Globalization and Sustainable Growth in China [1st ed.] 9789811598241, 9789811598258

Citation preview

Yanqing Jiang Yuwen Dai

Globalization and Sustainable Growth in China

Globalization and Sustainable Growth in China

Yanqing Jiang · Yuwen Dai

Globalization and Sustainable Growth in China

Yanqing Jiang School of Economics and Finance Shanghai International Studies University Shanghai, China

Yuwen Dai School of Economics and Finance Shanghai International Studies University Shanghai, China

ISBN 978-981-15-9824-1 ISBN 978-981-15-9825-8 (eBook) https://doi.org/10.1007/978-981-15-9825-8 © Springer Nature Singapore Pte Ltd. 2020 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, expressed or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Contents

1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

1

2 Technology and China’s Growth Trend . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 An Overview: Technology from a Development Perspective . . . . . . 2.3 Measuring Technology at the Macro Level . . . . . . . . . . . . . . . . . . . . . 2.4 Technology, Production and Socio-Economic Indicators . . . . . . . . . . 2.5 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

5 5 5 7 12 16 17

3 Spatial Effects of FDI and Growth in China . . . . . . . . . . . . . . . . . . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.3 The Basic Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4 The Variables and Data . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.7 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

19 19 20 23 25 28 32 33 34

4 Interregional Disparity in Productivity Growth . . . . . . . . . . . . . . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 The Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.3 Regression Results Without Human Capital . . . . . . . . . . . . . . . . . . . . 4.4 Including Human Capital . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.5 The Province Heterogeneity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

37 37 38 40 42 45 49 50

5 Trade Imbalance and Protectionist Policy on Imported Intermediate Inputs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51 51

v

vi

Contents

5.2 Theoretical Development . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.1 Exogenous Growth Models . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.2 Endogenous Growth Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 “Non-scale” Growth Models . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.3 Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Macroeconomic Equilibrium . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.1 Balanced Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.2 Consumption Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.3 Capital Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.4 Dynamic System . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.5 Non-scale Growth . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.6 Accumulation of Foreign Assets . . . . . . . . . . . . . . . . . . . . . . . 5.5 Shock Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Steady-State Changes . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.2 Impact Effects . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Conclusion and Policy Implication . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

51 52 52 54 61 65 65 66 67 67 70 71 72 72 73 74 75

6 When the US Sneezes, Will China Catch a Cold? . . . . . . . . . . . . . . . . . . 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Measurement of Business Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3 Synchronization of Business Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.1 Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.2 Single Synchronization Tests . . . . . . . . . . . . . . . . . . . . . . . . . . 6.3.3 Single Synchronization Tests—Extension . . . . . . . . . . . . . . . 6.3.4 Joint Synchronization Tests . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

77 77 78 83 83 85 86 88 89 90

7 Growth Cycles in the BRICS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 91 7.2 Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 92 7.2.1 Measurement of Growth Cycles . . . . . . . . . . . . . . . . . . . . . . . . 93 7.2.2 Correlation of Growth Cycles . . . . . . . . . . . . . . . . . . . . . . . . . . 94 7.3 Extension . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 101 7.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Reference . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 8 China’s Growth Targeting and Policy Implication for LDCs . . . . . . . . 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Background: Rule-Based Monetary Policy . . . . . . . . . . . . . . . . . . . . . 8.3 Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.4 Conclusion and Policy Implication . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

109 109 110 113 119 120

About the Authors

Yanqing Jiang, Ph.D. is a Professor of Economics with School of Economics and Finance at Shanghai International Studies University (SISU). Dr. Jiang’s research has a central focus on China’s opening up, growth, and development. He started his research in this area in 2004, when he was affiliated to the Hanken School of Economics and the Helsinki Center of Economic Research as a doctoral researcher in Helsinki, Finland. Dr. Jiang’s publications in this area include more than 50 published journal articles (all in English) and 10 published books (all in English) on China’s opening up, growth, and development. Yuwen Dai, Ph.D. CFA is an Associate Professor of Economics with School of Economics and Finance at Shanghai International Studies University (SISU). Dr. Dai’s research interests are in the fields of macroeconomics and international economics. She received her Ph.D. in Economics degree from the University of York. Prior to joining SISU, she had extensive teaching and research experience in Australia, United States, United Kingdom, and China. She has also worked as a consultant for the International Monetary Fund (IMF) and the Asian Development Bank (ADB). Dr. Dai is a CFA charterholder.

vii

Chapter 1

Introduction

Thanks to the market-oriented policy reforms that were initiated in the late 1970s, China has achieved spectacular growth over the past several decades, and its economic structure has experienced great transformation. In the meantime, in the general context of globalization, the whole country has also gradually opened up to foreign trade and foreign direct investment, transforming itself from a virtually completely closed economy into a major trading nation and the largest developing-country destination for foreign direct investment in the world. Uneven growth and development is one of the various major problems in the process of China’s industrial evolution under the new trends of globalization. Substantial disparities across different regions, especially the gaps in industrial development, and hence in incomes and living standards between coastal and inland areas, have been one of the most prominent features in post-reform China. In the next three chapters of this book, we investigate the potential and actual mechanisms and channels through which globalization, especially openness to foreign trade and foreign direct investment, affects industrial development and growth disparities in China. The results provide the readers with new facts and new findings that shed light on their understanding of important issues such as how trade openness and inflows of foreign direct investment shape the path of China’s industrial development and the evolution of the country’s growth disparities. Dr. Jiang wrote Chapter 2, Chapter 3, and Chapter 4. Chapter 2 is on technology and China’s growth trend. In this chapter, we conduct statistical analysis exploring the interrelationships between technology, production and various socio-economic variables in the context of China’s regional macroeconomy. We focus on empirically examining how technology at the macro level is interrelated to production, output, and various macro-level socio-economic indicators. Our statistical analysis shows that a higher level of technology is always associated with socio-economically desirable outcomes on the various chosen macro-level indicators. However, the causality may run bi-directionally. Our empirical results also show that even after the partial effects of all the other macro-level variables are netted out, that is, even when the indirect effects of technology are not considered, © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_1

1

2

1 Introduction

the remaining direct effect of technology is still shown to play a very crucial role in determining labor productivity in the production process at the macro level. Chapter 3 is on spatial effects of FDI and growth in China. In this chapter, we examine how the spatial distribution of foreign direct investment (FDI) affects productivity growth in less developed regions of China. A systematic theoretical framework is set up concerning the potential effects of FDI capital via different channels. Following the theoretical framework, this study presents empirical evidence showing that FDI spillovers from the developed regions play a crucial role in enhancing productivity growth in the less developed regions. Empirical results also show that local FDI spillovers in less developed regions promote local productivity growth. In addition, FDI spillovers from less developed regions are not shown to have a significant effect on productivity growth in other less developed regions. Chapter 4 is on interregional disparity in productivity growth. In this chapter, we present an empirical analysis to improve our understanding of the catch-up and convergence tendencies of total factor productivity (TFP) growth across the Chinese provinces. After controlling for the province heterogeneity, our regression results show that the Chinese provinces exhibit significant conditional convergence in TFP growth over the sample period. This indicates that province-specific factors play an important role in determining provincial TFP growth. Economic policies conducive to faster TFP growth should thus be directed to the relevant factors underlying the province heterogeneity. This chapter suggests that openness to international economic activities and human capital accumulation are two important factors that promote TFP growth, both of which rely on a salutary social infrastructure. The current wave of globalization has encouraged economic growth in the world economy and affected all sides of international economic involvement over the last few decades. Moving on from the examination of China’s economic growth, we next study China and the world economy in the next four chapters. Dr. Dai wrote Chapter 5, Chapter 6, Chapter 7, and Chapter 8. In Chapter 5, we join the line of research on the “trade war” between China and the US. In this chapter, we analyze changes in trade balance in response to protectionist policy on imported intermediate inputs. Our analysis is conducted in a non-scale growth model, which features endogenous labor supply, leisure choice, and an imported intermediate input in the domestic production. From our model results, we find that an increase in the price of the imported intermediate input will lead to lower productivity and lower output, which in turn will have an adverse effect on the trade balance. But the reduced consumption following the import tariff will have a positive effect on the trade balance. The net impact on the trade balance depends on the relative forces of these offsetting factors. Hence, trade protectionism on imported intermediate inputs will not necessarily reduce the trade deficit in the domestic economy. The policy implication is that President Trump’s trade protectionist policy on imported intermediate inputs from China will not work. Chapter 6 continues our study on China and the US, with a focus on their business cycles. In this chapter, we examine the synchronization of business cycles between China and the United States for the past few decades. To that end, we measure their

1 Introduction

3

business cycles and find the turning points in their cycles. Under a GMM analytical framework, we conduct both single synchronization tests and joint synchronization tests in the growth cycles and acceleration cycles between China and the US. From our synchronization tests, we find strong evidence of positive correlation in the growth cycles between China and the US, with spillover effects on each other. In addition, we find evidence of negative correlation in the acceleration cycles between China and the US across time, which implies that when the rate of US economic growth begins to slow down, the Chinese economy will be resilient. So when the US sneezes, China will not catch a cold. In Chapter 7, we apply the analytical framework as in Chapter 6 and examine growth cycles in the BRICS. It has been over a decade since the cooperation within the BRICS was formalized. In this chapter, we attempt to examine the behavior and the correlation of growth cycles among the BRICS. With this objective, we measure their growth cycles, locate the turning points in their cycles, and conduct contemporaneous and cross-time correlation tests in their growth cycles. From our empirical results, we find that the growth cycles of Brazil, Russia, and South Africa depend highly on the growth in the Chinese economy and the demand from the Chinese market. This common dependence on China drives growth cycle correlation among Brazil, Russia, and South Africa over time. As the BRICS are major economic forces in their respective regions, the findings from our study have important policy implication for future cooperation among the BRICS. A new “conventional wisdom” on globalization is that trade and financial openness do not lead to higher economic growth by themselves, in the absence of institutional reforms. So globalization needs to be complemented by institutional reforms, in both developed and developing countries, so as to fully reap its potential benefits. Developing and emerging market economies have introduced major policy reforms and experienced significant economic growth. As a developing country, China has set an annual growth target, which has driven the nation’s economic policies for decades. Motivated by the Chinese experience, we develop an analytical framework in Chapter 8 to examine the case of economic growth targeting. In our model economy, key features for developing countries include the degree of commitment by the central bank, semi-dependence between the monetary authority and the fiscal authority, and the risk of sovereign default. From our analytical results, we find that under a GDP growth targeting regime, when the central bank is more committed to its growth target, the actual growth rate becomes close to the target level, the inflation rate increases moderately for those countries with flatter supply curves, and the social welfare improves. Moreover, higher growth rate is associated with lower tax levied by the fiscal authority, and it is inversely related to the government’s default probability. Policy implication could be drawn from our model results for developing countries when they evaluate the policy option of growth targeting, in order to achieve positive and sustainable economic growth.

Chapter 2

Technology and China’s Growth Trend

2.1 Introduction The increasing role of technology in influencing socio-economic development can be seen in many aspects of a modern society. However, to date, socio-economic development in China still relies much more heavily on the advantages of low cost labor than on the utilization of advanced technology. This study focuses on a statistical analysis of the interrelations between technology, production and various socioeconomic variables in the context of China’s regional economy at the macro level. After designing a comprehensive macro-level measure for technology, we empirically explore how technology is interrelated to production, output, and various macro-level socio-economic variables through our statistical analysis. This study is structured as follows. Section 2.2 provides an overview of the important role of technology from a development perspective. In Sect. 2.3, we construct a feasible measure for technology at the macroeconomic level. In Sect. 2.4, based on the measure of technology we have constructed in the preceding section, we perform our statistical analysis to empirically explore the interrelationships between technology, production and various socio-economic variables in the context of China’s regional macroeconomy. Finally, Sect. 2.5 concludes this study.

