⁸ CEPD, 2015 Taiwan Statistical Data Book, 185.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

25.11 R S  P

.................................................................................................................................. As Taiwan’s economy approaches maturity, economic growth has been slowing in the last couple decades. At the height of post-war growth, in the 1960s and 1970s, the overall annual growth rate averaged close to 10 per cent. Since the new century began, growth has slowed significantly and averaged only 2.9 per cent from 2001 to 2014, signalling the arrival of a different stage of development. The years 2001 and 2009 were also the only years in the post-war period in which the economic growth rate was negative. And the average growth rate of investment has turned negative, 0.3 per cent, during this century (see Table 25.1). Taiwan’s economy has performed relatively well since embarking on its great economic, social, and political transformation in the late 1980s. After more than two decades, industry managed to continue to grow, and the unemployment rate remained at a moderate level. Although its labour-intensive production has moved offshore, its electronics industry persists in upgrading and expanding and maintaining its global competitiveness, thus becoming Taiwan’s pillar industry. Integration with the Chinese economy has provided growth momentum and has helped the second-movers expand in scale. The legacy of the developmental state has meant that the external economic liberalization has been a managed affair, allowing local services to grow. However, there remain many serious challenges. Overall growth is overly reliant on the old export-promotion regime, with an undervalued exchange rate and relatively low wage levels. There are many imbalances in the economy. The recent lack of investment growth is probably related to the fact that the flow across the Strait remains one-way, that the growth of domestic consumption lags behind overall growth, and that the dominant industry, electronics, has encountered greater competitive pressures as it continues to upgrade. Although economic integration with China continues to grow, political debate persists in hindering rational policy planning. At the same time, globalization has also brought an unprecedented increase in the degree of inequality. The new rules of political competition have not been conducive to addressing these challenges.²⁹ Nevertheless, the political transformation has been peaceful, and the framework of the developmental state has not been totally dismantled. But the conflict in political and economic directions remains unresolved, and society is yet to face up to the crucial question of how to fit China into Taiwan’s economic future. Thus, there is really no forward-looking development plan or vision for future development. Only if future political developments could promote more productive dialogue within Taiwan and across the Strait would Taiwan be able to formulate a new economic vision for its future development. ²⁹ Chu (2014).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

    



R Amsden, Alice H., 1979. ‘Taiwan’s Economic History: A Case of Etatisme and a Challenge to Dependency Theory’, Modern China, 5 (3), pp. 341–80. Amsden, Alice H., 1989. Asia’s Next Giant: South Korea and Late Industrialization, Oxford: Oxford University Press. Amsden, Alice H. and Wan-wen Chu, 2003. Beyond Late Development: Taiwan’s Upgrading Policies, Cambridge, MA: MIT Press. Chang, Ha-joon, 2005. ‘How Important Were the Initial Conditions for Economic Development? East Asia vs. Sub-Saharan Africa’, in Ha-Joon Chang, ed., The East Asian Development Experience: The Miracle, the Crisis, and the Future, London: Zed Press. Chao, Ji-chang, 1985. The Utilization of the US Aid in Taiwan (in Chinese), Taipei: Linking Press. Chen, Mei-hsia, 2011. ‘A Historical Analysis of Marketization of Taiwan’s Public Health System’ (in Chinese), Taiwan: A Radical Quarterly in Social Studies, 81, pp. 3–78. Chu, Wan-wen, 2001. ‘The Effects of Taiwan’s Industrial Policy: A Preliminary Evaluation’ (in Chinese), Taiwan: A Radical Quarterly in Social Studies, 42, pp. 67–117. Chu, Wan-wen, 2008. ‘The Early Development of Taiwan’s Cotton Textile Industry’ (in Chinese), New History, 19(1), pp. 167–227. Chu, Wan-wen, 2009. ‘Can Taiwan’s Second Movers Upgrade via Branding?’ Research Policy, 38, pp. 1054–65. Chu, Wan-wen, 2014. ‘Challenges for the Maturing Taiwan Economy’, in Larry Diamond and Gi-Wook Shin, eds, New Challenges for Maturing Democracies in Korea and Taiwan, Stanford: Stanford University Press, pp. 216–49. Chu, Wan-wen, 2017. The Causes of Taiwan’s Postwar Economic Growth: the Why and How of Late Development (in Chinese), Taipei: Academia Press and Linking Press. Chu, Wan-wen and Chia-yu Hung, 2002. ‘Business Groups in Taiwan’s Post-liberalization Economy’ (in Chinese), Taiwan: A Radical Quarterly in Social Studies, 47, pp. 33–83. Council on Economic Planning and Development (CEPD), various years. Taiwan Statistical Data Book, Taipei: CEPD. Cumings, Bruce, 1984. ‘The Origins and Development of the Northeast Asian Political Economy’, International Organization, 38 (1)’, pp. 1–40. Directorate General of Budget, Accounting, and Statistics (DGBAS), 2015. Available at: http:// www.dgbas.gov.tw Gold, T. B., 1981. ‘Dependent Development in Taiwan’, PhD dissertation, Harvard University. Ho, Samuel P. S., 1978. Economic Development of Taiwan, 1860–1970, New Haven, CT: Yale University Press. IMF, 2015. World Economic Outlook, April. Available at: http://www.imf.org/external/pubs/ ft/weo/2015/01/weodata/weoselgr.aspx Industrial Technology and Research Institute (ITRI), 2015. Semiconductor Industry Yearbook 2015, Hsinchu: ITRI. Lin, Ching-yuan, 1973. Industrialization in Taiwan, 1946–72: Trade and Import-substitution Policies for Developing Countries, New York: Praeger. Maddison, Angus, 2010. Historical Statistics of the World Economy: 1–2008 AD. Available at: http://www.ggdc.net/maddison/oriindex.htm

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



- 

Market Intelligence and Consulting Institute (MIC), various years. Information Industry Yearbook, Taipei: Institute for Information Industry. Ministry of Economic Affairs (MOEAa), 2015. Yearbook of Industrial Production Statistics, 2015, Taipei: MOEA. Ministry of Economic Affairs (MOEAb), various issues. Statistics on Overseas Chinese and Foreign Investment, Outward Investment, Outward Technical Cooperation, Indirect Mainland Investment, Guide of Mainland Industry Technology, ROC, Taipei: Investment Commission, MOEA. Myers, R. H. and M. R. Peattie (eds), 1984. The Japanese Colonial Empire, 1895–1945, Princeton, NJ: Princeton University Press. Wade, Robert, 1990. Governing the Market: Economic Theory and the Role of Government in East Asian Industrialization, Princeton, NJ: Princeton University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

 Lessons from an Experiment ......................................................................................................................

 

26.1 B B  B  A

.................................................................................................................................. T ‘African Rising’ or ‘African lions on the move’ narrative that has gained currency in recent years emphasizes the growth on the continent in the early part of the new millennium. External factors, especially prices and growing demand for primary commodities, have been favourable.¹ But while such growth in Africa has been celebrated, efforts to understand the structural drivers of longer term economic development in African countries have been inadequate. The swing between ‘Afro-pessimism’ tragedy and ‘Afro-euphoria’ hyperbole has been coupled with an erroneous ‘African dummy’ analytical approach that overlooks the continent’s diversities (Cramer and Chang 2015). These tropes are not only remote from reality, they also lack the perspective that growth should be underpinned by structural change. As global demand for commodities dropped after 2014, along with prices, various concerns were raised and valid insights were offered by observers, scholars, and policy makers. For instance, an article entitled ‘Africa’s Boom Is Over’ boldly proclaimed that ‘Africa was never going to get far without manufacturing’ (Foreign Policy 2015).² Africa obviously performed better in the early 2000s, but views have diverged on the drivers of this growth and on its sustainability, and on whether this growth will translate into structural change. The ‘Afro-euphoria’ of recent years was just as removed from reality as its predecessor, dismissive ‘Afro-pessimism’. ¹ For instance, see McKinsey Global Institute (2016). ² ‘Africa’s Boom Is Over’ by Rick Rowden, Foreign Policy, 31 December 2015. See also ‘Africa’s Boom’, African Business, IC Publication No. 428, March 2016.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

From a long-term perspective, a promising approach to understanding these dynamics is a structural transformation perspective, based on the view that the essence of economic development is economic transformation and structural change embedded in sectoral shifts, sustained productivity rises and constant technological advances, which are achieved through effective industrial policies and state activism (Johnson 1982; UNCTAD-UNIDO 2011; Mazzucato 2013; UNCTAD 2016).³ Although many sceptics have argued that industrial policies cannot work in Africa, it has become more fashionable to talk about such policy in recent years. Nevertheless, it is unclear what industrial policies entail in practical terms. Perhaps, it is time to examine and learn from the practice of industrial policies in African countries. Ethiopia is an ideal case study, since the country has achieved rapid economic growth over the past two decades, despite being located in a complex and challenging geopolitical region. This growth has not been fuelled by mineral exports and, while manufacturing remains small, Ethiopia has been engaged in industrial policies in key priority areas.⁴

26.2 I P  S T  E

..................................................................................................................................

26.2.1 Perspectives on Policy and Transformation Structural transformation is the prime driver of economic and social development. It involves the movement of people and outputs across sectors and within specific industries, and a shift from lower to higher productivity economic activities. It is argued that growth and structural changes can be sustained when driven by manufacturing. Manufacturing is an engine of growth because it is positively causally related to the growth of GDP and rises in productivity in the whole economy (Kaldor 1967;

³ For an in-depth understanding of structural transformation, see Kaldor (1967); Ocampo et al. (2009); and Thirlwall (2013). ⁴ The Ethiopian economy grew annually by 10.8 per cent between 2003–04 and 2014–15, driven by productive sectors and without any resource boom (MOFED 2010; NPC 2016). This annual growth rate is twice the average for sub-Saharan Africa. Average life expectancy increased by 19 years, from 45 to 64, between 1991 and 2014. The number of people living under the poverty line was halved from 50 to 25 per cent in the same period. The government has focused on long-term investment programmes, with a special focus on infrastructure development (especially in energy and modern transport) and university and technical schools. Despite rapid economic growth, the expansion of exports and manufacturing, and changes in their composition, have been inadequate.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



Thirlwall 2013).⁵ Sectoral shifts occur through diversification into new activities, development of domestic linkages and technological capabilities.⁶ This is because of increasing returns to scale, learning-by-doing, linkages (including intersectoral), innovation, and technological advancement. Historical experience suggests, further, that manufactured exports are particularly important, given the balance of payments constraint on growth. Nonetheless, structural transformation and catch-up is uneven, unpredictable, and compounded by political tensions (Cramer and Chang 2015; Whitfield et al. 2015). Growth cannot be sustained without the rapid expansion of exports and fundamental changes in the composition of those exports. Exports play a strategic role in structural transformation by expanding the limits on market demand, enabling productivity spill-over, driving technological advancement, loosening the balance of payment constraint, allowing the full utilization of domestic resources and nurturing import-substitution industrialization (ISI) (Ocampo et al. 2009; Thirlwall 2013). Ocampo et al. (2009: 152) highlights that the ‘major task of structural transformation policies is to facilitate a dynamic restructuring of production and trade toward activities with higher technological content’. Industrial policies have been used by forerunners and latecomers in the nineteenth and twentieth centuries for catching up and economic transformation (List 1856; Hamilton 1934; Chang 2003; Nayyar 2013).⁷ Such policies are ‘a strategy that involves a range of implicit or explicit policy instruments selectively focused on specific industrial sectors for shaping structural change in line with a broader national vision and strategy’ (Oqubay 2015: 18). Hence, industrial policies should serve as vehicles for structural transformation and catch-up. This chapter will review industrial policy in Ethiopia with the aim of extracting lessons from the comparative review of labour-intensive export-oriented sectors such as leather and leather products; capital-intensive, import-substitution industries such as the cement industry; and high productivity modern agriculture such as floriculture (Oqubay 2019a; Oqubay and Tesfachew 2019).⁸ These three sectors have different industrial structures and can collectively illustrate the practice of industrial policy and uneven outcomes in Ethiopia.⁹ ⁵ For Kaldor’s laws, see Kaldor (1967) and Thirlwall (2013). See also Szirmai et al. (2013) on current debates on structural transformation. ⁶ Ocampo et al. (2009) state that new activities involve new markets, products, processes, institutions, etc. New activities may be new to a country, but not necessarily to others. ⁷ The USA and Germany were among the nineteenth-century late-comers, while Japan, South Korea, and Taiwan are examples of twentieth century late-comers. Gerschenkron (1962) argued that latecomers can catch up by building on the advantage of backwardness and late development, which requires institutional innovations and an active role by the state. See also Johnson (1982), Amsden (1989) and Wade (2004). ⁸ This chapter is based on Oqubay, Made in Africa: Industrial Policy in Ethiopia (Oxford University Press 2015), which is in turn based on extensive original research conducted in the three sectors from 1991 to 2015. The research involved primary sources, qualitative and quantitative surveys of 150 firms, site observations of fifty firms, and more than 200 in-depth interviews. ⁹ See UNECA (2016a: 102–6) for a detailed review of Ethiopian industrial policies.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

26.2.2 A Strategic Import-substitution Industry 26.2.2.1 The Industry’s Context Cement manufacture is considered a basic industry that plays a strategic role in late industrialization, and is characterized by high capital intensity and process production. The global cement industry has lately been dominated by Chinese consumption, supply, and equipment provision.¹⁰ It has strong linkages with the construction industry and transport sector. The industry features significant economies of scale and is dominated by large firms, and its expansion is driven by capital deepening rather than capital widening).¹¹ The African cement industry is highly fragmented, with underdeveloped economies of scale and technology. In Ethiopia, the cement industry emerged in the mid-1960s, and until 2000 it was dominated by a single state-owned enterprise (SOE).¹² Demand was sluggish until the 1990s, but rapidly increased in the first decade of the new millennium. The 1.7 million tons produced became insufficient when government-sponsored infrastructure and integrated housing programmes were expanded. Cement shortages became a binding constraint, almost paralysing the booming construction industry and hindering the development of manufacturing plants. At this time, less than half the demand was being met.

26.2.2.2 Policy Instruments The government designed an ambitious and comprehensive policy to develop the cement industry while also trying to contain the damage resulting from the cement shortage. First, the government stimulated demand directly and indirectly by adopting various measures. As a provisional solution, close to 4 million tons annually were imported between 2006 and 2011. Cement is highly dependent on transportation, and the government had to import about 1,500 heavy trucks to increase the uplift capacity from ports, contributing to the modernization of the transport fleet in the process. The increasing volume of imports and the high profits served as strong signals for new investments in cement. Arguably, this is a typical example of Hirschman’s import-swallowing concept, in which imports play creative roles (through demand formation and demand reconnaissance) by stimulating new domestic manufacturing and spurring import-substitution (Hirschman 1958).

¹⁰ Global production in 2012 was 3.7 billion tons, and China accounted for more than half of it. Moreover, cement is a non-tradable commodity with less than 10 per cent internationally traded (Oqubay 2015). ¹¹ Capital deepening features a ‘rise in the capital/labour ratio’ and may serve as a basis for new industries (Amsden 1989: 268). On the scale and scope of such industries, see Amsden (1989); Penrose (1995); and Chandler (2004). ¹² Mugher Cement Enterprise dominated the industry for almost four decades and continues as a major player.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



Investment incentives were introduced to encourage and induce new investment and productive capacity. For instance, 101 new investment projects were registered between 2003 and 2012, while there were only two projects in the preceding decade. To boost the industry, the government instituted a three-year zero-income tax incentive, while factory land and raw material quarries (limestone, gypsum, etc.) were made available at nominal prices. The government furnished long-term investment financing at a subsidized interest rate to large-scale producers specifically for optimum productivity and economy of scale gains.¹³ The Development Bank of Ethiopia (DBE) provided about a quarter of its total loans to the cement industry and additional financing mechanisms were accommodated, including allowing foreign equity financing to many firms and assisting the SOE through the industrial developing fund (IDF).¹⁴ Moreover, the industry was afforded priority in the allocation of foreign exchange, not only for importing equipment and capital goods, but also for cement imports. The government then banned all imports once the new capacity was sufficient to meet domestic demand. As to electricity supplies, the government maintained low electricity costs for cement and other manufacturers and gave the cement industry priority because of its critical effect on public investment programmes such as housing and new energy supplies.¹⁵ Productivity and energy efficiency were far below the competitive cement industries in, for example, China or Pakistan. Gradually, government forced the cement factories to upgrade to coal-fired technology (instead of the more expensive furnace oil), by organizing loan facilities and bulk coal imports through a government agency. Apart from achieving savings through bulk purchases, this assistance reduced logistics complexity and pressure on working capital. Because of the adoption of coal-burning technology, the industry has experienced substantial efficiency gains (40–50 per cent of the cement industry’s total expenses stem from energy consumption).

26.2.2.3 Policy Outcomes Ethiopia’s installed cement-producing capacity has increased to 15 million tons, a fivefold growth between 2005 and 2016, making Ethiopia one of the top three producers in sub-Saharan Africa. The rate of expansion was three times faster than the average global growth rate for cement production. The industry has had significant spill-over effects for the economy. It is capital-intensive and employs fewer than 15,000 employees directly, although it generates jobs for skilled workers. In addition, the industry has strong employment linkages with the cement-products industry and the construction and transport sectors. For instance, the construction sector has become one of the largest employers, contributing 8.5 per cent of GDP in 2015. The price of cement has fallen with the increase in installed capacity, and has remained stable since 2012, thereby helping to fuel the construction industry. Domestic manufacturers continue ¹³ The kiln capacity increased by 250 per cent, and the new cement plants had a capacity of 2.3–2.5 million tons, in contrast to the prior 600–850,000 tons per annum. ¹⁴ IDF is a special fund organized by the government to finance expansion of SOEs. ¹⁵ The electricity tariff was 3 US cents per kilowatt hour in 2010–16, which is the lowest in Africa.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

to play a leading role in the Ethiopian cement industry, in contrast to other African countries where multinational corporations usually dominate. The narrow latitude for performance standards in cement production, the perishability of the product, the seasonal nature of the industry and strong pressure from the construction industry have induced the industry to improve capacity utilization, skills development, and production management. But there were serious limitations and tensions in government policies, and these were not without high costs. The 20 per cent devaluation of exchange rates in 2010 had a negative impact on the industry. Energy supplies could not keep pace with expanding demand in the industry, leading to major losses from downtime. Prioritization in financing and foreign exchange allocations starved other sectors. Many small cement factories vanished as an effect of a policy that favoured larger firms and the latest technology, which offered productivity gains. A new industrial structure has evolved involving new actors that will henceforth shape the game. These include shifts in the state–industry relationship, which, as noted already, plays a key role in the industry and has significant relationships with its industrial partners. Whereas in South Korea and China the cement industry served as a basis for developing technological capabilities, Ethiopia has missed out on this opportunity as there were no effective policy instruments to encourage domestic manufacturing of equipment, local content, and local capabilities.¹⁶ This strategic industry would have slowed without the foreign exchange provided by export earnings, which demonstrates the role and impetus from the export sector in supporting an import-substitution industry. Nevertheless, the development of domestic manufacturing may allow for significant savings on foreign exchange requirements. This highlights how export-led industrialization can complement importsubstitution (Amsden 1989). In conclusion, although the industrial policy in relation to cement production has had its drawbacks and costs, the net benefit to the overall economy and structural transformation has been irrefutable (Oqubay 2019a). The state played a critical role particularly through the public enterprise, which contributed expertise and production skills and had a demonstration effect. Government policies were the key drivers in the transformation of the cement industry, and its expansion was not based on factor endowments. The government has been able to learn from its mistakes and the new difficulties that arose. While the experience highlights the tensions, trade-offs, hard choices, diverging interests, and complexity of industrial policy, it also shows that an ¹⁶ Technological capabilities would include in investment project execution, initial plant erection, and manufacturing of less technologically complex fabrication works. Amsden (1989: 266–7) highlights that ‘cement-making never became one of Hyundai’s major enterprises. . . . The mill, however, was critical for Hyundai’s internal development’, and was the first manufacturing affiliate and first project execution. This process was supported by Fuller Company of the United States [emphasis added]. By 1974, Hyundai had developed capabilities except ‘basic engineering’ that was left to cement-process specialists. Likewise, in China, the government adopted policies to foster domestic manufacturing capacity, so that the country now accounts for up to 40 per cent of cement manufacturing technology worldwide.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



activist state can through effective industrial policies transform an industry that is strategic to industrial catch-up.

26.2.3 The Tale of Two Export Industries

(%)

The aims of industrial policy are the development of manufacturing industries and new activities, and the diversification and expansion of exports. In the Ethiopian context, the leather and leather-products sector has existed for almost a century and is among the government’s priority industries. It is labour-intensive, export-oriented, tradable, and strongly linked with agriculture. Despite the government’s focus on this industry, and despite countless international consultancy studies, the outcomes of industrial policy have not been satisfactory in terms of employment, output, export, and value addition (Oqubay 2019a). Meanwhile, the newer agro-industrial floriculture sector has demonstrated the gains that may be generated by industrial policy. The rise of this industry since 2004 is a shining example of how industrial policy should not be confined to traditional manufacturing, but is also applicable to high-productivity agricultural activities (UNCTAD-UNIDO 2011).¹⁷ The floriculture sector (like the leather sector) has benefited from Ethiopia’s natural endowments and competitive labour costs. Floriculture has also benefited from Ethiopia’s geographic location, climate and water, altitude and soil. Between 2004 and 2012, the floriculture sector generated more than $1 billion in export earnings (Figures 26.1 and 26.2). More than sixty firms operated in the sector, creating direct employment for 23.00 21.00 19.00 17.00 15.00 13.00 11.00 9.00 7.00 5.00 3.00 1.00 (1.00) (3.00) (5.00)

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 /99 /00 /01 /02 /03 /04 /05 /06 /07 /08 /09 /10 /11 /12 /13 /14 /15 /16

Manufactured goods (%) 9.88 9.91 21.05 14.73 17.86 14.46 13.12 13.65 12.27 12.11 8.98 6.70 8.88 9.08 10.23 10.21 11.40 10.97 Floriculture sector (%)

1.58 0.92

1.01 1.18 1.27 1.22 3.14 3.49 6.99 8.56 9.84 10.17 7.28 7.67

7.48 7.62 8.38 9.71

Leather sector (%)

6.97 7.45 17.58 12.62 13.66 9.37 7.99 7.53 7.86 6.77 5.20 2.85 4.01 3.49

3.93 4.81 4.42 4.06

Textile sector (%)

0.58 0.53

3.16 3.42 3.29 2.74

0.76 0.88 1.08 1.43 0.81 1.12 1.09 1.04 0.95 1.16 2.10 2.68

 . Export shares of manufacturing sector by export value (per cent) Source: ERCA (2016).

¹⁷ Transformation of agriculture is at the centre of the structural transformation inherent in late-late industrialization. See also Ocampo et al. (2009), and Thirlwall (2013).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Million $

50,000 people and indirect employment in the wider horticulture sector for 130,000 people. In 2012, the annual direct export volume reached 50,000 tons, worth $200 million in export earnings, thereby further diversifying Ethiopian exports and becoming an important contributor to Ethiopia’s tight balance of payments.¹⁸ There has been significant productivity growth, with Ethiopia emerging as one of the top five cutflower players globally, even if it has a long way to go to catch-up with the Kenyan horticulture industry, which has had forty years of experience. Learning-by-doing has been significant in the industry, and the sector (both foreign and domestic firms) relies on local skills in production management. By comparison, between 1992 and 2015, the growth of manufactured outputs in the leather sector was sluggish and showed erratic fluctuations.¹⁹ For instance, tanning production between 1992 and 2012 increased from 101 million to 160 million square feet, a very low growth rate. Between 1992 and 2009, footwear production increased from 874,000 to 2.2 million pairs. By contrast, Morocco and Tunisia alone exported more than 54 million pairs in 2010. Ethiopian export earnings from the leather sector rose from $61 to $110 million between 2002 and 2011, reflecting the sluggish expansion (Figures 26.1 and 26.2). The sector’s sixty factories employed about 20,000 people in 300.0 280.0 260.0 240.0 220.0 200.0 180.0 160.0 140.0 120.0 100.0 80.0 60.0 40.0 20.0 0.0 –20.0

1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 /99 /00 /01 /02 /03 /04 /05 /06 /07 /08 /09 /10 /11 /12 /13 /14 /15 /16

Floriculture

0.2

0.1

0.3

0.2

0.0

0.3

Non-floriculture

6.8

4.2

4.2

5.6

6.7

7.5 14.6 12.7 16.2

Total horticulture

7.0

4.3

4.5

5.7

6.8

7.9 22.3 34.7 79.8 125.4 142.6 201.9 188.5 241.7 230.5 245.7 249.3 275.5

Leather Leather products Total leather industry

7.7 22.0 63.6 111.8 130.7 170.2 175.3 197.0 186.7 199.7 203.1 225.3

31.1 34.7 78.4 61.2 71.5 60.1 55.4 72.9 83.0 0.1

0.1

0.1

0.1

1.0

0.2

1.4

1.9

6.6

31.2 34.7 78.5 61.3 72.5 60.3 56.8 74.9 89.6

13.7 11.9 31.7 13.2 44.7 43.9 45.9 91.7 67.8 50.2 94.3 99.2 100.9 123.1 7.5

7.6

6.3

9.5 10.8 20.2 32.0

46.2 50.1 92.2 73.1 39.4 42.2

99.2 75.3 56.5 103.8 109.9 121.1 155.1 131.5 115.3

 . Exports of floriculture and leather/leather products (million $) Note: Since 2012, the government has banned the export of semi-finished leather, contributing to value-addition. Non-floriculture includes herbs, vegetables and fruits. Source: ERCA (2016).

¹⁸ The export and employment performance of floriculture has been twice that of the longer established leather sectors. ¹⁹ And this was even though the livestock population in Ethiopia is the largest in Africa and among the top ten in terms of size worldwide.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



2011, averaging a 4.5 per cent annual growth rate between 1992 and 2011. Labour productivity growth has been erratic, with low productivity increases until 2011. Products have been low-value and progression has been very limited. Despite this, there have been new investments in recent years and, after decades of failure, there is some evidence that policy initiatives have finally begun to bear fruit.

26.2.3.1 Industrial Policy Instruments Although an active industrial policy was applied in both sectors, the outcomes diverged. Support given to floriculture was characterized by more effective coordination and commitment. The government engaged with a limited number of modern firms in the floriculture industry, while in the leather sector the engagement was relatively weak, due to the fragmentation of players and the large number of smallholders who are critical backward linkages. Foreign Direct Investment (FDI) played a critical role in floriculture, as the firms, although largely family owned, were equipped with technology and market capacity. Most domestic industrialists were keen to catch up, despite their limited share in the industry. The government provided suitable land to all firms at an affordable lease rate, primarily within a 200-km radius of Addis Ababa. Subsidized loans were provided to more than forty firms, both foreign and local, by the DBE. Aligning loan procedures to the specific nature of the industry and the firm’s situation, and linking financing to performance, were constraints. The risk for DBE was contained, however, as the predominant players had experience in the industry. Investment incentives and export promotion were applied, and the industry benefited from the devaluation of the currency in 2010. There were limitations in promoting linkages to input (chemicals and fertilizers) production, upgrading technology (especially new seeds demanded by customers) and expansion of greenhouses and irrigation systems. Air transport is the largest cost component. The products are perishable and require reliable and regular air transport, which was difficult to achieve in the earlier stages. A cool-chain logistics system and phyto-sanitary standards are also required. The government used the public enterprise Ethiopian Airlines (EAL) to develop its cargo capacity, expand its cool-chain storage, and serve the industry. This sector would not have grown without this strategic intervention. The government, together with industry players, also saved the industry when air freight costs shot up because of the twofold increase in fuel prices in 2008–09. A bold decision by government was for the treasury to subsidize a third and EAL another third of the fuel increase. All employment and export earnings would have been lost if this strategic decision had not been taken. Skills upgrading, environmental standards, and production codes were implemented by the Ethiopian Horticulture Development Agency (EHDA) and the Ethiopian Horticulture Producers and Exporters Association (EHPEA), and contributed to the improved performance of the industry. Moreover, this developmental partnership between industry and government was effective in ensuring collective learning.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Despite these positive interventions, major shortcomings and drawbacks are observable in the government’s industrial policy. For instance, the government failed to sustain the sector’s rapid growth through its failure to provide more land for expansion, a result of coordination failures and political factors. The share of Ethiopian domestic industrialists showed limited expansion because of a failure to introduce effective instruments, despite the existence of this opportunity to sustain growth. Technological upgrading was not sustained through the development of new and improved seeds, which are currently imported on a royalty basis. Moreover, government policies failed to support the development of non-floriculture exports (herbs, vegetables, fruits), whose production was minimal despite their huge export potential. The expansion from the central corridor to new corridors and clusters was very limited, although the airport logistics infrastructure was built. Finally, lessons have not been sufficiently learned from this sector to stimulate manufacturing and other agricultural sub-sectors. In short, a golden opportunity was lost because of the insufficient commitment by policy makers to provide the necessary support to sustain the sector’s growth and the concomitant failure to design appropriate policies for the sector’s growth stage. In the leather and leather-products sector, similar industrial policy instruments, especially investment and export-promotion incentives, development financing, and the privatizing of public enterprises (the major players until 2000) were put forth. New investment flows were dominated by the domestic industry until 2006. Tanneries predominated, with the leather-products sub-sector too weak to stimulate the sector and unable to sufficiently integrate into the global value chain. The industry faced a binding constraint in the supply of high-quality skins and hides, despite the large livestock population. Inadequate government focus on livestock development and an inability to transform the raw material value chain have been major strategic failures. This shows that an effective industrial policy must consider all the components in the value chain and focus on fostering linkage effects. The quality of skins and hides continued to fall, while prices tripled, magnifying the structural constraint. Industrial players are locked into low-value products, and there is major resistance to industrial upgrading. Lobbied by existing tanneries, the Ministry of Industry imposed a temporary ban on licensing new tanneries on the grounds of a shortage of raw materials. Domestic tanneries preferred exporting crusts, while shoe factories preferred producing for the domestic market because of their lack of competitiveness in the international market and limited technological capacity. In contrast to floriculture, the political economy constraints were significant, with the leather association failing to play a critical role in developing the industry. Moreover, the technological and economic characteristics were less favourable and the latitude for performance standards was wide, thus playing a minor role as a pressure device. In 2008, the government banned exports of raw skins and hides, favouring exports of semi-processed leather. Three years later, the government decided to ban exports of semi-processed leather (crust), to push finished goods. To this end, the National

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



Export Coordinating Committee (NECC) focused on developing a leather-products industry by attracting new foreign firms that are players in the global value chain. Since 2011, many large foreign footwear manufacturers have invested in Ethiopia, enhancing exports.²⁰ Although this was an important policy with strategic significance, the ban was not tied to a comprehensive package of financing, training, and technical assistance for the upgrading of tanning capacity to process finished goods. This was a major failure, as significant tanning capacity became idle, further shrinking exports. The policy decision was supported by footwear firms, while tanneries resisted. Another contradictory policy was the export of live animals, which negatively affected the development of the leather and meat-processing industry. The government faced the dilemma of choosing between foreign exchange earnings from live animals and supporting the manufacturing industry, a choice with structural and political implications.²¹ In conclusion, the industrial policy suffered from a combination of constraints, including the inability to develop backward linkages, political economy constraints, and wide latitude for performance. The leather sector also showed the pivotal role of agriculture and the level of complementarity between manufacturing and agriculture (Cramer and Sender 2015). Despite the apparent familiarization, insufficient understanding of the structure of the industry has contributed to weak policy design and execution. Policy instruments were not supported by reciprocal control mechanisms. Unlike floriculture, the state–industry relationship was weak.

26.2.3.2 Linkage Effects New industries emerged as an outcome of strong linkage dynamics from floriculture, namely packaging, air cargo, and new growth corridors (although these emerged rather slowly (Appendix 1)). The industry relies on packaging materials, and had to import all its requirements. This demand was a clear signal for new investments. Supports and policy inducements were used to develop a packaging industry, including incentives and standards, and facilitate coordination. Historically, air cargo was not a major business for EAL until 2005. As noted above, government policy led EAL to develop freight and cool chain capacity and to provide a reliable airfreight service, such that air cargo is now a strategic business for EAL. Research shows that the backward linkage potential in the leather and leatherproducts sector is strong, while forward linkage potential is weak.²² The transformation of the smallholder livestock sector was minimal and there are no large-scale ranches in

²⁰ For instance, Hua Jian, which employs 4,000 workers, and Gorge Shoe are building two footwear industrial parks. Leading foreign manufacturers continue to invest because of competitive labour costs, potential sources of inputs (from a longer-term perspective) and duty-free privileges in European and US markets. ²¹ The pastoralist community relies on selling live animals and the cross-border trade. ²² See international comparisons of sectoral interdependence (based on Italy, Japan, and the USA) by H. Chenery and T. Watanabe as quoted by Hirschman (1958: 106–7).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Ethiopia. Despite the huge backward linkage potential, this experience shows that linkage dynamics are not automatic and require effective policy responses.

26.2.4 Policy Instruments and Policy Organizations As already suggested, a range of policy instruments has been introduced to support selected sectors, such as subsidized development financing, export promotion incentives (devaluation, duty-drawback, voucher schemes, foreign currency retention), trade protection, investment incentives, foreign exchange allocations, privatization and the use of SOEs in strategic areas, and the establishment of a sectoral institute and national exports coordination mechanism. But the execution and monitoring of incentives were not uniform, partly because some incentives (for instance duty-drawback or voucher schemes) needed tighter monitoring and more advanced administrative capabilities than others. Devaluation did not require any administrative capacity, while investment incentives were easier to administer than export-promotion incentives. Performance criteria and ‘reciprocal control mechanisms’ were not used properly, highlighting the rudimentary nature of industrial policies. Despite these shortcomings, in floriculture performance did respond to incentives, to a significant degree because the narrow latitude for performance standards strengthened export discipline. This was not the case in the leather and leather-products industry, while monitoring the few large firms was not difficult in the cement industry. It was also evident that incentives had a varied impact on different sectors: for instance, devaluation benefited floriculture but weakened the financial position of capital investment projects, which had to import capital goods. The intensity, concentration, and coordination of support improved coherence and impact, as shown positively in floriculture and negatively in leather. The key lesson is that policy instruments should not be viewed as a menu to choose from, a common misconception. Policy instruments should be used creatively to stimulate the specific industry based on an understanding of the industrial structure, context, and the requirements for monitoring reciprocity. For instance, this means cultivating a cadre of highly trained (PhDs) staff in long-term development finance institutions, with specific knowledge and understanding of specific sectors. Readiness to adjust approaches during execution and to drop instruments when they fail to stimulate the industry is also essential. It also means that policy instruments will need to shift and be upgraded to meet the new demands of the industry (Appendix 26A).