2.2 An Overview: Technology from a Development Perspective Any country or region that creates, distributes, and applies technology and knowledge to generate social wealth gives rise to a network society in which the ability to access technology and join learning-intensive relations determines the socio-economic position of individuals and production units, as well as the general performance of the macroeconomy (Clarke 2001; Laszlo and Laszlo 2007). Technological advancements in society call for the incessant need to create and manage intangible assets © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_2

5

6

2 Technology and China’s Growth Trend

(technology and knowledge) that do not depreciate but increase in value over time. This incessant need has fostered the formation of technology management, which provides necessary means to generate, distribute, and utilizes technology (knowledge) in ways that increase value to business activities and open new opportunities for enterprises (Clarke 2001). Technology-led socio-economic development is an extension of the technology management agenda (Laszlo and Laszlo 2007). As technology management deepens, it is developing into a strategic management approach that is applied to purposeful social-economic organizations in general (Carrillo 2002). Technology-led development implies the application of technology management to development issues, and is therefore a powerful strategy for countries or regions to seek economic prosperity and social harmony. On the one hand, technical knowledge is directly useful in bring about economic returns (Lever 2002). On the other hand, technology-led development also aims to foster the human capital (knowledge and skills) of people as a means for individual and social development (Ovalle et al. 2004). Obviously, these two main functions of technology-led development are mutually supportive. Accumulation of knowledge and human skills brings about more creativity, innovation, and entrepreneurship that promote socio-economic outcomes while economic prosperity and social harmony offer individuals more opportunities to accumulate knowledge and skills. The growing role played by technology in issues associated with socio-economic development is seen in many aspects of a modern society. Investment in human capital and technology, such as expenditures on education, job training, and research, now accounts for over ten percent of total gross domestic product (GDP) for OECD countries as a whole. One key feature of technology-led development is the accelerated incorporation of knowledge into productive activities related to both tangible goods and intangible services: knowledge become embedded in production processes in a wide variety of ways, ranging from on-the-job “learning by doing” by workers to formal procedures of investment in advanced technology, knowledge application, and labor training. Similarly, embodied knowledge is playing a more and more important role in areas such as business consulting, services education and training, and medical diagnosis and treatment. The increasing technology intensity in socioeconomic activities is determined not only by the increasing technology intensity in individual goods and services, but also by the growing importance of goods and services that rely on embodied knowledge. In this sense, technology-led development relies on significant, and constant changes in the industrial structure of developing countries. The shift of the shares from the traditional labor-based goods industries to the technology and human capital based industries with respect to the composition of employment or output is an important feature of technology-led development (Sheehan 1999). To date, socio-economic development in China is still based much more heavily on the advantages of low cost labor than on the utilization of advanced technology. To China, one major challenge brought about by the global waves of technology revolution is how to build an industrial structure that could make fuller use of technology developed both abroad and within the country. In addition, development in an

2.2 An Overview: Technology from a Development Perspective

7

open, unified, technology-intensive global economy generates continuous pressures on the increasing regional and sectoral disparities within China. Inflows of foreign direct investment (FDI) have been widely seen as an indispensable element in accounting for China’s socio-economic development. The colossal success of China’s economic reform and the growth of the country’s innovation capacity are partly attributable to the policy of attracting and retaining FDI (Buckley et al. 2002; Liu and Wang 2003). However, some argue that with the huge influx of FDI China has become overly dependent on foreign technology (Gilboy 2004). Seeing this, the Chinese government, on the one hand, began to regulate FDI, and on the other hand, enhanced support to innovation activities in domestic enterprises (Huang and Soete 2007). To finance innovation, China also strived to build a wellfunctioning financial system, and especially a venture capital system, to support technology-based enterprises. Apart from FDI, trade is another mechanism through which Chinese enterprises can tap into global technology. Imported high-tech products and capital goods embody a tremendous amount of technology. Also, foreign trade brings salutary spillovers to Chinese enterprises and the Chinese economy. However, while China has been quite active in introducing foreign technology embodied in tangible goods, it has been less active in importing disembodied technology (such as patents, which normally incurs royalties or licensing fees). Low imports of disembodied technology may hamper the efficient use of technological knowledge. Although importing hightech products and capital goods is an important way of acquiring foreign technology, the accompanying knowledge support (disembodied technology) should also be acquired in order to maximize the efficiency of technology investment. All in all, China needs to fully exploit the rapidly growing global technology to accelerate its own socio-economic development and facilitate its transition toward a technology-based, sustainably growing economy. The challenge lies in striking a sensible balance between technology creation and technology acquisition (Dahlman and Aubert 2001). This in turn implies adapting foreign technology to the context of China’s socio-economic development and meanwhile budgeting for research and development (R&D) activities to create technology within China.

2.3 Measuring Technology at the Macro Level We need to design variables to measure “technology” at the macroeconomic level. Technology in this sense is not directly measurable because it does not refer to any specific technology in any specific industrial field. Instead, it is expected to refer to the general technological level of productive activities in a given economy (i.e. a Chinese region in the current study). Therefore, we need to find proxy or indicator variables for technology at the macro level. The variables, according to their functions, can be roughly divided into two groups, one measuring technology creation and the other measuring technology acquisition. Considering severe data constraints, we will only focus on a limited scope of chosen variables.

8

2 Technology and China’s Growth Trend

The group of variables we choose to measure technology creation is associated with regional statistics on R&D activities and patents of industrial enterprises above a designated size. These statistics include annual regional data on per capita full-time equivalent of R&D personnel (in person-years per capita), per capita expenditure of funds on R&D (in 10 thousand RMB yuan per capita), the number of per capita R&D projects (in units per capita), the number of applications for patents per capita (in pieces per capita) and of which the number of applications for patented inventions per capita (in pieces per capita), as well as the number of patented inventions per capita currently in force (in pieces per capita). R&D personnel refers to the number of people engaged in research, management and supporting activities of R&D, including people in project teams and in the management of related activities of enterprises as well as supporting staff providing direct service to the research projects. This indicator reflects the size of personnel engaged in R&D activities with independent intellectual property.1 Full-time equivalent of R&D personnel refers to the sum of the full-time staff and the full-time equivalent of part-time staff converted by workload.2 Expenditure of funds on R&D refers to the real expenses of surveyed units on their own R&D activities (such as basic research, application study, test and development) including direct expenses on R&D activities, indirect expenses of management and services on R&D activities, expenditure on capital construction and material processing by others (but excluding the expenditure on production activities, return of loan, and fees transferred to cooperated and entrusted agencies on R&D activities). The number of R&D projects (or subjects) refers to the number of R&D projects (subjects) set up and implemented at the reference year, and the number of R&D projects (subjects) set up in former years and under implementation, including the projects (subjects) finished and failed at the reference year, excluding the projects (subjects) implemented by others through entrustment. Patented inventions refer to new technical proposals to the products or methods or their modifications that have been granted the patent right. This is a universal core indicator reflecting technologies with independent intellectual property.3 To get a feel for the cross-sectional ranges and variations of the above-mentioned R&D indicators, we present the basic descriptive statistics for these indicators across 31 provincial-level regions in China for an arbitrarily chosen year (which is 2012). The descriptive statistics are summarized in Tables 2.1 and 2.2. In the upper section of Table 2.1, we present the means, standard deviations, and minimal and maximal values of the six indicators. Even a cursory look at the statistical characteristics of the variables in this table would reveal the great interregional disparity in R&D activities across the different Chinese regions. In the upper section of Table 2.2, we present 1 Definitions of this and the following statistical indicators come from the corresponding explanatory

notes provided by the officially published China Statistical Yearbook (various issues). 2 For instance, if there are three full-time researchers and two part-time workers with 20% and 70%

of working hours respectively on R&D activities, the full-time equivalent is then 3 + 0.2 + 0.7 = 3.9 person-years. This is an internationally comparable indicator of R&D manpower input. 3 Patent (patent right) refers to the exclusive right of ownership by the inventors or designers for the creation or invention, given from the patent office after due process of assessment and approval in accordance with the patent law. Patents are granted for inventions, utility models and designs.

2.3 Measuring Technology at the Macro Level

9

Table 2.1 Descriptive statistics of the variables in 2012 Variable

Obs

V1

31

Mean

Std. Dev

V2

31

V3

31

1.97288

2.051914

0.078018

V4

31

3.183859

3.674693

0.058514

V5

31

1.165881

1.403655

0.055263

4.986227

V6

31

1.767576

2.259219

0.209604

7.861053

X1

31

0.314293

0.368994

0.038587

1.440863

X2

31

0.290096

0.324504

0.027101

1.35797

14.16566 482.0095

Min

13.18542

Max

0.252584

467.9452

17.26741

43.21506 1810.625 8.535542 12.4161

The variables are defined in the text

Table 2.2 Pair-wise correlations between the variables in 2012 V1

V2

V3

V4

V5

V6

X1

V1

1

V2

0.9444

1

V3

0.9389

0.9551

1

V4

0.9467

0.9035

0.9277

1

V5

0.8777

0.8789

0.8393

0.9215

1

V6

0.8847

0.8631

0.8082

0.9074

0.9831

1

X1

0.7302

0.737

0.6664

0.8114

0.9135

0.9148

1

X2

0.812

0.7972

0.7152

0.7978

0.8196

0.8767

0.8653

X2

1

All the variables in this table are defined the same as in Table 2.1

the pair-wise correlations between the six indicators (temporarily ignoring the two variables X 1 and X 2 in the table, which will be explained later). Unsurprisingly, the six variables (V 1 through V 6 ) all exhibit very high positive correlation between one another. V 1 is the regional full-time equivalent of R&D personnel per capita (in personyears per capita). V 2 is regional expenditure of funds on R&D per capita (in 10 thousand RMB yuan per capita). V 3 is the regional number of R&D projects per capita (in units per capita). V 4 is the regional number of applications for patents per capita (in pieces per capita). V 5 is the regional number of applications for patented inventions per capita (in pieces per capita). V 6 is the regional number of patented inventions currently in force per capita (in pieces per capita). X 1 is regional trade-toGDP ratio calculated based on “Total Value of Imports and Exports by Location of China’s Foreign Trade Managing Units”. X 2 is regional trade-to-GDP ratio calculated based on “Imports Value of Commodities by the Places of Their Destination and Exports Value of Commodities by the Places of Their Origin in China”. The aforementioned variables are indicator variables for regional efforts and performances of technology creation. We now turn to a discussion of measures

10

2 Technology and China’s Growth Trend

for technology acquisition. Direct data on technology acquisition that are complete and consistent are hard to come by. Therefore, we resort to proxy and indicator variables for representing regional technology acquisition. We argue that regional openness to foreign trade and FDI may serve as a good proxy or indicator variable for regional technology acquisition. On the one hand, regional exposure to international economic activities (such as foreign trade and FDI) undoubtedly facilitates technology spillovers from abroad, which is in itself a crucial influencing factor of regional technology acquisition. On the other hand, we should note that social infrastructure is of critical importance in determining both technology creation and technology acquisition. This social infrastructure is in turn determined by underlying institutions and government policies. A favorable social infrastructure gets the price system right so that individuals capture the social returns to their actions as private returns. The ideal measure of social infrastructure would thus quantify the wedge between the private return to productive activities and the social return to such activities (Hall and Jones 1999). Seeing in practice we do not have a usable quantification of wedges between private and social returns, we need to come up with a proxy or indicator variable to represent social infrastructure. We argue that openness to international activities is an acceptable proxy variable for social infrastructure. Policies and practices concerned with international activities such as foreign trade and FDI are sensitive and meaningful indexes of social infrastructure (Jiang 2011). This is to say that in addition to the function of openness as a proxy variable for technology acquisition (as the former facilitates technology spillovers from other countries), openness can also serve as a proxy variable for social infrastructure where the latter is in itself a proxy variable for both technology acquisition and technology creation. However, we do not need to distinguish between the function of openness as a general proxy variable for social infrastructure and that of openness as a factor that facilitates technology spillovers due to freer trade and FDI inflows. This is because later on we will combine the openness variable with the aforementioned R&D variables into a single, unified macro technology index representing the overall regional level of technology, which takes account of both technology creation and technology acquisition combined. Owing to data constraints, we opt for the use of the regional trade-to-GDP ratio as a measure of regional openness, which is constructed as the ratio of total value of regional foreign trade (regional exports plus regional imports, converted from USD to RMB yuan) to regional GDP in RMB yuan. According to the China Statistical Yearbook, there are two types of “total values of foreign trade”: one is “Total Value of Imports and Exports by Location of China’s Foreign Trade Managing Units” and the other is “Imports Value of Commodities by the Places of Their Destination and Exports Value of Commodities by the Places of Their Origin in China”. Therefore, we construct two trade-to-GDP ratios based on the two types of “total values of foreign trade” provided by the Yearbook. In the lower section of Table 2.1, we present the basic descriptive statistics of two versions of the regional openness variables for the year 2012 (which is the midpoint of the period 2004–2020 when China’s innovative development is the most active): X 1 is regional trade-to-GDP ratio calculated based on “Total Value of Imports and Exports by Location of China’s Foreign Trade Managing

2.3 Measuring Technology at the Macro Level

11

Units” while X 2 is regional trade-to-GDP ratio calculated based on “Imports Value of Commodities by the Places of Their Destination and Exports Value of Commodities by the Places of Their Origin in China”. A quick look at Table 2.1 reveals that there exists significant disparity in the level of regional openness across the 31 Chinese regions. In the lower section of Table 2.2, we present the correlation between X 1 and X 1 , as well as all the pair-wise correlations of each of X 1 and X 1 with any of the six “V variables” (V 1 through V 6 ), where we see that the values of the correlation coefficients are all positive and practically large. This observation, in a sense, implies substitutability between the two groups of variables and thus justifies our merging the two groups into a single, unified technology index, which we do next. To construct the regional technology index, we normalize the values of all the eight variables so that all the values for one arbitrarily chosen region, Beijing, are unity in the specific year 2012. The normalized values are then averaged across the eight variables to obtain a single value of the technology index for each region in any given year, using weights as follows: T = V˜1 /8 + V˜2 /8 + V˜3 /8 + V˜4 /32 + V˜5 /32 + V˜6 /16 + Z˜ 1 /4 + Z˜ 2 /4

(2.1)

where T refers to the technology index and the values of the right-hand side variables (i.e. the “tilde variables”) are the corresponding normalized values.4 Heuristically, we expect that technology and people’s standard of living should be positively related. The scatterplot in Fig. 2.1, for example, reveals the positive correlation between our computed regional technology index and regional per capita GDP (the most widely and frequently used measure for the standard of living). The sample points in Fig. 2.1 are the 31 Chinese provincial-level regions (single cross-sectional data for 2012). The horizontal axis depicts the regional technology 10

GDP per capita

8 6 4 2 0 0

0.2

0.4

0.6

0.8

1

1.2

1.4

1.6

1.8

Technology index

Fig. 2.1 Regional technology index and GDP per capita in 2012

4 Values

of the technology index can be computed in the same way for any year other than 2012, so that a panel data set can be obtained.

12

2 Technology and China’s Growth Trend

index and vertical axis depicts regional per capita GDP (in 10 thousand RMB yuan per person). To get down to more detailed discussions, in the subsequent sections we set out exploring the relationships between the regional technology index and various other regional socio-economic indicators in the hope of finding useful clues regarding how technology progress may promote socio-economic development. Our focus is on the mutual linkage between technology and the level of socio-economic development in the Chinese regions.

2.4 Technology, Production and Socio-Economic Indicators Socio-economic development takes goods production activities as its foundation. Economy-wide production can be captured by an aggregate production function, which can be mathematically represented as the following general form: Y = A · F(K , H, Z , L)

(2.2)

where F denotes the functional form, Y is aggregate output (value added), K and H are two representative production inputs, namely, physical and capital and human capital, Z is the amount of environmental input, L is the amount of raw-labor input, and A is the so-called total factor productivity, which is a black box capturing anything in the (aggregate) production process that cannot be accounted for by the explicitly included production inputs (in this case, K, H and Z). Assuming the aggregate production function follows constant returns to scale, it can then be rewritten in its intensive form in per capita terms as: y = A · f (k, h, z)

(2.3)

where f denotes the functional relationship in the intensive form. The lowercase letters y, k, h and z denote per capita levels of output, physical and human capital, and environmental input, respectively. According to neoclassical growth models, on a balanced growth path over time, the levels of k, h and z, now specifically denoted with an asterisk, k*, h* and z*, are in themselves dependent on the contemporaneous level of A. Therefore, per capita output, denoted y*, can be written as a function of A: y∗ = A · f [k ∗ (A, wk ), h ∗ (A, wh ), z ∗ (A, wz )]

(2.4)

where the w’s are vectors of other factors influencing the levels of k*, h* and z* (respectively) on the balanced growth path. Equation (2.4) shows that the (contemporaneous) level of total factor productivity affects per capita output through two different channels: one is a direct effect as A is a direct argument in the aggregate

2.4 Technology, Production and Socio-Economic Indicators

13

production function while the other is an indirect effect as A is one of the ultimate influencing factors behind the three direct arguments k, h and z. Our analysis in this section is motivated by Eq. (2.4). We go one step further to consider what key factors should be contained in the w vectors. Growth theories tell us that the investment rate (denoted s hereinafter) and the population growth rate (denoted n hereinafter) are two important elements in wk . Labor economists argue that educational attainment (denoted u hereinafter) is an important determinant of human capital intensity (per capita human capital) h, that is, u should be an important element contained in wh . Environmental economics tells us that per capita pollution emission (denoted p hereinafter) is a primary factor influencing the per capita level of environmental input used for production, that is, p is a key element in the vector wz . In addition, the industry mix (the proportions of different industries in the aggregate economy in terms of either the labor shares or output shares) can be an indispensable element in any of wk , wh and wz . These selected (and many other) elements of the w vectors are highly likely to be related to A in a way or another. If we equate total factor productivity to technology and take our constructed technology index (see the preceding section) as a practical measure of total factor productivity A, we are now interested in finding out how technology (as represented by our technology index) is related to the various socio-economic indicators: the investment rate s, the population growth rate n, the (average) educational attainment u (for which we will use the share of the people who have finished a three-year college or above in the population aged six or above as a practical measure), the per capita level of pollution emission p (for which we will use the per capita level of sulphur dioxide emission in waste gas as a feasible measure), as well as the industry mix (for which we will use the output share of agriculture, i.e. the value of agricultural output divided by the total value of overall aggregate output, hereinafter denoted m, as a practical measure). For the 31 Chinese regions in the year 2012, we collect data on the abovementioned socio-economic indicators. The regional investment rate s is calculated by dividing regional gross capital formation by regional output (measured by gross regional product (GRP) by the expenditure approach), where gross capital formation refers to the fixed assets acquired less disposals and the net value of inventory, including gross fixed capital formation and changes in inventories, and gross regional product by the expenditure approach refers to the method of measuring the final results of production activities of the region during a given period (usually one year) from the perspective of final uses, which includes final consumption expenditure, gross capital formation and net export of goods and services.5 The regional population growth rate n is calculated as the regional natural growth rate of population, which refers to the ratio of the natural increase in population (the number of births minus the number of deaths) in a given period (usually one year) to the average population (or mid-period population) of the same period. The regional 5 According

to the China Statistical Yearbook, gross fixed capital formation refers to the value of acquisitions less those disposals of fixed assets during a given period (usually one year). Gross fixed capital formation can be categorized into total tangible fixed capital formation and total intangible fixed capital formation.

14

2 Technology and China’s Growth Trend

average educational attainment u, as was mentioned already, is measured by the ratio of the number of the regional residents who have finished a three-year college or above to the regional population aged six or above. The regional pollution emission intensity (i.e. per capita pollution emission) p, as mentioned earlier, is measured by the per capita level of regional sulphur dioxide emission in waste gas (in tons per person). The regional industry mix, as mentioned earlier too, is measured by the share of regional agricultural output in total regional output (where regional output is measured by GRP). The descriptive statistics of the variables are summarized in Table 2.3. To see the interrelationships among the indicators, we further present the pair-wise correlations between the variables in Table 2.4. The results show that a higher level of technology tends to be associated with a lower investment rate, a lower population growth rate, a higher level of human capital intensity (or average educational attainment), a lower level of per capita pollution emission, and a lower share of the agricultural sector in GRP (which indicates a higher degree of industrialization), respectively. The signs of the correlations are in general consistent with the predictions economic theory could offer regarding the interrelations between these indicators. A thorough discussion on this point is beyond the scope of the current analysis because it would take us too far afield into some involved areas of macroeconomics. However, one thing we can see from the results in that a higher level of technology is always associated with socio-economically desirable outcomes on the other variables. For example, a lower Table 2.3 Descriptive statistics of the variables in 2012 Variable

Obs

Mean

Std. Dev

Min

Max

T

31

0.441647

0.438888

0.070686

1.650108

s

31

0.64115

0.160914

0.380286

1.01087

n

31

0.005452

0.002684

u

31

0.114405

0.064935

0.042453

0.373503

p

31

0.018299

0.013838

0.00136

0.062831

m

31

0.105077

0.051974

0.006333

0.249179

−0.00039

0.01084

The variables in this table are defined the same as in the text

Table 2.4 Pair-wise correlations between the variables in 2012 T

s

n

u

p

T

1

s

−0.5657

1

n

−0.2392

0.3457

1

u

0.6174

−0.3845

−0.3388

1

p

−0.3301

0.4192

0.0578

−0.1096

1

m

−0.7073

0.3517

0.3623

−0.6049

−0.0637

All the variables in this table are defined the same as in Table 2.3

m

1

2.4 Technology, Production and Socio-Economic Indicators

15

investment rate is a socio-economically desirable outcome of physical investment in the sense that advancements in technologies (and management skills) can be regarded as a substitute for physical capital accumulation in the production process so that a higher level of technology can afford a lower rate of physical investment, leading to a higher consumption rate that is socially desirable. Similarly, a higher level of technology is associated with a lower rate of population growth, which is also a desirable goal in the context of China’s growth as China is currently so much overpopulated. In addition, a higher level of technology is associated with a higher level of human capital intensity (or educational attainment), a lower level of per capita pollution emission, and a higher level of industrialization, which are all desirable goals of the society. However, the correlations shown in Table 2.4 do not provide us with adequate information to imply any unidirectional causal relationship within each pair. The causality is likely to be bi-directional, running both ways from one variable to the other in each pair. For example, technological improvement may be a cause that contributes to the realization of desirable outcomes concerning the other variables. Or conversely, desirable outcomes of the other variables can be causes that, separately or collectively, contribute to advances in technology. Having done the correlation exercise, we now turn to a regression analysis. We regress regional per capita output on the six variables in Table 2.3. The results of the regressions are summarized in Table 2.5, where we opt to include four variant versions of essentially the same regression, of which the last one has the best fit. Table 2.5 Results of the regression analysis Dependent variable: lny Variable

Coef. (Std. Variable Coef. (Std. Variable Coef. (Std. Variable Coef. (Std. Err.) Err.) Err.) Err.)

T

0.626* (0.141)

T

0.601* (0.143)

T

0.593* (0.159)

lnT

0.345* (0.060)

s

0.328 (0.275)

s

0.443 (0.271)

lns

0.274 (0.184)

lns

0.366* (0.151)

n

−39.527* (14.181)

n

−34.865* (14.775)

lnn

−0.105 (0.055)

lnn

−0.093* (0.044)

u

1.929* (0.697)

u

2.036* (0.716)

lnu

0.337* (0.138)

lnu

0.239* (0.114)

p

5.250 (2.986)

lnp

0.070 (0.052)

lnp

0.051 (0.050)

lnp

0.084 (0.042)

m

−0.229 (1.091)

m

−0.518 (1.084)

lnm

−0.034 (0.086)

lnm

−0.092 (0.060)

_cons

0.812 (0.273)

_cons

1.134 (0.304)

_cons

1.578 (0.602)

_cons

2.145 (0.512)

Obs

31

31

30

30

R2

0.844

0.837

0.832

0.889

16

2 Technology and China’s Growth Trend

All of the estimated partial effects have the expected signs, but some of them are statistically insignificant. In all of the four regressions, the estimated partial effect of technology on per capita output is statistically significant and economically large. We should note that, based on Eq. (2.4) earlier, this partial effect of technology, by construction, captures only the direct effect of technology on per capita output while the indirect effects of technology via the other variables are absorbed into the partial effects of the other explanatory variables in the regression specification. Nevertheless, these results undoubtedly suggest that the level of technology plays a crucial role in determining labor productivity (per capita output) in the production process (at the macro level). The dependent variable is log regional per capita output measured by log per capita GRP. Four versions of essentially the same regression are included in this table. Owing to one negative value of n in the sample, the last two regressions lose one sample point. An asterisk * indicates statistical significance at the 5% level.