26.2.4.1 Understanding industrial structure Structural transformation is the shift towards new activities with higher productivity, and the industrial policies to achieve this aim (Ocampo et al. 2009). However, industrial policies cannot be designed without a sectoral approach and understanding of the industrial structure. Pressure devices can be used based on knowledge of technological

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



and economic characteristics. The promotion of linkage effects and the role of political economy also demand an understanding of the industry, while entering the international setting requires familiarity with the industry’s global value chain. Policy instruments should be wisely designed, monitored, and flexibly changed based on an in-depth grasp of the industry.

26.2.4.2 Policy Organizations The establishment of sectoral agencies and institutes is a critical component in the industrial policy observed in South Korea and Taiwan, where sectoral development institutes have played pivotal roles in supporting specific industries in terms of export promotion, skills development, research and development, and enhanced coordination. Ethiopia adopted this practice by establishing the Leather Industry Development Institute (LIDI) in 2010 and other institutes for textiles, food and beverages, pharmaceuticals and chemicals, etc. EHDA was established based on Kenya’s experience and with a push from the industry. Although the outcomes have been positive, major constraints were observed. Because of weak coordination among government offices, most of the institutes’ efforts have been directed to addressing short-term obstacles. Moreover, the institutes were unable to support industry fully because of low-level staffing in terms of expertise and experience. Twinning with foreign institutes has been promoted to develop capacity, with limited results. Linkages between institutes and universities and technical schools have been weak. Capacity-building of the institutes, with a focus on export promotion, skills development, and the development of technological capacity, is critical to increased participation by and nurturing of domestic industrialists. The strategic role of these sectoral organizations is crucial to effective industrial policy, and a single agency should serve as a focal point for each sector. The NECC, chaired by the prime minister and made up of relevant government agencies, was established after 2005. A regular full-day monthly meeting was held for almost a decade, and has played a critical role in addressing constraints in export performance. However, coordination has become the most binding constraint in the execution of industrial policies, despite multiple efforts. Multiple organizations serve industrial policy, such as the Ethiopian Investment Commission (EIC) spearheading investment promotion; the DBE, the development financing arm; the Industrial Parks Development Corporation (IPDC); major regulatory bodies such as the National Bank of Ethiopia (NBE), the Ministry of Finance and Economic Development, the Ethiopian Revenue and Customs Authority; and SOEs in strategic areas.²³ Developing intergovernmental cooperation mechanisms in very late industrialization requires relentless efforts and a more comprehensive approach.

²³ SOEs can play strategic roles if the state is selective and disciplined about fostering their competitiveness and developing technological capabilities.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

26.3 P L  I P

..................................................................................................................................

26.3.1 Learning-by-doing and Late Industrialization It is argued that learning-by-doing is the primary means of mastering production among late industrializers, and this is equally valid for policy making. Catch-up can thus be understood as a ‘process of learning how to compete’ in which the student plays a more central role, despite the importance of the teacher (Amsden 1989: v, vi, 3; Arrow 1962, 2014; Ohno and Oqubay 2019; Oqubay and Tesfachew 2019). From this perspective, the aim and role of industrial policies is to advance the pace of learning and to shape its direction. This is achieved by fostering the learning environment through instruments such as the reciprocity principle and export discipline, etc. Policy learning in Ethiopia illustrates the importance of policy independence and emulation in addition to learning-by-doing. As the three case studies demonstrate, policy making is often complex, full of tensions and conflicts, and policy learning was evident throughout the policy making process (Oqubay 2015; Oqubay and Tesfachew 2019). The Ethiopian government designed policies for the different sectors, which were neither a complete success nor a complete failure. The experimentation with policy making thus provided opportunities for new learning from mistakes and successes alike.

26.3.2 Policy Independence Policy learning is closely associated with policy independence. Despite its profound importance, policy independence may appear to be a blurred concept. Above all, it means: . . . the right, and political space, to make policy choices free of political pressure, or at any rate, without succumbing to particular [narrow] interests. From a slightly more unusual perspective, it means reserving the right to make mistakes and, in the process, to learn from them. Policy independence also means the freedom to make major policy decisions that entail risks and bold experiments. Without this dimension, policy decisions will sustain the status quo. (Oqubay 2015: 286)

Policy making in Ethiopia has been characterized by relative policy independence, including from donors and IFIs. The struggle to achieve this independence has been starkly outlined by Stiglitz (2002: 32) ‘[W]hen I arrived in 1997, Meles was engaged in a heated dispute with the IMF, and the Fund had suspended its lending program . . . Ethiopia resisted the IMF’s demand that it “open” its banking system [to foreign banks]’. Moreover, the government rejected uniform privatization of public enterprises, reforming public land ownership, and ‘crowding out’ of the private sector.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



It also focused on university expansion, despite the advice to focus on primary schools, and expansion of the energy sector. Notwithstanding these tensions, the government was an effective partner in mutually beneficial programmes and its implementation record has been remarkable. The decisions noted above were critical to structural transformation, despite the costs paid. Moreover, Ethiopia has consistently advocated that African countries should sit in the driver’s seat in respect of their national development agendas. Not all have followed this advice, even though a lack of policy independence has been a major obstacle for many of them. Curtailment of such independence may arise from many factors, including the Washington Consensus and prescriptions by IFIs and their shareholders. Moreover, the colonial legacy appears to play a role in influencing policy making in some African countries. Policy independence may also be undermined by interest groups where state power-holders lack legitimacy and the authority to ensure compliance with their decisions across the whole society. Policy independence does not come free of cost, but the government of Ethiopia could develop its own policies because of domestic political support and the country’s regional geopolitical importance. Clearly, political economy and international factors are at play here.

26.3.3 Emulation and Learning from Others While learning-by-doing is the prime means in policy learning, contacts with forerunners and emulation is also a source of successful catch-up for very late industrializers.²⁴ Emulation is observed in different production and policy areas. Another aspect of emulation, and one much emphasized by Amsden, is the importance of role models. East Asian economies (South Korea, Taiwan, and China), seeking to catch-up, looked to Japan as a role model, while many African countries have no concrete model to emulate but only other abstract theories, usually associated with the gurus of AngloSaxon capitalism.²⁵ Basic policy documents of the Ethiopian government show that East Asian experiences (South Korea, Taiwan, China) have been important sources of policy learning. There have been links with Japanese and South Korean scholars on industrial policy, the Kaizen production philosophy and export promotion (Ohno 2013; Oqubay 2015).²⁶ The transformation of universities, and the technical and vocational education system, were developed along German lines with the support of German specialists. The South Korean experience informed the development of science and technology universities and the ²⁴ Reinert (2010: xxiii) stresses that emulation is at the ‘heart of successful development’ and means ‘imitating to equal or excel’. ²⁵ See quote in Oqubay (2015: 289). ²⁶ For instance, Ohno (2013) observes that Ethiopia’s ‘active and responsive industrial policy, trialand-error attitude, and great attention to sectoral details’ are East Asian in origin.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

establishment of sectoral institutes. The government’s recent policies on industrial parks and clustering were primarily based on experiences from China, South Korea, and Singapore (Oqubay and Lin forthcoming). In terms of industrial policies, most— specifically the reciprocity principle, development financing, export discipline, targeted sectors, and the focus on manufacturing sector—are based on the experiences of East Asian countries (Amsden 1989; Amsden and Chu 2003; Ohno and Oqubay 2019; Oqubay 2019b; Oqubay and Tesfachew 2019). Emulation was not only a source of experience and knowledge, but also a source of optimism and motivation. But there are risks with emulation in terms of policy making. Emulation without a strategic perspective and long-term vision is most likely misdirected. Understanding the context is important, including the peculiarities of national or local conditions. That in turn requires an analytical mechanism, including both independent scholarly research and the perspectives of policy makers. It should be noted that emulation is not synonymous with international benchmarks, which may have limited relevance for the purpose in view. Emulation should, therefore, be viewed as complementary, conditioned to local circumstances to support learning-by-doing, and ultimately tested in experiments.

26.3.4 Learning-by-doing: Should a Country Take on Big and Complex Projects? The ability to make bold policy decisions and undertake complex projects has significant implications for structural transformation. Although such projects may face multiple constraints, they may also offer greater opportunities for learning. This approach contradicts the frequent paternalistic advice by development experts and aid organizations to stay away from big and complex projects. Hirschman (1968a: 129) highlights the conundrum: ‘ . . . how will the country ever learn about technology if it does not tackle technologically complex and problem-rich tasks?’ He adds that ‘a certain “unfitness” of the project for a country becomes an additional and strong argument for undertaking it; . . . if it is successful, [the project] will be valuable not only because of its physical output, but even more so because of the social and human changes it will have wrought’.²⁷ That this perspective has been relevant in the Ethiopian context is evidenced by several large public investment projects, such as in the sugar and chemical industries, expansion of universities and technical schools, railway and energy projects (including large hydro dams) and an integrated housing development programme. For instance, the Grand Ethiopian Renaissance Dam (GERD) is Africa’s largest hydro dam generating 6,000 MW at a cost of $5 billion, financed entirely domestically. It symbolizes the national aspiration to catch up, and it will boost domestic savings capacity by relaxing ²⁷ This point reminds us that late-comers may pursue new industries that may enhance their comparative advantage. See Lin and Chang (2009) and UNECA (2016a).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



the balance of external payments constraint.²⁸ Thus, the Ethiopian government has been undertaking extraordinarily challenging projects, in which Hirschman’s principle of ‘the hiding hand’ exerts strong pressures and inducements on the government and key players. This may boost the efforts made and the learning to ensure that projects do not fail, as the consequences of failure may have significant economic and political costs. Evidence also suggests that the government was ready to drop major projects when policy decisions were not effective, despite the political implications. One such example is the national condominium development programme, which was adopted in the wake of the successful integrated housing development programme launched in 2003 in Addis Ababa (UN-HABITAT 2010). Lack of demand and economical housing technology, and shortage of resources and its impact on inflation were among the major problems. The housing programme in sixty-five towns was terminated when it was evident it had failed in terms of employment creation, alleviating low- and middleincome housing shortages, development of the construction industry, fostering saving, and wealth creation. The scarcest resource for late late-comers is the ability to make development decisions, and development is often more complex than we realize (Hirschman 1958). This is a compelling reason to search for possible pressure mechanisms and inducement devices, such as the latitude for performance standards in different industries or development projects, to accelerate the pace of learning-by-doing (Hirschman 1968b). For instance, the floriculture sector has narrow latitude for standards of performance. The sector’s entire production is exported to the European market, which enforces high standards in terms of quality, timely delivery, etc. The perishability of flowers necessitates maximum care, and the industry requires intensive management. Similarly, the latitude for performance standards in the cement industry is narrow, although dislocation from exports allows wider latitude. Where possible, an understanding of the structure of the industry will help to focus on industries with strong linkage effects, thereby offering an additional impetus to catching up. Latitude for performance standards has been an essential instrument, especially given weak application of the reciprocal control mechanism (Amsden 1989). There have also been serious constraints in and challenges to learning. An important approach used in China’s catch-up (particularly after opening and reform) is effective piloting or experimentation, an approach that is well-captured in the dictums ‘feeling the stones to cross the river’ and ‘seeking truth from facts’. Open-minded pragmatism is critical for learning (Oqubay and Lin 2019; Cheru and Oqubay 2019). In Ethiopia, experimentation has seldom been practised, although it has been used effectively in a few instances. A one-year pilot programme preceded the initiation of the integrated housing programme, although the learning derived from it was inadequate. A more successful recent pilot project has been the development of Hawassa Eco-Industrial Park. It had multiple aims, from gaining experience in designing and building a ²⁸ GERD has stimulated savings and domestic mobilization of resources, which was facilitated by expansion of bank infrastructure. For instance, in 2011–15, branches expanded more than fourfold to 2,868.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

world-class industrial park in record time to applying and testing relevant principles and requirements (linking an investment pipeline with an industrial park, infrastructure development with clustering, environmental requirements, etc.). Much experience was gained and disseminated to new industrial parks, and the process was supported by cooperation between state and private sector and mutual learning to meet the relevant industrial standards and requirements (Oqubay 2019b; Ohno and Oqubay 2019; Oqubay and Tesfachew 2019). Piloting and experimentation should be used as the basis for almost all development projects, as its impact on policy learning is lasting. In addition, policy learning in Ethiopia has been constrained by ineffective institutional support for policy learning. Major constraints include the thin network of research organizations and inadequate experience of using research in policy making. Research institutions in government offices or at universities have weak links to industries. Yet how can there be a sufficiently long-term perspective for policy learning without reliable data and deep analyses? Policy makers should be encouraged to rely on research to enhance policy making and collective and mutual learning between government and industry should be developed. Finally, learning from one’s own successes or failures requires a positive attitude and environment.

26.4 T C  S T

..................................................................................................................................

26.4.1 Politics and the Political Economy of Industrial Policy Developmental states have played a leading role in catch-up by late industrializers, and are characterized by a grand vision, national mobilization, growth-enhancing management of rents, developmental politics, and embedded autonomy, as evidenced in twentieth- and twenty-first-century East Asia (Johnson 1982; Amsden 1989; Chang 1994, 2003, 2015; Evans 1997; Amsden and Chu 2003; Rodrik 2004; Zenawi 2012). In Ethiopia, politics and political economy have shaped policy outcomes at both the sectoral and national levels (Oqubay 2015). Whether a sector is dominated by larger firms or cohesive associations of industrialists or, by contrast, by dispersed smallholders has a substantial impact on the kinds and degree of political pressure that can be brought to bear on government and hence on policy making. Political pressure by social groups depends on their visible presence, strength and cohesiveness (Hirschman 1968a; Hall 1986).²⁹ The existing political economy has favoured speculative activities rather than productive investments in export-oriented manufacturing (Oqubay 2015). ²⁹ For instance, as observed in East Asian and Latin American contrasts, and in early eighteenthcentury US reforms.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



There have been variations in dynamism and the absorptive capacity of the private sector among the floriculture, leather and leather products, and cement sectors. In floriculture, government and industry were a good fit (picking each other, as it were), building trust and collective learning. In contrast, in leather and leather products, path dependency (low-value addition and a fixed mind-set) and internal fragmentation undermined collective learning (Oqubay 2015; Ohno and Oqubay 2019; Oqubay and Tesfachew 2019). Domestic floriculture firms view FDI firms as sources of technology and market capability, while mutual distrust characterizes the leather sector. There have been tensions in the floriculture sector, partly because of the largest firm’s logistical privileges, but conflicts have been resolved and changes have been negotiated.³⁰ Unlike in South Korea, for example, where the political economy allowed for a concentration of ‘intermediate assets’ among national champions, federalism, ethnic diversity and a commitment to equitable regional growth make such a concentration of rents, industrial clustering, or agglomeration more difficult in Ethiopia. The ruling party’s cohesive political and economic thinking has its roots not only in the disposition to learn from the rapid industrialization of East Asian economies, but also in its emergence as a wartime coalition fighting against the Derg’s military totalitarian rule (1975–91) (De Waal 2012). The government’s pursuit of developmental goals, embedded in the ‘Ethiopian Renaissance’, and focused on longer-term public investment, has also been facilitated by continuity of political rule. Its claim to legitimacy has been based above all on its support in rural Ethiopia, a legacy of the liberation struggle. This legitimacy is also tied to the country’s rapid economic growth and focus on more inclusive rural transformation, which has reduced rural poverty and improved economic empowerment.³¹ Despite widely recognized economic successes, there has been political discontent in urban and certain rural areas, especially after the contentious 2005 national election.³² The increased importance of political and economic inclusiveness, young people’s rising expectations and tensions within ethnic-based federalism remain significant challenges for government.³³ Considering the long history of political fragility, ethnic diversity and widespread and profound poverty, a commitment to equitable growth and federalism are essential.³⁴ The government could use this situation as an ‘internal ³⁰ There is also clearly a local political economy whereby—for example—large floriculture firms should do deals with local officials to ensure smooth operations. There have also been conflicts of interest over, for example, levels of compensation and the accuracy of compensation targeting, so that despite the many ‘winners’ (investors, the balance of payments, indirect beneficiaries such as service suppliers, and employees) there are also losers, including people who may have lost access to land or water, or who may not get cheap credit for other purposes because it is directed to floriculture, etc. ³¹ With a Gini coefficient of 30, ‘Ethiopia remains among the most egalitarian countries in the world’ (IMF 2015: 5). Despite reductions in poverty, food price inflation has impacted the poorest social groups. ³² But also, as recently as 2016. ³³ The expansion and transformation of university education and technical schools has given rise to the challenge of creating hundreds of thousands of professional and technical jobs for graduates. ³⁴ See Hirschman (2013: 74–90) ‘The changing tolerance for income inequality in the course of economic development’ for a discussion of how shifting expectations in segmented societies may lead to disappointment and alienation, and the role of the ‘hope factor’ and ‘tunnel effect’.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

threat’ to further foster its developmentalism and deepen structural transformation to meet popular demands (Doner et al. 2005).³⁵

26.4.2 Climate Change and Insertion into Global Value Chains In addition to internal structural constraints, there are significant global trends that impact the country’s policies and plans, to which Ethiopian industrial policy must also adapt. For instance, industrialization poses massive perils for the environment, as has been witnessed in advanced as well as emerging economies. In recent years, climate change has become a major global issue, and a consensus has emerged on how to tackle it, for instance the COP 21 Summit (UNCTAD 2016; UNECA 2016b).³⁶ Consumer tastes are shifting, putting pressure on firms and governments for increased environmental protection. Ethiopia has adopted a green economy strategy that aims to reduce greenhouse gas emissions by 64 per cent. To meet this objective, industrial policies will require the incorporation of measures to mitigate environmental damage and climate effects. Another key trend is the rise of global value chains, characterized by the expansion of global production networks (Ruigrok and Tulder 1995). The increasing internationalization and concentration of economic activities, in which multinational companies play a pivotal role, is referred to as the ‘global business revolution’ by Nolan (2014). This process has been accelerated by advances in ICT and space-shrinking transportation. Global value chains in different sectors are characterized by distinct characteristics (Schmitz 2007).³⁷ What matters is not openness to international trade but rather the mode of insertion into the global economy, and interconnectedness to domestic linkages (Ocampo et al. 2009).³⁸

26.4.3 Structural Transformation Constraints and the Way Forward Despite rapid economic growth in Ethiopia, it is evident that progress in terms of structural transformation has been inadequate. Rapid growth has not seen a ³⁵ See Doner et al. (2005), who argue for the positive role of threats in developmental states and catch-up. See Chang (1994: 123–7) on the politics of industrial policy in Korea and Evans (1997) on embedded autonomy. ³⁶ Climate change requires a global response and ‘international cooperation and coordination’ (Rodrik 2012: 248). ³⁷ For instance, some industries (such as apparel and footwear) are buyer-driven, while others (such as commercial aircraft production) are producer-driven (Schmitz 2007). ³⁸ Despite restrictions imposed by WTO and other bilateral regional agreements, developing countries have room to pursue industrial policies (Ocampo et al. 2009; Rodrik 2012: 198–200; UNECA 2016a: 115–42). Rodrik (2012: 248) argues that the destinies of such countries are ‘determined largely by what happens at home rather than abroad’.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



corresponding shift in the share of manufacturing in employment, output, and exports, and agriculture continues to employ three-quarters of the population and account for 37 per cent of GDP (NPC 2016). Moreover, the sluggish growth of exports has been dominated by low-value and primary commodities, a situation that has in turn become a binding structural constraint. The balance of payments constraint has increased as exports fall short of covering the surge in imports, pressing the country to rely on less preferable external resources. With 2.3 million youths entering the employment market annually, job creation is a strategic issue. The government has recognized that structural transformation is the route to catchin up and ensuring sustained growth. This is an enormous challenge and has profoundly shaped the development of a ten-year plan.³⁹ The government’s Vision 2025, which aims to make Ethiopia ‘the leading manufacturing hub in Africa’, puts greater emphasis on expanding manufacturing output and large-scale growth in industrial employment. This involves an annual growth rate in the manufacturing sector of 25 per cent, and a fourfold increase of manufacturing output (from 5 to 20 per cent of the GDP) and exports (from 12.5 to 50 per cent). Manufacturing is strongly associated not only with the creation of permanent jobs but also with strong employment linkages by stimulating indirect jobs. This requires the attraction of massive investment in key manufacturing industries, primarily in light and basic industries. To this end, a shift towards a proactive and targeted investment approach has become essential.⁴⁰ In addition, a better understanding of global value chains has resulted in a focus on attracting leading international buyers as anchors and related international manufacturers.⁴¹ New incentives and support schemes have been designed to support domestic industrialists. Another key policy initiative has been the vigorous and comprehensive promotion of industrial parks and industrial clusters (Oqubay 2019b; Oqubay and Tesfachew 2019).⁴² This policy approach contributes to effective environmental protection, rapid industrialization and the development of domestic linkages. Moreover, industrial parks will be specialized to promote linkages, vertical integration, and to learning and skills development.⁴³ As a learning model, the Hawassa Eco-Industrial Park has been built in ³⁹ Successive five-year plans, such as the Growth and Transformation Plan 2010/11–2014/5 (GTP I) have underscored these challenges (for instance, developing manufacturing and the exports sector) and offered lessons for Vision 2025, which will determine the catch-up trajectory of Ethiopia. ⁴⁰ Opportunities that may positively contribute to the success of the vision include the extension of AGOA until 2025 and the potential relocation of Chinese manufacturing in labour-intensive industries (Lin 2015). ⁴¹ Akitumi Kuchiki’s ‘flowchart approach to industrial clusters’ model emphasizes the initial agglomeration stage in which industrial park, anchor firm, related firms and capacity building are involved; later shifting to the innovation stage (Ohno 2013: 70). ⁴² Industrial parks and clusters are based on external economies, namely localization that focuses on specific industries and urban economies; and government policies can foster industrial clusters (Marshall 1920; Jacobs 1969; Krugman 1993; Porter 1998). See also Stein (2012), Ohno (2013), and Lin (2015) on industrial parks in Asia and Africa. ⁴³ Oqubay (2019b; Oqubay and Lin 2019).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

record time and to the highest environmental standards.⁴⁴ Based on this model, a dozen industrial parks are being built along the major railway corridor, thereby alleviating the logistical constraints manufacturers and exporters currently face and increasing productivity and profitability by cutting transportation times and costs. The envisaged ‘plug-and-play’ model of industrial parks serves as an incubator for new domestic industrialists, while working with major international manufacturers creates a learning ecosystem and facilitates learning-by-doing. Industrial parks have been developed by emulating Asia and to support structural transformation (Oqubay 2019b; Oqubay and Tesfachew 2019; Oqubay and Lin forthcoming). They will also enhance the prioritization of infrastructural projects and improve the business climate by providing a one-stop service. In the Ethiopian context, this new strategic approach to manufacturing investment and agglomeration is a distinctive feature of Vision 2025. Outcomes will depend on effective execution of the industrial policies and the pace and scope of learning. All these attempts further demonstrate policy learning and a pragmatic approach to industrial policy and industrialization, and a reenergized commitment to structural transformation and catch-up.

26.5 C: L   E

.................................................................................................................................. This chapter has discussed the Ethiopian experience with industrial policy and performance in the early part of this century. It has done so chiefly by comparing interventions and trajectories in three sectors. Important lessons can be learned from the Ethiopian experiment, and it is hoped that these experiences will add to a broader learning process throughout Africa, where there is increasing interest in researching, designing, and refining industrial policies. First, the Ethiopian experiment shows that structural transformation and industrial policy can work in Africa.⁴⁵ However, it also shows that structural transformation and catch-up are a colossal challenge. Next, it shows that industrial policies matter, and the state matters.⁴⁶ Destiny can be shaped by development paths and policies. Despite the dominant prescription that the state should play a minimal or ⁴⁴ A zero liquid discharge (ZLD) facility has been built. Specializing in apparel and textiles, the park will employ up to 60,000 workers and generate $1 billion in export earnings. A leading global retailer serves as anchor, the industrial cluster enjoys 100 per cent occupancy by both domestic and foreign manufacturers. See UNECA (2016b: 195–6) on Hawassa Eco-Industrial Park as an example of green industrialization. ⁴⁵ Ohno (2013: 36–9) emphasizes that proactive industrial policy is based on market forces (under globalization), a strong role for the state, vigorously developing skills, capacity, and technology, effective state–private sector partnerships, and a deep understanding of the industry. ⁴⁶ See Schwartz (2010); UNCTAD-UNIDO (2011); Mazzucato (2013); and UNCTAD (2016).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



at most a facilitating role, Ethiopian experience shows that structural transformation and catch-up require that the state play a pivotal strategic part. This includes formulating a vision and strategy, mobilizing the society and its resources around the vision and development projects, managing tensions, and nurturing developmental partnerships.⁴⁷ The chapter also shows that, despite growth in all three selected sectors under a single industrialization strategy, performance and policy outcomes were uneven. This highlights the importance for policy makers of understanding and engaging with the interactions and dynamics of specific industries and global value chains, maximizing linkage effects and having a deep understanding of politics/political economy. All policy decisions are determined through the political process, interest groups, and the state–society relationship (Hirschman 1958; Chang 2003; Rodrik 2004; Whitfield et al. 2015). This has significant repercussions for policy design and execution. Fourth, the Ethiopian experience has important implications for policy learning. As Amsden (1989: viii) stresses: ‘All late industrialisers have in common industrialisation on the basis of learning, which has conditioned how they behaved’ (emphasis added). This insight applies not only to industrial production, but also to policy making. The primary source of policy learning has been learning-by-doing, involving both successes and failures, and by ‘failing better’. Embracing bold experiments and grand projects has had positive learning implications. The Ethiopian experiment also shows that this was possible because of policy independence and the use of coping devices such as linkage pressures and latitude for performance standards. Emulation in the form of learning from role models was also used in industrial policy making. For instance, lessons can be learnt from East Asia, such as export discipline and a focus on manufacturing, reciprocal control mechanisms, and choice of priority sectors based on productive rather than political criteria. Industrial policy making in Ethiopia is a work-in-progress, but experience in that country does show that industrial policy can work and thrive in a low-income African country, and that the state can and should play an activist developmental role. For African countries, perhaps an important point of departure is the adoption of a structural transformation perspective, to enable understanding of the strategic importance of manufacturing and exports, as well as their complementarities with agriculture. For, among other things, structural transformation is squarely about the transformation of agriculture, rather than leaving it behind.

⁴⁷ See Cramer and Chang (2015), UNECA (2016a), and Rodrik (2004) on meta-structural arguments that view climate, geography, history, or culture as key determinants. The economic history of late-industrializers offers many examples of the development paths and policies that lead to catch-up.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

 26

..................................................................................................................................

S        Import-substitution industry: Cement

Export-oriented industry: Leather

Export-oriented industry: Floriculture

Capital intensive Big corporations Narrow FDI, SOE, Domestic Process production Capital deepening

Labour intensive Family business Wide Domestic, FDI Batch production Capital widening

Labour intensive Family business Exceptionally narrow FDI (2/3), Domestic

Weak Huge potential, weak outcome

Moderate (air cargo) Moderate (packaging)

Fiscal linkage Employment linkages History

Strong Strong (from construction industry) Strong Strong Founded in 1960s

Weak Weak Founded in 1920s

Moderate Strong Founded in 2000s

Political economy Industry players Industrial association State—private partnership

Nil Strong

Fragmented Weak Modest

Cohesive Dynamic Strong

Policy instruments Investment incentives Export promotion incentives Development financing Foreign exchange allocation Protection—Export ban

Yes NA Yes (large firms) Priority Nil

Yes Yes Yes NA NA

Protection—Import ban Five-year plan SOE/Privatization Sector-specific institute Skills & market support Reciprocity principle

High Yes Active SOE presence Established 2014 Insignificant Weak

Yes Yes Yes NA On semi-finished goods Insignificant Yes Privatized LIDI—in 2000 Moderate Weak

Policy outcomes Production output Export earnings Employment creation Competitiveness Total economic impact

High growth NA Moderate Moderate Significant

Slow growth

Moderate growth High growth Significant High Significant

Industrial structure Labour/capital intensity Ownership structure Latitude for performance Ownership-origin Technology Expansion approach Linkage dynamics Forward linkages Backward linkages

Weak Weak Weak

Capital widening

NA Yes NA EHDA—in 2009 Modest Weak

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



A The author thanks Christopher Cramer for insightful and useful comments, Tadesse Gurmu for data collection, and Peter Colenbrander and Binyam Arkebe for editing support.

R Amsden, Alice H., 1989. Asia’s Next Giant: South Korea and Late Industrializationí, Oxford: Oxford University Press. Amsden, Alice H. and Wan-Wen Chu, 2003. Beyond Late Development: Taiwan’s Upgrading Policies, Cambridge MA: MIT Press. Arrow, Kenneth J., 1962. ‘The Economic Implications of Learning by Doing’, The Review of Economic Studies, 29 (3), pp. 155–73. Arrow, Kenneth, 2014. ‘Commentary’, in Joseph E. Stiglitz and Bruce C. Greenwald, eds, Creating a Learning Society: A New Approach to Growth, Development and Social Progress, New York: Columbia University Press. Chandler, Alfred, 2004. Scale and Scope: The Dynamics of Industrial Capitalism, Cambridge MA: Harvard University Press. Chang, Ha-Joon, 1994. The Political Economy of Industrial Policy, London: Macmillan. Chang, Ha-Joon, 2003. Kicking Away the Ladder: Development Strategy in Historical Perspective, London: Anthem Press. Chang, Ha-Joon, 2015. ‘Industrial Policy: Can Africa do it?’, in Joseph E. Stiglitz, Justin Y. Lin, and Ebrahim Patel, eds, The Industrial Policy Revolution II: Africa in the 21 Century, New York: Palgrave Macmillan. Cheru, Fantu and Arkebe Oqubay, 2019. ‘Catalyzing China-Africa Ties for Africa’s Structural Transformation’, in Justin Lin and Arkebe Oqubay, eds, China-Africa and an Economic Transformation, Oxford: Oxford University Press. Cramer, Christopher and Ha-Joon Chang, 2015. ‘Tigers or Tiger Prawns? The Africa Growth “Tragedy” and “Renaissance” in Perspective’, in Celestin Monga and Justin Yifu Lin, eds, The Oxford Handbook of Africa and Economics. Volume I: Context and Concepts, Oxford: Oxford University Press. Cramer, Christopher and John Sender, 2015. Agro-processing, Wage Employment and Export Revenue: Opportunities for Strategic Intervention, Pretoria: TIPS. De Waal, Alex, 2012. ‘The Theory and Practice of Meles Zenawi’, African Affairs, 112 (446), pp. 148–55. Doner, Richard, Bryan Ritchie, and Dan Slater, 2005. ‘Systemic Vulnerability and the Origins of Developmental States: Northeast and Southeast Asia in Comparative Perspective’, International Organization, 59(2), pp. 327–61. ERCA (Ethiopian Revenue and Customs Authority), 2016. ‘Ethiopian Export Performance 1998–2015’, November, Addis Ababa, Ethiopia. Unpublished. Evans, Peter, 1997. Embedded Autonomy: States and Industrial Transformation, Princeton, NJ: Princeton University Press. Foreign Policy, 2015. ‘Africa’s Boom Is Over’. 31 December. Available at: http://foreignpolicy. com/2015/12/31/africas-boom-is-over/

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Gerschenkron, Alexander, 1962. Economic Backwardness in Historical Perspective, Cambridge MA: Harvard University Press. Hall, Peter, 1986. Governing the Economy: The Politics of State Intervention in Britain and France, Cambridge: Polity Press. Hamilton, Alexander, 1934. Papers on Public Credit, Commerce and Finance, ed. Samuel McKee, New York: Columbia University Press. Hirschman, Albert O., 1958. Strategy of Economic Development, New Haven, CT: Yale University Press. Hirschman, Albert O., 1968a. Development Projects Observed, Washington, DC: The Brookings Institution. Hirschman, Albert, 1968b. ‘The Political Economy of Import-Substituting Industrialization in Latin America’, Quarterly Journal of Economics, 82 (1), pp. 1–32. Hirschman, Albert, 2013. The Essential Hirschman, Princeton, NJ: Princeton University Press. IMF, 2015. The Federal Democratic Republic of Ethiopia. Selected Issues. IMF Country Report No. 15/326, 4 September. Available at: https://www.imf.org/external/pubs/ft/scr/ 2015/cr15326.pdf Jacobs, Jane, 1969. The Economy of Cities, New York: Vintage Books. Johnson, Chalmers, 1982. MITI and the Japanese Miracle: The Growth of Industrial Policy, 1925–1975, Stanford, CA: Stanford University Press. Kaldor, Nicholas., 1967. Strategic Factors in Economic Development, Ithaca, NY: Cornell University Press. Krugman, Paul, 1993. Geography and Trade, Cambridge, MA: MIT. Lin, Justin, 2015. ‘China’s Rise and Structural Transformation in Africa: Ideas and Opportunities’, in Célestin Monga and Justin Yifu Lin, eds, The Oxford Handbook of Africa and Economics. Volume II: Policies and Practices, Oxford: Oxford University Press. Lin, Justin and Ha-Joon Chang, 2009. ‘Should Industrial Policy in Developing Countries Confirm to Comparative Advantage or Defy it? A Debate between Justin Lin and Ha-Joon Chang’, Development Policy Review, 27 (5), pp. 483–502. Lin, Justin and Arkebe Oqubay, 2019. China-Africa and an Economic Transformation, Oxford: Oxford University Press. List, Friedrich, 1856. National System of Political Economy, Vol. I–IV, Memphis, TN: Lippincott. Marshall, Alfred, 1920. Principles of Economics, London: Palgrave Macmillan. Mazzucato, Mariana, 2013. The Entrepreneurial State: Debunking Public vs. Private Sector Myths, London: Anthem Press. McKinsey Global Institute, 2016. Lions on the Move II: Realizing the Potential of Africa’s Economies, McKinsey & Company, September. MOFED (Ministry of Finance and Economic Development), 2010. ‘Growth and Transformation Plan 2010/11–2014/15 (GTP I)’, Addis Ababa: MOFED. Nayyar, Deepak, 2013. Catch Up: Developing Countries in the World Economy, Oxford: Oxford University Press. Nolan, Peter, 2014. Chinese Firms, Global Firms: Industrial Policy in the Era of Globalization, New York: Routledge. NPC (National Plan Commission), 2016. Growth and Transformation Plan 2015/16–2019/20 (GTP II), Addis Ababa: NPC. Ocampo, José A., Codrina Rada, and Lance Taylor, 2009. Growth and Policy in Developing Countries: A Structuralist Approach, New York: Columbia University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

:    



Ohno, Kenichi, 2013. Learning to Industrialize: From Given Growth to Policy-aided Value Creation, New York: Routledge. Ohno, Kenichi and Arkebe Oqubay, 2019. How Nations Learn: Technological Learning, Industrial Policy, and Catch Up, Oxford: Oxford University Press. Oqubay, Arkebe, 2015. Made in Africa: Industrial Policy in Ethiopia, Oxford: Oxford University Press. Oqubay, Arkebe, 2019a. ‘Industrial Policy and Late Industrialization in Ethiopia’, The Oxford Handbook of the Ethiopian Economy, Oxford: Oxford University Press. Oqubay, Arkebe, 2019b. ‘The Structure and Performance of the Ethiopian Manufacturing Sector’, The Oxford Handbook of the Ethiopian Economy, Oxford: Oxford University Press. Oqubay, Arkebe and Taferre Tesfachew, 2019. ‘Learning to Catch Up in Africa’, in Ohno, Kenichi, and Arkebe Oqubay, eds, How Nations Learn, Oxford: Oxford University Press. Oqubay, Arkebe and Justin Lin, forthcoming. The Oxford Handbook of Industrial Hubs and Economic Development, Oxford: Oxford University Press. Penrose, Edith, 1995. The Theory of the Growth of the Firm, 3rd edn, Oxford: Oxford University Press. Porter, Michael, 1998. The Competitive Advantage of Nations, Basingstoke: Palgrave Macmillan Reinert, Erik S., 2010. How Rich Countries Got Rich and How Poor Countries Stay Poor, London: Constable. Rodrik, Dani, 2004. ‘Industrial Policy for the 21st Century’. September. Available at www.hks. harvard.edu Rodrik, Dani, 2012. The Globalization Paradox: Why Global Markets, States, and Democracy Can’t Coexist, Oxford: Oxford University Press. Ruigrok, Winfried and Rob van Tulder, 1995. The Logic of International Restructuring, New York: Routledge. Schmitz, Hubert, 2007. ‘Reducing Complexity in the Industrial Policy Debate’, Development Policy Review, 25 (4), pp. 417–28. Schwartz, Herman, 2010. States versus Markets: The Emergence of a Global Economy, New York: Palgrave. Stein, Howard, 2012. ‘Africa, Industrial Policy, and Export Processing Zones: Lessons from Asia’, in Akbar Noman, Kwesi Botchwey, Howard Stein, and Joseph E. Stiglitz, eds, Good Growth and Governance in Africa: Rethinking Development Strategies, Oxford: Oxford University Press. Stiglitz, Joepsh E., 2002. Globalization and its Discontents, London: Penguin. Szirmai, Adam, Wim Naudé, and Ludovico Alcorta, 2013. Pathways to Industrialization in the 21st century: New Challenges and Emerging Paradigms, Oxford: Oxford University Press. Thirlwall, Anthony, 2013. Economic Growth in an Open Developing Economy: The Role of Structure and Demand, Cheltenham: Edward Elgar. UNCTAD, 2016. ‘Structural Transformation for Inclusive and Sustained Growth’. Trade and Development Report, New York: UNCTAD. UNCTAD-UNIDO, 2011. ‘Fostering Industrial Development in Africa in the New Global Environment’. Economic Development in Africa Report, Geneva: UNCTAD-UNIDO. UNECA, 2016a. Transformative Industrial Policy, Addis Ababa: UNECA. UNECA, 2016b. ‘Greening Africa’s Industrialization’. Economic Report on Africa, Addis Ababa: UNECA. UN-HABITAT, 2010. The Ethiopia Case of Condominium Housing: The Integrated Housing Development Programme, Nairobi: UNHABITAT.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

Wade, Robert, 2004. Governing the Market: Economic Theory and the Role of the Government in East Asian Industrialization, Princeton, NJ: Princeton University Press. Whitfield, Lindsay, Ole Therkildsen, Lars Buur, and Anne Mette Kjær, 2015. The Politics of African Industrial Policy: A Comparative Perspective, Cambridge: Cambridge University Press. Zenawi, Meles, 2012. ‘States and Markets: Neoliberal Limitations and the Case for Developmental State’, in Akbar Noman, Kwesi Botchwey, Howard Stein, and Joseph E. Stiglitz, eds, Good Growth and Governance in Africa: Rethinking Development Strategies, Oxford: Oxford University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

        New Evidence from Tanzania ......................................................................................................................

 ,  ,     

27.1 W  T F  W D T D?

.................................................................................................................................. T Micro, Small and Medium Sized Enterprise Survey (MSME) 2010 is nationally representative and covers a little under 3 million businesses and around 5 million employees. It primarily covers firms in the informal sector. The sampling framework for this survey is households and the selection of households is based on the 2002 census. This poses at least two problems. First, because the survey is household based, it is representative of households and not businesses or firms. Thus, since Tanzania is still a very poor country, we are likely to miss some of the more productive businesses. Indeed, an analysis of the data reveals that medium-sized firms are under-represented in this dataset (MSME Report 2012). Second, because the sampling framework is 2002 census, it oversamples rural households. This is because there was a significant reduction in rural activity between 2002 and 2012 as documented in Section 27.2 of this chapter. Therefore, the reader should keep in mind that our analysis is likely to understate the contribution of small businesses to economy-wide productivity and employment and also to understate the importance of small businesses in urban areas. These firms operate in all twenty-six geographic regions of Tanzania. Unlike the distribution of firms, the distribution of employment is almost evenly split between

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

rural and urban areas indicating that firms in urban areas have, on average, more employees. Overall, the distribution of employment in small firms is roughly proportional to population size with the highest concentration of employment in MSMEs located in Dar es Salaam (17.32 per cent) and the lowest concentration of employment in Kusini Pemba (0.35 per cent). Apart from Dar es Salaam, other regions with a high concentration of MSMEs include Mwanza (10.7 per cent), Mbeya (8.02 per cent), and Morogoro (6.44 per cent). Although the MSMEs operate in a wide range of activities, the bulk of these activities can be classified into trade services (79.9 per cent) and manufacturing (16.7 per cent). Manufacturing enterprises operate in the following six sub-sectors: grain milling (1.7 per cent), beverages (8.3 per cent), textiles (3.4 per cent), wood (0.5 per cent), building materials (1.1 per cent), and furniture (1.6 per cent). Firms in the trade services sector operate primarily in retail (47.1 per cent), food services (22.3 per cent), and beverage services (7.9 per cent). Many of these activities appear to have strong links to agriculture but without further information, it is not possible to identify which ones and exactly how these linkages work. This is an important area for future research.

27.2 S- M  S B O

.................................................................................................................................. The MSME survey includes three questions designed to elicit the reasons for which businesses are opened. The first question is: what was your main occupation before you started this business? Responses revealed that 36.73 per cent of respondents were previously farming, 21.23 per cent of respondents were previously working in another business, and 19.99 per cent of respondents were previously housewives. Only 7.56 per cent of respondents reported that they were previously unemployed. The second question is: for what reason did you choose your line of business? A little under half of all business owners say that the reason they chose their line of business is because they saw a market opportunity. This response is more pronounced in trade services than in manufacturing. The second most common reason for operating in a particular line of business is that the owners’ capital could only finance that line of business. The third and most common reason for picking a line of business was having friends and family who operated a similar business. And the fourth reason for opening in a particular line of business was business experience. The third question is: if you were offered a full-time salary paying job, would you take it? Only 46.57 per cent of small business owners would leave their current business for a full time salaried position. Of the respondents who would prefer a full time salaried job, 63.92 per cent say they would like to work for the government. This is consistent with results reported in Banerjee and Duflo’s (2007) analysis of the

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



economic lives of the poor. Another 23.99 per cent of the respondents say they would prefer to work for a large private company. The reported reason for preferring a full time salaried position is better security of income.

27.3 T P H  J C  S F

.................................................................................................................................. We use kernel densities of the log of value added per worker to examine the productive heterogeneity of MSMEs. Value added is computed as the firm’s average monthly sales minus the firms’ average monthly costs of production. Firms in the MSME database report sales on a monthly basis and thus we are able to take seasonality into account. Our analysis of the productive heterogeneity of firms in the MSME sector reveals two important features of these firms. First, there is significant overlap in productivity between ‘formal’ and ‘informal’ firms. We use two definitions of formality. First, firms are considered formal if they are registered with BRELA, Tanzania’s business registration and licensing agency. Second, firms are considered formal if they have a tax identification number (Tax ID). The overlap in the distribution of productivity between formal and informal MSMEs is significant. Thus, it would be a mistake to classify all informal MSMEs as unproductive and all formal MSMEs as productive. Second, a little over half of the firms in the MSME sector have labour productivity levels higher than the economy-wide average labour productivity in agriculture. This is not surprising and is consistent with our previous work showing that average productivity in the sectors dominated by small firms is consistently higher than average productivity in agriculture—at least for now. What is more surprising is the fact that 15 per cent of the MSMEs have labour productivity higher than economy-wide manufacturing labour productivity. These firms account for 70 per cent of the total value-added generated by the MSME sector. By contrast, the remaining 85 per cent of the MSMEs account for only 30 per cent of the value added generated by the MSME sector. This is important because it underscores the productive heterogeneity of the informal sector. It is very important to keep in mind that firm size in the universe of firms that we are studying in this chapter is by definition very small. So we cannot make broad statements about employment growth among the firms in our sample relative to firms in the formal sector. Nevertheless, by examining employment growth among the MSMEs, we can begin to get a sense of whether any of the MSMEs have the potential to grow into larger firms generating employment for the scores of individuals who would rather have salaried jobs than work in a small business and at the same time take advantage of economies of scale.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

27.4 P D: W H W L?

.................................................................................................................................. The results presented in this chapter indicate that it is a mistake to lump all MSMEs together. Some MSMEs already contribute significantly to employment and productivity growth and have the potential to contribute significantly more. At the other end of the spectrum, the least productive MSMEs are typically owned by people who would rather not be in business. The implication is that blanket policies that offer assistance in the form of business training and access to credit with a view to growing these businesses is likely to be a waste of time and money. Instead, policies targeted at the MSMEs with potential for employment and productivity growth may have large payoffs. This is not the way MSME policy in Tanzania is done today. On the one hand, it would be easy to conclude that MSME policy in Tanzania is not well designed because the people at the top do not really care about MSMEs. For example, Tanzania’s proposed Integrated Industrial Development Strategy for the years 2016/2017 through 2021/22 accords a ‘special’ role to MSMEs in the industrialization of Tanzania. But it includes no details about why they are special or how they might be included in Tanzania’s development strategy (Banerjee 2015). On the other hand, Isaga et al. (2016) provide an up-to-date and excellent overview of the Tanzanian governments’ official initiatives vis-à-vis MSME development. They identify at least fourteen government or quasi-government institutions that deal with MSMEs in one way or another. There appears to be very little coordination between these institutions and none of these programmes have been evaluated in terms of their impact on MSMEs. Instead, they have been evaluated in terms of outputs such as how many training sessions they run. Tanzania is not special in this regard. One of the most ‘successful’ programmes to support MSMEs has been the Indian government’s small-scale reservation policy. It was successful in that it reached large numbers of small entrepreneurs by reserving products in the manufacturing sector for production only by small enterprises. But after nearly sixty years, the policy was widely perceived as a failure and was gradually dismantled beginning in 1997. An evaluation of the impact of the de-reservation by Harrison et al. (2014) finds that the elimination of small-scale industry promotion in India lead to higher growth in employment and wages in districts that were more exposed to the de-reservation policy. Instead of targeting products, the Brazilian government used credit as an instrument for the development of small firms. De Paula and Schneikman (2011) argue that the government of Brazil indiscriminately subsidized credit to small firms leading to an economically inefficient outcome. They develop a theoretical model which they use to show that a blanket policy of subsidized credit to informal firms would lead to an

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



economically inefficient outcome. They then apply their model to Brazil and make the case that subsidized credit lead to the proliferation of inefficient firms. On the other hand, there is some very recent evidence that targeting high potential firms through business plan competitions can have an impact on employment and productivity growth (McKenzie 2015). In 2011, Nigeria’s President Goodluck Jonathan launched a national business plan competition dubbed YouWiN! The programme cost roughly 36 million dollars, almost all of which was contributed by the Government of Nigeria. Although the programme was expensive, McKenzie (2015) compares it to a fiscal stimulus package in the UK and shows that the YouWiN! competition was actually more cost effective. In total, 24,000 applicants applied and 1,204 were eventually awarded an average of around $50,000 each. Importantly, more than half of the winners were randomly selected from a pool of successful applicants which allowed for an evaluation of the causal impact of the programme. McKenzie (2015) finds that three years after applying for the grants, winners were more likely to survive and more likely to have businesses with ten or more employees. However, while the business plan competition seems to have been quite successful in Nigeria, the World Bank has supported competitions in a number of other countries across Africa and it is very unclear whether or not these have been equally successful. Although quite different, Banerjee et al. (2015) find that a micro-credit programme in India that had no significant ‘average’ effects, did have an impact on a specific groups of firms which they call the ‘gung-ho’ firms. They study residents of urban Hyderabad, India six years after a randomly rolled out micro-credit intervention designed to lower the cost of credit and spur business creation among borrowers. They find that a specific group of entrepreneurs whom they label ‘gung-ho entrepreneurs’ reaped significant business benefits from the intervention relative to ‘reluctant entrepreneurs’. Importantly, the ‘gung-ho entrepreneurs’ had already been in business prior to accessing micro-credit so they had some kind of a proven track record. They interpret these results as evidence that heterogeneity in entrepreneurial ability is important and persistent; and that lenders entering a new market may be better off by focusing on borrowers at the intensive rather than extensive margin. In Tanzania, some interesting and very recent programmes have been started by young entrepreneurs whose goal is to target and assist businesses with growth potential. For example, Darecha Limited—a start-up founded by Julius James Shirima and incorporated on 11 August 2014—is bridging the gap between some of Tanzania’s wealthiest firms and MSMEs through micro-venture capital and mentorship. For example, on 25 January 2016, Darecha partnered with Mohammed Dewji—Forbes Africa’s Businessman of the Year 2015—to pilot a competition for entrepreneurs. The competition is designed to identify young (aged 18 to 30years old) entrepreneurs who are already in business. Winners of the competition will be supported along three dimensions: access to finance, access to networks, and mentorship by Dewji himself. Darecha also partnered with STATOIL—Norway’s state-owned oil company—to launch the ‘Heroes of Tomorrow (HoT)’ competition in the Mtwara region of Tanzania with the aim of supporting youth entrepreneurship.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

MerseyTel is following a different model but also with the aim of targeting financial assistance to young entrepreneurs with promising business ideas. MerseyTel runs a bit like a co-operative and does not offer loans per se but offers micro-venture capital. One of the most interesting aspects of MerseyTel is that it is Sharia compliant.¹ The reason this is so interesting is that around a quarter of the MSME owners surveyed in Tanzania report that they do not believe in interest. All of these programmes are very new so it is obviously too soon to tell what kind of impact—if any—they will have. The good news is that all of the programme sponsors reported that they are open to randomization and impact evaluation.² Taken together, the most important policy implication of the evidence presented in this chapter is that if the goal is to grow MSMEs with the potential to contribute to productive employment, policies must be targeted at the most promising firms. Some of the information we already used to classify firms according to productive potential might be used for the purposes of targeting. For example, it should be possible to verify whether or not a firm keeps written accounts. However, verifying the accuracy of the accounts would be cumbersome and time consuming. And for obvious reasons, the MSME data cannot be used to track down high potential firms. Instead, in Subsections 27.4.1 and 27.4.2, we investigate the possibility that there are other salient observable differences between the MSMEs in the in-between sector³ and the rest of the MSMEs. We organize this investigation into two parts: individual and business characteristics and obstacles to doing business as reported by the business owners.

27.4.1 Can Individual and Business Characteristics be Targeted? A comparison of means across high potential MSMEs (our Group 3) and the rest of the MSMEs is presented in Table 27.1. Table 27.1 is divided into two parts: individual characteristics and business characteristics. The individual characteristics include socioeconomic characteristics of the business owners such as age and gender. The business characteristics we examine include whether or not a firm started with a business plan and whether or not a firm is registered as well as the following subgroups: (i) business attributes; (ii) the business owners employment and training history; (iii) infrastructure and technology used by the business; (iv) access to financial services; and (v)) measures of business practices. The results in Panel A that describe the individual characteristics of MSME owners reveal a number of significant differences between the in-between firms and the rest of the MSMEs. Owners of the Group 3 in-between firms are significantly more educated, they are much less likely to be female, they are much less likely to be rural, and they are ¹ Interview with Mohamed H. Kassango, 30 January 2016. ² Interviews with Dewji, 27 January 2016 and Julius Shirima the founder of Darecha, 2 February 2016. ³ The concept of ‘in-between sector’ is developed in Diao and McMillan (2018).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



Table 27.1 How high performing firms compare to the rest: individual and business characteristics Mean (not in- Mean (inDifference in between firms) between firms) means t-test Panel A: Individual characteristics Education (above primary level) Marital status Owner’s age Female Rural Owner is not poor Owner is moderately poor Owner has started other businesses Main source of income is the business Main source of income is farming Owner is member of business savings club Owner is member of a business association Owner has taken expert advice Saw business as a market opportunity Views business as growing

0.194 0.904 36.932 0.521 0.745 0.446 0.354 1.121 0.402 0.195 0.089 0.053 0.170 0.478 0.573

0.478 0.935 38.850 0.304 0.413 0.717 0.196 1.261 0.480 0.022 0.130 0.152 0.152 0.674 0.870

4.844 0.700 1.218 2.935 5.136 3.682 2.234 2.245 1.056 2.960 0.987 2.959 6.943 2.645 4.055

Panel B: Business characteristics Firm age Business runs full time Business is run out of the household Firm has market access Business near similar businesses Firm keeps written accounts Firm maintains business budget Firm started with a business plan Firm has some licence Firm pays income tax Firm advertises Firm pays workers in cash Workers received technical training Workforce increased for the business in the past year Firm has regional customers Number of daily customers is more than 20 Firm’s suppliers are individuals Firm’s suppliers are small traders Firm’s suppliers are nationwide Business registered with BRELA Business has a tax ID Business gets inputs on credit Business has rental agreement for business premises

6.323 0.798 0.517 0.589 0.721 0.417 0.078 0.007 0.164 0.044 0.017 0.098 0.161 0.074 0.174 0.288 0.461 0.507 0.041 0.028 0.049 0.101 0.081

10.350 0.891 0.174 0.761 0.761 1.000 0.174 0.130 0.652 0.435 0.261 0.935 0.587 0.130 0.326 0.391 0.326 0.304 0.391 0.240 0.478 0.130 0.413

4.529 1.570 4.642 2.365 0.595 8.026 2.400 9.417 8.868 12.642 12.758 19.074 7.800 1.462 2.690 1.540 1.832 2.736 11.722 8.505 13.102 0.654 8.164

Panel C: Labour history Previously unemployed Previously a home maker Previously worked in the education sector

0.068 0.173 0.040

0.196 0.152 0.022

3.424 0.368 0.623 (continued )

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

Table 27.1 Continued Mean (not in- Mean (inDifference in between firms) between firms) means t-test Previously employed in large private enterprise Previously employed in similar sized private enterprise Previously ran a similar sized enterprise Previously a civil servant Previously engaged in farming/rearing of livestock Owner was trained on previous job Owner was trained in a course Panel D: Infrastructure and technology Firm owner has a mobile phone Owner uses mobile to conduct business Firm owner has a calculator Business has office equipment Business owns a cooling facility Firm has received legal services Firm has received technical services Firm has security services Business uses electricity to light business Panel E: Access to financial services Loan from government Owner has borrowed for business Firm has received financial services Owner has formal bank account Owner uses debit card for business Owner saves money in a bank account Owner saves in secret hiding place Owner uses a Sacco Owner regularly sends and receives money for business Panel F: Measures of ability/financial literacy Owner does not believe in interest Owner uses profits to expand business Owner uses profits to buy stocks in advance Owner uses profits to invest in business Owner uses profits to invest in buildings and land

0.030 0.178

0.087 0.000

2.198 0.913

0.108 0.030 0.476 0.042 0.018

0.087 0.087 0.130 0.109 0.043

0.463 2.600 4.681 2.250 1.260

0.509 0.419 0.154 0.168 0.040 0.006 0.030 0.142 0.160

0.913 0.891 0.522 0.304 0.196 0.043 0.152 0.370 0.652

5.477 6.486 6.833 2.460 5.358 3.105 4.830 4.392 9.140

0.014 0.176 0.048 0.065 0.510 0.073 0.694 0.016 0.152

0.000 0.304 0.304 0.543 0.500 0.522 0.349 0.022 0.413

0.336 2.260 7.981 12.999 13.546 11.542 4.884 0.320 4.896

0.196 0.183 0.436 0.183 0.054

0.261 0.348 0.500 0.348 0.174

1.094 2.873 0.868 2.873 3.518

Source: Authors’ calculations based on the 2010 MSME survey. Firms in the ‘in-between’ category satisfy the following conditions: (i) owner wouldn’t leave the firm for a full time salaried job; (ii) keeps written accounts; (iii) has paid employees and; (iv) labour productivity is higher than economy-wide labour productivity in manufacturing.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



much less likely to come from poor households. Only 2 per cent of these businesses report that their main source of income is farming compared to roughly 20 per cent of the rest of the businesses. The owners of these businesses are also much more likely to view their businesses as growing and they are more likely to have started their business because they saw it as a business opportunity (as opposed to having done it out of necessity). The results in Panel B that describe business attributes also reveal some important differences between the Group 3 in-between firms and the rest of the MSMEs. The inbetween firms have been in business longer (10.35 years compared to 6.32 years), they are much less likely to be run out of the household, they are more likely to have some sort of licence, they are more likely to pay income taxes, they are more likely to pay their workers in cash, the workers in these businesses are more likely to have received technical training, they are more likely to have increased their workforce over the past year, they have many more customers, they are more likely to have regional (as opposed to local) input suppliers and customers, they are more likely to be registered with BRELA and to have a Tax ID number, they are more likely to get inputs on credit, and they are more likely to have a rental agreement for their business premises. Anecdotal evidence suggests that having regional customers is likely to be associated with exporting across borders into Kenya and Rwanda to the north, Rwanda, Burundi, and the Democratic Republic of Congo to the west and Zambia, Malawi, and Mozambique to the south. Unfortunately the information in the survey is not detailed enough to sort out the importance of this trade. In short, the firms in the in-between sector look more like formal firms than do the firms in the rest of the MSME sector. The results in Panel C describe the labour history of business owners in the MSME sector by firm type. While there are some statistically significant differences in labour histories—the magnitudes of the differences are not as large as the magnitudes of the differences in the individual and business attributes. Owners of the Group 3 in-between firms are more likely to have previously worked for a large private enterprise and they are more likely to have been trained on a previous job. Practically none of the firms report that they were trained in a public or private programme. Instead, our interviews with small business owners indicate that they have typically learned as paid or unpaid apprentices. The results in Panel D reveal significant differences in the use of technology. Almost all of the in-between business owners own mobile phones while only half of the rest of the MSME owners own mobile phones. Similarly, 89 per cent of the owners of the inbetween businesses use their mobile phones to conduct business while only 42 per cent of the remaining firms do so. The firms in the in-between sector are also much more likely to own a calculator, own office equipment, own a cooling facility, have security services, and use electricity in their businesses. The results in Panel E indicate that owners of the in-between MSMEs are much more likely to participate in the formal banking sector although not necessarily through borrowing. One of the most striking differences in financial practices is that in-between sector firms are much more likely to save in a formal bank account

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

(52.2 per cent vs. 7.3 per cent). However, although the vast majority of firms say they are in need of a loan to invest in machinery and equipment, expand their businesses and/or purchase stocks, most firms have not taken loans for these purposes (30.4 per cent of the in-between firms and 17.7 per cent of the rest of the MSMEs). Of the firms that do borrow, roughly 50 per cent of the in-between firms say they borrow from formal banks while only 8 per cent of the remaining firms report borrowing from formal banks. Only 15 per cent of the in-between firms report borrowing from friends and family; by contrast 48.2 per cent of the remaining firms report borrowing from friends and family. Around 20 per cent of both types of firms report borrowing from microfinance institutions. Finally, the results in Panel E of Table 27.1 are meant to capture other financial practices. Between 19 and 26 per cent of the business owners report that they do not believe in interest. Given that Tanzania is roughly 50 per cent Muslim, this is not that surprising. Firms in the in-between sector are significantly more likely to use profits to expand their businesses and/or invest in buildings in land.

27.4.2 Obstacles to Doing Business In Table 27.2, we present the results of tabulating answers to questions about what firms say they view as obstacles to doing business (Panel A) and the type of assistance they would like to see provided by the government (Panel B). The questionnaire itself lists twenty-one possible obstacles to doing business. If less than 1 per cent of the firms reported one of the items as a significant obstacle to doing business, we do not report it. Apart from access to working capital, there are no significant differences between the in-between firms and the rest of the firms in terms of obstacles to doing business. And even in the case of insufficient working capital, the percentages of business owners who report that insufficient working capital is an obstacle is relatively low compared to the share of firms that report that they could use a business loan. This is because when firms are asked what they would use the loans for, they almost never say they need it for working capital. Instead, they report that they would use loans to invest in buildings and machinery and business expansion. In Panel B of Table 27.2, we report means and differences in means for areas in which firms would like to see government intervention. Access to finance and investment in infrastructure are at the top of the list. The in-between firms also report that they would like to see more business services provided by the government such as information and consulting. But without further information, it is difficult to know what this means. Notably, very few firms of either type report that they would like to see the government provide more training. This does not appear to be because firms are already getting training from the government since in a separate section of the survey almost no firms report that they received training from the government. But it is consistent with what we heard from small business owners about receiving training on the job in both large and small firms.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



Table 27.2 Obstacles to business Mean (not Mean Difference in in-between firms) (in-between firms) means t-test Panel A: Obstacles to doing business Insufficient working capital Insufficient market access Low demand for products/services Customers taking products on credit and not paying High competition from other businesses High cost of inputs Crime, theft, disorder Access or costs of finance/credit Shortage of inputs Poor roads/access to business

0.273 0.096 0.072 0.059

0.152 0.087 0.109 0.00

1.990 0.203 0.949 1.697

0.056 0.054 0.054 0.049 0.029 0.029

0.065 0.087 0.065 0.043 0.00 0.022

0.274 0.972 0.349 0.159 1.174 0.084

0.326 0.087

2.286 0.735

0.239

2.563

0.087 0.022 0.022 0.065 0.065 0.000 0.000

0.383 0.237 0.429 1.333 2.595 0.465 0.446

Panel B: Issues firms would like to see addressed by the government Providing access to finance 0.495 Providing/improving infrastructure (e.g. energy, 0.123 telecoms, transport, water) 0.117 Providing business services (e.g. information, consulting) Creating markets for products 0.072 Improving skills and training 0.067 Simplify the loan conditions 0.033 Reforming tax system 0.031 Easing the regulations controlling business 0.016 Reduce product prices 0.005 Provide funds and working facilities 0.004

Source: Authors calculations based on the 2010 MSME survey. Firms in the ‘in-between’ category satisfy the following conditions: (i) owner wouldn’t leave the firm for a full time salaried job; (ii) keeps written accounts; (iii) has paid employees and; (iv) labour productivity is higher than economy-wide labour productivity in manufacturing.

An especially interesting example of this type of training was reported to us by Godwin H. Makayo the CEO of MAKTECH.⁴ Mr Makayo was selected for one year of technical training by NORAD which he did partly at the private company Vodacom and partly at the University of Dar es Salaam. Mr Makayo started working for Vodacom in Tanzania in 1998. In 2000, he left Vodacom and set up his own firm contracting out his services to Vodacom. When the company started, it was not registered and it had only three employees—Mr Makayo and two relatively uneducated technicians whom he was able to train to do installations of Vodacom equipment. As his company grew, he realized that he had strong technical skills but not the kind of ⁴ Interview with Godwin H. Makyao, Dar es Salaam, 25 January 2016.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 ,  ,     

skills he would need to manage a growing business. He enrolled on an online business course at the University of Liverpool and received his Master’s Degree in Business Administration in 2007. Between 2001 and 2011, his company grew to 150 employees. Today, MAKTECH is a leading service and solution provider in the telecom infrastructure contracting industry. We heard a similar story about training on the job from a small shop owner in Kariakoo Market—the heart of Dar es Salaam’s small business activity. Mr Eliya Gomezulu told us that he learned everything he knows about metalworking from watching an Indian metalworker who used to be in the business but who has since moved back to India. Now, Mr Gomezulu runs a small shop in Gherazani—the metalworking district of Kariakoo Market—that makes steel products from scrap metal including ovens and machines for grinding coconut shells. Although Mr Gomezulu does not have any employees per se, he has the tools needed to do metalworking and he ‘rents’ out his space and tools to other metalworkers and in return he gets 10 per cent of their profits.

27.5 C

.................................................................................................................................. Using Tanzania’s first nationally representative survey of micro, small, and mediumsized enterprises, we have shown that there is an enormous amount of heterogeneity among these MSMEs. The owners of the most productive of the enterprises share the following characteristics: (i) they wouldn’t leave their business for a full time salaried position; (ii) they keep written accounts; (iii) they hire paid employees; and (iv) a relatively small group of these firms—10 per cent—already contribute roughly half a percentage point to annual labour productivity growth in Tanzania. This group of firms operates primarily in the manufacturing and trade services sectors and employs roughly the same number of employees as Tanzania’s formal manufacturing and trade services firms. The group has the potential to contribute 1.3 percentage points to annual labour productivity growth; to put this in perspective the formal modern sector contributed 1.58 percentage points to annual labour productivity growth over the past decade. To achieve this result, Tanzania needs an MSME policy that is targeted at the most productive MSMEs. Although targeting is difficult, we have outlined some important differences between the most productive MSMEs and the least productive MSMEs. In particular, the owners of the most productive MSMEs are better educated and they are significantly more likely to save in a formal banking account. Thus, one way of reaching these MSMEs is likely to be through the formal banking sector. This does not mean that the rest of the MSMEs serve no purpose. On the contrary, they provide sorely needed extra income to some of Tanzania’s poorest families. But we should not expect these MSMEs to be a source of labour productivity growth or within firm employment growth.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

       



We have also described some innovative programmes in Tanzania designed to target the more productive MSMEs. These programmes are operating exclusively in the private sector and are in their infancy. If some of the programmes are successful, scaling them up may require public–private partnerships. Scaling up will require targeting. Targeting the MSMEs with the potential for productivity and employment growth will require careful design of a product(s) that is attractive to the MSMEs with potential but that is too costly for the MSMEs without the potential for employment and productivity growth.

R Banerjee, Abhijit V. and Esther Duflo, 2007. ‘The Economic Lives of the Poor’, The Journal of Economic Perspectives: A Journal of the American Economic Association, 21 (1), p. 141. Banerjee, Abhijit V., Emily Breza, Esther Duflo, and Cynthia Kinnan, 2015. ‘Do Credit Constraints Limit Entrepreneurship? Heterogeneity in the Returns to Microfinance.’ Working Paper. Dewji, Mohammed, 2016. CEO MeTL Group Tanzania and Chairman Mo Dewji Foundation. Personal interview, 27 January. De Paula, Aureo and Jose Scheinkman, 2011. ‘The Informal Sector: An Equilibrium Model and Some Empirical Evidence from Brazil’, Review of Income and Wealth, 57 (s1), pp.s8–s26. Diao, Xinshen and Margaret McMillan. 2018. ‘Toward an Understanding of Economic Growth in Africa: A Reinterpretation of the Lewis Model’, World Development, 109 (September), p. 511. Financial Sector Deepening Trust, 2012. ‘National Baseline Survey Report for Micro, Small and Medium Enterprises in Tanzania’. Available at: http://www.fsdt.or.tz/data/msmebaseline-survey/ Harrison, Ann, Leslie A. Martin, and Shanti Nataraj, 2015. ‘In with the Big, Out with the Small: Removing Small-scale Reservations in India’. NBER Working Paper, 19942. Available at: http://www.nber.org/papers/w19942 accessed 24 July 2015. Isaga, N., L. Mwagike, and G. Rasheli, 2016. ‘Review of Interventions and Their Impact of Furniture Enterprises in Tanzania’ Draft, 31 January. Kassongo, Mohamed H., 2016. CEO UMI, Personal interview, 28 January. McKenzie, David J., 2015. ‘Identifying and Spurring High-growth Entrepreneurship: Experimental Evidence from a Business Plan Competition.’ World Bank Policy Research Working Paper No. 7391.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

  ......................................................................................................................

        ......................................................................................................................