2.5 Concluding Remarks The growing role that is being played by technology in shaping socio-economic development is seen in many aspects of a modern society. A society that creates, distributes, and applies technology to generate social wealth gives rise to a networked macro production procedure in which the ability to access and join technology and learning intensive relations determines the general performance of the macroeconomy. Technological advancements in society call for the continuous need to create and manage intangible assets that do not depreciate but increase in value over time. This continuous need, in turn, calls for necessary means to generate, distribute, and utilizes technology in ways that increase value to business activities and open new opportunities for enterprises. To date, socio-economic development in China is still based much more heavily on the advantages of low cost labor than on the utilization of advanced technology. To China, one major challenge is how to make fuller use of technology developed both abroad and domestically. Development in an open, unified, technology-intensive global economy generates continuous pressures on the increasing regional and sectoral disparities within China. This study focuses on a discussion of the interrelationships between technology, production and various socio-economic variables in the context of China’s regional macroeconomy. Our discussion is based on a statistical analysis. After designing a macro-level measure for technology, we move on to explore how this macrolevel technology is interrelated to production, output, and various macro-level socioeconomic variables. Through our statistical analysis, we find that a higher level of technology is always associated with socio-economically desirable outcomes on the various macro-level indicators. The causality may run bi-directionally: technological improvement may be a cause that contributes to the realization of desirable outcomes concerning the various macro-level indicators, or conversely, desirable outcomes of the macro-level indicators can be causes that, separately or collectively, contribute to

2.5 Concluding Remarks

17

advancements in technology. Finally, our regression analysis shows that even after the partial effects of all the other macro-level variables are netted out, that is, even when the indirect effects of technology are not considered, the remaining direct effect of technology is still shown to play a very crucial role in determining labor productivity in the production process at the macro level. The sample points are 31 Chinese provincial-level regions in the year 2012. The horizontal axis depicts the regional technology index and vertical axis depicts regional per capita GDP (measured in 10 thousand RMB yuan per person).

References Buckley, P.J., J. Clegg, and C. Wang. 2002. The Impact of Inward FDI on the Performance of Chinese Manufacturing Firms. Journal of International Business Studies 33 (4): 637–655. Carrillo, F.J. 2002. Capital Systems: Implications for a Global Knowledge Agenda. Journal of Knowledge Management 6 (4): 379–399. Clarke, Thomas .2001. The Knowledge Economy. Education + Training, 43(4/5): 189–196. Dahlman, Carl J., and Jean-Eric. Aubert. 2001. China and the Knowledge Economy: Seizing the 21st Century. The World Bank, Washington, D. C.: WBI Development Studies. Gilboy, George. 2004. The Myth behind China’s Miracle. Foreign Affairs 83 (4): 33–48. 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): 83–116. Huang, C. and L. Soete. 2007. The Global Challenges of the Knowledge Economy: China and the EU. UNU-MERIT Working Paper Series No. 2007–28, United Nations University, Maastricht Economic and Social Research and Training Centre on Innovation and Technology. Jiang, Yanqing. 2011. Economic Environment, Technology Diffusion, and Growth of Regional Total Factor Productivity in China. Journal of Chinese Economic and Business Studies 9 (2): 151–161. Laszlo, K.C., and A. Laszlo. 2007. Fostering a Sustainable Learning Society through Knowledge Based Development. Systems Research and Behavioral Science 24 (5): 493–503. Lever, W.F. 2002. Correlating the Knowledge-Base of Cities with Economic Growth. Urban Studies 39 (5/6): 859–870. Liu, Xiaohui, and Chenggang Wang. 2003. Does Foreign Direct Investment Facilitate Technological Progress? Evidence from Chinese Industries. Research Policy 32 (6): 945–953. Ovalle, M.D.R.G., J.A.A. Marquez, and S.D.M. Salomon. 2004. A Compilation of Resources on Knowledge Cities and Knowledge-Based Development. Journal of Knowledge Management 8 (5): 107–127. Sheehan, Peter. 1999. The Global Knowledge Economy: Challenges for China’s Development. CSES Working Paper No. 15, Victoria University, Melbourne, Australia.

Chapter 3

Spatial Effects of FDI and Growth in China

3.1 Introduction Many developing countries exhibit continued enthusiasm for attracting foreign direct investment (FDI) from abroad. For China, the spectacular economic take-off is closely dependent on the country’s opening up and increasing inflows of FDI.1 Before 1978, there were virtually no foreign-owned firms operating in China. In 1978 China initiated its economic reform and embraced the open-door policy, and since then the country has received large inflows of FDI, shifting from restrictive policies to permissive policies in the early 1980s, then to policies encouraging FDI in general in the mid-1980s, and to policies encouraging more high-tech and more capital-intensive FDI projects in the mid-1990s (Fung et al. 2004). The take-off of FDI inflows in China occurred in 1993, since when China has the largest amount of FDI inflows among the developing countries. By 2004, China had accumulated 500 billion US dollars in FDI inflows, which largely occurred via solely foreign-owned enterprises, joint ventures, and cooperative enterprises. However, FDI inflows in China are highly unevenly distributed across different regions (Yin 2011; Zhu et al. 2008). One prominent feature of FDI spatial distribution is that the coastal regions have by far the larger share of total FDI, compared with China’s interior regions (Cheung and Lin 2004). This broad spatial pattern of FDI distribution has remained fairly stable over time, with the share of the coastal regions being as large as 85%. China’s remarkably high growth rates since the 1980s and the huge influx of FDI have stimulated a lot of discussion in recent literature. Many studies have highlighted the role of FDI in promoting China’s economic growth. However, while there has been an increasing body of literature, systematic treatments of the effects of FDI on 1 China’s

increasing inflows of FDI has been closely related to its opening up to foreign trade. Foreign-invested firms can account for up to 50% of China’s exports and 60% of its imports. See, for example, Whalley and Xin (2010) for a recent discussion.

© Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_3

19

20

3 Spatial Effects of FDI and Growth in China

China’s development are quite limited. Specifically, few studies have been devoted to analyzing the spatial spillover effects of FDI on China’s regional development and interregional inequality (Zhang 2006; Ouyang and Fu 2012).2 Analysis of the spatial pattern of FDI spillover effects contributes to a more comprehensive understanding of the role of FDI in China’s growth and development. For a big developing country like China, it is important to determine whether the spatial distribution of FDI follows a substitution or a complementary pattern across different regions. While striking a balance between fast growth and inequality reduction is crucial for policy stability, the desirability of a liberal FDI policy hinges on the spatial pattern of the spillover effects of concentrated FDI on FDI-scarce regions Therefore, analysis of the spatial pattern of FDI spillover effects improves our understanding of the tension between reduction of interregional inequality and overall economic growth in China. To close a gap in the literature, in the present study we investigate how the spatial distribution of FDI affects productivity growth in less developed regions of China. We set up a systematic theoretical framework to distinguish the potential effects of FDI through different channels. Under such a theoretical framework, empirical work is then carried out. Among other findings, our results show that FDI spillovers from the relatively more developed regions play an important role in promoting productivity growth in the less developed regions of China. This study is structured as follows. In Sect. 3.2, we provide a brief review of related literature. In Sect. 3.3, we set up the theoretical framework and empirical model, on which our subsequent regression analysis will be based. In Sect. 3.4, we discuss various issues concerning the sample, variables and data. In Sect. 3.5, we carry out our regression analysis and present our estimation results. In Sect. 3.6, we discuss further issues and provide robustness checks. Section 3.7 concludes.

3.2 Literature Review Empirical studies so far, at either country or firm level, usually support the view that FDI promotes economic growth by bring physical capital, advanced technology, as well as management expertise into the host country (Ouyang and Fu 2012). A representative country-level study, Borensztein et al. (1998), examines the impact of FDI on economic growth in a cross-country regression framework. The study shows that a strong positive interaction exists between FDI and human capital, while the same interaction is insignificant in the case of domestic investment. Thus according to Borensztein et al. (1998), FDI promotes economic growth only when the absorptive capacity of advanced technologies exceeds a certain threshold in the host economy. Likewise, another study, Kuo and Yang (2008), when investigating how and to what 2 Two

exceptions are Madariaga and Poncet (2007) and Ouyang and Fu (2012).

3.2 Literature Review

21

extent knowledge capital and technology spillovers contribute to regional growth in China, finds that human capital is an important aspect of the capacity to absorb spillovers from FDI, and that the level of a region’s absorptive ability plays a crucial role in absorbing external knowledge sources embodied in FDI and imports.3 Micro-level mechanisms of the growth effects of FDI are the focus of firm-level studies. Some of such studies often do not find a significant contribution of FDI to productivity growth of domestic firms in the same industry. For example, Aitken and Harrison (1999) find no evidence supporting the existence of technology spillovers from foreign firms to domestic ones. Djankov and Hoekman (2000) show that FDI has the predicted positive impact on productivity growth in recipient firms, but the magnitude of the spillover effect is quite small. Keller and Yeaple (2009), in contrast to earlier work, find that FDI leads to significant productivity gains in domestic firms.4 They provide a detailed account of why their results are different from those found in previous work. A considerable number of studies support the existence of backward and forward externalities, which occur as foreign firms spread advanced technology via buying intermediate goods from upstream domestic suppliers or selling intermediate goods to downstream domestic firms (Ouyang and Fu 2012). Representative studies include Javorcik (2004), Kugler (2006) and Liu (2008). Focusing on effects operating across industries, Javorcik (2004) presents evidence consistent with positive productivity spillovers from FDI taking place through interactions between foreign affiliates and local suppliers in upstream sectors. Kugler (2006), by examining both technological and linkage externalities, shows that FDI substitutes within-sector domestic investment but complements the latter across sectors, and therefore, the net effect on aggregate capital formation by host-country producers hinges on the interaction between linkages and spillovers. Liu (2008), using a large panel of Chinese manufacturing firms, finds that FDI lowers the short-term productivity level but enhances long-term productivity growth in domestic firms of the same industry. The study also finds that externalities through backward and forward linkages have similar effects on productivity in domestic firms. Some studies take account of the technology gap between domestic and foreign firms. Representative studies include Kokko (1996), Glass and Saggi (1998), Girma et al. (2001), Girma and Görg (2002), and Girma (2005). Kokko (1996) finds that the technology spillovers are not determined solely by foreign presence, but rather by the interactions between foreign and domestic firms. Glass and Saggi (1998) show how 3 Concerning

absorptive capability, some studies focus on other aspects of the economy. Alfaro et al. (2004), for example, examine the various links among FDI, financial markets, and economic growth, and show that financial market development determines the extent to which a country can absorb spillovers from FDI. Similarly, Durham (2004) shows that the effects of FDI are contingent on the absorptive capacity of the host country, which is closely related to financial and institutional development. 4 They show that FDI spillovers accounted for about 14% of productivity growth in U.S. firms during the period 1987–1996. They argue that their results are likely to generalize to other countries and periods. In addition, there is also evidence for imports-related spillovers, but it is weaker than for FDI.