ˊ ˊ ’

28.1 I

.................................................................................................................................. S after gaining their independence in the 1960s, the countries of the West African Economic and Monetary Union (WAEMU), like others in sub-Saharan Africa, achieved significant economic performance in terms of economic growth (Severino and Ray 2010). This was essentially based on the exportation of natural resources, rents on commodities, and the good performance of commodity prices.¹ Due to the limits of these explanatory variables (the reversibility of commodity prices, a dependence on external resources, etc.), the economic growth it generated was not able to continue beyond 1979, nor did it encourage the transformation of the Union’s economies. This failure was particularly felt from 1980 to 2000, a period during which the countries of the WAEMU had the bitter experience of no longer enjoying economic growth.² Efforts undertaken to right the situation necessitated important initiatives. These included Structural Adjustment Programmes, and the Heavily Indebted Poor Countries (HIPC) Initiative which made it possible to improve the situation of public finances. New factors came together to favour the return of economic growth beginning in 2000. This new growth is based on complementary characteristics: the consolidation of domestic demand, demographic growth, growing urbanization, etc. ¹ For a presentation of the development strategies of African countries in the past, present, and future, see Fosu and Ogunleye (2015). ² See N’Guessan (2012).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



The new growth set in motion in 2000 and which is still ongoing for all the WAEMU countries is certainly welcome, but simply accepting the fact is not enough. We must go further and ask whether this strong growth has led to the structural transformation of the Union’s economies. The problem of structural transformation is vital for a country because it has been established that an economy may have strong growth without necessarily experiencing development. According to Duarte and Restuccia (2010) and Herrendorf et al. (2013), structural transformation is a fundamental driver of economic development. In fact, it is now known that economic growth can be sustained and inclusive only if it encourages the structural transformation of the economy. Furthermore, structural transformation is one of the chief sources of growth (Ros 2000; Ocampo 2005; Hausmann and Rodrik 2006). By reallocating the resources of traditional activities (such as subsistence agriculture) to highly productive sectors (such as manufacturing and modern services), the change in productive structure can not only stimulate economic growth but also make it more inclusive and lasting. Thus structural transformation influences not just the level of growth, but at the same time, in a number of ways, its quality. The aim of this chapter is to answer the following question: in what measure has the new economic growth begun in 2000 in the countries of the West African Economic and Monetary Union (WAEMU) led to the structural transformation of the Union’s member countries? To answer this question, in terms of the methodological approach, a number of indicators are available. Among these, I have selected four: the share of final consumption in the gross national product, the share of each sector in total production, the decline of agricultural employment, and the diversification of the exports of WAEMU member countries. The remainder of the chapter is organized around the following sections: (1) a presentation of the measures used to evaluate structural transformation; (2) a presentation and analysis of the growth of the WAEMU economies; (3) the structural transformation of the WAEMU economies: and (4) a SWOT analysis and recommendations for removing obstacles to the structural transformation of the WAEMU economies.

28.2 M  S T

.................................................................................................................................. Several statistical measure indicators are used to grasp the dynamics of structural transformation. These include the share of final consumption expenditure, the movement of employment between sectors, a consideration of the contribution of each sector to the total value added of the countries concerned (Vergne and Ausseur 2015), and the diversification of the production systems and/or the exports.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

28.2.1 Final Consumption as a Measure of Structural Transformation The structural transformation of an economy can be evaluated by means of the change in structure of the final demand. With the level of development and the increase in income per capita, the share of private consumption demand in the GNP tends to decrease, which makes possible the increase of the investment share.This is shown in Figure 28.1 concerning four emerging countries that can serve as references for those of the WAEMU: China, India, Indonesia, and Korea. This methodological approach is used by Moshe (1988) to assess structural change in Colombia between 1950 and 1988 and by Herrendorf (2013) to study transformation in OECD countries. The decrease of consumption’s share in the GNP in the course of the development process enables the increase of the share of investments in economic infrastructures and human capital, sources of productivity improvement, growth of the GDP, and per capita income. The composition of the final demand is only one of the measures of structural transformation. The composition of production can also be used to judge the degree of transformation of an economy’s structure.

China

80.0

100.0

70.0

80.0

60.0 50.0

60.0

40.0

40.0

30.0 20.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014 India

Korea

120.0 100.0 80.0 60.0 40.0 20.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0

0.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

20.0

10.0 0.0

Indonesia

120.0

Consumption

0.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

90.0

Investment

 . Evolution of the demand structure in several emerging countries

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



28.2.2 Composition of Production as a Measure of Structural Transformation The utilization of sector contribution (agriculture, industry, services) in the total value added as an indicator is based on the reasoning that structural transformation can be considered to take place when the increase in per capita GDP is accompanied by a decrease in both the shares of employment and the value added in agriculture and by an increase in the shares of employment and of value added in industry and services (McMillan and Rodrik 2011; Herrendorf et al. 2013; Timmer et al. 2013). Thus an analysis of the sectoral composition of a country’s production (in particular, of the share of manufacturing value added in the GDP) makes it possible to grasp the sectoral dynamics. The increase in income in the course of the structural transformation process is initially accompanied by relative stagnation in the weight of the agricultural sector at reasonably high levels, followed at first by a rapid and then by a slower decrease, and finally by stagnation at relatively low levels. This significant outcome is found in practically all the studies proposing a theoretical or empiric model of structural transformation (Clarke 1940; Herrendorf et al. 2013). The diminishing economic role of agriculture can also be observed through an examination of the sectoral allocation of resources, notably that of employment.

28.2.3 Movement of Employment from the Agricultural Sector to the Other Sectors The decrease in the weight of agriculture in the total value added is accompanied by a redistribution of the workforce to the detriment of the agricultural sector, which eventually loses its role of primary supplier of employment in the economy. Indeed, as income levels rise, the share of employment utilized in the agricultural sector decreases. The dynamics of the sectoral distribution of the workforce and of the value added suggests that the process of structural transformation is most often accompanied at a given time by a relatively rapid increase of labour productivity in the agricultural sector as income rises, an increase that could possibly benefit the rest of the economy as suggested by the dual model of Lewis (1954). The increase in labour productivity in the economy comes from within-sector performance and a better allocation of the workforce, which is an indicator of the sectoral transformation of the economy. The overall performance of the economy and structural transformation are linked, due to a combination of gains of sectoral productivity and an intersectoral reallocation of the workforce, according to the following formula: X si;t  TΔyi;t þ yi;t  TΔsi;t ΔYt ¼ i

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

Where: ΔYt represents the variation of labour productivity in the economy over the time period T, yi the productivity of labour in sector i and si the share of sector i in overall employment. The economy’s gain in productivity is thus the result of increased efficiency within sectors (first term) and a better allocation of the workforce between sectors (second term). The latter shows the full potential of the structural transformation of the economy, which gives information on the factor movement underlying a more efficient sectoral redistribution of economic activity. Changes regarding employment are the method of choice to measure sectoral shifts. It is used by Timmer et al. (2013) to measure the structural transformation of the African economies in the Groningen Growth and Development Centre’s database (GGDC). Unfortunately, the data concerning sectoral employment in low-income countries are of poor quality and unequal significance. Consequently, this measure is generally supplemented by other measures such as production and export diversification.

28.2.4 Diversification of Production and Exports as a Measure of Structural Transformation When countries are at low levels of development, they generally specialize in products corresponding to their endowments in natural resources. This type of specialization strengthens the vulnerability associated with a high degree of concentration, for the price, conditions of production, and demand for this type of goods are unstable. Imbs and Wacziarg (2003), Klinger and Lederman (2004), and Cadot et al. (2011) have shown that economic development goes through a stage of diversification of the productive structure and of exports. The less dependent a country’s exports are on a limited number of goods, the more it is considered to be diversified. Conversely, when one or several goods represent a large share of a country’s exports, these are considered to be concentrated and constitute a source of vulnerability. Work based on transnational analyses and case studies (IMF 2011; Papageorgiou et al. 2012) shows that the diversification of production, exports, and trading partners generally plays a large role in growth. Furthermore, Cadot et al. (2011) show that an increase in per capita income is initially accompanied by a diversification, and subsequently by a re-concentration of production and employment. Diversification is also closely linked to structural transformation, particularly in countries that are in the first stages of their economic development. Theory as well as empiric work (Lin 2012; McMillan and Harttgen 2014) indicates that structural transformation—that is, the dynamic reallocation of resources from less productive to more productive sectors is an essential ingredient of economic development, as was shown in the 1970s and 1980s by the Asian tigers which transformed themselves from agrarian into manufacturing economies. Because low-income countries are very specialized in a small number of agricultural and other activities linked to natural resources, structural transformation by resource

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



reallocation will almost inevitably lead to diversification and a more balanced production structure. The economy then moves to a more varied production structure, with the introduction of new products or the development of pre-existing products, particularly products of better quality. An increase in per capita income is strongly associated with a greater diversification of export products. This general relationship proves true at least until a country reaches the status of an advanced economy (IMF 2014). Theoretically, there are several ways in which export diversification impacts economic growth. Herzer and Lehman (2006) show that it has a positive effect on economic growth by reducing the country’s dependence on a limited number of primary products. Thus, according to structuralist economists, developing countries should move from exporting primary products to manufactured products to ensure stable economic growth (Chenery 1979; Moshe 1988). The measures most currently used to evaluate the degree of export diversification are the Herfindahl, Gini, Theil, and entropy indices (Cadot et al. 2011). Of the three indicators, the International Monetary Fund (2014) has used the Theil index to evaluate the export diversification of low-income countries. Its results are used in this chapter. This index makes it possible to distinguish between two kinds of diversification: extensive and intensive margin diversification. The index is calculated as follows:   n n 1X xk xk 1X ln where μ ¼ xk T ¼ n k¼1 μ n k¼1 μ n = the total number of exported goods xk = the value of the exports of product k µ = the average value of the exports The West African Economic and Monetary Union (WAEMU) has experienced strong growth in the past few years. What is the nature of this growth and what effect has it had on the transformation of the productive structures of the countries of the subregion? To answer this question, the four approaches described above will be used. But before evaluating the degree of structural transformation of their economies, Section 28.3 will present the countries of the West African Economic and Monetary Union and analyse the growth of their economies.

28.3 P  A   G   WAEMU E

.................................................................................................................................. A better understanding of the results of an evaluation of the structural transformation of the WAEMU economies demands a brief presentation of the Union, as well as an examination of the evolution of economic growth in the various member countries.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

28.3.1 Presentation of the West African Economic and Monetary Union (WAEMU) The birth of the WAEMU in January 1994 is part of the broad reforms undertaken to reconnect with economic growth, because the policies of structural adjustment put in place in the 1980s were slow to bear fruit.³ The Union is composed of eight States which together cover an area of 3,500,000 km² (1,351,358 mi.²) and had some 113 million inhabitants in 2015. This population is projected to reach 131.82 million by 2020 and 174.15 million by 2030.⁴ The basic indicators of the Union’s member countries are presented in Table 28.1. These eight West African countries use the CFA Franc,⁵ pegged to the French Franc (now the euro). However, they have different institutional profiles. From 1960 to the mid 1980s, they were faced with similar problems: loss of competitiveness; increasing budget deficits; loss of the State’s ability to meet its obligations; decrease in standard of living of the urban population and pauperization of the rural population; dependence on imports; etc. These problems were worsened by monetary mechanisms that led to an appreciation of the CFA Franc. In 1994 the heads of State decided to devalue their common currency, the CFA Franc and also to create a new structure. Thus was born the WAEMU, after the West African Monetary Union (WAMU). WAEMU was designed to enable the eight States to go beyond

Table 28.1 Basic indicators of the economies of the WAEMU, 2015

Benin Burkina Faso Côte d’Ivoire Guinea-Bissau Mali Niger Senegal Togo WAEMU Africa

Population (thousands)

Area (thousands of km2)

Population (density per km2)

Gross domestic product (millions $ PPP)

Per capita GDP in PPP

10,850 18,106 22,702 1,844 17,215 19,899 15,129 7,305 113,050 1,184,501

115 274 322 36 1,240 1,267 197 57 3,508 30,066

95 66 70 51 14 16 77 129 64.7 39.0

21,156 31,184 78,335 2,676 29,151 18,960 36,300 10,818 228,580 5,768,932

1,945 1,722 3,451 1,451 1,656 953 2,399 1,481 1,882 4,870

Source: ADB: African Development Outlook, 2016.

³ See Severino and Ray (2010). ⁴ United Nations Conference on Trade and Development (UNCTAD) statistics. ⁵ In the Central African Franc Zone, CFA stands for Coopération Financière d’Afrique (Financial Cooperation in Central Africa) and in the West African Franc Zone, Communauté Financière d’Afrique (Financial Community of Africa).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



the limits of the WAMU mechanism by adding an economic dimension to the Union in order to strengthen and deepen the creation of an integrated regional economic area.⁶ The track record of the WAEMU, with regard to the goals it was assigned at its creation, is measured in terms of economic growth, trade within the zone, and convergence criteria. The results for economic growth to date are encouraging. This growth should, however be analysed in order to tease out the realities it covers.

28.3.2 Analysis of the Growth of the WAEMU Economies Like the African continent as a whole, the subregion has experienced two periods of strong growth: from 1960 to 1979 and from 2000 to the present. Several factors have contributed to the renewal of economic growth during this second period. Among these there are, certainly, old factors such as the strong demand for commodities owing to the worldwide economic recovery, the good performance of international prices for export products, etc. However, this time new factors must be taken into account. Indeed, the implementation of better economic policies, investments in infrastructure, the consolidation of domestic demand linked to increased income, favourable agricultural campaigns, the cessation of conflicts . . . all encouraged a noticeable and durable renewal of average growth, as can be seen by the growth curve of the real GDP in Figures 28.2 and 28.3, representing the evolution of real GDPs respectively of subSaharan Africa and of the WAEMU. 10.0 8.0 6.0 4.0 2.0

–2.0

1961 1963 1965 1967 1969 1971 1973 1975 1977 1979 1981 1983 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013

0.0

 . Evolution of the growth of the real GDP of sub-Saharan Africa (per cent)

⁶ For more information on CFA franc Zone, see N’Guessan (2008) and Tchatchouang (2015).

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’ 10.00 8.00 6.00 4.00 2.00 0.00 –2.00 –4.00 1961

1968

1975

1982

1989

1996

2003

2010

WAEMU

 . Evolution of the growth of the real GDP of UEMOA (per cent) Source: World Bank, World Development Indicators (2016).

Table 28.2 clearly shows the two periods of strong growth. It is obvious that after 1978, Africa experienced no true growth until 2000, the year in which growth resumed. Sub-Saharan Africa again registered a clear acceleration of its growth rate, making it the world’s second most dynamic region after East Asia (ECA 2013). The average annual real GDP growth rate was 6.8 per cent between 2004 and 2008 (IMF 2016). In 2009, it regressed and came out again at 4 per cent due to the global financial crisis of 2008 and the sudden rise in food and fuel prices. It then picked up again to reach an annual average of 5.2 per cent between 2010 and 2014. The GDP increase in 2015 is estimated at 3.4 per cent, down from the 5 per cent progression of 2014. The current fall of commodities prices has slowed growth but has not checked it. The real GDP growth of the West African Economic and Monetary Union experienced the same evolution as that of sub-Saharan Africa between 1960 and 1980: a period of strong growth between 1960 and 1980, then a loss of steam and finally an acceleration beginning in 2000, as can be seen in Figure 28.3. To draw out all the lessons of growth evolution in the WAEMU countries from 2000 to 2015, the entire period must be analysed, and then three sub-periods: 2000–04, 2005–10, and 2011–15 (Table 28.2). The period from 2000 to 2004 corresponds to the start of strong growth for the WAEMU countries. The interval from 2005 to 2010 is considered as the time during which growth reached its maturity, and finally the sub-period 2011–15 is selected because it was impacted by external factors, socio-political crises, conflicts, etc. The analysis cannot be limited to average annual growth rates. The study of levels of average economic growth within different sub-periods is also done by country. The figures in Table 28.2 show, for the entire period 2000–15, differences between average growth rates that reached 3.3 percentage points between the least well performing country (Guinea-Bissau) and the country whose rate was the highest (Burkina Faso). During this period only two countries, Burkina Faso and Niger, achieved better performances, with growth rates above 5 per cent. Average growth rates were modest in most of the countries between 2000 and 2004. With the exception of Burkina Faso, which achieved a growth rate slightly above

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Table 28.2 Average annual growth rates of the real GDP Benin Burkina Faso Côte d’Ivoire Guinea-Bissau Mali Niger Senegal Togo WAEMU

2000–04

2005–10

2011–15

2000–15

4.5 5.3 3.6 1.1 4.6 3.4 4.2 0.8 3.2

3.6 5.7 1.9 4.7 5.2 5.8 3.9 2.9 4.2

5.0 5.8 6.5 2.9 3.1 6.0 4.1 5.0 4.8

4.3 5.6 3.9 2.3 4.3 5.1 4.1 2.9 4.1

Source: World Bank (2016).

5 per cent, the other countries had rates below this figure. The rate was even negative in the case of Guinea-Bissau. The average growth rates of this sub-period are relatively weak because this was only the beginning of the period during which the countries of the Union resumed growth. The sub-period 2005–10 is the time when the growth of the WAEMU economies showed a relatively better performance. Five of the eight members of the Union improved their average annual growth rates over those achieved between 2000 and 2004. Moreover, three countries, Burkina Faso, Mali, and Niger achieved rates above 5 per cent, while only one country (Côte d’Ivoire) had an average annual growth rate below 2 per cent during this sub-period. With the exception of Guinea-Bissau, Mali, and Senegal, the average annual GDP growth rate was above 5 per cent in the countries of the Union between 2011 and 2015. Despite the unfavourable international business cycle (slowing of activities in partner emerging countries, fall in the prices of commodities, etc.), the resilience of the WAEMU countries enabled them to achieve significant growth rates during this period. Did this significant, sustained economic growth favour the economic transformation of the WAEMU countries?

28.4 S T   WAEMU E

.................................................................................................................................. The countries of the WAEMU have admittedly experienced relatively good economic performances over the past fifteen years, but has this growth resulted in a significant transformation of productive structures, as might be expected? To answer this question, I shall analyse the structural transformations of the economies of the WAEMU countries since 2000, using the indicators given in Section 28.2.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

28.4.1 Analysis of the Demand Structure The transformation of the composition of the final demand is one of the most uniform characteristics of the development process. Once again, one would expect consumption to decrease in the course of the process of economic development. What about the WAEMU countries? There has been a decrease in consumption between 2000 and 2015, but at a slow rate. The decrease in the consumption share in favour of investment, as can be seen by the curves in Figures 28.4 and 28.5 is weak compared with those of Figure 28.1 that represents the emerging countries of China, India, Indonesia, and Korea, presented in Section 28.2. It is true that in Benin, Burkina Faso, and Niger the share of consumption in the GDP decreased noticeably and that of investment rose. This dynamic is important in favouring the structural transformation of the economies of these three countries. This is not the case for Côte d’Ivoire, Guinea-Bissau, Mali, and Togo where, on the contrary, the consumption share increased, weakening the proportion of income devoted to investments that would improve agricultural and industrial productivity.

28.4.2 Sectoral Contribution to the Total Value Added The most typical element of structural transformation is the movement from the production of primary products to manufactured production. The value added of agriculture decreases significantly during this transition. What about the production of the countries of the West African Economic and Monetary Union, whose growth has been sustained since 2000? The composition of each country’s production is analysed to ascertain whether there has been a change after fifteen years of growth. Three years have been selected for this: 2000, the year when sustained growth began in the WAEMU countries; 2007, an intermediate year; and 2014, the most recent year for which data are available. The results are illustrated in Figures 28.4 and 28.5. The decline in agricultural value added is seen in only four countries of the Union: Benin, Côte d’Ivoire, Niger, and Senegal. In Benin, it went from 25.8 per cent in 2000 to about 23.4 per cent in 2014, a decrease of 2.4 percentage points. The decline was 2.6 per cent and 1 per cent in Côte d’Ivoire and Niger respectively. The greatest change took place in Senegal, where the contribution of agriculture to the value added fell from about 19 per cent in 2000 to about 16 per cent in 2014, a decrease of more than 3 per cent. In the other countries of the Union (Burkina Faso, Guinea-Bissau, Mali, and Togo) the contribution of agriculture in 2014 was higher than in 2000 and 2007. When the Union is considered in its totality, the contribution of the agricultural sector to the GDP increased by 0.36 percentage points between 2000 and 2014. This increase in the share of agricultural activity in total value added indicates that undertaking structural transformation is a difficult struggle for the countries of the WAEMU.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

       Benin 100.0

100.0

80.0

80.0

60.0

60.0

40.0

40.0

20.0

20.0

0.0

0.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

120.0

1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015

Mali

120.0

Burkina Faso 100.0

100.0

80.0

80.0

60.0

60.0

40.0

40.0

20.0

20.0

0.0

0.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

120.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

Niger

120.0

Senegal

Côte d’Ivoire 100.0 90.0 80.0 70.0 60.0 50.0 40.0 30.0 20.0 10.0 0.0

120.0 100.0 80.0 60.0 40.0 0.0

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

20.0

Togo

Guinea-Bissau 140.0

120.0

120.0

100.0

100.0

80.0

80.0

60.0

60.0

40.0

40.0

20.0

20.0

CONSUMPTION

1960 1962 1964 1966 1968 1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

0.0

1970 1972 1974 1976 1978 1980 1982 1984 1986 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006 2008 2010 2012 2014

0.0



INVESTMENT

 . Evolution of the demand structure in the countries of the WAEMU Source: World Bank, World Development Indicators (2016).

At the end of 2014, only Benin (14.4 per cent), Côte d’Ivoire (12.6 per cent), and Senegal (13.1 per cent) had a share of more than 10 per cent in the manufacturing sector of the GDP. This shows the weakness of the manufacturing sector, whose development is a guarantee of job creation, increasing incomes, and demand. In all the countries for which data are available, it is obvious not only that the contribution of manufacturing production to the GDP is weak, but also that it has not experienced the

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’ Sectoral GDP composition (Mali)

Sectoral GDP compostion (Benin) 25.8 27.62 23.45 31.82 27.58 23.16 23.68 18.81 14.36

Agriculture Industry Manufacturing Services

Agriculture Industry Manufacturing Services

Agriculture Industry Manufacturing

Agriculture

Manufacturing 42.38 44.8

38.08 Sectoral GDP composition (Niger)

Agriculture 17.76 13.18 19.6

Industry

13.16 11.04

Manufacturing 45.66 48.38 42.74

Sectoral GDP Composition (Côte d’Ivoire) 24.99 21.99 22.37 21.5 23.27 21.1 17.21 14.56 12.62

Industry Manufacturing 53.51 54.74 56.53

Sectoral GDP composition (Guinea-Bissau) 42.74 44.33 Agriculture 43.92 14.45 13.34 Industry 13.56

6.79 5.18 6.18 44.4 45.85 43.69 Sectoral GDP compostition (Senegal) 19.14 13.77 15.83 23.23 24.12 23.52 14.65 14.47 13.11 57.63 62.11 60.65

Services

Sectoral GDP composition (Togo) 35.14 35.82 41.7

Agriculture 18.32 18.69 17.16

Industry

Manufacturing

Manufacturing 42.81 42.33 42.52

8.58 9.18 5.73

Services 2000

37.84 40.97 36.71

Services

Agriculture

Services

Services

42.57 41.09

Services

53.39

Sectoral GDP Composition (Burkino Faso) 32.81 32.67 34.2 21.53 18.95 23.06 6.39

20.98 22.71 22.37

Industry

36.45 36.2 39.55

2007

46.54 45.49 41.14

2014

 . Sectoral composition of the WAEMU countries’ GDP Source: World Bank, World Development Indicators (2016).

increase one would have expected. Despite the relatively good economic performances of the Union economies between 2000 and 2014, manufacturing production decreased, thus confirming ‘the theory of the deindustrialization of Africa’. In the WAEMU taken as a whole, the manufacturing sector’s share in the GDP fell from 14 per cent in 2000 to 12.2 per cent in 2007, then to 9.7 per cent in 2014, a decrease of about four percentage points in 15 years. This evolution is partly the consequence of the privatization of public companies during the period of structural adjustments, the decrease in investment rates, etc. The manufacturing sector is thus, at present, far from playing its role as a sector that creates jobs, acts as a driver of upstream and downstream activities, and enables production and export diversification.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Table 28.3 Evolution of the agricultural workforce (in percentage of the total workforce) Benin Burkina Faso Côte d’Ivoire Guinea-Bissau Mali Niger Senegal Togo

2000

2005

2010

2015

57.8 85.8 44.9 80.7 70.7 88.4 76.9 52.1

52.9 86.7 39.5 76.4 68.0 85.6 74.7 48.4

47.7 87.3 34.8 72.2 64.7 84.9 73.9 45.8

42.4 88.7 30.2 70.5 61.6 84.7 74.2 44.1

Sources: CNUCED, UNCTADstat.

28.4.3 Decrease of Employment in the Agricultural Sector The fall in the percentage of the workforce employed in agriculture has important consequences on the pattern of the relative workforce productivity between sectors. In the WAEMU countries, the share of agricultural labour in the total workforce decreased significantly in terms of percentage points between 2000 and 2014 (Table 28.3). The decrease was 15 percentage points in Benin, 14.7 in Côte d’Ivoire, 10 in Guinea-Bissau, 9 in Mali, and 8 in Togo. It was very modest in Niger (4 points) and Senegal (3 points). In Burkina Faso, on the other hand, the share of agriculture in the workforce increased by about 3 per cent between 2000 and 2014. As the secondary sector was not able to absorb the workforce freed by the agricultural sector, it found itself in the service sector and generally operates in the informal economy.

28.4.4 Diversification of the Production and Export System The concentration of production due to the limited development of the manufacturing sector leads to a weak diversification of exports. Indeed, the export profile of the West African Economic and Monetary Union has not really evolved in two decades. It remains characterized by a dependence on basic products with weak value added. Exports depend on a narrow range of products, in particular agricultural products, hydrocarbons, and minerals. In five of the Union countries, oil and mineral products are the principal export products. Agricultural commodities constitute the principal goods exported by the other three. With the exception of Côte d’Ivoire, Senegal, and Togo, less than five products represent 75 per cent of the countries’ exports.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

Table 28.4 Indices of export diversification of WAEMU countries Benin Burkina Faso Côte d’Ivoire Guinea-Bissau Mali Niger Senegal Togo WAEMU Sub-Saharan Africa Developing economies

1995

2000

2005

2010

2014

0.77 0.80 0.82 0.69 0.76 0.77 0.81 0.74 0.78 0.53 0.28

0.81 0.75 0.81 0.67 0.81 0.85 0.77 0.75 0.75 0.61 0.26

0.79 0.82 0.73 0.66 0.82 0.78 0.69 0.72 0.72 0.61 0.25

0.75 0.83 0.73 0.76 0.84 0.79 0.76 0.72 0.69 0.58 0.21

0.76 0.76 0.74 0.77 0.84 0.83 0.73 0.69 0.71 0.53 0.18

Source: CNUCED, UNCTADstat, 2015.

The weak export diversification of the WAEMU countries is confirmed by the evolution of the export diversification index, which remains above 0.60 for the 1995–2014 period while it has decreased in sub-Saharan Africa and the body of developing economies (Table 28.4). The share of manufactured products in the Union’s exports is modest. It is only 15.42 per cent of the total exports of goods and services as opposed to 56.79 per cent for the imports. Côte d’Ivoire, Mali, and Senegal are the three countries of the Union whose manufactured products represent, on average, more than 10 per cent of the exports of goods between 2010 and 2014. With the exception of Benin and Senegal, more than 50 per cent of the imports of the other countries consists of manufactured products. Given the fact that the countries of the WAEMU zone produce the same goods, intra-union trade remains weak. In 2014, exports within the WAEMU represented only 19.4 per cent of the Union’s total exports, and 19.0 per cent of the imports. The exports and imports within Africa represented respectively 21.9 per cent and 23.6 per cent of the Union’s total exports and imports. Besides the similarity of products, the low level of trade is due to the weakness of infrastructures, particularly those of transport. In conclusion, the good macroeconomic performances of the countries of the West African Economic and Monetary Union have not had a significant impact on structural change. This weak structural transformation has harmful consequences on the socioeconomic indicators of the WAEMU economies. Thus, in 2013, the Human Development Index (HDI) in all the countries of the Union was inferior to the average in Africa and the world (UNDP 2014). Between 2002 and 2010, an average of 50.4 per cent of the total population was living below the poverty level. The incidence of poverty ranged from 36.2 per cent (Benin) to 69.3 per cent (Guinea-Bissau). In four of the countries of the Union, the mortality rate of children under the age of 5 was more than 100 per 1,000, whereas it was only 47 per 1,000 at the world level.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Beyond the elevated incidence of poverty, most other social indicators are unsatisfactory. Life expectancy in the Union at birth is 56.7 years compared to 70.8 years on a world scale. With the exception of Senegal, Benin, and Niger, life expectancy in the Union countries is below the average in Africa and other developing States, particularly in Southeast Asia, Latin America, and the Caribbean. This is all the more worrisome because if structural transformation does not take place, these indicators can endure in the coming years. This counter-performance with regard to the structural transformation of the WAEMU countries can be explained by various obstacles that must be identified and removed.

28.5 SWOT A  R  R  O   S T   WAEMU E

.................................................................................................................................. Since 2000 the States of the West African Economic and Monetary Union have experienced appreciable rates of economic growth. Nevertheless, the structural transformation of the Union’s economies that ought to result from this is almost imperceptible. The sectoral composition of production has remained stable with little diversification; the service sector accounts for more than 50 per cent of the economic activity, about 30 per cent comes from agriculture, which continues to be the chief job provider, and less than 20 per cent from the industrial sector. The degree of production and export diversification is also weak and stagnant. And yet, the WAEMU economies have assets that would enable them to achieve their structural transformation. What then are the weaknesses that are slowing down change to the production structures, the opportunities that are available and can be exploited against these weaknesses, and the threats that are likely to wipe out efforts at structural transformation? The SWOT analysis in Subsection 28.5.1 emphasizes these economies’ principal strengths, weaknesses, and opportunities, and the threats to them. In the light of this analysis, recommendations will be formulated in order to help accelerate the structural transformation process in the WAEMU economies.

28.5.1 SWOT Analysis Table 28.5 summarizes the SWOT analysis for the WAEMU countries. They have at their disposal a certain number of assets for achieving the structural transformation of their economies: specifically political stability and social peace, demographic growth

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

Table 28.5 SWOT analysis of the structural transformation of the economies of the WAEMU Strengths

Weaknesses

-Political stability and social peace -A young population -Implementation of sectoral reforms (National development programmes) -Availability of diversified natural resources

-Weakness of human capital accumulation and of total factor productivity -Weak production and export diversification -Insufficient job creation -Market weakness

Opportunities

Threats

-Regional integration and globalization -WAEMU restructuring and upgrading programme -Community Development Program (CDP) of the ECOWAS

-Demographic growth -Growth slowdown in emerging countries -Terrorist threat -Climate change

and the youth of the population, structural reforms and the availability and diversity of natural resources. Political stability and social peace in the WAEMU zone create an environment of security that offers investors a reassuring climate. They can go about their business without fear and uncertainty, work calmly, and move about in the region without being bothered. This situation is an asset for the WAEMU economies, for it could favour the industrial investment necessary for diversifying production and improving productivity. Another asset of the WAEMU countries is the demographic transition and the youth of the population, which offers them the possibility to benefit from a growth dividend. Indeed, the demographic transition could be characterized by an increase in the economically productive population that can generate income. This will make it possible to reduce budgetary expenditures as regards services such as health and education and, consequently, to invest more in the agricultural and industrial sectors to make them more productive. Furthermore, the arrival of a great number of supplementary workers could favour the structural transformation and diversification of the economies because young people are more inclined than the existing workforce to participate in new economic activities. Almost all the WAEMU countries want to become emerging economies. National development programmes have been drawn up and constitute coherent frameworks of economic policy and reform. The leaders’ visions to make their homelands emerging nations and the sectoral reforms in progress to reach this goal constitute an asset which could accelerate the process of structural transformation of the WAEMU economies. Alongside the assets, there are weaknesses that slow down the structural transformation of the WAEMU countries’ economies. Among them are the weakness of human capital and of total factor productivity, the weak diversification of production and exports, insufficient job creation, and the narrow market.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Economic growth essentially depends on human capital and on productivity, which are factors in which the WAEMU countries lags behind the emerging countries of Asia and America. At the human capital level, basic education rates in the WAEMU area are decidedly inferior to those of Asian countries (IMF 2015). Now, a better trained workforce is more able to adapt and adopt the new technologies that enable a country to engage in diversification and development. It can also help maximize the cross-over effects with investments in physical capital. For not only are skilled workers needed to run sophisticated machinery and equipment, but better technologies also increase.the output of these workers. Regarding total factor productivity, that of work remains weak due, in part, to the lack in quality of human capital and to the structure of employment. The major proportion of the working population is in the agricultural sector where productivity is very weak because of the insufficiency of physical as well as human capital. The employment structure predisposes a low rate of total productivity because the dominant sector, agriculture, is characterized by the use of low-level techniques. In the States of the WAEMU, an important share of production is agricultural in origin. The primary sector is the largest provider of jobs and foreign trade depends on it. It is probable that in the medium term this situation will persist, even if the manufacturing sector develops. Given its importance, the transformation of the agricultural sector through gains in the productivity of existing activities and the improvement of results outside the sector should constitute fundamental goals of the development policy of the WAEMU countries. Improvement of factor productivity in this sector, which is still weak, could then stimulate growth. Among the weaknesses that are slowing the structural transformation of the WAEMU economies, there is also weak export diversification. Diversification is necessary to increase the resilience of the Union countries against external shocks, in particular variations in commodity prices. Beyond risk management, diversification is an instrument for improving the quality of growth by promoting technological innovation and workforce productivity. It is also useful in increasing values added within the Union’s economies and thus creating productive jobs. The leaders of the WAEMU countries should set their sights on diversifying their agriculture, while broadening their industrial bases and encouraging the rapid expansion of a new service economy. Poor job creation is another weakness of the WAEMU economies. The formal sector, consisting chiefly of small businesses and a small number of large firms, continues to provide few jobs. The insufficiency of decent jobs, especially in the industrial sector, limits the base on which social security contributions are assessed as well as the improvement of productivity. It is important to promote job creation by giving primacy to industrializing transformation, particularly in the food processing and textile industries, which are generally competitive and which create jobs and wealth. Moreover, the WAEMU countries are faced, individually and collectively, with the challenge of a narrow market. This is an obstacle to the adoption of new lines of economic activities. Market weaknesses include providing infrastructure and financing, establishing trade networks and the functioning of factor markets, the regulatory and

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

institutional framework. Measures taken to promote structural transformation and diversification should aim to remedy these weaknesses. For example, lowering obstacles to entering a market can encourage entrepreneurs to broaden the scope of their activities. There are indeed weaknesses forming obstacles to the structural transformation of the WAEMU economies, but they are also presented with opportunities that, if well exploited, can enable them to face up to these weaknesses. The WAEMU countries benefit from the integration measures of the Union and those of the Economic Community of West African States (ECOWAS). Vintage products and artisanal products are able to move freely within the two organizations, as well as processed products that are in conformity with the rules of origin and are certified. The integrated framework thus offers a great opportunity for confronting the weakness of the countries’ internal markets, enabling them to diversify their exports. Furthermore, with globalization market boundaries are receding. In the context of its common industrial policy, the WAEMU has set out to institute a regional restructuring and updating programme in each of the eight member countries. The global objective aims to revive industrial production, promote investment, employment, and the competitiveness of firms and economies at the regional and international levels. The specific objectives are (i) to strengthen the capacities of firms so that they can follow and master technological evolution and adapt to the demands of regional integration and international competition, and (ii) to facilitate the emergence of support and consulting services that would offer firms the necessary competences and qualifications. The achievement of these objectives will make it possible for firms to develop, which can have a driving effect on the other sectors of activity and favour structural transformation. This restructuring is an opportunity to develop the industrial sector, create jobs, improve factor productivity, and diversify production. Good exploitation of the opportunities described above could make it possible to begin the structural transformation process of the WAEMU economies. However, the change will be successful only if demographic growth is brought under control, the terrorist threat is eradicated, and climatic conditions remain relatively favourable. Fertility rates in the WAEMU continue to be the world’s highest, despite the rapid decrease in infant and child mortality. The population could double in the course of the next twenty years, going from about 100 to 200 million people, which would mean a net annual population increase of the working population of some 1.3 million persons (IMF 2016). If fertility rates do not decrease from their present levels, the proportion of the working age population will remain constant, and neither the demographic transition nor the corresponding growth dividend will materialize in the medium term. What is more, due to its rapid increase, the population will continue to exert pressure on public services and infrastructures, and the surplus workforce will have to seek informal jobs in weakly productive agriculture or find itself unemployed. This situation will present risks for the growth of total productivity, poverty levels, and social cohesion.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



Beyond the risks of very rapid demographic growth, the structural transformation of the WAEMU economies can be influenced by a slowdown in the growth of partner countries, particularly emerging countries, even if they account for only a limited share of the Union countries’ international trade. A decline in economic activity would reduce the diversification of trading partners and exported goods, which could negatively affect the diversification of production. The upsurge in terrorist risks in West Africa could affect the WAEMU economies and delay the process of structural transformation. Threats of attack force governments to use a large share of their resources to ensure security. This reduces expenditures on infrastructure and the number of jobs created, which are factors of structural transformation. Finally, apart from the uncertainties linked to demographic growth, the evolution of partner economies, and terrorist threats, the performance of the economies of WAEMU countries remains associated with climatic conditions. A rapid fall in production of the principal export crops could have negative consequences on the productivity of the very concentrated workforce in the agricultural sector.

28.5.2 Recommendations for Removing the Obstacles to Structural Transformation Boosting the structural transformation of the WAEMU economies will require three measures. It will be necessary to modify the structure of the final demand, work on the factors of change and sectoral productivity, and encourage the diversification of production and exports.

28.5.2.1 Modifying the Structure of the Final Demand In order to modify the demand structure, the States’ intervention in industry growth will be essential. They must finance manufacturing, regulate production, and control its quality. Investments must be made in the structural transformation of the countries of the WAEMU space, in particular in the weak link, the manufacturing sector, and its bottlenecks (such as high transport and energy costs which cause them). Moreover, the economic growth of the WAEMU countries is not inclusive. Efforts should be made for a more equitable distribution of the wealth produced. Thus, the improvement of income could lead to an increasing demand for new products. This would result in stimulating investment in new industrial sectors, thus leading to a diversification of production and therefore of exports. The WAEMU countries have an opportunity to modify the structure of consumption. The restructuring programme, in particular, is likely to develop the industrial sector, making it possible to create jobs, improve incomes, and stimulate the demand for new goods and services, thereby fostering the diversification of production and

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

factor mobility, principally moving the workforce from less productive sectors to those that are more productive.

28.5.2.2 Encouraging Factor Mobility and Sectoral Productivity Policies and institutions have a direct impact on the extent to which resources move between sectors (value added or employment). Sectoral changes occur in reaction to the regulatory and infrastructure context, which has an impact on employment mobility. An improvement in education goes hand in hand with an increase in the share of value added in the manufacturing sector as well as in services. Gains in productivity continue to be reduced by the low level of investment in the factors of production, among them human resources. The abilities of the working population are underexploited, which weighs on productivity. Investments are rarely made in technology instruction and innovation and in the development of skills apt to stimulate productivity and respond to market needs. This situation is worrisome because it is an accepted fact in economic literature that if movements of productivity toward high end service activities take place, it is due to the implementation of services based on knowledge, innovation, and entrepreneurship and to the tools of electronic governance. To improve productivity, investments in physical capital, but also in R&D activities should be implemented. Support measures in the form of fiscal and financial incentives can be granted to stimulate firms that invest in R&D. Furthermore, the strengthening of human resource capacities must be emphasized. By improving productivity, the WAEMU countries will become competitive and will be able to move toward freer trade, following the example of India which has, by easing economic and trade restrictions, succeeded in accelerating economic growth accompanied by structural changes that are admittedly slow, but tangible. The countries of the WAEMU, particularly Côte d’Ivoire, have invested heavily in infrastructure, but the deficit in this domain remains large with regard to built-up backlogs. The States of the Union remain weakly equipped with basic infrastructure, particularly as regards energy and transportation. Infrastructure constraints are common to the States ofWAEMU, and the investment needed to remove them is great. It would therefore be wise for these countries to use the resources available to invest, in a targeted manner, in energy, transport, and port services and in the upkeep of these infrastructures so as to maximize and maintain the effectiveness of the investments made. In addition to these physical investments, the governance of the public sector must be strengthened, notably in managing stateowned companies and public investments. Regional integration and the ECOWAS Community Development Program provide for the implementation of structuring infrastructures. Moreover, carrying out the WAEMU’s restructuring programme affords an opportunity to develop the industrial sector, which will make it possible to improve productivity. The West African Economic and Monetary Union is, furthermore, still facing a sizeable challenge in improving the quality of technical and higher education. In the

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



sub-region as a whole, the number of students enrolled in professional and technical courses is well below the minimal norm of 20 per cent prescribed by UNESCO. As skilled human resources are vital to economic transformation, the countries of the WAEMU should accelerate educational reforms, particularly in technical, professional, and scientific fields, to improve the quality of the workforce and increase its productivity. The emphasis should be on the acquisition of essential skills targeted to fields promising diversification and the development of technical and entrepreneurial capabilities. Training should focus on jobs being created by industry and become a genuine springboard to good career prospects. Investment in science and technology should also be significantly increased. A regional approach in domains where national efforts would be more costly is desirable. India provides an instructive example, for the adoption of modern agricultural technologies has helped the country to eradicate famine, improve its growth rate (that had never been higher than 2 per cent), and achieve growth rates on the order of 10 per cent, a rate that has helped improve the structural transformation of the economy. Moreover, adopting technologies has made it possible to diversify production and exports.

28.5.2.3 Encouraging Export Diversification Diversification and structural transformation build on general measures and reforms. Even if all countries do not follow the same path of diversification, the similarity of results seems to indicate that successful diversification is based on general policy measures and economic factors. Investment in infrastructure is an essential ingredient of diversification, for it reduces firms’ operational costs. This means that the State has an important role to play in supporting diversification as a producer and/or regulator of infrastructures. Investment in human resources is fundamental to progressing on the quality scale. A better level of education in the workforce is also more likely to deliver effective entrepreneurs wishing to embark on new activities and improve their existing products. The business environment is an important hindrance to transforming the WAEMU economies. Between 2005 and 2015, the countries of the Union undertook reforms in business regulation, notably in the areas of company creation, obtaining loans, and tax payments. However most of the States of the Union are still at the bottom of the World Bank’s classification of ease of doing business. The best-rated Union country in 2015 was Côte d’Ivoire, classified 142nd out of 189 countries. The extent of deficiencies in doing business varies by country, but four urgent areas of common reforms can be highlighted: business creation, access to credit, the legal environment, and cross-border trade. Transparency and the effectiveness of institutions and governance are particularly important for the Union countries because of their natural resources. In this regard Botswana’s experience can serve as an example for the leaders of the WAEMU countries. They should emphasize (i) strengthening the struggle against corruption

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

and the control and responsibility mechanisms that are essential to maintaining a high degree of accountability and transparency; (ii) strengthening human and institutional capacities to improve government efficiency and the quality of public expenditure; and (iii) improving the quality and availability of financial information on management and practices relative to the natural resource sector, including publishing contracts. Improving the environment of the private sector is necessary and urgent for the WAEMU countries, whose business climate has very few incentives. The example of Mauritius, considered the best African country in which to do business, shows that creating a business-friendly regulatory environment and incentives to attract foreign firms helped lay the foundation for the country’s economic transformation. The example of Mauritius shows that public action ought to give priority to (i) robust protection of property rights by strengthening the capacities and performance of the legal sector; (ii) implementation of a simple tax system with more incentives; and (iii) simplification of business creation.

28.6 C

.................................................................................................................................. The countries of the WAEMU have, from 2000 to the present, registered a marked acceleration in growth rate. Since economic growth and structural transformation are interdependent, it would be expected that the economies of these States experienced a modification of their production structures. But an analysis of the composition of the final demand and of intersectoral dynamics based on production, factor movement, and the diversification of exports has shown that this sub-region’s economies did not undergo a significant structural transformation in the course of the 2000–15 period. And yet the States of the West African Economic and Monetary Union have assets that could enable them to make changes in their production structures. These are, among others, political stability and social peace, a young population, sectoral reforms, and natural resources. To best exploit these assets and succeed at structural transformation, weaknesses slowing this process must be corrected, notably the weakness of human capital and factor productivity, insufficient diversification of production and exports, deficient job creation, and a narrow market. Opportunities to correct these deficiencies are available to the WAEMU States in order to modify the structure of the final demand, favour the intersectoral dynamics of factors of production and stimulate the diversification of production and exports. Indeed, regional integration and globalization offer the WAEMU countries not only access to a larger market, thus solving the problem of the narrowness of the subregional market, but also the opportunity to adopt new product lines for export, therefore diversifying production and exports. Furthermore, the WAEMU’s restructuring and upgrading programme in the context of its common industrial policy is an opportunity for each country of the Union to

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi

      



further broaden its industrial base. The expansion of this sector can foster job creation and therefore income distribution. This will lead to changes in the final demand structure. Development of the industrial sector will also favour the growth of global productivity by stimulating the movement of the agricultural workforce to manufacturing. One of the obstacles to the structural transformation of the economies of the West African Economic and Monetary Union is the low level and poor quality of infrastructures. Targeted investments in infrastructure are needed to remedy this. The resources devoted to this sector should first and foremost be put into energy and transportation infrastructure and its maintenance in order to maximize the return on the investments made. This will best be done in the context of the ECOWAS Community Development Program. All these opportunities available to the countries of the Union can be developed only if they bring their demographic growth rates under control and manage to establish security in the WAEMU area. While this chapter focuses on the WAEMU, the problems it identifies are practically the same in most African countries. This being the case, with only a few reservations, the recommendations made here can be useful for a good many African countries in effecting the structural transformation of their economies in the years to come.

R African Development Bank, 2011. Africa in 50 Years’ Time. The Road towards Inclusive Growth, September, Tunis: African Development Bank. Cadot, O., C. Carrere, and V. Strauss-Khan, 2011. ‘Export Diversification: What’s Behind the Hump?’ The Review of Economics and Statistics, 93 (2), pp. 590–605. Chenery, H., 1979. Structural Change and Development Policy, Oxford: Oxford University Press. Clarke, C., 1940. The Conditions of Economic Growth, London: Macmillan & Co. Duarte, M. and D. Restuccia, 2010. ‘The Role of Structural Transformation in Aggregate Productivity’, The Quarterly Journal of Economics, 125 (1), pp. 129–173. ECA, 2013. Economic Transformation for Africa’s Development. UN Commission for Africa, Macroeconomic Policy Division, Washington, DC: ECA. Fosu, A. K. and E. K. Ogunleye, 2015. ‘African Growth Strategies: The Past, Present, and Future’, in Célestin Monga and Justin Yifu Lin, eds, The Oxford Handbook of Africa and Economics, Volume II: Policies and Practices, Oxford: Oxford University Press, pp. 23–38. Hausmann, R. and D. Rodrik, 2006. ‘Doomed to Choose: Industrial Policy as a Predicament’. Mimeo, Cambridge, MA: Harvard University. Available at: http://www.ricardohausmann.com Herrendorf, B., R. Rogerson, and A. Valentinyi, 2013. ‘Two Perspectives on Preferences and Structural Transformation’, American Economic Review, 103 (7), pp. 2752–89. Herzer, D. and F. Lehman, 2006. ‘Export-Led Growth in Chile: Assessing the Role of Export Composition in Productivity Growth’, The Developing Economies, 44 (3), pp. 306–328. Imbs, J. and R. Wacziarg, 2003. ‘Stages of Diversification’, American Economic Review, 93 (1), pp. 63–86.

OUP CORRECTED PROOF – FINAL, 31/12/2018, SPi



ˊ ˊ ’

International Monetary Fund (IMF), 2011. ‘Changing Africa: Rise of a Middle Class’, Finance & Development, 48 (4). International Monetary Fund (IMF), 2014. ‘Sustaining Long-Run Growth and Macroeconomic Stability in Low-Income Countries—The Role of Structural Transformation and Diversification’, Washington, DC: IMF. International Monetary Fund (IMF), 2015. West African Economic and Monetary Union—Staff Report on Common Policies of Member Countries. Report 15/100, Washington, DC: IMF. International Monetary Fund (IMF), 2016. Regional Economic Outlook: Sub-Saharan Africa. Time for a Policy Reset, Washington, DC: IMF. Klinger, B. and D. Lederman, 2004. ‘Discovery and Development: an Empirical Exploration of New Products’. Policy Research Working Paper No. 3450, World Bank. Lewis, W. A. 1954. ‘Economic Development with Unlimited Supplies of Labor’, Manchester School of Economic and Social Studies 22, pp. 139–91. Lin, J. Y., 2012. ‘Structural Change in Africa’, in Ernest Aryeetey, Shantayanan Devarajan, Ravi Kanbur, and Louis Kasekende, eds, The Oxford Companion to the Economics of Africa, Oxford: Oxford University Press, pp. 296–303. McMillan, M. and K. Harttgen, 2014. ‘What Is Driving the “African Growth Miracle”?’ NBER Working Paper No. 2007. McMillan, M. and D. Rodrik, 2011. ‘Globalization, Structural Change and Productivity Growth’. NBER Working Paper No. 17143. Moshe, S., 1988. ‘Croissance économique, changement structurel en Colombie: une comparaison internationale’, Revue Tiers Monde, XXIX (115). N’Guessan, T., 2008. Gouvernance collégiale de la Banque Centrale et politique monétaire: enjeux, fondement et modalités pour les pays africains de la Zone franc, Paris: l’Harmattan. N’Guessan, T., 2012. ‘Côte d’Ivoire, Economic Reversibility’, in Ernest Aryeetey, Shantayanan Devarajan, Ravi Kanbur, and Louis Kasekende, eds, The Oxford Companion to the Economics of Africa, Oxford: Oxford University Press. Ocampo, J., 2005. ‘The Quest for Dynamic Efficiency: Structural Dynamics and Economic Growth in Developing Countries’, in J.A. Ocampo, ed., Beyond Reforms: Structural Dynamics and Macroeconomic Vulnerability, Palo Alto, CA: Stanford University Press, ECLAC, and World Bank. Papageorgiou, C., F. Perez-Sebastian, and N. Spatafora, 2012. ‘Structural Change through Diversification: A Conceptual Framework’. Working Paper. Ros, J., 2000. Development Theory and the Economics of Growth. Ann Arbor: University of Michigan Press. Sévérino, J. M. and O. Ray, 2010. Le temps de l’Afrique, Paris: Odile Jaco. Tchatchouang, J. C., 2015. ‘The CFA Franc Zone: A Biography’, in Célestin Monga and Justin Yifu Lin, eds, Oxford Handbook of Africa and Economics, Volume 2: Policies and Practices, Oxford: Oxford University Press. Timmer, M., G. de Vries, and K. de Vries, 2013. ‘Structural Transformation in Africa: Static Gains, Dynamic Losses’. Research Memorandum No. 136. Groningen Growth and Development Center. UNDP, 2014. Human Development Report. Sustaining Human Progress: Reducing Vulnerabilities and Building Resilience, Washington, DC: UNDP. Vergne, C. and A. Ausseur, 2015. ‘La croissance de l’Afrique subsaharienne: diversité des trajectoires et des processus de transformations structurelles’. AFD, Macroéconomie et Développement, May. World Bank, 2016. World Development Indicators, Washington, DC: World Bank.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

P A R T V .............................................................................................................

CONCLUDING THOUGHTS .............................................................................................................

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

        ......................................................................................................................

                 A Reassessment of Development Economics ......................................................................................................................

ˊ  

29.1 S, B,  R

.................................................................................................................................. G pride themselves on being the first African country to reclaim its independence from colonialists. The pride and the boasting reached new heights in 2017, as they celebrate the sixtieth year of their freedom. But behind the chanting, dancing, talking, and conferencing, the question often discussed mezza voce is the true length of the road travelled since then, and the size and value of the dividends of freedom. Many Ghanaians in the poor slums of Accra and Kumasi, and in the miserable rural areas of the country, are blaming their political leaders of the past six decades for being mainly sinners who should be held responsible for their plight. They don’t think that independence has yielded its promise. In fact, it is an appropriate moment for Ghanaians, African leaders, and development economists to step aside from the dancing and the chanting to reflect on progress—and on what went wrong in the quest for prosperity, not just in this country, but throughout the developing world. And perhaps to seek redemption. Back in 1957 when Osagyefo Kwame Nkrumah successfully led Ghana’s quest for independence, I was not yet born but I know from reading history books that this country was expected to quickly become the beacon of hope for the continent and for the Third World. There was a lot of excitement. Since then, Ghana has gone through its ups and downs but has mostly underachieved—like most countries elegantly branded as ‘developing’. Nkrumah seemed to have all that was needed to convert freedom into collective welfare for his people: he had the vision, the passion, the inner strength, the popular support, and the basic wit. He even recruited as one of his senior advisors, Sir Arthur Lewis, one of the greatest economic minds of the time and to this day, and the only Black person to ever win the Nobel Prize in economics.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

Yet, great economic thinking did not seem to help policy making. Six decades after independence and despite experimenting with almost all types of modern forms of government and the rebasing and the statistical revisions of gross domestic product, Nkrumah’s country is still not doing very well. Growth has been insufficient to reduce poverty, and most of the labour force—whether highly skilled or not—is still underemployed if not unemployed. What happened? Like many developing country leaders since independence, Nkrumah was a bright young man, well educated, and a true hero to his people. But he lacked some basic wisdom—which manifests itself in the humble search for the truth and the willingness to ignore ideological jargon and focus instead on economic theories and policy practices that provide indisputable, quick, and sustainable results. Compare Nkrumah to Deng Xiaoping or Lee Kuan Yew, who rose to the challenge of leadership and presided over successful experiments in politically difficult circumstances. What did they know that Nkrumah di not? What can we learn from their stories? It would be unfair to pick on Ghana. In fact, Ghana’s story has largely been Africa’s story, and that of many other developing countries. I will argue in this think piece that Ghana’s failure to rise to its potential and achieve its goals has mainly been the failure not of politics, the weakness of institutions, or the lack of initial conditions, but the failure of economic thinking and policy making. Therefore, I believe that the failure of development was not—and still is not—caused by insufficient human capital, by inadequate access to finance, or by such popular notions as infrastructure deficits or corruption. These factors made matters worse but they are consequences of bad initial strategic choices. But they are not the causes of the problem. I will also argue that Ghana’s failures and developing countries’ failures may reflect poor leadership but not in the traditional, popular, and vague sense. Few political leaders ascend to power and devote their efforts to destroying their country or impoverishing their people. In fact, politicians everywhere often have the same utility function: when their political regime is secure, safe, and stable—as Nkrumah’s was in 1957—they are typically interested in staying in power as long as they can, and in having a good name in history after they are gone. This is true in democratic and autocratic regimes. Barring some extremist and crazy characters, they all have these same basic motivations. The major difference among political leaders across political systems is indeed their leadership ability, but leadership defined more narrowly as the capacity and willingness to articulate bold but credible visions, choose good economic advisers, and lead teams that can implement plans with discipline and humility to achieve results rapidly—while correcting the unavoidable mistakes along the way. Most African politicians have, on average, been bad leaders. But the failure of economic development is primarily due to the pervasiveness of bad ideas, which translate into bad advice by influential economists—those in the position to shape or influence policy making. Poverty and underemployment in Africa today (at a time when resources of all kinds and opportunities have never been so numerous and so cheaply available everywhere) primarily reflect the imagination deficit among

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



development economists—not the incompetence of politicians. After all, politicians everywhere do what they are supposed to: politics, a selfish and often dreadful sport. It is the responsibility of economists to put great ideas on the table and find ways to positively influence the public discourse and the policies that are implemented. Retreating to their ivory towers, and complaining that political leaders do not solicit their wisdom and knowledge, is an easy escape route. It does not absolve them from their duty as elites, intellectuals, social leaders, and minority members of a community in which they represent what W. E. B. Du Bois called ‘the talented tenth’. They do not seem to fully grasp the burden of the responsibility that comes with being highly educated in a society that craves for knowledge and learning. To repeat: What exactly happened to Ghana and to other African and developing countries? In my view, two basic dynamics negatively reinforced each other over decades. First, national leaders gained political power, often in unconventional ways, and spent most of their energy either trying to replicate the governance models from colonial times, or to imitate policy frameworks in vogue in advanced countries but inappropriate for theirs. Then, they quickly ended up succumbing to the intoxicating lustre of authority and struggling to maintain their grip on the levers of state political and financial power. Many of them failed to understand that the most effective way to remain at the top and retain control was to deliver quick, tangible results, and to improve the lives of their people. Most of the time, they consistently picked and implemented bad ideas, often from well-meaning economists and development experts. The combination of the two processes gradually made things worse and worse. The main results were economic failures, massive unemployment, poverty, despair, distorted belief systems, social and political chaos, and the generalization of witchcraft. Frustrated by the course of events, many good economists gave up—in fact, the field of development economics was pushed to the fringes of the discipline for much of the 1960s and 1970s. Others opted for intellectual laziness, spending their time in intellectual mimicry, and simply transferring whatever concepts or theories were in fashion in Latin America and Asia to their analyses of African countries. It is not surprising that a lot of development economics has been dominated by the identification of the sins committed and the search for who is to blame. The good news is that time has gone by. Lessons have been learned. Spectacular successes such as the rise of China, Taiwan-China, Singapore, South Korea, Dubai, and the United Arab Emirates, and unfolding ongoing successes in Mauritius, or Vietnam, have shed light on what should be done or avoided. Despite many failed experiments, economic history now provides enough good stories that do not have to be copied but that can certainly inspire both researchers and policy makers. There are possibilities of redemption for political leaders and development economists in Ghana and elsewhere. This note summarizes some of the key elements of the knowledge accumulated in development economics. It is obviously biased toward my own work with Justin Yifu Lin (under the label New Structural Economics), and Joseph Stiglitz (on the rethinking of industrial policy)—a selected reading list is provided in the references. Section 29.2 challenges the dominant paradigms of development thinking, from a fundamental,

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

philosophical perspective. Section 29.3 sums up the key recommendations for a more appropriate approach. Section 29.4 offers concluding thoughts.

29.2 E   P: A C

.................................................................................................................................. Years ago, while studying in Boston, I asked Robert Solow why many great economists like him had stayed away from African issues. I was trying to provoke him because the truth is that many great economic minds had worked on Africa. In his typical direct style, Solow responded that he avoided development economics simply because it was ‘too hard! Most of my colleagues who ventured in that area did not do too well.’ He also explained that the kind of macroeconomics he did would apply only in socioeconomic and political environments where the institutions required to make it work were already in place. He felt that developing countries were intrinsically different from advanced economies. In 2008 while an economist at the World Bank, I tried to convince Olivier Blanchard, then the chair of the Economics department at MIT, to express interest in the World Bank Chief Economist position, which had just become vacant. I felt that his stellar intellectual contributions to the analysis of employment, unemployment, and labour market issues would bring them to the centre of the global agenda. I always believed that job creation should be at the core of our poverty reduction strategies. Olivier’s response was swift and clear: ‘I think I know a bit about macroeconomics, but I know very little if anything about economic development. I certainly would not qualify for that job.’ Later, after he became the chief Economist of the IMF, I asked how he felt about the research his institution was doing on Africa. He praised his team but told me very candidly that he was not satisfied with some of the analytical tools used for policy analysis and simulations on low income countries. He indicated that some of the work was too often small variations from dynamic stochastic general equilibrium models designed for industrialized economies. The day we discussed this, he had just reviewed the econometric model for Togo and could tell that the assumptions about the functioning of the labour market there were probably inappropriate for a low income African country. He also expressed frustration at not being able to provide a suitable econometric model for effective policy making in Togo. Beyond their remarkable humility and their refusal to venture outside their field of expertise, Solow and Blanchard were highlighting one of the biggest intellectual problems in development economics: the disconnect between high income and low income economies, and the refusal of some economists to acknowledge the structural differences that this implies or to draw the analytical implications from that central fact. Solow and Blanchard intuitively recognized that Burundi is not Switzerland even

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



though both are small and landlocked, and that Madagascar is not Japan, even though both are sea-locked. Structure matters. Neglecting that intrinsic truth has been the biggest of all analytical sins by development economists. That sin has led to two major strategic mistakes, which in turn have invalidated much of the excellent work done since the emergence of development economics as a subdiscipline of economics after World War II. First, in thinking about convergence and the task of transforming low-income countries into industrialized economies, we have too often selected the wrong comparators and benchmarks, and set the wrong objectives for our work—and obviously that of the policy makers who follow our advice. The wrong choice of model economies has led many theoreticians and empiricists to adopt some profoundly misleading assumptions that no sensitivity analyses could correct. Second, these compounding mistakes have generated a false economics of preconditions, and the wrong policy prescriptions. Let me briefly discuss these two mistakes in turn.

29.2.1 The Wrong Model Economy and Reference It is obviously legitimate for Burundi to try to be like Switzerland, and there is indeed no reason for the many hardworking citizens of Bujumbura to expect to live the supposedly good life of the not so hardworking people of Zurich. But even leaving aside history, the current endowment structures of two economies are so different that it would not make much analytical sense to study them with the same tools and to derive policy recommendations from similar econometric models. It is legitimate for political leaders in Burundi to try to emulate the success of others in advanced economies. It is strange for economists to rigidly and mechanically apply methods and tools designed for capital-intensive Switzerland to capital-poor, labour-intensive Burundi. It has been reported that Nkrumah wanted the Ghanaian economy to surpass that of England almost the day after independence. A noble political goal perhaps, but a foolish predicament for economic policy. The model economy that Burundi 2017 or Ghana 2017 may want to ‘emulate’ should not be Switzerland 2017 or the United Kingdom 2017 but, perhaps, Mauritius 1974. There is a sequencing of strategies that low income economies ignore only at their own peril. Let me stress one point: I am not advocating a teleological view of economic development. Of course, one can move faster from one step to the other, and Walt Rostow was wrong in suggesting that there is an almost linear process with a rigid timing in the various ‘stages of economic development’. The acceleration of economic history is the evidence that he was wrong. Prior to the eighteenth century, it took 1,400 years to double income in the Western world. In the nineteenth century, it took only about seventy years. In the twentieth century, it required only about thirty-five years. The changing pace of performance among individual countries is even more encouraging. It took 150 years for Great Britain to initially double its income. The USA

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

needed fifty years to do the same. And without the natural resources, excellent infrastructure, or human capital of Britain, the USA, or the Scandinavian countries, China did it in just twelve years. Some economists draw the wrong inference from these facts: they mistakenly conclude that low income economies can achieve prosperity without developing manufacturing, by just launching high value added industries or encouraging the emergence of tradable services. Such strategies, which violate comparative advantage, would not be sustainable. Trying to develop sophisticated, capital-intensive industries and services in a $500 per capita economy whose comparative advantage is still in labour-intensive industries is the problem: with a labour force of some 600 million people, most of them low-skilled workers, Africa puts less than 20 per cent of its labour force in the formal sector. The lesson here is not that poor economies should try to circumvent the steps of industrialization and jump from low-productivity agriculture to high-tech industries and services when they are capital-scarce and have poor business environments. Instead, they should try to put the largest possible fractions of their labour force to work by developing industries that are consistent with their existing (and changeable) comparative advantage. The dynamics of more people working in the formal sector, earning gradually decent incomes and developing soft skills and their human capital, eventually moves the economy into more sophisticated, high value added industries and sectors. Of course, good public policies can help speed up the evolution of an economy’s endowment structure and compress the timing of structural transformation. Unfortunately, some development economists continue to neglect the need for industrial upgrading in low income countries. They still advocate benchmarking poor economies, with high income economies as the model to emulate. That has been a recipe for disappointment for a long time.

29.2.2 The Wrong Assumptions and Preconditions Choosing the wrong model or reference economy to copy carries some heavy implications and leads to risky optical errors. Instead of seeing the resources already in place in each poor country and focusing the intellectual and policy resources to maximizing the existing assets and building on successes to accelerate reforms, one focuses only on the missing ingredients for growth and prosperity, and quickly becomes obsessed with the lengthy list of preconditions to be fulfilled—before the growth process can be ignited. This is a disturbing trend, and a counter-productive intellectual attitude. That mindset of ‘missing elements’ and ‘gaps’ has translated into a peculiar intellectual posture: economists have confined themselves to the role of detectives, if not prosecutors. Without fully realizing it, many development experts behave like detectives in Arthur Conan Doyle and Agatha Christie novels. They have converted themselves into a Sherlock Holmes or Hercule Poirot whose divine mission is to find what is wrong with low income countries, not what may be right and sufficient there to start

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



something good. The ‘gap’ mentality among researchers has therefore stimulated the emergence of a dominant brand of development policy that is basically an obsessive (if not compulsive) quest to correct the ‘deficiencies’. Of course, the search has produced long lists of true or false deficiencies, real or imaginary gaps, and truly missing or illusory ingredients for development recipes, which are presented as necessary conditions for economic progress. But the whole exercise has not paid off. Indeed, the search has created more problems for economists and policy makers than it has brought solutions. Instead of focusing on what each country—even the poorest—already has to build a viable development strategy, the compulsive search for the missing gap has validated and legitimized the notion that little can be done in poor countries unless they meet a long list of preconditions. Instead of adopting the mindset of how to maximize whatever few production factors are in place, development economists have too often devoted their energy and creativity to what must be done as preconditions for growth and prosperity. Instead of looking at the glass as half full, they have consistently seen it as almost completely empty. Yet, we know from the history of development that no single successful economy in the world started out with ideal country conditions. Successful development processes always emerge from average if not very poor institutional and policy environments. Neither the United States nor Great Britain had ‘excellent’ infrastructure stocks prior to the Industrial Revolution, or even in the years and decades after that. China did not have ‘adequate’ levels of human capital when Deng Xiaoping launched the shocking economic journey of growing the economy by nearly 10 per cent a year for three decades and lifting some 600 million people out of poverty. By succumbing to the Sherlock Holmes syndrome, some researchers miss what should be the focus of economic policy in low income countries—and that is structural transformation, the transfer of human resources and capital to the most productive activities. This may be because many of us limited the development agenda to the Washington Consensus policies, mainly designed to address macroeconomic imbalances that developing countries experienced in the 1980s. Macro stabilization and structural reforms were necessary but not sufficient conditions for prosperity. They were not necessarily recipes for creating employment. Rwanda President Paul Kagame has declared that his country scores among the top-performers in almost all categories of the Doing Business Indicators but has not created enough employment in the formal sector. That is true: his country has done remarkably well in improving the business environment, but it must still do even more to generate the kind of good labourintensive industries that will eventually bring prosperity to the people. Another issue not always carefully studied is good governance. We all know how important that is and can see the correlation with GDP per capita in cross-country regressions. But we need to dig deeper, define it more precisely in our research, and customize what ‘good institutions’ might look like in different country contexts. Let me just give one example to illustrate the point. Late Presidents Félix Houphouët-Boigny and Julius Nyerere are among the most admired in recent African political history. Yet, the former—who once said at a press conference that any ‘serious individual’ should

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

have a Swiss bank account and urged people not to dwell on high levels of corruption in Côte d’Ivoire—was able to propel his country on a long path of high growth. Compare his development performance with that of President Nyerere, widely perceived as a near saint for running arguably the least corrupt African government of the post-independence era, who apologized for failing his people with low growth and high poverty. Clearly, poor governance is a grave issue to be tacked energetically. But even as this is still under way, smart growth strategies can be implemented and yield satisfactory results. Development thinking should aim at providing more actionable sets of policies to political leaders, and avoid offering laundry lists of reforms that are politically difficult to implement and may not immediately yield the intended results. Leaving economic development to the market is taking a bet on what I call the painful economics of chance, approaching economics as a prayer that may or may not be answered. Different industries require distinct types of infrastructure. And since low income country governments do not have the financial resources to accommodate all industries at once, it is best to work with the private sector to identify industries where the economy has a comparative advantage, and to focus on providing specific infrastructure and transparent, limited incentives that would allow these industries to grow. Look at the list of recent success stories in Africa to understand the role of industrial policies. Textiles in Mauritius, apparel in Lesotho, cotton in Burkina Faso, cut flowers in Ethiopia, mangos in Mali, and gorilla tourism in Rwanda all required that governments provide different types of infrastructure. The refrigeration facilities needed at the airport and the regular flights to ship Ethiopia’s cut flowers to the auctions in Europe are obviously quite different from the improvements required at the port facilities for textile exports in Mauritius. Similarly, the type of infrastructure for the garment industry in Lesotho is distinct from the one for mango production and export in Mali or for attracting gorilla tourism in Rwanda. Because fiscal resources and implementation capacity are limited, the government in each of those countries had to prioritize and decide which specific infrastructure they should improve or where to optimally locate the public services to make those success stories happen.