22

3 Spatial Effects of FDI and Growth in China

the quality of technology transferred through FDI is linked, through the technology gap, to innovation and imitation when the absorptive capacity of less developed countries is limited. Girma et al. (2001) find that firms with low productivity relative to the sector average, or in low-skill, low foreign competition sectors gain less through FDI spillovers from foreign firms. Girma and Görg (2002) show evidence for a Ushaped relationship between domestic productivity growth and FDI interacted with absorptive capacity. Girma (2005) finds that there is a minimum absorptive capacity threshold level below which productivity spillovers from FDI are negligible or even negative.5 Many studies focus on spatial spillover effects of FDI. Girma and Wakelin (2002) find that domestic firms gain from the presence of foreign firms in the same sector and region, but “loose out” if the firms are located in a different region but in the same sector. Hale and Long (2006) find that FDI promotes the performance of private domestic firms but not that of state-owned enterprises in China. Girma and Gong (2008) examine whether state-owned enterprises in China have benefited from the technical and managerial skills possessed by multinational firms operating in the country, and find that the evidence for positive spillovers is not overwhelming. Madariaga and Poncet (2007) and Ouyang and Fu (2012) investigate potential mechanisms and channels through which FDI may promote regional growth in China. Madariaga and Poncet (2007) find that regional growth responds positively to FDI received locally as well as to FDI inflows into proximate cities. Ouyang and Fu (2012) focuses on coastal-inland FDI spillovers and highlights the exploration of related spillover channels. However, regional FDI studies often show that within the host country, the spatial concentration of FDI may aggravate unbalanced development, widening the income gap between developed and under-developed regions (Ouyang and Fu 2012). For example, Brun et al. (2002) find that coastal development benefits in an unequal way to inland regions. Fujita and Hu (2001) find that the spatial pattern of FDI has had a significant impact on the widening interregional income gap in China. Similarly, Zhang and Zhang (2003) show that even after netting out the effects of many other important factors, foreign trade and FDI are still two important factors contributing to the widening interregional inequality in China. Nunnenkamp and Stracke (2007) investigate the location choices of foreign investors and find that the concentration of FDI in a few relatively developed regions may prevent the effects of FDI from spreading to the less developed regions of the host country.

5 Regarding

firm-level absorptive capacity, some studies focus on other aspects of the firm for an explanation. For example, Kinoshita (2000) finds that R&D expenditure is another determinant of a firm’s ability to absorb spillovers from FDI. That is, the learning effect of R&D is far more important than the innovative effect in explaining the productivity growth of a firm.

3.3 The Basic Model

23

3.3 The Basic Model Based on existing literature, our present study empirically examines the role of the interregional pattern of FDI distribution in shaping productivity growth in less developed regions in China. Specifically, this study contributes to the literature by presenting empirical evidence supporting the existence of a significant spillover effect of FDI inflows in the more developed coastal regions on productivity growth of the less developed non-coastal regions in China. In this section, we build our basic theoretical model, on which subsequent empirical analysis will be based. We adopt a Cobb–Douglas aggregate production function of the form Yi (t) = Ai (t)K i (t)α Hi (t)1−α

(3.1)

where i indexes a less developed region (i.e. a non-coastal region) in China.6 Yi is regional output, Ai is the level of regional total factor productivity (TFP), K i is regional stock of physical capital (including domestic and foreign-invested physical capital together), and Hi is our measure of regional human capital stock.7 Let L i be the total number of regional workers and define per worker human capital stock as h i ≡ Hi /L i . Equation (3.1) can then be written intensively as yi (t) = Ai (t)ki (t)α h i (t)1−α

(3.2)

for which we define yi ≡ Yi /L i and ki ≡ K i /L i as per worker output and per worker physical capital stock, respectively. Equation (3.2) immediately leads to ln yi (t) = ln Ai (t) + α ln ki (t) + (1 − α) ln h i (t)

(3.3)

by which we can calculate the level of regional TFP, Ai , as a residual once data on yi , ki and h i are obtained. Following the central idea of Nelson and Phelps (1966), Aiyar and Feyrer (2002), Lucas (2009) and Jiang (2011), we hypothesize that TFP growth in a less developed Chinese region is positively related to the size of the gap between its actual TFP level and its potential (target) TFP level at any given point in time. Technically, we assume d ln Ai (t) = λ[ln Ai∗ (t) − ln Ai (t)] dt 6 In

(3.4)

this study, a less developed region (province) in China refers to a non-coastal region (province) in China. 7 See, for example, Hall and Jones (1999), for a justification for the way of incorporating human capital into the production function.

24

3 Spatial Effects of FDI and Growth in China

where A∗ (t) refers to the potential level of regional TFP at time t, and λ is the rate of convergence in regional TFP. To model the effects of FDI on regional productivity growth, we further assume that potential TFP is governed by Ai∗ (t) = X i f i (t)π Si (t)μ h i (t)ω T (t)

(3.5)

where X i encompasses a set of time-invariant, region-specific factors affecting potential TFP in region i, f i (t) measures the degree of direct exposure of region i to FDI at time t while Si (t) measures the degree of indirect exposure of region i to FDI at time t via other regions j in China ( j ∈ Z , where Z is some subset of all the other Chinese regions, to be defined later), h i (t) measures regional per worker human capital stock as defined earlier, and T (t) denotes the world frontier level of TFP, which is assumed to grow exogenously over time. Plugging (3.5) into (3.4), we obtain d ln Ai (t) + λ ln Ai (t) = λ[ln X i + π ln f i (t) + μ ln Si (t) + ω ln h i (t) + ln T (t)] dt (3.6) Multiplying (3.6) throughout by eλt gives 

 d ln Ai (t) + λ ln Ai (t) dt e dt t1  t2  t2 λt λe dt + π λeλt ln f i (t)dt = ln X i



t2

λt

t1 t2

 + μ  +

t1

λeλt ln Si (t)dt + ω

t1 t2



t2

λeλt ln h i (t)dt

t1

λeλt ln T (t)dt

(3.7)

t1

Assuming f i (t), Si (t) and h i (t) remain constant within the small interval [t1 , t2 ], we perform the integration in (3.7) and multiply throughout by e−λt2 to obtain the following ln Ai (t2 ) = e−λτ ln Ai (t1 ) + π(1 − e−λτ ) ln f i (t1 ) + μ(1 − e−λτ ) ln Si (t1 )  t2 λeλt ln T (t)dt + ω(1 − e−λτ ) ln h i (t1 ) + (1 − e−λτ ) ln X i + e−λt2 t1

(3.8) where τ ≡ t2 − t1 . Rearranging (3.8) gives ln Ai (t2 ) − ln Ai (t1 ) = −ρ ln Ai (t1 ) + πρ ln f i (t1 ) + μρ ln Si (t1 )

3.3 The Basic Model

25

+ ωρ ln h i (t1 ) + ρ ln X i + e−λt2



t2

λeλt ln T (t)dt

t1

(3.9) where ρ ≡ (1 − e−λτ ). Using (3.3), (3.9) implies ln yi (t2 ) − ln yi (t1 ) = α[ln ki (t2 ) − ln ki (t1 )] + (1 − α)[ln h i (t2 ) − ln h i (t1 )] − ρ ln yi (t1 ) + ρα ln ki (t1 ) + ρ(1 − α + ω) ln h i (t1 )  t2 λeλt ln T (t)dt + πρ ln f i (t1 ) + μρ ln Si (t1 ) + ρ ln X i + e−λt2

(3.10)

t1

Equation (3.10) can be rewritten as a panel data regression specification in its conventional notations as follows  ln yit = α ln kit + (1 − α) ln h it − ρ ln yit + ρα ln kit + ρ(1 − α + ω) ln h it + πρ ln f it + μρ ln Sit + xi + ηt + εit

(3.11)

where the sign  is associated with  t the difference between the levels at the two time points t2 and t1 . The term e−λt2 t12 λeλt ln T (t)dt is subsumed into the time intercept ηt . The time-invariant term ρ ln X i is written as the region heterogeneity xi . We add εit to the equation as the zero-mean idiosyncratic error term. Equation (3.11) constitutes the foundation for our empirical analysis in subsequent sections.

3.4 The Variables and Data Our sample consists of 18 provincial-level non-coastal Chinese regions over the period 1997–2012. These non-coastal regions are Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Guangxi, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia and Xinjiang.8 The coastal regions concerned in this study (a possible choice of the set Z ) are 10 provincial-level coastal Chinese regions, which are Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong and Guangdong.9 Before we obtain data on all the variables in (3.11), the two “FDI variables” in our regression specification, f it and Sit , should be properly defined. The direct exposure variable, f it , is defined in a quiet straightforward way as follows

8 Owing 9 One

to missing data, two inland regions, Tibet and Chongqing, are not included. coastal province, Hainan, is not included owing to missing data.

26

3 Spatial Effects of FDI and Growth in China

f it ≡

Fit Yit

(3.12)

where the numerator Fit denotes the capital stock of FDI of region i at time t. The indirect exposure variable, Sit , is defined in a more complicated way as follows ln Sit ≡





φ

ln wi j,t

j∈Z

F jt

 (3.13)

Yitσ

where F jt denotes the capital stock of FDI in region j ( j = i, j ∈ Z , where Z is to be defined later) at time t. The wi j,t stands for nonnegative weights that are specified a priori. These weights measure the relative importance of the regions j in the process of domestic spatial FDI spillovers for any non-coastal region i at time t. Rearranging (3.13) directly leads to ln Sit =



ln wi j,t + φ

j∈Z



lnF jt − σ



j∈Z

lnYit

(3.14)

j∈Z

Inserting (3.12) and (3.14) back into (3.11) yields  ln yit = α ln kit + (1 − α) ln h it − ρ ln yit + ρα ln kit + ρ(1 − α + ω) ln h it    + πρ ln(Fit /Yit ) + μρ ln wi j,t + μρφ ln F jt − μρσ ln Yit j∈Z

+ xi + ηt + εit

j∈Z

j∈Z

(3.15)

which can be estimated using a nonlinear least squares method once data on the relevant variables are obtained. We need to obtain data on the variables in (3.15), which are regional output Yit and regional per worker output yit , regional per worker physical capital stock kit , regional per worker human capital stock h it , and regional FDI capital stock Fit and F jt . Panels of the size of regional employment (i.e. the number of workers), regional nominal Gross Regional Product (GRP) and GRP indices are directly available from the official publications of the National Bureau of Statistics of China, so that values of Yit (i.e. real GRP) and yit (i.e. real per worker GRP) can be calculated. Data on domestic and foreign investments are also available from the various official publications of the National Bureau of Statistics of China. However, these publications do not directly record capital stock data. Therefore, we use the perpetual inventory method (PIM) to calculate the relevant levels of domestic and FDI capital stock. In doing so, we follow Zhang (2008) and adopt a universal annual depreciation rate of 9.6% for both domestic and FDI capital for all the regions throughout our sample period. By applying such a PIM procedure, data on kit , Fit and F jt can be obtained.

3.4 The Variables and Data

27

We follow the basic method of Hall and Jones (1999) in obtaining data on regional per worker human capital stock h it . We assume that h it is related to educational attainment by the function ln h it = ψ(E it ) with ψ(0) = 0, where E it is the average number of years of education attained by a worker in the regional labor force. The derivative dψ(E it )/d E it is the return to schooling estimated in a Mincerian wage regression (Mincer 1974). According to evidence of Psacharopoulos (1994) from many countries on return-to-schooling estimates, ψ(E it ) is piecewise linear, with the rate of return being 13.4%, 10.1% and 6.8% respectively for schooling of the first four years, the second four years, and that beyond the eighth year. Therefore, with this evidence, we can calculate the values of regional per worker human capital stock h it based on regional data on educational attainment, which are also found in official publications of the National Bureau of Statistics of China.10 The weights wi j,t in (3.13)–(3.15) indicate the relative importance of the regions j in the process of domestic interregional spillover of FDI for a less developed region i at time t. In this study, we use the values of a domestic market integration index as the weights wi j,t . The idea behind constructing the index is that the dispersion (across goods) of price differentials of identical goods between two regions can be an inverse indicator of the degree of market integration between the two regions (Parsley and Wei 2001). To construct the market integration index by this price-based method, we follow Sheng and Mao (2011) and let pitm and p mjt be the price of good m respectively in regions i and j at time t. Then we can define  Dimj,t

≡ ln

pitm m pi,t−1



 − ln

p mjt p mj,t−1

 (3.16)

which measures the difference in the percentage change of the price of good m between regions i and j during the time interval (t − 1, t). The dispersion of Dimj,t across a chosen set of goods can be an inverse indicator of the degree of market integration between regions i and j at time t.11 We calculate the variance of Dimj,t across all the chosen goods for each region pair (i, j) at time t. Any specific weight wi j,t in (3.13) can then be defined as wi j,t ≡ [var(Di j,t )]−1/2

(3.17)

where we denote the variance of Dimj,t as var(Di j,t ). We need to choose a specific set of goods to construct (3.16) and (3.17). Considering data availability, the types of goods we choose are: grain; oil and fat; meat, poultry and related processed products; eggs; fish and shellfish; vegetables; fresh and dried fruit; tobacco; liquor; garments, clothing fabric; footwear and hats; durable consumer goods; daily use household articles; and cosmetics. 10 For a recent discussion of human capital and economic growth in China, see, for example, Fleisher et al. (2010). 11 See, for example, Parsley and Wei (2001).