29.3 E  S D T

.................................................................................................................................. Where do we go from here? Well, back to the basics. If we believe the long-run statistics put together by Angus Maddison, we must accept that for 1,400 years of economic history all countries in all regions of the world were low income. Then the Industrial Revolution separated regions and countries. The divergence did not happen by chance: clever ideas generated prosperity in some places while bad ideas took hold in others. If we have learned anything since the Industrial Revolution, it is the basic truth that modern economic growth is a process of industrial, technological, and institutional upgrading

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



that reflects the changing dynamics of comparative advantage and endowment structure. Ignoring it leads to policy prescriptions that defy the laws of economics and lead inevitably to disappointments, and to what Ernest Aryeetey has called ‘negative diversification’ (moving labour from low-productivity, subsistence agriculture, to low-productivity, informal services). The challenge to development economists is to continue building richer and more credible economic models that also offer policy makers clearer road maps for implementing a key principle of the discipline: allocating scarce resources to industries, sectors, and geographical areas with the highest possible payoffs. Inevitably, this would imply being more rigorous in selecting where to put the money and other resources, and being more thoughtful in targeting reform efforts. As Ricardo Hausman, Dani Rodrik, and Andres Velasco have shown, not all binding constraints are equal—or deserve the same amount of attention from high-level policy makers. The very words ‘selection’ and ‘targeting’ tend to immediately raise concerns about ‘industrial policies’ by activist governments. So, let me clarify what I have in mind. All governments are ‘activist’ in one sense or the other. The question is whether their ‘activism’ is devoted to creating the optimal conditions for agents to strive, and to addressing coordination and externalities that prevent the private sector from flourishing and generating jobs. All governments in the world are constantly engaged in various forms of industrial policies—they take actions that favour certain industries more than others and therefore shape sector allocations in the economy. In all countries, some industries, sectors, and even firms are favoured within the legal framework and often heavily subsidized, often in opaque ways. Bankruptcy laws that put derivatives first in line in the event of bankruptcy effectively give preference to the financial sector. Most countries’ tax codes are riddled with tax expenditures that provide hidden subsidies to some industries. But even in the absence of such ‘special’ provisions, the design of depreciation allowances will affect industries with different capital lifespans differently. Budget policies also inevitably have impacts on industrial structure: where governments locate roads and ports affects different industries and firms differently. In short, one cannot escape thinking about the different impacts of different policies on different sectors. True, the record of most activist governments is mixed. Critics of the industrial policies in many African countries just after independence argue that they introduced profound distortions: they used limited public resources to pursue unsustainable importsubstitution policies. To reduce the burden of public subsidies, governments sometimes resorted to administrative measures—granting the non-viable enterprises in priority industries a market monopoly, suppressing interest rates, overvaluing domestic currency, and controlling prices for raw materials. Such interventions introduced further distortions, sometimes even causing shortages in foreign exchange and raw materials. Preferential access to credit deprived others of resources. There was a high opportunity cost. While industrial policies were often blamed for these disappointing outcomes, failures in macroeconomic policies and governance often played a role—and often were the real source of the problem.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

The standard argument was that markets were efficient, so there was no need for government to intervene either in the sector allocation of resources or in the choices of technique. And even if markets were not efficient, governments were not likely to improve matters. But the 2008–09 global crisis has shown that markets are not necessarily efficient. Indeed, there was a broad consensus that without strong government intervention—which included lifelines to certain firms and certain industries— the market economies of the USA and Europe may have collapsed. It is a mistake to trust markets blindly. Some of the most important national and global policy objectives (such as equality of opportunity for all citizens, pollution control, and climate change) are simply not reflected in market prices. Even economists who oppose sectoral industrial policy (the ‘vertical’ policies to support specific industries) acknowledge the need for broad, neutral, ‘horizontal’ industrial policy (one that does not target specific industries). Yet the lines between the two can be blurry. Everything governments do or choose not to do benefits or can be captured by vested interests. An exchange rate policy could be presented as ‘neutral’ and ‘broad-based’. Yet, we know that some sectors, industries, social groups, and even regions are always favoured or penalized by any stance on the exchange rate. Even when there is no change, some benefit while others lose. Likewise, infrastructure development is often presented as a suitable tool of economic policy because of its perceived ‘neutrality’. Yet there is nothing neutral about the choice of infrastructure that a country needs at any given time, or where and when it should be built. These decisions always involve some political judgement about priorities, and therefore represent industrial policies. The same is true for education, which often is mistakenly presented as ‘neutral’. Therefore, the question is not whether any government should engage in industrial policy—it is how to do it right. All governments in the world, regardless of their politics, engage in industrial policies every single day. In fact, the entire budget exercise—which consists of submitting to parliament a law that grants different tax rates and different programmes, projects, and expenditure levels to different industries, sectors, and regions—is itself industrial policy. In sum, the intellectual challenge for development economists is not to engage in semantic debates about what industrial policy is. It is to come up with better guiding principles on how ‘best’ any society should move its human, capital, and financial resources out of subsistence sectors. For the process to be efficient, coordination issues and externality issues must be addressed. Markets typically do not manage such structural transformations on their own well. And governments must play no more but no less of their facilitating role in the process. A general rule may be to encourage only industries in which the economy has a clear comparative advantage—and the private sector usually identifies these industries and sector easily. When that is done, the government can come in and help foster learning among firms, within firms, and with the economy. Solow’s work helped us understand how most increases in standards of living are related to the acquisition of knowledge, to ‘learning’. Most increases in per capita income arise from advances in

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



technology—about 70 per cent of growth comes from sources other than factor accumulation. In developing countries, a substantial part of growth arises from closing the technology (or knowledge) gap with those at the frontier. And within any country, there is enormous scope for productivity improvement simply by closing the gap between best practices and average practices. If improvements in standards of living come mainly from diffusing knowledge, learning strategies must be at the heart of the development strategies.

29.4 C T: P L A

.................................................................................................................................. Surveys of students in economics departments around the world still indicate that the subdiscipline of development is not a preferred field for young researchers. No one can blame them: the quest for prosperity has so far been a difficult and frustrating academic endeavour. Yet, it is in my view the most exciting area of research—as noted by Robert Lucas in his famous 1988 paper on the mechanics of economic development. Development economics is also a high-risk, high-reward domain. Joseph Stiglitz said in his Nobel Prize lecture that his thinking about the economics of information started when he was working in the 1960s in Nairobi, Kenya. Several other future Nobel Prize laureates were there at the same time: James Tobin and Peter Diamond. Many other very successful and influential economists—such as Gary Fields and John Harris—also started their careers in Nairobi. Africa has attracted in recent years many of the best minds in the business—from Roger Myerson (another Nobel laureate) to Timothy Besley, Paul Collier, Lord Nicholas Stern, Dani Rodrik, Kaushik Basu, Justin Yifu Lin, and many others. Perhaps Robert Solow was too pessimistic: there is a promised land for economists working on Africa and other developing areas. I can even foresee a few Nobel Prize winners in this field in the next decade. We have recently completed the two-volume, 90-chapter Oxford Handbook of Africa and Economics, which features their work and highlights some the insights that the study of Africa brings to economic science. The World Bank’s research department has released a World Development Report on ‘Minds and Culture’, which takes behavioural economics to new heights and shows how some of the findings in the African context can enlighten our understanding of economic development. My exhortation to all economists interested in development is never to abandon the arduous work it entails. In fact, for all economists in the world, Africa is the intellectual deal of the century (to paraphrase a recent statement by African Development Bank President Akinwumi Adesina). There are many jackpots to be won there if one remembers the words of civil rights leader John Lewis: ‘Don’t give up, don’t give in, don’t give out!’ Yes, there is a promised land ahead.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



ˊ  

R Lin, J. Y., 2009. Economic Development and Transition: Thought, Strategy and Viability (Marshall Lectures), New York: Cambridge University Press. Lin, J. Y., 2012a. New Structural Economics: A Framework for Rethinking Development and Policy, Washington, DC: World Bank. Lin, J. Y., 2012b. The Quest for Prosperity: How Developing Economies Can Take Off, Princeton, NJ: Princeton University Press. Lin, J. Y., 2012c. ‘From Flying Geese to Leading Dragons: New Opportunities and Strategies for Structural Transformation in Developing Countries’, Global Policy, 3 (4), pp. 397–409. Lin, J. Y. and C. Monga, 2010. ‘Growth Identification and Facilitation: The Role of the State in the Dynamics of Structural Change’. Policy Research Working Paper No. 5313, Washington DC: World Bank. Lin, J. Y. and C. Monga, 2011. ‘Growth Identification and Facilitation: The Role of the State in the Dynamics of Structural Change’, Development Policy Review, 29 (3), pp. 264–90. Lin, J. Y. and C. Monga, 2012. ‘Solving the Mystery of African Governance’, New Political Economy, 17(5), pp. 659–66. Lin, J. Y. and C. Monga, 2013. ‘Comparative Advantage: The Silver Bullet of Industrial Policy’, in Joseph E. Stiglitz and Justin Yifu Lin, eds, The Industrial Policy Revolution I: The Role of Government Beyond Ideology, New York: Palgrave Macmillan, pp. 19–39. Lin, J. Y. and C. Monga, 2014. ‘The Evolving Paradigms of Structural Change’, in Bruce Currie-Adler, Ravi Kanbur, David Malone, and Rohinton Medhora, eds, International Development: Ideas, Experience, and Prospects, New York: Oxford University Press, pp. 277–94. Lin, J. Y. and C. Monga, 2017. Beating the Odds: Jump-Starting Developing Countries, Princeton, NJ: Princeton University Press. Lin, J. Y. and D. Rosenblatt, 2012. ‘Shifting Patterns of Economic Growth and Rethinking Development’, Journal of Economic Policy Reform, 13 (3), pp. 1–24. Lin, J. Y. and G. Tan, 1999. ‘Policy Burdens, Accountability, and Soft Budget Constraints’, American Economic Review, 89 (2), pp. 426–31. Lin, J. Y., X. Sun, and Y. Jiang, 2013. ‘Endowment, Industrial Structure and Appropriate Financial Structure: A New Structural Economics Perspective’, Journal of Economic Policy Reform, 16 (2), pp. 1–14. Monga, C., 1997a. ‘A Currency Reform Index for Western and Central Africa’, The World Economy, 20 (1), pp. 103–25. Monga, C., 1997b. L’argent des autres: Banques et petites entreprises en Afrique—le cas du Cameroun, Paris: LGDJ. Monga, C., 2006. ‘Commodities, Mercedes-Benz, and Adjustment: An Episode in West African History’, in E. K. Akyeampong, ed., Themes in West Africa’s History, Oxford: James Currey, pp. 227–64. Monga, C., 2011. ‘Post-Macroeconomics: Lessons from the Crisis and Strategic Directions Ahead’, Journal of International Commerce, Economics and Policy, 2 (2), pp. 1–28. Monga, C., 2012a. ‘The Hegelian Dialectics of Global Imbalances’, The Journal of Philosophical Economics, 6 (1) (Autumn), pp. 2–51. Monga, C., 2012b. ‘Shifting Gears: Igniting Structural Transformation in Africa’, Journal of African Economies, 21 (Supplement 2), pp. ii19–ii54.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



Monga, C., 2013. ‘Winning the Jackpot: Jobs Dividends in a Multipolar World’, in Joseph E. Stiglitz, Justin Yifu Lin, and Ebrahim Patel, eds, The Industrial Policy Revolution II— Africa in the 21st Century, New York: Palgrave Macmillan, pp. 135–71. Monga, C., 2014. ‘The False Economics of Preconditions Policymaking in the African Context’, Keynote Address at the Annual Conference of the African Finance and Economic Association (AFEA), Journal of African Development, 16, (2), pp. 121–40. Monga, C., 2015a. ‘Principles of Economics: African Counter-narratives’, in C. Monga and J. Y. Lin, eds, The Oxford Handbook of Africa and Economics, vol. 1: Context and Concepts, New York: Oxford University Press. Monga, C., 2015b. ‘Measuring Democracy: An Economic Approach’, in C. Monga and J. Y. Lin, eds, The Oxford Handbook of Africa and Economics, vol. 1: Context and Concepts, New York: Oxford University Press. Monga, C., 2017. ‘The Macroeconomics of Marginal Gains: Africa’s Lessons to Social Theorists’, in W. Adebanwi, eds, The Political Economy of Everyday Life in Africa, Melton: James Currey, pp. 115–131. Stiglitz, J., J. Y. Lin, and C. Monga, 2013a. ‘Rejuvenation of Industrial Policy’, in J. Stiglitz and J. Y. Lin, eds, The Industrial Policy Revolution I: The Role of Government Beyond Ideology, New York: Palgrave Macmillan, pp. 1–18. Stiglitz, J., J. Y. Lin, C. Monga, and E. Patel, 2013b. ‘Industrial Policy in the African Context’, in J. Stiglitz, J. Y. Lin, and E. Patel, eds, The Industrial Policy Revolution II: Africa in the 21st Century, New York: Palgrave Macmillan, pp. 1–24.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

              Lessons for Global Development ......................................................................................................................

 . 

The institutions received from England were admirably calculated to lay the foundation for temperate and rational republics. The materials in possession of the people, as well as their habits of thinking, were adapted only to governments in all respects representative; and such governments were universally adopted. The provincial assemblies, under the influence of Congress, took up the question of independence; and many declared themselves in favor of an immediate and total separation from Great Britain. —John Marshall (1844)¹

30.1 A N E  T P A

.................................................................................................................................. A cluster of small English settlements on the eastern coast of North America grew over three centuries to become the richest and most powerful nation on Earth. This extraordinary development ultimately depended on the deep strengths of the political system that was introduced early in the history of these colonies. Several fundamental

¹ Marshall (1844: ch. 4).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



principles remained remarkably constant in American history thereafter, even as the nation’s territory expanded vastly, its wealth spectacularly multiplied, and its population was augmented by immigrants from every part of the world. But reformers who sought to apply American political principles in other countries often found that something essential could be lost in their translation abroad. Such reformers generally focused on the ideals of representative democracy and human rights, but often neglected the decentralized federal nature of American democracy. We should recognize, first and foremost, that the United States of America was established as an independent nation by a congress of delegates from thirteen provincial assemblies, each of which consisted of representatives elected by their communities. The strength of the American republic is deeply rooted in its unique political origin, created not by an army or a tribe, but by the locally elected members of thirteen separate assemblies. This point is clearly expressed in America’s 1776 Declaration of Independence, once we read beyond the long introductory sentence about human rights. The broad statement of universal human rights has been inspirational, but it is so lacking in substantive details as to be compatible with the ownership of slaves by many signers of the Declaration. A different focus emerges after the text of the Declaration asserts that ‘governments long established should not be changed for light and transient causes’. The first interpretation of these words may be that the political connections between America and Britain should not be broken without good cause. But a second and more forceful interpretation of this point emerges as the main focus of the Declaration, as it accuses the King of Great Britain of acting in many ways to subvert the traditional rights of the elected legislative assemblies in the thirteen colonies. In fact, the largest part of the Declaration of Independence is a list of complaints of legislators. The charge that the king has fatigued legislators by making them meet in unusual and uncomfortable places gets more discussion than some burning of towns. The king has repeatedly dissolved the provincial assemblies, has prevented them from passing necessary laws, has undermined their ability to supervise local courts, and has imposed new taxes without their consent. When this usurpation of their traditional rights was resisted by the colonists, the king unleashed military forces against them. The Declaration of Independence expresses a clear view that the form of government long established in colonial America was one where British-appointed officials could act only with the cooperation and approval of the colonists’ elected representatives. When this cooperation broke down after the Stamp Act of 1765, the elected assemblies felt compelled to exercise power on their own. The American Revolution was fought to enforce this claim of sovereign power for the thirteen provincial assemblies, which then reconstituted themselves as state governments and sent delegates to form a Congress to coordinate their revolutionary efforts.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

30.2 C O

.................................................................................................................................. We may ask why Britain permitted the development of such institutions of representative government in its North American colonies. Of course, colonists from England were accustomed to a government there that included locally elected representatives in Parliament. From 1620 in Virginia, institutions of local self-government were introduced to induce English settlers to come to America and offer loyal service in local militias, which were essential to defend the colonies’ long frontier. When representative assemblies were lacking or became ineffective, many settlers would lose confidence in the willingness of the government to protect their land, and then the militias could rebel, as they did in Virginia (Bacon’s Rebellion) in 1676 and in Massachusetts in 1689. Bacon’s rebellion is of particular interest because it offers a perspective on how different political decisions could have put America on a path to become a poor, lessdeveloped country. In 1676, William Berkeley had been governor of Virginia for most of the time since 1641. He stopped calling popular elections for representatives to Virginia’s House of Burgesses after 1661, and his appointed councillors and sheriffs ruled Virginia as an autonomous oligarchy. The militias rebelled under Nathaniel Bacon, but were ultimately suppressed by naval forces from England. Then, as Governor Berkeley initiated punitive expropriations against anyone suspected of supporting the rebellion, a mass of settlers rushed to take their moveable property out of Virginia.² The prospect of Virginia being impoverished by such disinvestment and capital flight was recognized by royal commissioners from England, and they acted to reconstitute the government of the colony. Governor Berkeley was dismissed, and his autonomous oligarchic regime was replaced by a new system in which power was divided between the locally elected assembly and a governor appointed from outside the colony. Thus, the British imperial government effectively supported the rights of the elected assembly, after Bacon’s rebellion showed the dangers of concentrating power under a strong local governor. However, political gains for enfranchised citizens also separated them from the enslaved. Poor whites and blacks had fought side by side during Bacon’s rebellion, but thereafter Virginia’s assembly passed racist laws denying basic legal rights to negroes and Indians. The 1689 uprising of the militias in Massachusetts occurred after England’s Glorious Revolution, but also after the royal governor had suspended the representative assembly. The settlers’ worst fears were realized in one district where appointed commissioners began requiring bribes to re-confirm settlers’ land claims.³ A multitude of small independent farmers could not develop individual plots of land securely without any political voice or representation. Thus, Thomas Jefferson’s ideal of a society populated by small independent farmers implicitly depended on representative assemblies.

² Webb (1984: 140).

³ Lustig (2002: ch. 7).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



The American colonies were essential partners in the triumph of British forces in North America during the Seven Years’ War (1754–63). The strategic turning point in the war occurred in 1757, when William Pitt decisively realized that the Americans would provide greater resources for the conquest of French Canada if the British government treated its colonial assemblies more like its European allies, offering subsidies to encourage their contributions to the war effort, instead of treating them like subordinates whose resources could be simply commandeered.⁴ When the colonial assemblies were confident of their autonomous rights within the British Empire, they were willing to contribute generously to an imperial effort in which they shared a common interest. This confidence disappeared rapidly, however, after the successful conclusion of the Seven Years’ War. The British Parliament, facing heavy war debts, asserted its right to impose taxes in America, but Americans resisted any taxation that was not approved by their own representatives. As the conflict intensified, some royal governors acted to dissolve the representative assemblies, whose members reconvened in revolutionary conventions. Then people in every colony felt an urgent need to defend their local political privileges against such attacks, and they elected representatives to a new Congress to devise a common strategy. Thus thirteen autonomous colonies came together in 1776 to declare their independence, which John Adams considered to be the great unprecedented achievement of the American Revolution.⁵

30.3 B  D D

.................................................................................................................................. Thus, as John Marshall noted (in the opening quote), English colonial rule laid the foundations for representative government in America. Having claimed power as the elected representatives of the people in the Revolution, the leaders of the new republic could not subsequently assert any other basis for holding power without elections. From 1776 to 1788, the thirteen states were joined together under Articles of Confederation which decentralized almost all power to the separate state governments. A need for stronger national coordination became increasingly evident, however, both during and after the Revolutionary War. Thus in 1788 the states adopted a new Constitution which allocated substantial powers to the national government under the Congress and an elected President. But the federal Constitution still left the separate states with primary responsibility for local government. The leaders of the new national government never had any option to integrate all sovereign power under their centralized control, as the thirteen separate state governments had long traditions of exercising power themselves for over ⁴ Anderson (2000: ch. 21).

⁵ Jensen (1968: 33).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

a century before the first election of any national officials. America had to be a federal democracy. Indeed, a balanced sharing of power among different levels of government has been the general rule throughout America’s history. Before 1776, the elected provincial assemblies shared power with imperial agents from Britain. After 1788, the elected state officials shared power with the national government. Much in America changed over subsequent generations, as new states were added, slavery was ended, the franchise was extended, and the federal budget grew proportionally larger. But America’s growth and development has always been guided by the basic principles of representative democracy and a federal division of power between national and subnational governments. These principles were vital both for establishing the new nation and making it durably democratic. The decentralization of power admittedly created difficulties for financing the war effort during the American Revolution, but decentralization also gave the revolutionary movement a broadly distributed political strength that was essential to its ultimate success. In 1776, every community had at least one widely respected leader—its local assembly representative—who had a substantial vested interest in defending the new regime. The British forces could not hope to mobilize such broad political support without essentially recreating the colonial assemblies. The federal division of power between national and local levels of government helped to sustain the constitutional limits on national officials, as James Madison predicted in Federalist 51.⁶ Most notably, when the national government acted to suppress political opposition via the Sedition Act of 1798, it was strongly opposed by state governments, including those of Kentucky and Virginia. Decentralized democracy made the new nation rich in local leaders who had held elected office. We should understand that successful democracy requires more than just elections. For democratic competition to effectively benefit the public, voters must have a choice among candidates with proven records of public service who have developed good reputations for exercising power responsibly in elected office. When such trusted leadership is lacking, democracy is inevitably fragile. This essential supply of trusted democratic leadership can develop best in responsible institutions of local government, where successful local leaders can prove their qualifications to become strong competitive candidates for higher office. From this perspective, even a small nation can benefit from having some decentralization of power to autonomous subnational governments, where future candidates for national leadership can demonstrate their ability to serve the people. The successful establishment of strong competitive democracy at the national level in America after 1788 depended on the large supply of potential candidates with proven records of public service in the thirteen former colonies. For example, Thomas Jefferson had served as a local representative in the House of Burgesses and as

⁶ Hamilton et al. (2003).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



Governor of Virginia long before he ran against John Adams for President of the United States. Throughout American history, competitive politicians have climbed such a ladder of democratic advancement from local to national office.⁷ Thus, local democracy in America has helped to strengthen democratic competition at the national level. Democracy at the national level has also helped to strengthen local democratic competition, as America developed a strong system of national parties that endorsed candidates for local offices. Indeed, James Madison (imagining a Rhode Island separated from the United States) also argued in Federalist 51 that politics in a small unitary state can be unstable and vulnerable to local autocracy. But with federal democracy, a local political boss who has lost popular approval cannot suppress competitive democratic challengers who are supported by a rival national party.

30.4 A D A  D

.................................................................................................................................. Americans have always believed that their political system should be an example to the world, and admirers abroad have sought to strengthen their own nations by applying principles of American democracy. Leaders of the French Revolution looked to the American Revolution as their model, turning to Thomas Jefferson for advice when he was serving as America’s minister to France. In August 1789, a group from the French National Assembly met at Jefferson’s Paris home to draft a new revolutionary constitution for France. After this meeting, Jefferson wrote to John Jay and James Madison that leaders in France were planning a constitutional monarchy with an elected unicameral legislature, but that there would also be provincial assemblies which would send delegates to an advisory Senate.⁸ Many points in Jefferson’s letters at this time proved to be accurate predictions of the Constitution that France actually adopted in 1791, but provincial governments were not included. Under France’s 1791 Constitution, the provinces were replaced by smaller departments, and local officials were not allowed to tax or borrow without the National Assembly’s approval.⁹ The National Assembly’s power was constitutionally checked only by the king, and the system broke down when the king was accused of treason. ⁷ Before 2016, Americans never elected a President without some prior record of serving responsibly in a public office at the local or national level. ⁸ See Jefferson’s letters to Jay and Madison on 27 and 28 August 1789 in Papers of Thomas Jefferson, Volume 15, ed. Boyd (1958), pp. 356–69. See also Lafayette’s prior letter to Jefferson on 25 August 1789 (p. 354), and Jefferson’s subsequent letters to Paine and Price (pp. 424–5). ⁹ Cobban (1943: 13–31). The formation of local revolutionary committees in towns throughout France did force the National Assembly to recognize the election of municipal officials at the lowest (commune) level of local government.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

Thus, in contrast to America, the establishment of democratic government in France was accompanied by a strong centralization of power. Regional parliaments, which had provided significant political decentralization in France’s Old Regime, were viewed as vestiges of hereditary privilege and were suppressed by the Revolution. The underfunded and chaotic system of revolutionary local governments was ultimately brought under central direction by appointed national agents. As Alexis de Tocqueville noted after visiting America in the 1830s, the French Revolution was the enemy both of royalty and provincial institutions.¹⁰ It should not be surprising that the members of the French National Assembly would choose to create a new constitutional structure which concentrated power in their own chamber. They could have been prevented from doing so only by other institutions that could effectively mobilize popular support for a share of power, but traditional institutions were generally discredited in revolutionary France. Under such circumstances, it is hard to see who could have compelled the National Assembly to create new provincial assemblies which would then have competed with it for a share of power. The potential benefits of a federal division of power for France might have been evident to a few idealists when they talked with Thomas Jefferson, but nobody in France’s revolutionary leadership had any real interest in advocating a federal decentralization of power. Decades later, when Jefferson wrote in his Autobiography (1821) about the French constitutional discussions at his home, he did not even mention the suggestion of provincial assemblies in revolutionary France.¹¹

30.5 E D  H  F D

.................................................................................................................................. While suggestions of federal decentralization in France came to nothing in the National Assembly, plans to create new institutions of decentralized political power in America’s unsettled western territory won strong support in Congress soon after the Revolution. Before going to France in 1784, Thomas Jefferson helped to formulate Congressional plans to divide the new territory into smaller districts in which settlers would organize local governments that could ultimately join the Union as equal partners with the original states.¹² The considerations that induced Congress to decentralize power in the west certainly depended on the fact that democratic decentralization already existed in the original states. Post-revolutionary America was characterized by a sensitive balance among the thirteen states, each suspicious of the others and of central authority. So the alternative of accepting some states’ claims to vast western territories would be objectionable to the ¹⁰ de Tocqueville (1945: 100. ¹² Berkhofer (1972: 231–62).

¹¹ Jefferson (1854: 104–5).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



other states, and all the states would resist the alternative of permanent centralized Congressional control over the territories. Having fought a war to defend the sovereignty of the representative assemblies in their thirteen states, Americans readily accepted a belief that their form of representative government could not endure in larger states where many voters would live too far from the state capital. Thus, decentralized federal democracy, once established in America, created political forces for sustaining and extending itself. The Northwest Ordinance of 1787, which defined the structure of new territorial governments, presents America’s idealized reconstruction of the colonial institutions from which it developed. The territorial governments included a federally appointed governor and a house of locally elected representatives, as well as a council whose members were federally appointed from a list of candidates designated by the house of local representatives. Acts of law in a territorial district required approval both by the governor and by majorities in the house and council. The basic principle was that a territorial government should do nothing that was not approved by a majority of local representatives, but any action by a territorial government could also be blocked by the governor, who represented the federal government. This broad supervision by a resident federal agent with veto power would be withdrawn when the territorial district was admitted as a state in the Union, although, of course, state government officials remained liable for compliance with federal laws under the Constitution. But even during the period of territorial status, the fact that local elites would be sending representatives to Congress after statehood implied that no national party wanted a reputation for sending oppressive territorial governors who would alienate local leaders of the new states.

30.6 E I  D D

.................................................................................................................................. When de Tocqueville wrote Democracy in America in 1835, the number of American states had grown from 13 to 24, and the American population had more than tripled in the fifty years since the Revolution. The force of this expansion depended on a steady flow of immigrants from many countries, not just England. Regardless of the different political traditions in these countries, immigrants were readily assimilated into America’s federal democracy as they acquired the right to vote. In a large centralized republic, a small immigrant group might find itself lacking any effective political influence; but in a decentralized federal democracy, a cluster of new immigrant voters could realistically expect that at least some in their group would be elected to local offices.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

Thus, the American political system became an engine that could attract new citizens from around the world, and, with them, could populate new communities and states across the entire continent. Compared to other countries with more centralized or less democratic political systems, common citizens in America could generally feel more confident about basic legal protection for their personal investments in the new country. Responsible officials of state and municipal governments, being accountable to their communities, had both the power and the motivation to undertake the local public investments that are essential for developing a prosperous community. And so the United States inexorably expanded across the continent on which it was founded, and it grew to become the richest nation in the world.

30.7 T I  D  N E

.................................................................................................................................. Although decentralized federal democracy has proved politically stable and economically beneficial in America, many nations since the French Revolution have instead been drawn to centralized democracy. One basic reason for a bias against federal decentralization is that members of the national political elite naturally acquire an interest in a centralization of power. To anyone who frequents the central offices of government, a redistribution of powers to unknown provincial elites may seem risky and inconvenient. Even if a redistribution of power to autonomous local governments would be beneficial for most of the population, the individuals who would expect to lose from such decentralization may include some of the most powerful people in the country. In a nation where local governments are run by centrally appointed officials, a national leader can use powerful local offices as highly valued patronage rewards for key supporters. Then a reform permitting provincial voters to elect their own local officials would be a costly disappointment for some of the leader’s most important supporters. So an incumbent national leader is unlikely to advocate political decentralization in a country where it has not previously existed. Thus, when a transition to democracy begins with elections only at the national level, the winners can acquire a compelling interest in centralizing power around themselves. As the only faction to have been entrusted with power by a vote of the people, the new national leaders will have a strong mandate to guide any further process of constitutional reform, and they are unlikely to promote reforms that would transfer some of their power to others. So if democracy is introduced in the national government before it is introduced in local governments, political decentralization will become much more difficult thereafter.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



In Egypt, for example, popular demands for democratically accountable government in 2011 were followed by elections to form a new democratic national government. In these elections, only one party could win the prize of centralized national power. The elected government then wrote a constitution which allowed the new national leaders to retain control over all local administration, offering only a vague promise to introduce elected local governments sometime in the future. Such centralization might have seemed convenient for the short-term interests of national leaders, but it left Egypt’s new democracy perilously vulnerable to fears of another autocracy. Empowerment of trusted local leaders in local governments throughout the country could have done much to reduce such fears. The downfall of Egypt’s elected president in 2013 led to questions about what went wrong in the process of building a new democracy. Many have asked whether Egypt might have moved too quickly into a presidential election, but few have asked whether the move to introduce democratic local government was too slow. Diplomats and leaders of major international organizations are accustomed to working with national governments, and so it may not be surprising that they sometimes overlook the potential role of subnational governments in democratic development. Even American diplomats have sometimes focused more on the advantages of having a strong partner in a centralized government than on the developmental benefits of political decentralization. In 2002, for example, America supported the creation of a new centralized presidential government in Afghanistan, a country which had a long tradition of decentralizing substantial power to traditional tribal leaders. In subsequent years, Americans paid a heavy price to support the regime. With power concentrated in the capital, there were many rural districts where nobody felt any personal political stake in the government, and so its authority could be maintained only with support from foreign forces. In an account of the struggle for one district in Afghanistan, Carter Malkasian described a successful counterinsurgency strategy in which the essential key was to offer some real authority to selected local leaders.¹³ But with no constitutionally protected autonomy for local governments, such locally negotiated political settlements could be nullified by manipulation in the capital, and hard-won gains were lost.

30.8 F   P

.................................................................................................................................. We have seen that federal decentralization has benefited Americans in many ways. Democratic competition in America’s federal system has been strengthened by the ability of successful local leaders to become competitive candidates for higher offices, and by the ability of national parties to sponsor competitive candidates for local offices. ¹³ Malkasian (2013: 178).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 . 

Responsible local governments have had the power and incentive to make the local public investments that are essential for developing a prosperous community, and the security of democratic rights has encouraged common citizens to make their own investments in this prosperity. For people who are unfamiliar with federalism or democracy, however, the benefits of federalism may be harder to appreciate than those of democracy. People everywhere can readily appreciate the democratic ideal that they should have a choice about who will lead them, and that authority in government should depend on broad popular approval. But federal democracy asks more from voters. When power is divided among local and federal officials, there will inevitably be disagreements about the line dividing local from federal authority. In a democracy, the resolution of such constitutional disputes must ultimately depend on judgements by the voters. Voters must understand that their elected leaders at the national level need supreme power to serve and protect the broad interests of the entire country, but that their elected leaders at the local level also need some autonomous power to provide public services for their communities. An official at either level who acts to undermine the other level’s legitimate constitutional powers should be distrusted and rejected by voters in the future. It may take some years of experience with federal democracy for voters to develop a broad understanding about what should be an appropriate balance between the different levels of government. America was indeed greatly blessed in that its local governments were elected long before its first national government. Even before the American republic was established, its citizens had decades of experience of powersharing between their locally elected provincial assemblies and the greater government of the British Empire. The key lesson from American history is that those who would promote vigorous democratic development should appreciate the vital benefits of a balanced federal system—one in which the people can elect responsible local governments, as well as their sovereign national government.

R Anderson, Fred, 2000. Crucible of War: The Seven Years’ War and the Fate of Empire in British North America 1754–1766, New York: Alfred A. Knopf. Berkhofer, Robert F., Jr., 1972. ‘Jefferson, the Ordinance of 1784, and the Origins of the American Territorial System’, William and Mary Quarterly, 29 (2), pp. 231–62. Cobban, Alfred, 1943. ‘Local Government during the French Revolution’, English Historical Review, 58 (229), pp. 13–31. de Tocqueville, Alexis, 1945. Democracy in America, Volume 1, New York: Knopf. Hamilton, Alexander, James Madison, and John Jay, 2003. The Federalist, Cambridge: Cambridge University Press. Jefferson, Thomas, 1854. Writings of Thomas Jefferson, Volume 1, Washington, DC: Taylor and Maury.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

     



Jefferson, Thomas, 1958. Papers of Thomas Jefferson, Volume 15, edited by Julian P. Boyd, Princeton, NJ: Princeton University Press. Jensen, Merrill, 1968. The Founding of a Nation: a History of the American Revolution 1763–1776, Indianapolis: Hackett. Lustig, Mary Lou, 2002. The Imperial Executive in America: Sir Edmund Andros 1637–1714, Madison, WI: Fairleigh Dickinson University Press. Malkasian, Carter, 2013. War Comes to Garmser: Thirty Years of Conflict on the Afghan Frontier, New York: Oxford University Press. Marshall, John, 1844. The Life of George Washington: Written for the Use of Schools, Philadelphia: James Crissey. Webb, Stephen Saunders, 1984. 1676: The End of American Independence, New York: Knopf.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

  ......................................................................................................................

     ......................................................................................................................

 

T idea of a ‘structural transformation’ of present-day economies raises questions: What is harmful in the existing structures? What goals do we want any new structures to serve? And what structures would serve the chosen goals?

31.1 H   P S

.................................................................................................................................. For several decades now, the advanced economies of the world have been described as having—by historical standards, at any rate—weak investment, depressed labour-force participation and slow growth of productivity.¹ This syndrome fits to a tee what Alvin Hansen in 1938 dubbed ‘secular stagnation’. Unemployment also worsened appreciably in Germany, Italy, and France from 1965 to 1995, although it has since largely recovered in Germany. But much of these rises reflect the upward trend of income and wealth: In both the USA and the UK, where income growth was modest, there has been little upward trend in the unemployment of men and none at all of women, aged 20 and over, since 1965. To that litany of complaints one must add the discontents resulting in the advanced economies when competition from overseas—notably Asia—drove down prices in their export and import-competing industries. In the USA, some three-quarters of jobs in manufacturing were lost since the 1970s. Although most of the displaced workers ¹ Giles (2016: 2). The writer quotes Eswar Prasad of the Brookings Institution.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



found jobs elsewhere and many are now of retirement age, those jobs pay markedly less. That might not be regarded as very important in a world where all work is far better rewarded than it was 200 or even 100 years ago—although not 50 years ago. But among men, at any rate, the lower earning power has resulted in a serious loss of self-respect—in John Rawls’s terms: They feel they can no longer fulfil their duty as husbands and fathers.² They may well feel that an injustice has been done—or increased injustice. The jobs that displaced workers have been forced to take also offer generally less meaningful work—less job satisfaction.³ In recent years, I have been exploring elements, or aspects, of this satisfaction.⁴ For one thing, these workers keenly feel they are not prospering. By prospering I mean coming to know the satisfactions of ‘succeeding’—called ‘getting on’ in the nineteenth century and ‘getting ahead’ in the twentieth; but more of this later. They, like others in the labour force, want jobs with some agency. And even in manufacturing, a worker may be allowed enough autonomy and initiative to have a sense of agency. Also, these workers are not flourishing—the philosophers’ term for a lifelong journey of inquiry, exploration, and, in the modern age, creation of the new and being transformed in the process. I have argued that, from the 1820s well into the 1960s, many ordinary people in Britain and America, later Germany and France, were fortunate enough to flourish in their careers—over much of them, at any rate. Another element of job satisfaction that these displaced workers may have lost is their sense of importance to society and particularly to the national economy, which may be regarded as society’s central project. Such conditions would be less difficult to bear if these workers had hopes that the future would offer substantially improved opportunities. But they have no such hopes—nor are there grounds for expecting an improvement. The sociologists’ notion of upward mobility is a related idea. They suppose that a worker making a normal effort can move up the ladder with some probability. And this prospect of more satisfying work in the future can be a powerful source of satisfaction in the present— powerful enough to offset to some degree the hardships of current conditions. (A colleague of mine in the 1970s remarked that he and his wife were happier looking forward to an expected prize than they would have been if they had unexpectedly received it.) However, sociological studies do not find a larger proportion of the working class managing to climb to higher positions in the distribution of wages. Are these aspects of job satisfaction deeply felt? The rise of suicide and drug-related deaths noticed by Anne Case and Angus Deaton might be in part a result of the frustration and alienation felt by many in the working class.⁵ One cause of these developments could be a loss of the ‘agency’ workers need in order to express their opinion and exercise their judgement and imagination.⁶ Another cause could be the anger many in the working class feel at the ‘protection’ accorded to companies, banks, ² See Cherlin (2014). ³ On this phenomenon, see Weil (2014). ⁴ See, for example, Phelps, Mass Flourishing: How Grassroots Innovation Created Jobs, Challenge and Change, Princeton University Press, 2013. ⁵ See Case and Deaton (2015). ⁶ See Sennett (2006).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

and other special interests while no such protection is accorded them—only low tax rates, which do nothing to raise job satisfaction and thus do not lift the human spirit. Are some of these harms caused by forces exogenous to these economies rather than by the structures of the advanced economies? It might be that a radical change in prices brought about by the new supply of goods coming from overseas—Latin America, Eastern Europe, and Asia—is the cause of the frustration and alienation amid the working class, not a change in economic policies or institutions. It might be that a loss in the dynamism that sparked the age of innovation is the cause of the slowdown of productivity and wage rates. However, the working class in the advanced economies would not have suffered from the change in relative prices wrought by international trade were it not for the prevalence of values approving and even seeking trade with the rest of the world—notwithstanding the likely effects on the distribution of income— and such values are part of the structure of a society and its economy. Likewise, the economy might not have suffered a loss of dynamism were it not for a revival of traditional values or a weakening of modern values—values that, to repeat, are part of an economy’s structure. In the subsequent pages, I will describe how values are a part of the ‘structure’.

31.2 G W W  S  S

.................................................................................................................................. In the nations with relatively advanced economies, there are movements calling for the economy to be directed toward new goals. In the economic policy area, advisors and critics have proposed the need for shifts in resources that are not being achieved by free markets under capitalism: shifts from investment to consumption, from heavy industry to ‘services’. Many of America’s millenials—people entering the labour force in 2000 or later—resonate with these shifts. They are rebelling against economic growth: they do not seek ever-rising wealth nor a desire to entrepreneur a business venture or to innovate a new product. In the thinking of these proponents, it would be natural to implement these shifts by extending the reach of the public sector, thus narrowing the remit of the private sector—thus a shift from private to public. This thinking draws on the feeling that the capitalist system will have a diminished role to play in organizing advanced economies. What needs to be said about these proposals? First, something crucial is missing here. The explorers found in poetry and the heroes of the sagas would have little feeling for these proposals. The transformation they call for is expressed in terms of resource allocation, not human satisfactions and aspirations. There is no concern about the experience of working life—only increased satisfactions from outputs are sought. One would have expected to hear proposals to make the economy into a place for the exercise of imagination and insight, thus

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



leading, possibly, to new methods of producing and new products to produce. One would have thought that the time has come to structure economies so as to lift up the satisfactions of work itself—not just the outputs. Economists from Adam Smith and Karl Marx to Alfred Marshall and Gunnar Myrdal stressed the importance of the experiences in the workplace, positive and negative. But there is no such movement. On the contrary, a belief has grown up—very clearly in Western continental Europe—that optimal allocation of resources and well-functioning institutions are sufficient for a satisfactory economy. Italians, Germans, and French put great value in leisure, particularly their famously long vacation. They work hard when at work although only for a small number of hours per year. That is one reason why hourly productivity and hourly wage rates in Europe have been very high—higher than in the USA and the UK. I do not subscribe to that belief nor do I think the Europeans are wise to have that belief. My belief is that redirecting resources is far from all that nations must do—at least those in the West and the economically advanced nations in Asia. An approximately optimal resource allocation (of which efficiency is a part) may be helpful for a good economy but it is not sufficient for a ‘good economy’. Indeed, single-minded attention to allocation could get in the way: The focus on a shift to neglected consumption possibilities is likely to distract a nation from other policy moves that are essential for a good economy. Which belief is supported by the evidence? It must be recognized that the continental Europeans do not seem very gratified with the organization of their economy— and perhaps of their society. There is circumstantial evidence of dissatisfaction in the fact that continental Europeans have relatively low participation in the labour force—markedly lower than the Americans until recent years. There is more direct evidence in the data on job satisfaction. Among the large Western nations, the lowest job satisfaction levels reported in household surveys are those in continental Europe.⁷ It might be said, as many have, that job satisfaction is only one kind of satisfaction. Fortunately, the World Values Survey—the invaluable project founded at the University of Michigan has been sporadically collecting data on life satisfaction in this or that county starting in 1981. These data clearly show a downhill slide of life satisfaction from the start—putting aside the giddy years of the internet boom. This can reasonably be attributed to a loss in the rate of innovation that took hold in the late 1960s, as argued in my book Mass Flourishing (2013). But even in recent times—from 2001 to 2005 and from 2011 to 2015—America and Germany, with their relatively innovative economies—report much higher levels of life satisfaction than the levels reported in Spain and the Netherlands as well as Japan. Life satisfaction in Italy and France is estimated only once: in the years of the financial crisis. That these satisfaction levels were no higher than in the other countries—they were lower than almost all the

⁷ See the data in Phelps 2013: 200, table 8.1, ‘Mean Nonmaterial Reward in the G10+2’.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

others—suggests that Italy and France’s relatively low satisfaction levels result from their relatively low levels of indigenous innovation.⁸ Such findings of low job and life satisfaction ought not to surprise us. The Continent’s companies are generally not places where new stimuli and new challenges are engaging the mind: Decades ago, the few once-dynamic economies in Europe virtually abandoned innovating. Of course, in the social sciences, there are always exceptions. It should not be astonishing that Switzerland shows a level of life satisfaction consistently far above the levels in America. But it might not have been expected that Switzerland shows a rate of innovation that is comparable to the American rate, although not quite as high. It would be reasonable, then, to infer from the disparity of the performance indicators between the more innovative countries and the less innovative that a nation had better avoid the mistake of adopting an economic model that settles for efficiency, such as the Continental model. But what is the right model? I argue in my book Mass Flourishing that the right model is the good economy, and the good economy is the kind of economy offering the good life. Here we may differ. As you may know, some economists—including my dear friends Joseph Stiglitz, Jean-Paul Fitoussi, and Vladimir Kvint—are proponents of a concept that has come to be known as the quality of life. By a life of high quality they mean mainly ample consumption and ample leisure. In recent years, they have stressed several public goods: clean air, safe food and safe streets; also civic amenities such as municipal parks and sports stadiums. This concept is a fleshed-out version of the European ideal, traceable to ancient Rome. Of course, I don’t dislike these services—and I won’t argue against their provision by the state. But they are not what the philosophers’ concept of the ‘good life’ is about. (Aristotle, founder of the concept, joked that we sometimes need to go to the theatre or the arena to recharge our batteries for the next day’s work.) The philosopher-economist Amartya Sen—another dear friend—points out that all this consuming leaves out something: the need of people to ‘do things’.⁹ That widens the concept of human wants. But it appears not to go far enough. People want more than to be enmeshed in a programme of work in which they have no autonomy. For a good life, people need in their work to have an adequate degree of agency. What exactly do they want to do with it? At the very least, they want lives having some meaning and work that is engaging. As some philosophers have said, people value having some room to express themselves—to have the initiative to voice their thoughts, show their talents, or save this ship. As many observers have said, people value attaining things through their own efforts and insight. (Tony Blair and Mitt Romney say these things.) Recently, I have used the word prospering (from the Latin pro spere, meaning ‘as hoped, or expected’) to refer to ⁸ The fact that the financial crisis was less severe in Italy and France than in the US and UK suggests that some part of the low life satisfaction in Italy and France is their relatively low rates of indigenous innovation. ⁹ See Sen (1985).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



the experience of succeeding in one’s work: a craftsman’s gratification at seeing his hard-earned mastery result in greater demand or better terms for the work he does, a prospector’s delight at finding that the nuggets his expertise has led him to have ‘panned out’, a merchant’s pleasure at seeing ‘his ship come in’, or a scholar’s sense of validation from being awarded an honorary professorship.¹⁰ In the history of the West, a great many people have also valued the experience of exercising one’s imagination and the exploration or experimentation to which the imaginings lead. Many have had entire careers of that sort. Following others, I use the word flourishing to refer to the satisfaction from a journey into the unknown: the excitement of the challenges, the gratification of overcoming obstacles, the fascination of the uncertainties, and the excitement of ‘acting on the world’. I would note that prospering, flourishing and self-expression are generally experiential rewards, not material rewards—though money often comes with those rewards.

31.3 S  O  D G

.................................................................................................................................. How, in general, have economies offering the good life been structured? History books suggest an economy full of entrepreneurial people—people alert to unnoticed opportunities, who search for better ways of doing things, and exercise their initiative to try out new things—and an economy full of innovative people—people imagining new things, developing new concepts into commercial products and methods, and marketing them to potential. California in the late 1990s, New York around the 1940s, Paris in the 1920s, Berlin in the 1890s, London in the mid-1800s, perhaps Shanghai around 1700 and Venice in the 1300s—all had a good economy. In such economies, entrepreneurship and innovating in the business sector were pervasive across most industries and inclusive from the grassroots of society to the most advantaged. Of course, I have not forgotten the parallel concept of a just economy. Economic justice requires maximum economic inclusion of the disadvantaged. But we make a terrible mistake if we focus on the just economy so single-mindedly that we forget the desirability of a good economy. In fact, we cannot determine in any substantive way how an economy can be just until we decide what is good. The above sort of economy is the kind I hope the once-dynamic nations of continental Europe will regain and the nations of Asia will gain over the next decades. Of course, there are desirable things besides prospering and flourishing. In a time of hardship, a nation may not be able to afford a good economy. Of course, people want not only the experience of a good economy—they want air safe enough to breathe and food safe enough to eat. Rightly so. What I would say is that satisfying in full all the ¹⁰ For earlier iterations of this idea, see Phelps (2016 and 2017).

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi



 

myriad demands for public services and other public programmes would require a public sector of a size that would significantly crowd out adaptive and innovative activities in the private sector. Furthermore, the private sector is not inferior to the public sector in providing many services now taken over by the public sector. (Underground railways were once the creation of private entrepreneurs, not city governments.) Even today, the most radical step in urban transportation is Uber. And the most radical change that appears to be coming in the near future is the self-driving car. Both of these are creations of the private sector. It is asked whether in some nations the business people possess the temperament and the sophistication to be innovators. Some observers doubt China can be a prolific producer of indigenous innovations. Others doubt that the land-locked or simply inward-looking nations such as Austria, Hungary, and Slovakia or even Italy, Spain, Portugal, and the Netherlands have the temperament and spirit to be innovative. The most crucial question relates to capitalism. Of course, no economy is entirely capitalist nor could it be. There must be vast economic activity going on in legislating, regulating, public health, education, national defence, administration, law enforcement, and much else. Capitalism can co-exist with extensive social insurance and social assistance—indeed, it is argued that these social inventions serve to encourage people to take a chance on venturesome projects and innovative visions. But genuine capitalism leaves it entirely or at least largely up to owners of wealth (or those who can get their hands on other people’s wealth) to make the basic decisions in the enterprise sector: the entry and the closing of firms, their location, their employment, their investment and their direction—their entrepreneurial ventures and innovative projects. It is crucial that those working in an economy voyaging into the unknown place the control of these decisions in the hands of the owners, since they will have the most to gain or lose; so they will make it their business to make the best decisions they can. Now, in several countries, actions have been taken to reduce the number of firms under profit-seeking private owners. In America, an increasing number of firms are operating under the ownership of entities in the non-profit sector, such as charitable foundations. In Britain, prime minister Theresa May announced a broad agenda of state interventions in the business sector, such as a plan to put ‘worker and consumer representatives’ on corporate boards—a signature measure taken by the Weimar Republic in 1922 that became emblematic of German socialism. The alternative to genuine capitalism that has become by far the most important in most of the world—perhaps all of it—is not socialism: it is corporatism. In a corporatist economy, the government exercises ultimate authority over all enterprises, including those under private ownership. The objective is to keep the direction of the economy and the progress of the various interest groups under central control. When an industry runs into difficulties, it can appeal for help and expect to receive it from the government. A key principle is social protection: the state’s obligation to attend to the eligible interest groups. Another principle is marching together: the state is free to share the gains of winners with others. It is an attack on individualism. As is well-known, corporatism arose in Italy in the 1920s and spread in the 1930s to Germany, Vichy

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

    



France, and various countries in Eastern Europe and Latin America. A more limited and more informal corporatism also reached Britain and America in those years. Today, the influence of corporatist ideas remains strong in Western Europe and the USA. Its effects on innovation are important: it blocks potential newcomers from entering industries or it stymies them by protecting the incumbents. It has created a multitude of protections of companies and workforces through cartels, patents, a thicket of regulations on banks and businesses, teachers, and doctors. In America, the main blow delivered by this corporatism has been an end to innovation in the traditional industries, most of which are located in the heartland of the country. What then happened is that the only course for innovators was to create new industries. But the new industries created by innovators in Silicon Valley, while achieving great results initially, were steadily driving down the prices of what they produced, with the result that wage rates and the national income were slowed almost to a crawl. Silicon Valley is right that innovative input continues unabated. But, owing to falling prices, its contribution is waning. The system of economic dynamism is the only system seen so far that is capable of delivering the good life for the many. But this system has proved to be unable to function when it is imbedded in a society wedded to corporatist principles.

R Case, Anne and Angus Deaton, 2015. ‘Rising Morbidity and Mortality in Midlife Among White Non-Hispanic Americans in the 21st Century’, PNAS, 112 (29). Cherlin, Andrew J., 2014. Labor’s Love Lost: The Rise and Fall of the Working-Class Family in America, New York: Russell Sage Foundation. Giles, Chris, 2016. ‘Global Growth Sliding into the Morass’, Financial Times, 3 October, p. 2. Phelps, Edmund, 2013. Mass Flourishing: How Grassroots Innovation Created Jobs, Challenge and Change, Princeton, NJ: Princeton University Press. Phelps, Edmund, 2016. ‘Europe’s Losses of Innovation: The Individual as well as Societal Harms’, Working Paper No. 89, Center on Capitalism and Society, Columbia University. Phelps, Edmund, 2017. ‘The Dynamism of Nations: Toward a Theory of Indigenous Innovation’, Capitalism and Society, 12 (1), Article 2. Sen, Amartya, 1985. Commodities and Capabilities, Amsterdam: North Holland Publishing Co. Sennett, Richard, 2006. The Culture of New Capitalism, New Haven, CT: Yale University Press. Weil, David, 2014. The Fissured Workplace, Cambridge, MA: Harvard University Press.

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 .......................

Introductory Note: References such as ‘–’ indicate (not necessarily continuous) discussion of a topic across a range of pages. Wherever possible in the case of topics with many references, these have either been divided into sub-topics or only the most significant discussions of the topic are listed. Because the entire work is about ‘structural transformation’, the use of this term (and certain others which occur constantly throughout the book) as an entry point has been minimized. Information will be found under the corresponding detailed topics.  crisis – absolute poverty ,  absorption of surplus labour –, ,  accounting standards , –, –, ,  accumulation , , , , –, , , – of capabilities  capital, see capital accumulation of factor endowments , , ,  Acemoglu, D. , –, –, , , ,  ADB, see Asian Development Bank advanced economies –, –, , –, , –, –,  long-run behaviour of ,  AfDB, see African Development Bank Afghanistan ,  Africa , –, –, –, –, –, –, ; see also individual countries bottom-up economic transformation – implications of Korean experience – sub-Saharan , –, , –, –, –, –,  African Development Bank (AfDB) ,  ageing , , , –, ,  agglomeration , , –, , , , –, 

continuous index ,  economies – descriptive statistics – indices –, , –, , –,  levels –, – measurement , – of MNCs –, , , , , , – offshore –, , , , ,  patterns , –, , –, , , , – process , ,  quantification methodology – regional –,  aggregate capital to labour ratio , , , ,  aggregate consumption –, –,  aggregate demand , , –, ,  aggregate employment , ,  aggregate geographic levels –, , ,  aggregate growth , , , ,  aggregate labour productivity –,  aggregate production functions , , , , , , ,  aggregate productivity growth –, –,  aggregates, macroeconomic ,  agrarian economies , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





agricultural development –, , ,  agricultural exports ,  agricultural goods –, , ,  agricultural innovations – agricultural labour , , –, –,  agricultural output , –, , –,  agricultural prices , – agricultural production –, , –, , , , –,  agricultural productivity , , , , –, , ,  growth , ,  agricultural terms of trade – agricultural transformation , , , , ,  agriculture –, –, –, –, –, –, –, –; see also farming analytical perspectives and policy approaches since  – Asia – in the broader economy – output growth – prospects – shifting locus of global production – transformation –, ,  air cargo ,  air transport  airports , , ,  Akamatsu, Kaname –, –, ,  Akkus, S. , , , ,  allocation , , , , , , ,  asset ,  capital ,  component – credit , ,  efficient , ,  foreign exchange ,  labour –, –, , ,  optimal –, ,  resources , , , , , – sectoral , , , – American federal democracy –

Amsden, Alice , –, –, –, –, , –,  Angola  apparel industry –, , , –, –, , –,  Argentina –, –, , , –, , , – Asia , , –, –, , , –, – agriculture – East , –, –, –, , , –, – South , , , , ,  Southeast , ,  Asian Development Bank (ADB) , , , , ,  assemblies, colonial – asset allocation ,  asset bases – asset classes , , , –,  asset-backed security ,  asset-liability management (ALM)  assets , , –, –, , , –,  financial ,  foreign currency reserve – high-quality liquid – infrastructure ,  life insurance ,  profitable operating – asymmetries , , – of information , ,  Australia , , , , , –, –, – Austria , , ,  automation , , , , – automobile industry , , , , , , , – autonomy , , ,  embedded ,  average growth –, –, –, , , , – backward linkages , –, , , –,  balance sheets –, –, , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 balanced growth –, –, –,  generalized – in one-sector growth model – Balassa, B.  Baldwin, R. , , , –, , ,  bananas , –,  Banerjee, A. , – Bangladesh , , , , , – banks –, , , –, –, , ,  central –,  commercial –, , , –, ,  foreign , ,  bargaining power , , , – Basel II – Basel III –, , – basic metals industry –, –, ,  baskets, export –, , –, ,  Baumol, W. , ,  behavioural economics ,  Belgium , – benchmark models –, –, ,  Benin , – Bolivia , , –, ,  bonds , –, –, , , –, –,  corporate –,  foreign currency – infrastructure , – long-term ,  booms , , , , , , ,  commodity , ,  Boppart, T. , , – borrowers , ,  bottleneck technologies , , – bottlenecks , , , ,  infrastructure ,  brawn –, –,  Brazil –, , , –, –, , –, – bubbles , , , , –,  budget deficits ,  budgetary expenditures , ,  budgets , , , , –, ,  Buera, F.J. , ,  Burkina Faso , –, –, 



Burundi , – business environment , –, , –, –, , ,  business owners , , – small , – business services , , , –, –, , –, – Canada , , , , , , –, – French  capital –, –, –, –, –, –, –, – charges –,  costs , , , – flight –,  foreign , , –,  intangible ,  marginal product –,  micro-venture – organizational ,  physical , , , ,  rental price , ,  substitution with labour – venture ,  working , –, ,  capital accumulation , –, , , , , ,  human ,  capital allocation ,  capital formation, gross –, ,  capital goods –, , , , , –, , – exports  imports ,  capital income share –,  capital intensities –, , –, , , –, –,  heterogeneous –, – capital markets , , , –, , , ,  international , ,  capital requirements , , –,  solvency – capital stock , , , –, , , ,  aggregate ,  human 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





capital to labour ratio –, –, ,  aggregate , , , ,  capital-good demand – capital-good markets , , –, , ,  correlation –, ,  externalities , –, , , , – scale economies  capital-intensive sectors/activities , , , , , , ,  capitalism , , ,  Anglo-Saxon  crony  capitalists , , , ,  Caribbean –, , –, ,  cash flows , –, , , ,  cashmere cluster ,  catch-up growth –, , –,  customized strategies  government and market roles – causation , , , –,  reverse ,  CDF (Conditional Density Function) , – cement industry , , , , –, , ,  censuses , , , ,  Central America , , –, –, ,  central banks –,  central government , , –,  cereals , , – change structural, see structural change technological , , , , –,  charges, capital –,  chemicals industry , , , , –, –, –, – Chenery, H. , , , , , , ,  children , ,  Chile , , , –, –, , –, – China –, –, –, –, –, –, –, –; see also Hong Kong; Taiwan

– – – – economic reform and structural change – exports , , – factor endowments ,  industrial upgrading – industries – Jiangsu – service sector – shift to service-led economy – Third Plenum , – classical patterns , , ,  classifications , –, , , , , ,  climate change , , –, , , , ,  and gender – clustering , , , –, , , – clusters , , , , , –, , – building industrial clusters – cashmere ,  hometown-based , – service , – coagglomeration , ,  coalitions –, ,  Cobb–Douglas function –, – coke , , , , , ,  collective learning ,  Colombia –, , , , , ,  colonial assemblies – colonies , , , , – Comin, D. –, – commercial banks –, , , –, ,  commodities , , , , , , ,  boom , ,  exports , , , ,  prices , –, , , –, , ,  primary , ,  commodity dependence , – common component of labour productivity growth , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 communication services , , , , ,  communication technologies , , , , –,  comparative advantages –, , , , , –, –,  competition , , , , , , ,  democratic ,  foreign ,  imperfect  international ,  market , ,  competitive advantage –, , , ,  competitive equilibrium , , –, , , ,  competitive labour costs ,  competitive pressures ,  competitiveness – conditions – definitions ,  drivers –,  and economic development – export ,  framework –, –, –, –,  fundamentals , – global ,  international ,  macroeconomic  microeconomic –,  and structural transformation – upgrading –, ,  competitors , , , ,  complementarities , , , ,  complex products –,  complexity , , , , –, , ,  economic , –, , – levels , – composite indices – composition of employment , –, ,  of output , –,  sectoral , , –, , , , 



concentration , , , , , –, ,  export – industrial –, , ,  spatial –,  Conditional Density Function (CDF) , – Congo , ,  constant growth , , ,  constant market share analysis , ,  construction –, –, –, –, –, –, –, – consumer goods , , – consumption –, –, , , –, , ,  aggregate –, –,  final , , , , – goods , – investment ,  contingent liabilities – continuity , ,  control centralized ,  discretionary ,  entry ,  financial –,  mechanisms, reciprocal –, ,  variables , ,  convergence , –, , , , , ,  unconditional  coordination , , , , , , –,  failures ,  copper , , ,  corporate bonds –,  corporatism – Costa Rica , , –, , , –, –, – costs , , , , , –, –,  capital , , , – energy ,  factor , ,  labour , ,  production ,  social , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





costs (cont.) trade , , , , –, , –, – transaction , , , , , ,  transportation , –, , ,  wage  Côte d’Ivoire , –, –,  cotton , , , ,  counterfactuals , , –, – creative destruction , ,  credit –, , , –, –, –, ,  enhancement , ,  private – as proxy for financial depth – subsidized – credit ratings ,  public , , , – crises  – financial , , , , , –, –, – food , –, , –, ,  crony capitalism  crop production –,  cross-section data –, , ,  Cuba , , , –, , , ,  currencies , , –, , , , ; see also foreign exchange cyclical policies – Czech Republic , –,  Dar es Salaam , – data missing , , – sets –, , , , , –, –, – sources , , , ,  debt , , –, , ,  crisis , –, ,  external ,  infrastructure , – decentralization , –, , – decomposition , –, –, , , , , – of change in employment  of change in GDP , 

of change in world openness – Olley-Pakes  of productivity growth , ,  of SAM multipliers ,  of structural change –, , – deficits , , , , , , , – trade , ,  deflators , , , ,  implicit , –,  MVA  de-industrialization , –, , –, , , ,  premature , ,  demand –, –, –, –, –, –, –, – aggregate , , –, ,  capital-good – investment , ,  labour , ,  market , , ,  democracy –, –,  American federal – representative ,  demographic change –,  demographic dividend , , ,  demographic growth , –, –,  demographic transition , , ,  Denmark , , , –, ,  dependence, commodity , – dependency path ,  ratios –, ,  dependent variables –, , , ,  despair –,  destruction, creative , ,  devaluations , –, , , ,  developed economies –, , –, –, , –, –,  developing economies , –, –, , –, , ,  development economics , , , –, , –, ,  reassessment – economists , , , , , –, –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 financial –, , , , , , ,  financing , , , –, ,  industrial , , , , , , ,  infrastructure , , , , , , ,  manufacturing , , , –, ,  patterns , –, , – stages , –, , ,  development finance institutions (DFIs) – DFIs (development finance institutions) – difference-in-difference regression approach , – differential capital intensities ,  diffusion of knowledge –, –, , , ,  digital technologies –,  digital TV , – disaggregation , , ,  discretionary control ,  displaced workers – distance , , , –, , ,  thresholds , –, ,  distribution –, , , –, , , ,  geographic , , ,  income , –, , , –, , –, – intra-sector ,  sectoral , ,  divergence , , , , , ,  diversification –, –, –, –, , –, –,  exports –, –, , , –, –, –, – stages of ,  domestic linkages , , – domestic markets , –, –, , , –, –,  domestic savings , –, , –,  dominance , , , , – Dominican Republic , , , , 



double-helix ladder of development –, ,  downturns , , , ,  Duranton, G. , , –, ,  early industries –, –, , –, –,  East Asia , –, –, –, , , –, – Eastern Europe , –, , , , , ,  ECI (Economic Complexity Index) , , , – ECLAC (Economic Commission for Latin America) –, , , –,  Economic Community of West African States (ECOWAS) , , ,  economic complexity , –, , – Economic Complexity Index, see ECI economic development , , , –, , –, –,  and competitiveness – sustained ,  economic empowerment ,  economic forces –, ,  economic geography , , , , –, –, ,  global , – economic growth –, –, –, –, –, –, –, – Egypt – engines of ,  models , ,  modern , , , , , , ,  rapid , , , , , , , – sustained , , ,  economic history , , , , , , ,  economic integration –, , ,  economic liberalization , ,  economic participation , ,  economic performance , , , , , , ,  economic policies , , , , , , ,  economic power , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





economic reform , –, , –, , , ,  Economic Reform and Structural Adjustment Program (ERSAP) ,  economic rents ,  economic structuralism – economic structures , , , , –, –, , – economic upgrading –,  economics behavioural ,  development –, –, , –, , , ,  of ideas – of information –,  structural , , , , , –, ,  economy-wide productivity , ,  ECOWAS (Economic Community of West African States) , , ,  Ecuador , , , ,  education , –, –, –, –, –, –,  higher , , , , ,  secondary ,  vocational ,  efficient markets – Egypt , , –,  economic growth – exports – external sector – labour market – political economy of structural reform – population ,  public finances – EHDA (Ethiopian Horticulture Development Agency) , ,  Eicher, C.K. – El Salvador –,  elasticities , , , , , , – employment , , , ,  growth – income , , , , , ,  of substitution , –, , ,  elections , , –, –

electrical machinery and apparatus industry , , –, –, , ,  electricity , , , , , , – electronics , , –, –, –, , , – Ellison, G. , , –, – embedded autonomy ,  embeddedness  EMDCs (emerging market and developing countries) , , ,  EMDEs (emerging market and developing economies) –, , , , – emerging countries –, –, , , , , –, – emerging market and developing countries, see EMDCs emerging market and developing economies, see EMDEs emerging market economies , , –, , , – emerging markets , , , , , , ,  empirical analyses –, , , , , , , – employees , , , , –, , , – employers, largest , ,  employment –, –, –, –, –, –, –, – aggregate , ,  composition , –, ,  distribution of – levels , –, – manufacturing , , , –,  opportunities , ,  quality of ,  shares , –, –, –, , –, –,  total , , , , , , , – employment–population ratio, see EP ratio emulation –, – endogeneity , –, , , –,  endogenous structural change –,  endowment structures –, , –, –, , –, 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 endowments, see factor endowments energy , , , , , , ,  costs ,  subsidies ,  Engel effects – Enlightenment –,  entrepreneurs , –, , , , , , – young – entry barriers , , , , , , , – control ,  EP ratio (employment–population ratio) , , –,  equilibrium , , , –, –, ,  competitive , , –, , , ,  market ,  models , ,  new –, ,  static – equipment industries , , , –, –, , , – equity , , , , , , , – common – investment – markets , – ERSAP (Economic Reform and Structural Adjustment Program) ,  Ethiopia , , , , –,  challenges of structural transformation – floriculture , –, , , ,  government –,  Grand Ethiopian Renaissance Dam (GERD) – Hawassa Eco-Industrial Park , – industrial policy and structural transformation – lessons to be learned – linkage effects –,  policy instruments – policy learning – policy organizations – policy outcomes –



Ethiopian Horticulture Development Agency (EHDA) , ,  European Union , , , , – exchange rates , –, , , , ,  multiple , , – overvalued ,  real effective – expenditure , , , , , , –,  final consumption ,  shares , –,  expenditures, household –,  experimentation , , , –,  explanatory variables , –,  export baskets –, , –, ,  export competitiveness ,  export concentration – export earnings , , –,  export markets , , , , – export processing zones ,  export promotion , , –, , , –, ,  incentives ,  export sophistication , – export structures , , , , , , –,  exported goods , , ,  exporters –, , –, , , , ,  Export–Import Bank of Korea – exporting countries –, ,  export-led growth , , –,  export-led industrialization , ,  exports –, –, –, –, –, –, –, – agricultural ,  China , , – commodity , , , ,  diversification –, –, , , –, –, –, – Egypt – and GVCs – manufactured –, , , ,  manufacturing , , , ,  Mexico , – natural resource-based , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





exports (cont.) processing , ,  service , ,  structural transformation through – Taiwan –, – total , , , –, , , ,  world –, , , –, ,  external debt ,  external resources –, ,  external scale economies ,  externalities , , , , –, , ,  capital-good market , –, , , , – labour market , ,  Marshallian  negative ,  positive , , , , ,  fabricated metals , , , –, , , – fabricated metals industry , –, ,  factor allocations , ,  factor costs , ,  factor endowments –, , , –, –, –, ,  factor markets , –, , , , ,  factor productivity , , , , , –, –,  total , , , , , –, , – factor substitution, see substitution factories , , –, ,  families , , , , , ,  family workshops , ,  farmers , , , –, , , , – farming , –, , , , ; see also agriculture farms –, , –, – FDI (foreign direct investment) –, , –, , –, , , – feasibility , –, ,  federal decentralization , – federalism , – feedback effects –

fees, termination , ,  female labour force participation –, –,  fertility rates , ,  fertilizers –, , , ,  final consumption , , , , – expenditure ,  finance –, –, , , –, –, , – development , , , –, ,  infrastructure – financial assets ,  financial control –,  financial crisis , , , , , –, –, – financial depth , , ,  financial institutions , , , , , –, ,  financial institutions (IFIs), international , –, –, , – financial institutions, international , –, –, , – financial liberalization , , , ,  financial liberalization index –, ,  financial markets , –, , , –, , ,  global –,  financial reform index –, ,  financial reforms –,  effect , , – financial resources , , ,  financial services , , , , , , ,  financial status quo –, –, –,  financial structure , , – Finland , , ,  First Industrial Revolution ,  fiscal policies , , , , –,  Fisher, P.S. , , – fishing , , –, –, ,  floriculture , –, , , ,  flows capital –, , –, ,  cash , –, , , ,  trade ,  flying-geese theory –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 double-helix ladder of development –, ,  Friedrich List’s legacy  government and market roles in catch-up growth – Hegelian dialectics  MNCs as instrument of catch-up – and national ecosystem reform – Schumpeteriean ladder and s-shaped growth trajectory – types of FG-style structural transformation – United States as leader – follower economies , ,  food and beverages industry –, , , –, –, , ,  food crises , –, , –, ,  food prices, instability , – food processing , , , , –, ,  food security –, , –, ,  macro –,  micro ,  food supplies , ,  footwear –, , –, , –, –, ,  forces agglomeration –, , , –, , –, , – economic –, ,  market , , , ,  Marshallian ,  foreign aid , , , ,  foreign capital , , –,  foreign currency bonds – foreign currency reserve assets – foreign direct investment, see FDI foreign exchange , , , , –, , –,  allocation ,  foreign subsidiaries , –, , – foreign trade , ,  forestry , , , –, –, ,  formal models , , ,  formal sector , , , , , – former Soviet Union , , –, , –, , 



forward linkages , –, , , ,  Fourth Industrial Revolution ,  fragmentation , , , –, , , ,  global – vertical ,  France –, , , –, –, , –, – free trade , , , , , ,  French Canada  French Revolution ,  frictional labour market – Friedrich List’s legacy  fuel prices ,  functional upgrading ,  funding stability – furniture , , , , ,  Furtado, C. ,  gap in labour productivity , , ,  GDP (gross domestic product) –, –, –, –, –, –, –, – composition –,  shares , , –, , , , –,  gender , –,  demographic change –,  inequality – macroeconomics of – policy options – and technological change – generalized balanced growth – genetically modified organisms (GMOs) ,  geographic concentration ,  geographic distribution , , ,  geographic levels, aggregate –, , ,  geography of multinational firms – literature – GERD (Grand Ethiopian Renaissance Dam) – Germany –, , –, , , , –, –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