28

3 Spatial Effects of FDI and Growth in China

3.5 Empirical Results To implement our econometric exercise based on (3.15), we divide our entire sample period 1997–2012 into five three-year sub-periods, which are 1997–2000, 2000– 2003, 2003–2006, 2006–2009 and 2009–2012, respectively. Therefore in (3.15) the sign  now pertains to a time interval of three years. For example, when t in yit indexes the calendar year 1997,  ln yit then refers to growth during the time interval 1997–2000 (and so on). As we divide the whole sample length into five sub-periods, the time intercept ηt in (3.15) can then be practically replaced by four time dummy variables (each for one sub-period) plus a common intercept in our regression exercise. In addition, to take account of the unobserved region heterogeneity xi in (3.15), we include the full set of individual region dummies (i.e. 17 dummies for the 17 regions other than the arbitrarily chosen base region) in our regression. We apply a nonlinear least squares method to estimate the parameters α, ρ (hence λ), π , μ, ω, φ and σ in (3.15). The regression results are listed in Table 3.1. In this regression, we define that the set Z (the regions j) includes all the coastal and inland regions other than region i in China (27 regions altogether). In Table 3.1, all of the estimated parameters have the expected positive sign and are significant (at the one-percent level or at the usual five-percent level). The estimated coefficient on  ln kit , i.e. the estimate of the output elasticity of physical capital α, is 0.501. This estimate of α falls in the very vicinity of its traditionally accepted values in the case of China, which are exactly around 0.50–0.60 (Chow and Li 2002; Chow 2008; Zheng et al. 2008; Brandt and Zhu 2010). The estimated coefficient on ln yit , the parameter ρ, is 0.290, which is significantly lower than unity (so that e−λτ is properly defined, remembering ρ ≡ (1−e−λτ )). The estimate implies that the annual convergence rate λ is estimated to be 0.114 (where ρ ≡ (1 − e−λτ ), in which τ = 3 in the current case). This result shows that the non-coastal regions exhibit significant Table 3.1 Estimated Parameters Based on Eq. (3.15). (Regions j are 27 coastal and inland regions in China) Parameter

Estimate

Std. Err.

p-value

[95% Conf. Interval]

α

0.501

0.119

0.000

0.267

0.736

ρ

0.290

0.071

0.000

0.150

0.430

π

0.160

0.067

0.019

0.027

0.293

μ

0.053

0.009

0.000

0.035

0.071

ω

0.217

0.058

0.000

0.102

0.331

φ

0.535

0.260

0.043

0.020

1.050

σ

1.593

0.211

0.000

1.172

2.014

Implied λ: 0.114 The sample includes 18 non-coastal province-level regions in China. The sample period is 1997– 2012, divided into five equal-length sub-periods. The number of observations is 90. To save space, the estimated intercepts, i.e. the common intercept, the coefficients on the time dummies, and the coefficients on the region dummies, are not reported in the table

3.5 Empirical Results

29

conditional convergence in regional productivity over the sample period. Conditional convergence in TFP does not mean that all the regions concerned converge to a common TFP growth path, but that each region converges to its respective TFP growth path, which is captured by (3.5). Therefore, our result shows that each non-coastal region converges to its own target TFP level at an annual rate of 11.4%. The estimates of the parameters π , μ and ω, which are the elasticities of the potential TFP, Ai∗ (t), with respect to f i (t), Si (t) and h i (t), are 0.160, 0.053 and 0.217 respectively, all of which have the expected positive sign and are significant. These results imply that FDI contributes to regional output through three distinct channels. One channel is that FDI capital stock affects regional output as a direct, accumulable factor of production. In our current study, FDI capital is included as part of K i (t) in the production function in (3.1). Therefore, FDI contributes to regional output growth as an accumulable production input through the “ln kit ” and “ ln kit ” terms in our regression model in (3.15). The second channel is that, besides the direct, static effect of local FDI as an accumulable production input, local FDI also exerts a spillover effect on regional TFP growth (by affecting the potential level of regional TFP as indicated by (3.5)). This spillover effect of local FDI is captured by the “ f i (t)” term in (3.5) and the “ln f it ” term in (3.11). Our results (i.e. the estimated magnitudes of ρ and π ) in Table 3.1 show that the spillover effect of an increase in local FDI capital stock by, say, 10% would induce a faster local productivity growth by about 0.46% over a three-year interval, ceteris paribus (see (3.15)). Third, besides the two channels above, FDI in other regions (the regions j) also exerts a spatial spillover effect on regional TFP growth of a non-coastal region (also by affecting the potential level of regional TFP as indicated by (3.5)). This interregional spillover effect of FDI is captured in the “Si (t)” term in (3.5) and the “ln Sit ” term in (3.11). Our regression exercise thus provides a decomposition of the three channels of effects of FDI on output growth in non-coastal regions in China. From these results, we can also see that human capital contributes to regional output through two different channels. One is that human capital affects output directly as an accumulable factor of production. This direct, static effect of human capital can be seen from the production function in (3.1) or its intensive form in (3.2). The other effect of human capital, which is an indirect, dynamic effect on output growth, is realized through its impact on regional TFP growth (as captured by (3.4) and (3.5) combined).12 According to our regression results, a 10% increase in the initial level of regional per worker human capital would lead to an instantaneous increase in regional per worker output of 4.99% (see (3.2)). However, combining (3.2) and (3.5), we see that in the long run, after the indirect, dynamic effect of a 10% one-time shock in regional per worker human capital has fully exhibited itself, regional per worker output would then increase by a total of 7.16%, which is larger than the static, an instantaneous increase in regional per worker output of 4.99% mentioned above.

12 For

example, better-educated workers have a comparative advantage in implementing new technologies. See Benhabib and Spiegel (1994) and Prescott (1998).

30

3 Spatial Effects of FDI and Growth in China

The estimates of the parameters φ and σ also have the expected positive sign and are significant. The estimated magnitude of the former, combined with those of μ and ρ, shows that higher FDI capital stock levels in all the regions j (27 regions in the current case), say, by 10%, would, ceteris paribus, lead to faster growth of per worker output of an inland region (over a three-year interval) by about 2.2% (see (3.15)). The regression in Table 3.1 shows that the interregional spillover effect of FDI is an important channel through which the overall spatial distribution of FDI inflows in China affects productivity growth in the less developed regions of China. In order to further investigate the spatial characteristics of the interregional FDI spillover effect, we rerun regressions according to (3.15), redefining the set of spillover source regions, i.e. the set Z of all the regions j. By comparing estimation results obtained from variant regressions based on differentially defined sets of spillover source regions, we can hope to find out some important characteristics of the spatial pattern of the interregional FDI spillover effect on productivity growth in the less developed regions. The results of our second regression, which is also run based on (3.15), are contained in Table 3.2. In this regression, we define that the set Z (the spillover source regions j) includes only the coastal regions (10 of them altogether). In Table 3.2, again, all of the estimated parameters have the expected positive sign and are significant. The estimate of α, which is 0.482, is lower than that in Table 3.1, but is still very close to its empirically accepted values in the case of China. The estimate of ρ is 0.316, a bit higher than that in Table 3.1, resulting in a slightly higher implied value of the annual convergence rate λ, which is now 0.127. The estimates of π , μ and ω are 0.181, 0.064 and 0.223, all of which are slightly higher than those in Table 3.1, respectively. For φ and σ , the estimates are fairly close to their previous estimates in Table 3.1 too. The estimated magnitudes of ρ and π in Table 3.2 show Table 3.2 Estimated Parameters Based on Eq. (3.15) (Regions j are 10 coastal regions in China) Parameter

Estimate

Std. Err.

p-value

[95% Conf. Interval]

α

0.482 0.482

0.114

0.000

0.254

0.709

ρ

0.316

0.078

0.000

0.162

0.470

π

0.181

0.073

0.015

0.036

0.326

μ

0.064

0.011

0.000

0.043

0.086

ω

0.223

0.060

0.000

0.103

0.343

φ

0.557

0.274

0.046

0.011

1.103

σ

1.388

0.188

0.000

1.014

1.762

Implied λ: 0.127 The sample includes 18 non-coastal province-level regions in China. The sample period is 1997– 2012, divided into five equal-length sub-periods. The number of observations is 90. To save space, the estimated intercepts, i.e. the common intercept, the coefficients on the time dummies, and the coefficients on the region dummies, are not reported in the table

3.5 Empirical Results

31

Table 3.3 Estimated Parameters Based on Eq. (3.15) (Regions j are 17 inland regions in China) Parameter

Estimate

Std. Err.

p-value

α

0.514

0.115

0.000

[95% Conf. Interval] 0.285

0.743

ρ

0.285

0.073

0.000

0.140

0.430

π

0.177

0.068

0.011

0.041

0.313

μ

0.024

0.012

0.058

– 0.001

0.048

ω

0.209

0.055

0.000

0.100

0.318

φ

0.483

0.244

0.051

– 0.002

0.968

σ

1.646

0.213

0.000

1.222

2.070

Implied λ: 0.112 The sample includes 18 non-coastal province-level regions in China. The sample period is 1997– 2012, divided into five equal-length sub-periods. The number of observations is 90. To save space, the estimated intercepts, i.e. the common intercept, the coefficients on the time dummies, and the coefficients on the region dummies, are not reported in the table

that the spillover effect of an increase in local FDI capital stock by 10% would induce a faster local productivity growth by about 0.57% over a three-year interval, ceteris paribus (see (3.15) again). The estimated magnitudes of μ, ρ and φ together suggests that higher FDI capital stock levels in all the regions j (10 coastal regions now) by 10% would, holding other factors constant, lead to faster growth of per worker output of an inland region (over a three-year interval) by about 1.15% points. In sum, none of the results presented by Table 3.2 and their implications are significantly different from those given by Table 3.1. The results of our third regression based on (3.15) are summarized in Table 3.3. In this third regression, the spillover source regions (i.e. the set Z ) are defined, instead, as all the other non-coastal regions in China (17 regions altogether). In Table 3.3, all the estimates of the parameters have the expected positive sign and most of them are significant. The estimate of α is 0.514, very close to those values estimated from the previous two regressions. The estimate of ρ is 0.285, which is only slightly lower than those in Tables 3.1 and 3.2, resulting in an implied annual rate of convergence of 0.112. The estimates of π , μ and ω are 0.177, 0.024 and 0.209, of which the estimate of μ is much lower than those in Tables 3.1 and 3.2, and is not significantly greater than zero at the usual five-percent significance level. The estimates of φ and σ are not far from those in Tables 3.1 and 3.2, but that of φ is not significantly positive at the 5% level. Different from the earlier regressions, this regression does not produce significantly positive estimates of μ and φ, both of which are associated with the interregional spillover effect of FDI. In sum, seen from the results of the three regressions in Tables 3.1, 3.2, and 3.3, which use differentially defined sets of the spillover source regions j, one important characteristic of the spatial pattern of the interregional FDI spillover effect in China is that FDI capital in the coastal regions exerts a significant spillover effect on productivity growth of the interior regions, but FDI capital in other interior regions does not show a significant spillover effect on productivity growth of any interior region.