GFC (Global Financial Crisis), see financial crisis Ghana , –, , –,  Ghani, E. – Gini coefficient ,  girls , – Glaeser, E. ,  global confidence bands –,  global economy , , , , , –, ,  global financial market –,  global markets , , –, , , , –,  global population –,  global supply chains , , , ,  global value chains, see GVCs globalization –, , –, –, –, –, –, – and gender – and Taiwan – two globalizations – GMOs (genetically modified organisms) ,  GNI (gross national income), per capita –,  gold , , ,  Golden Age , , ,  Gonzales, C. –,  good life , –,  goods , –, –, , –, –, , – consumer , , – exported , , ,  intermediate , –, , , , , –,  manufactured , , , –, , , , – primary , –,  public –, , , , –,  governance , , –, – good ,  indicators , , ,  structure ,  government central , , –,  local –, , –, , , –, , –

national , –, – ownership –,  policies , –, , , , , –,  revenues , ,  role , , ,  support , –,  Grand Ethiopian Renaissance Dam (GERD) – Great Britain, see United Kingdom Great Depression , , , –, –, , ,  interpretation – Great Recession , , , , , ,  gross capital formation –, ,  gross domestic product, see GDP growth accounting ,  aggregate , , , ,  average –, –, –, , , , – balanced –, –, –,  catch-up, see catch-up growth constant , , ,  demographic , –, –,  dividends , ,  elasticity – employment , –, , –, , –, , – export , , , , , , ,  export-led , , –,  generalized balanced – high , , , , , , ,  inclusive , ,  income , , ,  industrial , , , , , , ,  labour productivity –, –, , , –, , , – labour-intensive ,  long-run , , , ,  manufacturing ,  models, one-sector , , – in openness – output , , , –, , ,  patterns , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 population , , , ,  process , , , ,  productivity , –, , , –, , –, – prosperity –, ,  rapid , –, , –, –, , , – rates , , –, , –, –, –, – services-led –, , ,  slow , , , ,  steady , , – sustainable , –,  total factor productivity , – guarantees , , , –, ,  minimum revenue ,  Guatemala , ,  Guerrieri, V. –, –,  Guinea  Guinea-Bissau , – GVCs (global value chains) , –, , –, , –,  and exports – participation in – policy discussions – risks and opportunities for firms and people – special challenges for less developed countries – transformation of global economy – Hamiltonian system – harvesting services , – Hausmann, R. , –, , , , –, –,  Hawassa Eco-Industrial Park , – Hayami, Y. –,  HDI (Human Development Index) , ,  health –, –, , , –, , –,  outcomes ,  public , ,  Hegelian dialectics  Helpman, E. , , , – Herrendorf, B. , , , , , , , –



heterogeneity , , , , , , ,  structural ,  heterogeneous capital intensities –, – Hidalgo, C. , – hidden subsidies ,  higher education , , , , ,  high-income economies –, , , –, –, , , – high-income status , , , ,  high-income trap  high-quality liquid assets – high-tech industry , , –, ,  high-tech products , ,  Hirschman, A.O. , , , , , , –,  history –, , , –, –, , ,  economic , , , , , , ,  hometown-based clusters , ,  Honduras , –,  Hong Kong , , , , , –, –,  host countries , , , , , , ,  household expenditures –,  household income , , –, – household tasks ,  households –, –, , , , , –,  rural , , , –, ,  urban – housing –, , , ,  human capital accumulation ,  Human Development Index (HDI) , ,  human resources , , , –,  human-capital-intensive service sectors –,  Hungary –,  hunger –, , , , ; see also food security IBK (Industrial Bank of Korea) , – IBTDPs (Industrial Base Technology Development Projects) –,  ideas, economics of –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





IFIs, see international financial institutions IFRS (International Financial Reporting Standards) , –,  imitation , , , , , , ,  implicit deflators , –,  import substitution , –, , , , – secondary , ,  imported intermediates –,  imports , –, , –, –, , –,  import-substitution industrialization, see ISI imputation, multiple , –, – incentives –, , , , , , –,  export-promotion ,  investment , , ,  price ,  tax , , , ,  income effects , , , –, ,  income elasticity , , , , , ,  income inequality –, ,  incomes –, –, –, –, –, –, –, – distribution , –, , , –, , –, – Kuznets beyond Kuznets – Kuznetsian basics – simple analytics – high , , , , , , –,  household , , –, – low , , , , , , ,  national , , , , , , ,  per capita –, –, –, –, , –, , – independent variables ,  India –, , , –, –, , –, – contours of structural change – different path to transformation – patterns of change – rural – structural transformation – stylized facts and analytical constructs –

Indonesia –, , , , –, , ,  Industrial Bank of Korea, see IBK Industrial Base Technology Development Projects, see IBTDPs industrial clusters, see clusters industrial concentration –, , ,  industrial development , , , , , , ,  industrial dynamics , , – industrial growth , , , , , , ,  industrial parks , , –, , –, , , – industrial policies , –, –, –, –, –, –, – active ,  common ,  as cyclical policies – definitions ,  role , , ,  traditional ,  industrial prices – industrial production , , , , , , ,  Industrial Revolution , –, , , , , , – First ,  Second  Third  industrial robots ,  industrial sector –, , –, –, –, , , – industrial structures , –, , , , , –,  optimal –, ,  industrial transformation ,  industrial upgrading , , –, –, , , ,  industrialization –, –, –, –, –, –, –, – export-led , ,  late , – processes , , , ,  rapid , , , –, , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 state-led , , –, –, –, , –,  as unavoidable feature of change – industrialized countries –, , , , , , , – industry pairs –, –, , –, , ,  inequality , –, , –, ,  income –, ,  increasing , ,  national –,  sectoral , – infinite industries –,  inflation rates , , , ,  informal sector , , , , , , ,  information, economics of –,  infrastructure –, , –, , –, –, –,  changes in regulation and accounting standards – current climate of funding and finance – debt , – finance – funding gap – listed – loans –, , –,  physical , ,  projects –, , , ,  proposals for innovative financial products – sectors , , –, – services , – social ,  targets for EMDE infra finance – transport , , , ,  unlisted – innovation , –, –, –, , , –,  agricultural – indigenous ,  technological , , , , ,  input factors – inputs , , –, –, , , –, 



intermediate , , , –, , , ,  instability , , , , ,  food prices , – institutional investors –, –, , , , , –,  international ,  institutional reforms , ,  institutions , , –, –, –, –, ,  insurance , , ,  contracts – liabilities – life , –, , , , , , – risks ,  social ,  insurers –, –, , , – integrated steel mills , ,  integration , , –, –, , , ,  economic –, , ,  regional –, –, , , , , ,  intensive margin , , –,  interest rates , , , , –, , , – real , ,  inter-industry interaction , –, , – inter-industry ladders  intermediate goods/products , –, , , , , –,  intermediate inputs , , , –, , , ,  International Accounting Standard Board (IASB)  international competitiveness ,  international division of labour , , ,  international financial institutions (IFIs) , –, –, , – International Financial Reporting Standards, see IFRS international institutional investors ,  international markets , , , , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





International Public Sector Accounting Standards, see IPSAS International Standard Industrial Classification, see ISIC international trade , , , –, –, , ,  intervention, government , –, , , , –, –,  intra-industry side-ladders  intra-industry trade , – inventions –, ,  social  investment –, –, –, –, –, –, , – consumption ,  decisions , , , ,  foreign, see FDI infrastructure –, –, , , , , ,  long-term , , , , – private –, , , ,  public , , , , , , ,  strategies , ,  investors , , , –, , , –, – long-term , –, ,  IPSAS (International Public Sector Accounting Standards) , , , – irrigation , –, ,  IS-EP progression , – ISI (import-substitution industrialization) –, , ,  ISIC (International Standard Industrial Classification) , –,  Israel –,  Italy –, –, , , –, , –,  Japan , , –, , –, –, ,  Jarreau, J. , – Jefferson, Thomas , ,  JLW model –, – extensions – job satisfaction –

job-seekers , ,  Johnston, B.F. – Jordan ,  Kaboski, J.P. , ,  Kaldor facts , , , , ,  KDB (Korea Development Bank) , –, ,  Kenya , , , –, , , ,  Keynes, J.M. –,  Keynesian policies , , –,  Klinger, B. , , , –,  knowledge , –, –, –, –, , , – spillovers , ,  transfers , ,  knowledge-intensive work , –,  Kongsamut, P. , –, , ,  Korea , –, –, –, –, –, –,  digital TV , – financial control and industrial policy – government , –, – industrial policy and financing – North  Pohang – roles and evolution of development banks – Korea Development Bank, see KDB Korean War , ,  Krugman, P. , , , , –, , ,  Kuwait ,  Kuznets, Simon , , , , –, –, ,  Kuznetsian basics – labour –, –, –, –, –, –, –, – absorption , , , , , –,  agricultural , , –, –,  allocation –, –, , ,  demand , ,  intensity , –, , ,  international division of , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 mobility ,  reallocation –, , –, , , , ,  skilled , –, , , , , ,  supply , – surplus , , , , –, , ,  unskilled , , , ,  labour costs , ,  competitive ,  labour force –, –, –, –, –, , , – participation – female –, –,  labour markets –, , , , –, –, ,  correlation –,  Egypt – frictional – perfect  labour productivity –, –, –, –, –, –, –, – aggregate –,  economy-wide , , ,  gap , , ,  growth –, –, , , –, , , – common component , ,  sectoral ,  labour-intensive growth ,  labour-intensive industries , , –, , , , , – labour-intensive manufacturing , , , –, – labour-intensive production , , , ,  labour-intensive sectors , , , ,  labour-intensive services , , , – ladder of development, double-helix , , ,  LAFTA (Latin American Free Trade Association)  Lagrange multipliers ,  land –, , –, , –, –, ,  productivity ,  reform –, 



late industrializers –, ,  late industries –, –, –, , – late-comers , , , , ,  Latin America –, , –, , –, –, ,  from commodity exports to State-led industrialization – market reforms and premature de-industrialization – structural transformation patterns , , , , , , ,  Latin American Free Trade Association (LAFTA)  Latin American Integration Association (LAIA)  latitude for performance standards , ,  LCR (liquidity coverage ratio) – leaders , –, , , –, , , – national , – political , –, ,  leapfrogging , ,  learning , –, –, –, –, –, ,  collective ,  policy –, , – processes , , , ,  learning-by-doing , , , –, – leather , , , –, , , –,  products , , , , , ,  Lederman, D. , –, ,  legitimacy , , , ,  Lesotho  Lewis, W. Arthur , –, , , –, , , – liabilities –, –, , – long-term –, –, –,  liability costs ,  liberalization , –, , , , –, ,  economic , ,  financial , , , ,  trade , , , , , , ,  liberalized trade , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





Liberia ,  Libya ,  life expectancy , , , –, ,  life insurance , –, , , , , , – assets ,  light industries , ,  light manufacturing , ,  linkage dynamics –,  linkage effects –,  linkages , , –, –, , , ,  backward , –, , , –,  domestic , , – output , , , –, – production , , , , –, –, –,  liquid assets, high-quality – liquidity , – liquidity coverage ratio, see LCR listed infrastructure – literacy rates , – livestock , –, , , , ,  living standards –, , , , , , – loans , , –, , , , ,  infrastructure –, , –,  policy –,  total , ,  local governments –, , –, , , –, , – location –, , –, –, , , –,  decisions , , – effect – fundamentals –, , –, , , –, – and agglomeration economies –, , , ,  measurement – MP – role , , – patterns , ,  long-run growth , , , ,  long-term bonds ,  long-term institutional investors , ,  long-term investments , , , –

long-term investors , –, ,  long-term liabilities –, –, –,  losses , , –, , , , , – low productivity sectors , , , , –, , ,  low-end manufacturing –,  low-income countries , –, , , –, , –, – low-income trap , – macro food security –,  macroeconomic aggregates ,  macroeconomic policies , , ,  macroeconomic stability , ,  macroeconomics , , , ,  of gender – Maddison, A. , , , , , ,  Madison, James  Malawi , , ,  Malaysia , –, –, –, , –, ,  Mali , , –,  Malta ,  manufactured exports –, , , ,  manufactured goods , , , –, , , , – manufacturing –, –, , –, –, –, –, – analysis – data, variables, and estimations – economies , , , ,  employment , , , –,  growth ,  industries , , –, –, , –, –,  labour-intensive , , , –, – light , ,  low-end –,  output , , ,  results – value added , , , –, –, , ,  market competition , ,  market demand , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 market economies , –, , , ,  market equilibrium ,  market failures , –, , , , , ,  market forces , , , ,  market prices , , , ,  market reforms , –, , , , –, , – market risks , –,  market size –, , –, , , , ,  market transition –,  marketing , , , , , , ,  marketization , ,  markets –, –, –, , , –, –,  capital-good , , –, , ,  domestic , –, –, , , –, –,  efficient – emerging , , , , , , ,  equity , – export , , , , – factor , –, , , , ,  failure to manage structural transformations – financial, see financial markets global , , –, , , , –,  international , , , , , ,  labour, see labour markets narrow –,  new , , , ,  secondary , ,  securities ,  US , , ,  world , , , , –, , ,  Markusen, A. –, ,  Marshall, A. , , , , –, , , – Marshall, John ,  Marshallian externality  Marshallian forces , 



Martinique  Marx, Karl ,  mass poverty  materials-intensive industries ,  Matsuyama, K. –,  Mauritania  Mauritius , –, , , ,  megatrends – MENA (Middle East and North Africa) –, , –, ,  metals basic , , , –, , ,  fabricated , , , –, , , – methodologies –, –, , , , , ,  empirical , ,  Mexico –, , –, –, –, , –, – exports , – micro, small and medium sized enterprises, see MSMEs micro food security ,  microeconomic competitiveness –,  microeconomic factors – microenterprises ,  middle and late industries –, , – Middle East and North Africa, see MENA middle industries –, , –, , , – middle-income countries , –, , , , –, ,  middle-income status –, –, ,  middle-income trap , , –, , , , ,  MIGA (Multilateral Investment Guarantee Agency) , , – migration –, , , ,  rural-urban , , ,  minerals, non-metallic , , , –, –, –, –,  minimum revenue guarantees ,  minimum wages , ,  mining , , , –, , –, , – coal  economies , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





missing data , , – MNCs (multinational corporations) –, –, , , , –, –,  agglomeration –, , , , , , – headquarters –,  offshore –, – offshore and domestic compared – establishment data – geography – mobile phones , – mobility , , ,  labour ,  models formal , , ,  growth –, , , –, , , ,  JLW –, – stages , ,  static , ,  modern economic growth , , , , , , ,  modern services , , , , , , –,  modernization –, , , , ,  monetary policies , , , , –, , ,  Morocco , ,  motivation/motives , , –, , , , ,  motor vehicles , , –, –, , – Mozambique  MSMEs (micro, small and medium sized enterprises) –, – Multilateral Investment Guarantee Agency, see MIGA multilateral trade agreements ,  multinational corporations, see MNCs multinational foreign subsidiaries , ,  multinational production –, – multinational subsidiaries , – multinationals, see MNCs multiple exchange rates , , – multiple imputation , –, –

multipliers , , –, , , –,  backward linkage –, ,  Lagrange ,  MVA deflators  Myrdal, Gunnar , ,  Namibia  Nashashibi, K. – national ecosystem reform – national government , –, – national income , , , , , , ,  national leaders , – natural resource-based exports , ,  natural resources –, –, –, , –, –, , – negative externalities ,  negative spillovers  neo-liberalism –,  net stable funding ratio, see NSFR Netherlands , , –, , –, , ,  New Structural Economics (NSE) , –, , –, –, , , – benchmark model – and competitiveness framework – dynamics – empirical relevance – and flying-geese theory – insights for structural change measurement – literature – market equilibrium – model environment – policy implications  New Zealand , , , – Ngai, L. R. , , , –, , ,  Nicaragua , , –,  NIEs , ,  Niger , , – Nigeria , , , ,  Nkrumah, Osagyefo Kwame –,  Nobel Prizes , , , , ,  non-agricultural sectors –, –, ,  non-competitive market structures – non-financial corporate debt –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 non-homotheticities , –, –, –, –, – non-metallic minerals , , , –, –, –, –,  non-tradables , –, , , ,  non-uniform technical change – normalization , – North Atlantic –, ,  North–South trade – Norway , , –,  NSE, see New Structural Economics NSFR (net stable funding ratio) – numerical simulations –,  Nyerere, Julius – Oceania ,  O’Connell, S.D. – OECD (Organization for Economic Cooperation and Development) –, –, , –, , , ,  offshore agglomeration –, , , , ,  offshore production , ,  Ohno, K. , –, –, – oil , –, , , –, –, , ; see also petroleum Olley-Pakes decomposition  Oman  one-sector growth model , , –, – openness , –, , –, –, –, ,  growth – total , –,  and trade – world , , , – operational risks ,  optimal allocation –, ,  optimal industrial structures –, ,  Organization for Economic Cooperation and Development, see OECD original industries –,  Ottoman Empire ,  output , , –, –, –, –, , – agricultural , –, , –,  composition , –, 



growth , , –, , ,  agriculture – relationships , , – total –, , , , , , ,  outsourcing , , , ,  Overman, H. , –, ,  ownership , , –, , , , ,  private , , , ,  state , ,  pairwise industries –, , –,  Pakistan , ,  Panama , , , , – Papua New Guinea  Paraguay , , ,  participation economic , ,  female –, –,  path dependency ,  PBAM (Peixian Bureau of Agricultural Mechanization) – PCI (Product Complexity Index) –,  Peixian Bureau of Agricultural Mechanization (PBAM) – pension funds , , –, –, , , , – per capita income –, –, –, –, , –, , – performance standards ,  latitude for , ,  personal services ,  Peru , , –, , , –, – petroleum , , , , , –, , ; see also oil refined , , , , ,  Philippines , , , –, , ,  physical infrastructure , ,  Piketty, T. , –,  plastics , , , , –, ,  Pohang Steel – policy frameworks , , ,  policy independence –,  policy indexes –, ,  policy learning –, , – policy loans –, 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





policy makers –, , , , , –, –, – political decentralization , – political decisions ,  political economy , , –, , , , –,  political factors ,  political institutions , ,  political leaders , –, ,  political pressures , , , ,  political stability , , , –,  political tensions , , ,  politics –, , , , –,  Poncet, S. , – pooling, labour market , , ,  poor countries , , , –, –, –, , – population –, –, –, –, –, , ,  global –,  rural –, ,  shift , – total , –, ,  urban –, , , ,  working , –,  working age , , ,  Porter, Michael , –, –, , ,  portfolio building ,  portfolio management –,  portfolios , , , –, , –,  ports , , ,  positive externalities , , , , ,  poverty –, , –, , –, , –, – absolute ,  mass  power –, , , , , –, , – bargaining , , , – economic , , ,  PPP contracts , ,  preferences , –, –, –, , , ,  homothetic , ,  non-homothetic , , , –, –

premature de-industrialization , ,  pressure , , , , , , –,  competitive ,  political , , , ,  price incentives ,  price signals , , ,  prices , –, , , –, –, –, – agricultural , – commodity , –, , , –, , ,  market , , , ,  relative , –, , , , , , – rice , – volatility – primary commodities , ,  primary goods , –,  primary products , , ,  primary sector , , , , –, , –,  printing and publishing , , , –, ,  Pritchett, L. , ,  private credit – private financing institutions –, ,  private firms , –, , –, , , ,  private investments –, , , ,  private ownership , , , ,  private sector –, , –, , , –, –,  privatization , , , , , , ,  processing exports , ,  producers , , , , , , ,  agricultural , , ,  local ,  Product Complexity Index, see PCI product quality , ,  production agricultural –, , –, , , , –,  costs ,  crops –, 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 food ,  functions , –, –, , , –,  aggregate , , , , , , ,  industrial , , , , , , ,  labour-intensive , , , ,  linkages , , , , –, –, –,  vertical , , , , –, –,  multinational –, – productive resources , ,  productive sectors –, , , , , , ,  productive structures , , –, –, , , –,  productive transformation –,  productivity , –, –, –, –, –, , – agricultural, see agricultural productivity differences , ,  economy-wide , ,  factor , , , , , –, –,  growth , –, , , –, , –, – aggregate –, –,  agricultural , ,  decomposition , ,  high , , –, , , , ,  higher , , –, , , , –,  improvements , , , , , , , – labour, see labour productivity levels , ,  low , , , , –, , ,  relative –,  rural ,  sectoral , – products agricultural , , ,  complex –,  intermediate , , ,  leather , , , , ,  new , , –, , , , , –



primary , , ,  wood , , , , ,  pro-employment path , –,  profitability , , , , , , ,  profitable operating assets – profits , , , –, , , ,  progress, technological , , –, , , –, ,  promotion, export , , –, , , –, ,  prosperity –, –, –, , –, –, –, – differences , , – growth –, ,  proxies , , , –, , , ,  proximity , , –, –, , ,  prudential regulation  public credit ratings , , , – public goods –, , , , –,  public health , ,  public investment , , , , , , ,  public sector , , , , , , ,  public services , , , , , , ,  Puga, D. ,  Puyuan ,  Qatar –, – qualitative variables , ,  quality –, –, , –, , , , – control  of employment ,  high , ,  poor ,  standards ,  upgrades – quantitative easing (QE) ,  quantitative variables , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





Radwan, S. , – rapid industrialization , , , –, , , ,  Ravallion, M. , –,  raw materials , , , , , , ,  R&D , , , , , , ,  real estate , , , – reallocation –, , –, , ,  dynamic ,  of economic activity ,  labour –, , , –, , , ,  resource , , , ,  recessions –, –, , , , , ,  reciprocal control mechanisms –, ,  redistribution , , –,  refinancing risks –, , ,  refined petroleum , , , , ,  reform strategies –, – reforms –, –, , –, , –, , – financial –,  institutional , ,  market , –, , , , –, , – policy , ,  sectoral ,  structural , , , , , , ,  regional agglomeration –,  regional integration –, –, , , , , ,  regressions , , , , –, , ,  growth ,  results , ,  relatedness , ,  relative prices , –, , , , , , – renewable energies ,  rents , , , , –, –, ,  economic ,  natural monopoly  representative assemblies –, 

representative democracy ,  re-primarization –, – resource allocation , , , , , – resource reallocation , , , ,  resources , , –, , –, –, –, – domestic , ,  external –, ,  financial , , ,  productive , ,  scarce , ,  Restuccia, D. , ,  retail , , , , , ,  retailers ,  revenues , , – government , ,  reverse causation ,  rice , , , , , –, , – rich countries , , , , , , ,  rigidities, structural , – risk, credit , ,  risk appetites –, , – risk levels , ,  risk management ,  risks –, –, –, , , , –, – financial ,  market , –,  operational ,  robots , , ,  Rodrik, Dani –, –, , , , , ,  Romania  Rosenblatt, D. , ,  Rosenstein-Rodan, P.N. – Roy’s identity ,  rubber and plastics , , , , – rural areas , , , , , , ,  rural households , , , –, ,  rural population –, ,  rural productivity ,  rural sector –, , ,  rural-urban migration , , ,  Russia , , , , –, , ,  Ruttan, V.W. –, ,  Rwanda , , –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 salaried jobs –, , – samples –, –, , , , , ,  SAMs (Social Accounting Matrices) –, , , –,  Saudi Arabia , , – savings , , , –, , , , – domestic , –, , –,  gap ,  scale economies , , –, , , ,  Schmitz, H. –,  Schultz, T.W. –, , –,  Schumpeter, J. –, , , , , ,  sclerosis –,  SDGs (Sustainable Development Goals) , , , ,  Second Industrial Revolution  secondary import substitution , ,  secondary markets , ,  secondary sector , , , –, , –, ,  sectoral allocation , , , – sectoral composition , , –, , , ,  sectoral inequalities , – sectoral productivity , – sectoral reforms ,  securities markets ,  Sen, A. , ,  Senegal , –, , , – service clusters , – service economy , –,  service exports , ,  service-led economies –,  service-led growth –, , ,  services –, , –, –, –, –, –, – business , , , –, –, , –, – communication , , , , ,  financial , , , , , , ,  harvesting , – human-capital-intensive – infrastructure , –



labour-intensive , , , – modern , , , , , , –,  personal ,  public , , , , , , ,  share , , , , , , ,  Taiwan – trade , ,  transport , , , ,  SEZs (special economic zones) , , , , – shipbuilding , –, , – shortages , , , ,  signals , –, ,  price , , ,  Singapore , –, , , –, –, –,  skill levels , , , , – skilled labour , –, , , , , ,  skills , , , , , , , – development , ,  upgrading , ,  slowdowns , –, –, , , , ,  small business owners –, – SMEs –, , , , , , , – Smith, Adam , , , , , ,  Social Accounting Matrices, see SAMs social costs , ,  social groups , –,  social infrastructure ,  social insurance ,  social peace –,  Social Planner’s problem , , , , , , ,  social upgrading –,  SOEs (state-owned enterprises) –, , –, , , , –, – soils , –,  Solow, R. , –, , , , – Solvency II Directive , , –, ,  sophistication , , , , –, –, ,  exports , , –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





sourcing , , , , , , ,  South Africa , , , , , , ,  South Asia , , , , ,  South Korea, see Korea Southeast Asia , ,  Southern Cone , , –,  sovereign power ,  sovereign wealth funds , –, , –, ,  Soviet Union, former , , –, , –, ,  soybeans –, ,  Spain , , –, , –, , ,  spatial concentrations –,  special economic zones, see SEZs special purpose vehicles, see SPVs specialization , –, –, , –, , ,  spillovers , , , , , –, – knowledge , ,  negative  technology , , , –,  spin-offs ,  sponsors , , , –,  SPVs (special purpose vehicles) , ,  SSA, see sub-Saharan Africa s-shaped growth trajectory – stability , , ,  financial  funding – institutional  legal  macroeconomic , ,  political , , , –,  stages, development , –, , ,  stages models , ,  caveat – stagnation , , , , , , ,  standards , , , , , , – state ownership , ,  state-led industrialization , , –, –, –, , –,  state-owned enterprises, see SOEs

status quo , –, –, , ,  financial –, –, –,  steel , , , , , , ,  industry , , , – steel-consuming industries ,  stock markets , ,  structural breaks , ,  structural change, see also Introductory Note decomposition –, , – drivers –, , ,  endogenous –,  financial reforms and financial development – and heterogeneous capital intensities – index composition – index construction – index reliability – measurement , – traditional approaches – patterns , , –, , ,  processes –, , , , , ,  remodelling – sustainable , , , , , , ,  and trade – via non-uniform technical change – via uniform technical progress – structural economics , , , , , –, ,  structural factors ,  structural heterogeneity ,  structural reforms , , , , , , ,  structural rigidities , – structural transformation, see also Introductory Note and balanced growth – benchmark model – and competitiveness – evolutionary process , ,  and growth – and income distribution – income effects , , , –, ,  indexes , , ,  indicators –,  key elements –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 measurement, new approache – as organizing framework – processes , –, , , –, , –, – sustainability – qualitative analysis of driving forces – relative price effects , – successful , –, –,  sustainable ,  tensions along path – through exports – structural transmigration , ,  structural upgrading ,  structuralism , –, , , ,  economic – structureless framework – structures , , –, –, –, , ,  economic , , , , –, –, , – export , , , , , , –,  importance for development – industrial , –, , , , , –,  production , , –, –, , , ,  productive , , –, –, , , –,  sub-Saharan Africa (SSA) , –, , –, –, –, –,  subsidiaries , , , , ,  foreign , –, , – multinational , – subsidies , , , , , , ,  energy ,  hidden ,  subsistence sectors , ,  substitutability ,  perfect ,  substitution , , , , –,  elasticities , –, , ,  import, see import substitution Sudan  Suez Canal , , ,  suppliers , , , –, 



supply chains , , , , –, –,  labour , – stable , , ,  surplus labour , , , , –, , ,  surpluses, trade , , ,  sustainability , , , ,  structural transformation processes – Sustainable Development Goals, see SDGs sustainable growth , –,  sustainable structural change – Sweden –, , –, , , – Switzerland , , , , –, , –,  SWOT analysis , – Syrquin, M. , , , ,  Taiwan –, –, , , –, –, ,  entry into high tech – exports –, – and globalization – import substitution – industrial structural transformation – liberalization and delayed adjustment – policy reform and start of export-led growth – secondary import substitution  services – slowdown and prospects  social transformation – tanneries –; see also leather Tanzania – lessons learned – productive heterogeneity and job creation of small firms  small business owners –, – tariffs , , , –, , , ,  tax incentives , , , ,  taxes , , , , , , ,  technical change, non-uniform – technical progress , , , , –,  uniform –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





technical training , , ,  technological capabilities , , , , , , ,  technological change , , , ,  and gender – technological innovation , , , , ,  technological progress , , –, , , –, ,  technological upgrading , , , –,  technologies –, , –, , –, –, , – bottleneck , , – communication , , , , –,  digital –,  technology, spillovers , , , –,  tensions , , , –,  political , , ,  termination fees , ,  terms of trade, agricultural – terrorist threats , – tertiary sector , –, –, , –, –,  textiles , , , –, –, –, –,  TFP, see total factor productivity Thailand –, , , –, , , ,  Third Industrial Revolution  Third Plenum , – threshold distances , ,  tobacco , , , , , , ,  Tocqueville, Alexis de – Togo , –,  total employment , , , , , , , – total factor productivity , , , , , –, , – growth , – tourism , , –,  township and village enterprises, see TVEs tradables , , –,  trade –, , , , –, –, –,  barriers , , , , 

costs , , , , –, , –, – flows ,  foreign , ,  free , , , , , ,  global , , – growth from  – international , , , –, –, , ,  intra-industry , – liberalization , , , , , , ,  liberalized ,  North–South – openness – services , ,  and structural change – surpluses , , ,  two globalizations – world , –, , , –, , , – traditional industries , , , ,  training , , , , , , , – technical , , ,  transaction costs , , , , , ,  transformations agricultural , , , , ,  economic –, , , , , , , – industrial ,  processes , , , , , , ,  productive –,  structural, see structural transformation transition –, –, , –, , , –, – demographic , , ,  economies , ,  transmigration, structural , ,  transparency , , – transport –, , , , , –, –, – air  equipment , , –, , – infrastructure , , , ,  services , , , , 

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi

 transportation costs , –, , ,  Tunisia  Turkey ,  TVEs (township and village enterprises) –,  Uganda – UNCTAD (United Nations Conference on Trade and Development) , , , , , , ,  undervaluation , – unemployment , , –, ,  high ,  persistent  rates , ,  structural  uniform technical progress – United Arab Emirates , –,  United Kingdom , , –, –, –, –, –, – United Nations , , , – United Nations Conference on Trade and Development, see UNCTAD United States –, –, –, –, , , , – as leader – markets , , ,  universities , , , , , –, – unpaid work  unskilled labour , , , ,  upgrading –, , , , , , –,  competitiveness –, ,  economic –,  functional ,  industrial , , –, –, , , ,  social –,  structural ,  technological , , , –,  urban areas –, , , , , , , – urban households – urban population –, , , ,  urban sector –, , ,  urbanization , , , , –, , –, 



Uruguay , , , , , –, ,  pension funds ,  Uruguay Round  utilities –, , , , , –, –, – utility functions , –, , –, , –, ,  value added, manufacturing , , , –, –, , ,  value chains , , –, , , –, –,  variables control , ,  dependent –, , , ,  explanatory , –,  independent ,  outcome ,  qualitative , ,  quantitative ,  vegetables , , , , ,  Venables, A.J. –, , ,  Venezuela –, , , , ,  vertical fragmentation ,  vertical production linkages , , , , –, –,  Viet Nam , , , , , , –,  backward and forward linkage multipliers – decomposition of SAM multipliers – decomposition of structural change – head-line ratios – Virginia ,  vocational education ,  volatility , , , , , –, ,  price – vulnerabilities , , , ,  Wacziarg, R. , , , ,  WAEMU (West African Economic and Monetary Union) , – .growth of economies – measures of structural transformation –

OUP CORRECTED PROOF – FINAL, 27/12/2018, SPi





WAEMU (West African Economic and Monetary Union) (cont.) structural transformations of economies – SWOT analysis and recommendations for removal of obstacles to structural transformation – wage costs  wages , , , , , –, –,  levels –, , ,  lower , , , , ,  minimum , ,  rates , , , , , , –,  WAMU (West African Monetary Union)  Washington Consensus , , , ,  WB, see World Bank wearing apparel industries , , –, –, – welfare , , , , , , ,  Wenzhou , , – West African Economic and Monetary Union, see WAEMU West African Monetary Union (WAMU)  Western Europe , , , , , , , ; see also European Union

winners , , , , , , ,  women , , , , –,  wood products , , , , ,  workers –, –, –, –, –, –, –,  displaced – workforce , , , –, , , –,  working age, population , , ,  working capital , –, ,  working population , –,  World Bank (WB) –, , , –, –, –, –, – world food crisis , , – world trade , –, , , –, , , – World Trade Organization, see WTO WorldBase , – Worldwide Governance Indicators , ,  WTO (World Trade Organization) , , , , –, , , – Yemen 