32

3 Spatial Effects of FDI and Growth in China

3.6 Robustness Checks The regression exercise in the preceding section is critically dependent on the nonlinear least squares method we have used, which is in turn applied based on the regression specification in (3.15). We are interested in seeing whether alternative regression methods would lead to essentially the same empirical results as those obtained above. Following the nonlinear least squares method, the three regressions in Tables 3.1, 3.2, and 3.3 generate estimates of the output elasticity of physical capital α that are very close to its empirically accepted values (around 0.50) for China. Therefore, we can now assume α = 0.50 a priori, and estimate the parameters involved in our model by applying various regression methods based on the following the specification  ln Ait = −ρ ln Ait + ρω ln h it + πρ ln(Fit /Yit ) + μρ + μρφ

 j∈Z

ln F jt − μρσ





ln wi j,t

j∈Z

ln Yit + xi + ηt + εit

(3.18)

j∈Z

which is derived directly from (3.15) by applying (3.3), where the values of TFP are calculated as a residual assuming α = 0.50 a priori. To check the robustness of our earlier empirical results, we run regressions based on (3.18) and compare the new results with our earlier ones. We perform a least squares dummy variables (LSDV) method to estimate the parameters according to (3.18).13 The LSDV regressions produce results that are fairly close to those summarized in Tables 3.1, 3.2, and 3.3. Next, we shift the assumed value of α from α = 0.50 to α = 0.45 and α = 0.55 (respectively) to see whether such a shift leads to regression results that are significantly different from our prior ones. By following the same LSDV procedure, we find that these changes in the assumed value of α do not alter our prior empirical results in any significant ways. Further, we apply an alternative regression method, which is an extended GMM procedure proposed by Blundell and Bond (2000), to see whether it generates results that deviates significantly from those obtained from the LSDV method used above.14 It turns out that the GMM regressions do not produce very precise estimates of the parameters. Many of the estimates are statistically insignificant, though most of them have the expected positive sign.15 13 As the model in (6.18) is dynamic in nature, the LSDV estimator is inconsistent when the asymp-

totic properties are considered in the cross-sectional direction. However, the LSDV method is considered acceptable in the current case. See, for example, Amemiya (1967), Chamberlain (1982), Islam (1995) and Yao and Zhang (2001). 14 This extended GMM, in which lagged first differences are also used as instruments for the levels, works better than the usual first-differenced GMM when the variables are highly persistent so that lagged levels are only weakly correlated with subsequent first differences. 15 Different variants of the GMM regressions are tried, depending on differently assumed endogeneity status of the explanatory variables.

3.6 Robustness Checks

33

One issue our models in (3.15) and (3.18) leave out is the potential interaction between FDI and human capital. The effects of FDI (local or in other regions) on local productivity growth may well be dependent on the level of local (per worker) human capital. To check the existence of the interactive effects of FDI and human capital, we run regressions based on the following specification  ln Ait = β1 ln Ait + β2 ln f it + β3 ln Sit + β4 ln h it + β5 ln Ait ln h it + β6 ln f it ln h it + β7 ln f it ln h it + xi + ηt + εit

(3.19)

in which the variable f it is constructed according to (3.12), Sit is constructed according to (3.13) with the estimated values of φ and σ in Table 3.2 inserted, and the values of Ait are calculated in the same way as above assuming α = 0.50. To take account of the potential interactive effects, we include various interaction terms associated with human capital in (3.19). LSDV and GMM regressions are run based on (3.19). Our results show that the effects of the interaction terms are insignificant, either separately or jointly. This finding supports the appropriateness of our regression specifications in (3.15) and (3.18). Another issue with our models in (3.15) and (3.18) is that the models do not explicitly consider regional institutional environment as a potential factor affecting regional productivity growth (and therefore do not explicitly include it as a control variable in the specifications). This regional institutional environment may include factors such as preferential policies from the central government, regulations of the local governments, and access to financial institutions, all of which are likely to influence local productivity growth. The time-constant region heterogeneity term xi in (3.15) and (3.18) captures only “permanent” regional features; it does not capture time-variant factors that constitute the changing regional institutional environment. We are thus interested in seeing whether our empirical results are sensitive to the inclusion of time-varying institutional factors in our regression model. Owing to the lack of direct measures, we use the relative size of the non-state-owned sector (in terms of both the labor share and the investment share) as a proxy variable for institution quality, hoping that this measure is able to reflect the level of flexibility of government policy and accessibility to financial institutions in any certain region.16 It turns out that the inclusion of the institution quality variable does not significantly affect our prior regression results.

3.7 Concluding Remarks Analysis of the spatial pattern of FDI spillover effects provides a better understanding of the role of FDI in China’s growth and development. Specifically, it is important 16 Data needed for computing the relative size of the non-state-owned sector can be found in official

publications of the National Bureau of Statistics of China.

34

3 Spatial Effects of FDI and Growth in China

to see whether the spatial distribution of FDI follows a substitution or a complementary pattern across different Chinese regions. The desirability of a liberal FDI policy is dependent on the spatial pattern of the spillover effects of concentrated FDI on FDI-scarce regions Therefore, analysis of the spatial pattern of FDI spillover effects improves our understanding of the tension between reduction of interregional inequality and overall economic growth in China. This study investigates how the spatial distribution of FDI affects productivity growth in less developed regions of China. In this study, we set up a systematic theoretical framework concerning the potential effects of FDI through different channels. Under such a theoretical framework, relevant empirical work is then carried out. The major contributions of this study are threefold. First, this study fills a lacuna in the literature by providing empirical evidence showing that the FDI spillover effect of the relatively more developed coastal regions plays a critical role in promoting productivity growth in the less developed interior regions of China. Our regression results suggest that, over a three-year interval, a 10% increase in FDI capital levels in the coastal regions would accelerate productivity growth in the interior regions by about 1.15%. Second, empirical results of this study confirm that local FDI inflows in less developed regions facilitate local productivity growth. It is estimated (in Table 3.2) that the spillover effect of an increase in local FDI capital stock by 10% would accelerate local productivity growth by about 0.57% over a three-year interval, ceteris paribus. Third, our empirical results show that FDI capital located in any less developed region does not produce a significant spillover effect on productivity growth in other less developed regions in China.

References Aitken, Brian J., and Ann E. Harrison. 1999. Do Domestic Firms Benefit from Direct Foreign Investment? Evidence from Venezuela. American Economic Review 89 (3): 605–618. Aiyar, Shekhar, and James Feyrer. 2002. A Contribution to the Empirics of Total Factor Productivity. Dartmouth College Working Paper No. 02-09. 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. Amemiya, Takeshi. 1967. A Note on the Estimation of Balestra-Nerlove Models. Technical Report No. 4, Institute for Mathematical Studies in Social Sciences, Stanford University. Benhabib, Jess, and Mark M. Spiegel. 1994. The Role of Human Capital in Economic Development: Evidence from Aggregate Cross-Country Data. Journal of Monetary Economics 34 (2): 143–173. Blundell, Richard, and Steve Bond. 2000. GMM Estimation with Persistent Panel Data: An Application to Production Functions. Econometric Reviews 19 (3): 321–340. Borensztein, Eduardo, Jose De Gregorio, and Jong-Wha Lee. 1998. How Does Foreign Direct Investment Affect Economic Growth? Journal of International Economics 45 (1): 115–135. Brandt, Loren, and Xiaodong Zhu. 2010. Accounting for China’s Growth. Working Papers tecipa394, Department of Economics, University of Toronto. Brun, Jean-Frana, Jean-Louis Combes, and Mary-Frana Renard. 2002. Are There Spillover Effects Between Coastal and Non-coastal Regions in China? China Economic Review 13 (2–3): 161–169.

References

35

Chamberlain, Gary. 1982. Multivariate Regression Models for Panel Data. Journal of Econometrics 38: 5–46. Cheung, K., and Ping Lin. 2004. Spillover Effects of FDI on Innovation in China: Evidence from the Provincial Data. China Economic Review 15 (1): 25–44. Chow, Gregory C. 2008. Another Look at the Rate of Increase in TFP in China. Journal of Chinese Economic and Business Studies 6 (2): 219–224. Chow, Gregory C., and Kui-Wai. Li. 2002. China’s Economic Growth: 1952–2010. Economic Development and Cultural Change 51 (1): 247–256. Djankov, Simeon, and Bernard Hoekman. 2000. Foreign Investment and Productivity Growth in Czech Enterprises. World Bank Economic Review 14 (1): 49–64. Durham, J. Benson. 2004. Absorptive Capacity and the Effects of Foreign Direct Investment and Equity Foreign Portfolio Investment on Economic Growth. European Economic Review 48 (2): 285–306. Fleisher, B., H. Li, and M.Q. Zhao. 2010. Human Capital, Economic Growth, and Regional Inequality in China. Journal of Development Economics 92 (2): 215–231. Fujita, Masahisa, and Hu. Dapeng. 2001. Regional Disparity in China 1985–1994: The Effects of Globalization and Economic Liberalization. Annals of Regional Science 35 (1): 3–37. Fung, K.C., H. Iizaka, and S.Y. Tong. 2004. FDI in China: Policy, Recent Trend and Impact. Global Economic Review 33 (2): 99–130. Girma, Sourafel. 2005. Absorptive Capacity and Productivity Spillovers from FDI: A Threshold Regression Analysis. Oxford Bulletin of Economics and Statistics 67 (3): 281–306. Girma, Sourafel, and Holger Görg. 2002. Foreign Direct Investment, Spillovers and Absorptive Capacity: Evidence from Quantile Regressions. GEP Research Paper 02/14, University of Nottingham. Girma, Sourafel, David Greenaway, and Katharine Wakelin. 2001. Who benefits from Foreign Direct Investment in the UK? Scottish Journal of Political Economy 48 (2): 119–133. Girma, Sourafel, and Yundan Gong. 2008. FDI, Linkages and the Efficiency of State-Owned Enterprises in China. Journal of Development Studies 44 (5): 728–749. Girma, Sourafel, and Katharine Wakelin. 2002. Are There Regional Spillovers from FDI in the UK?” GEP Research Paper, No. 16, University of Nottingham. Glass, Amy Jocelyn, and Kamal Saggi. 1998. International Technology Transfer and the Technology Gap. Journal of Development Economics 55 (2): 369–398. Hale, Galina, and Cheryl Long. 2006. FDI Spillovers and Firm Ownership in China: Labor Markets and Backward Linkages. Federal Reserve Bank of San Francisco working paper series No. 25. Hall, Robert E., and Charles I. Jones. 1999. Why do Some Countries Produce So Much More Output Per Worker Than Others? Quarterly Journal of Economics 114 (1): 83–116. Islam, Nazrul. 1995. Growth Empirics: A Panel Data Approach. Quarterly Journal of Economics 110: 1127–1170. Javorcik, Beata Smarzynska. 2004. Does Foreign Direct Investment Increase the Productivity Of Domestic Firms? In Search of Spillovers Through Backward Linkages. American Economic Review 94 (3): 605–627 Jiang, Yanqing. 2011. Understanding Openness and Productivity Growth in China: An Empirical Study of the Chinese Provinces. China Economic Review 22 (3): 290–298. Keller, Wolfgang, and Stephen R. Yeaple. 2009. Multinational Enterprises, International Trade, and Productivity Growth: Firm-Level Evidence from the United States. Review of Economics and Statistics 91 (4): 821–831. Kinoshita, Yuko. 2000. R&D and Technology Spillovers Through FDI: Innovation and Absorptive Capacity. IMF Working Paper, No. 349. Kokko, Ari. 1996. Productivity Spillovers from Competition Between Local Firms and Foreign Affiliates. Journal of International Development 8 (4): 517–530. Kugler, Maurice. 2006. Spillover from Foreign Direct Investment: Within or Between Industries. Journal of Development Economics 80 (2): 444–477.

36

3 Spatial Effects of FDI and Growth in China

Kuo, Chun-Chien, and Chih-Hai Yang. 2008. Knowledge Capital and Spillover on Regional Economic Growth: Evidence from China. China Economic Review 19 (4): 594–604. Liu, Zhiqiang. 2008. Foreign Direct Investment and Technology Spillovers: Theory and Evidence. Journal of Development Economics 85 (1–2): 176–193. Lucas, Robert E. 2009. Trade and the Diffusion of the Industrial Revolution. American Economic Journal: Macroeconomics 1 (1): 1–25. Madariaga, Nicole, and Sandra Poncet. 2007. FDI in Chinese Cities: Spillovers and Impact on Growth. The World Economy 30 (5): 837–862. Mincer, Jacob. 1974. Schooling, Experience, and Earnings. New York: Columbia University Press. Nelson, Richard, and Edmund Phelps. 1966. Investment in Humans, Technological Diffusion, and Economic Growth. American Economics Review 56: 69–75. Nunnenkamp, Peter, and Rudi Stracke. 2007. Foreign Direct Investment in Post-reform India: Likely to Work Wonders for Regional Development? Kiel Institute of World Economics Working Paper No. 1375. Ouyang, Puman, and Fu. Shihe. 2012. Economic Growth, Local Industrial Development and Interregional Spillovers from Foreign Direct Investment: Evidence from China. China Economic Review 23 (2): 445–460. Parsley, David C., and Shang-Jin Wei. 2001. Limiting Currency Volatility to Stimulate Goods Markets Integration: A Price Based Approach. NBER Working Paper No. 8468, National Bureau of Economic Research. Prescott, Edward C. 1998. Lawrence R. Klein Lecture 1997: Needed: A Theory of Total Factor Productivity. International Economic Review 39 (3): 525–551. Psacharopoulos, George. 1994. Returns to Investment in Education: A Global Update. World Development 22: 1325–1343. Sheng, Bin, and Qilin Mao. 2011. Trade Openness, Domestic Market Integration, and Provincial Economic Growth in China: 1985–2008. The Journal of World Economy, November 2011, 44–66. Whalley, John, and Xian Xin. 2010. China’s FDI and non-FDI Economies and the Sustainability of Future High Chinese Growth. China Economic Review 21 (1): 123–135. Yao, Shujie, and Zhang Zongyi. 2001. Regional Growth in China Under Economic Reforms. Journal of Development Studies 38 (2): 167–186. Yin, H. 2011. Characteristics of Inter-regional Income Disparities in China. Social Sciences in China 32 (3): 123–144. Zhang, Jun. 2008. Estimation of China’s Provincial Capital Stock (1952–2004) with Applications. Journal of Chinese Economic and Business Studies 6 (2): 177–196. Zhang, Kevin H. 2006. Foreign Direct Investment and Economic Growth in China: A Panel Data Study for 1992–2004. In Conference of “WTO, China and Asian Economies”. Beijing, China: University of International Business and Economics. Zhang, Xiaobo, and Kevin Zhang. 2003. How does Globalisation Affect Regional Inequality within a Developing Country? Evidence from China. Journal of Development Studies 39 (4): 47–67. Zheng, Jinghai, Angang Hu, and Arne Bigsten. 2009. Measuring Potential Output in a Rapidly Developing Economy: The Case of China in Comparison with the US and EU. Federal Reserve Bank of St. Louis Review, July/August 2009, 91 (4): 317–342. Zhu, S., M. Lai, and X. Fu. 2008. Spatial Characteristics and Dynamics of Provincial Total Factor Productivity in China. Journal of Chinese Economic and Business Studies 6 (2): 197–217.

Chapter 4

Interregional Disparity in Productivity Growth

4.1 Introduction China has been one of the fastest growing economies in the world in the past 35 years. However, different regions in China have markedly different growth rates, and as a result, show substantial disparities in per capita income levels.1 What are the key driving forces behind the uneven growth? Is it growth of total factor productivity (TFP hereinafter) or factor accumulation that has mainly shaped the uneven regional growth? A finding of an important role of TFP in promoting regional growth, for example, suggests that the mere channeling of capital investment into lagging provinces may not ensure their faster growth unless it is associated with TFP growth (Islam and Dai 2007). It is important for policymakers to gauge the relative contributions of capital accumulation and TFP growth to income growth as this information is useful in making necessary policies to counteract the rising trend of interregional disparity in China. The main objective of this chapter is to explore the characteristics of interregional TFP disparity in China and to provide an empirical analysis that will enrich our understanding of the catch-up and convergence processes of the Chinese regions regarding TFP. Our empirical analysis shows that province-specific factors play an important role in determining provincial TFP growth. After controlling for the effects of province-specific factors, our regression results show that the Chinese provinces exhibited significant conditional convergence in TFP growth. Economic policies beneficial to faster TFP growth should thus be directed to relevant factors underlying the province heterogeneity term in our regression equation. Our analysis also suggests that openness to international economic activities and human capital accumulation

1 Many

studies have explored various driving forces behind economic growth and interregional inequality in China. See, for example, Jian et al. (1996), DaCosta and Carroll (2001), Demurger (2001), Demurger et al. (2002), Huang et al. (2003), Zhang and Zhang (2003), Kanbur and Zhang (2005), Zhu et al. (2008), Jiang (2011), and Fleisher et al. (2010). © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_4

37

38

4 Interregional Disparity in Productivity Growth

are two important factors that promote TFP growth, both of which rely on a salutary social infrastructure. The rest of this chapter is organized as follows. In Sect. 4.2, we present the model and derive the baseline regression equation on which our later empirical analysis will be based. In Sect. 4.3, we present and analyze results from regressions based on a simplified version of our baseline regression equation with no human capital variables appearing as explanatory variables. In Sect. 4.4, we calculate regional per worker human capital stocks and run regressions controlling for human capital variables. Regression results are presented and analyzed accordingly. In Sect. 4.5 we provide an analysis of the province heterogeneity that affects provincial TFP growth. Section 4.6 contains a summary and conclusion.

4.2 The Model In this section we derive the baseline regression specification on which our subsequent empirical analysis will be based. To account for growth in per worker output, we assume a Cobb–Douglas aggregate production function with Hicks-neutral TFP. For province i at time t we have Yi (t) = Ai (t)K i (t)α Hi (t)β L i (t)1−α−β

(4.1)

where Y is total output, K is the stock of physical capital, H is the stock of human capital, L is the number of workers, and A is the Hicks-neutral TFP. The intensive form in per worker terms is then yi (t) = Ai (t)ki (t)α h i (t)β

(4.2)

where y ≡ Y/L, k ≡ K /L, and h ≡ H/L are output per worker, physical capital per worker, and human capital per worker respectively. It follows that for province i at time t growth in per worker output can be written as d ln yi (t) d ln ki (t) d ln h i (t) d ln Ai (t) =α +β + dt dt dt dt

(4.3)

We further assume that growth of TFP for province i at time t is determined by d ln Ai (t) = λ[ln A F (t) − ln Ai (t)] + ξi dt

(4.4)

A F (t) denotes China’s frontier level of TFP at time t. ξi denotes an unobserved time-constant province heterogeneity that affects TFP growth. Therefore, Eq. (4.4) describes the convergence tendency of the Chinese provinces in the process of TFP

4.2 The Model

39

growth, and the parameter λ measures the speed of (conditional) convergence of provincial TFP: the farther the provincial TFP lags behind the national frontier level of TFP at time t, the faster the provincial TFP tends to grow at time t. It then follows that  t2 λeλt ln A F (t)dt (4.5) ln Ai (t2 ) = e−λτ ln Ai (t1 ) + (1 − e−λτ )ξi /λ + e−λt2 t1

where τ = t2 − t1 . For a variable x, define x(t1 ) ≡ x(t2 ) − x(t1 ), then  ln y i (t1 ) = α ln k i (t1 ) + β ln h i (t1 ) +  ln Ai (t1 )

(4.6)

Inserting Eq. (4.5) into (4.6) gives us  ln y i (t1 ) = α ln k i (t1 ) + β ln h i (t1 ) − (1 − e−λτ ) ln Ai (t1 )  t2 −λτ −λt2 λeλt ln A F (t)dt + (1 − e )ξi /λ + e t1

= α ln k i (t1 ) + β ln h i (t1 ) − (1 − e−λτ )[ln yi (t1 ) − α ln ki (t1 ) − β ln h i (t1 )]  t2 −λτ −λt2 λeλt ln A F (t)dt (4.7) + (1 − e )ξi /λ + e t1

The second equality immediately follows the production function in (4.2). In traditional panel data notations, Eq. (4.7) can be rewritten as the following regression model: qit = αm it + βbit − ϕ(qit − αm it − βbit ) + ci + ηt + εit

(4.8)

where the time subscript t = 1, …, (10 − τ ). Letters q, m, and b denote logs of variables y, k and h respectively. α, β, and ϕ are parameters to be estimated with ϕ ≡ (1 − e−λτ ). ci is a time-constant province-specific latent variable, ηt is the time intercept, and εit is the zero-mean idiosyncratic error term. In the next two sections, we will use nonlinear least squares methods to estimate the values of the parameters α, β, and ϕ based on different versions of Eq. (4.8). It should be noted that an alternative approach to examining TFP growth and convergence of the Chinese provinces within the framework of our model can also be applied by running regressions that have the TFP growth  ln Ait directly as the left-hand side variable. This method involves explicitly calculating TFP as a residual from the production function, which requires an assumed value of the output elasticity of capital α a priori. In contrast, our current method based on Eq. (4.8) has the advantage of being able to circumvent the somewhat difficult assumptions to be

40

4 Interregional Disparity in Productivity Growth

made on the likely values of α. Instead, our current method provides an estimated value for the output elasticity of capital α as a byproduct through a regression based on Eq. (4.8).

4.3 Regression Results Without Human Capital Since in growth empirics calculating human capital has always been a weak spot fraught with measurement difficulties, in this section we first run regressions based on a simplified version of Eq. (4.8) without including human capital in the equation (i.e. assuming β = 0). We will postpone including human capital in the regression equation until the next section. Our sample is 29 Chinese province-level regions over the period 2003–2012.2 We obtain from the officially published Chinese Statistical Yearbooks series of nominal Gross Regional Product (GRP), GRP indices, and numbers of total employed people for each province, based on which we calculate the values of real GRP for each province. We calculate real per worker output as real GRP divided by the number of total employed people. We obtain annual data on real provincial capital stocks, by following the method of Zhang et al. (2007), and the provincial real per worker capital stocks can thus be calculated. We are first interested in whether there exists absolute convergence in TFP across the Chinese regions over the sample period. To check this, we run a single cross section regression based on Eq. (4.8) by setting the time interval τ = 9 (years). This is to say the explained variable on the left-hand side is now growth of real regional per worker output over the entire sample period 2003–2012. Major results from the nonlinear least squares method of this single cross section regression is summarized in the upper half of Table 4.1 (Reg. 1–1). The estimated value of the output elasticity of (physical) capital, α, is 0.447, with a 95% interval estimate of [0.273, 0.620]. The estimated value of ϕ is about –0.1, which is significantly negative at the 5% significance level. Therefore, we fail to find absolute convergence in TFP across the Chinese regions over 2003–2012. Instead, there exists absolute divergence in TFP across the Chinese regions over the sample period. Next, we are also interested in whether there exists ‘club convergence’ in TFP across the Chinese regions over the sample period. We divide the Mainland China into three zones: the eastern coastal zone, the central zone, and the western zone. The three big zones exhibit systematic differences not only in aspects such as climate and resource endowment, but also in aspects such as culture, policy and exposure to foreign trade and foreign direct investment. In the latter half of Table 4.1 (Reg. 1–2), we present results of a regression that includes two zone dummy variables, E 2 This

time period is chosen because this is the period of time that economic convergence across the Chinese regions shows interesting features. These regions include provinces, ethnic minority autonomous regions, and province-level municipalities, but for convenience we call all of them “provinces”. Owing to missing data municipality Chongqing and province Hainan are not included in our sample.

4.3 Regression Results Without Human Capital

41

Table 4.1 Single cross sectional regressions τ = 9 (Obs: 29)

Reg. 1–1

Parameter

Estimate

Std. Err.

α

0.447

0.084

0.273

0.620

ϕ

– 0.099

0.040

– 0.181

– 0.018

τ = 9 (Obs: 29)

Reg. 1–2

Parameter

R2 = 0.5700 95% conf. interval

R2 = 0.5799

Estimate

Std. Err.

α

0.455

0.089

95% conf. interval 0.271

0.640

ϕ

– 0.073

0.056

– 0.189

0.043

Coef. on E

0.022

0.051

– 0.083

0.127

Coef. on W

– 0.017

0.041

– 0.102

0.067

(for ‘east’) and W (for ‘west’). E = 1 whenever the region is located in the eastern coastal zone and E = 0 otherwise, and W = 1 whenever the region belongs to the western zone and W = 0 otherwise. The estimated value of α from this regression is 0.455, with a 95% interval estimate of [0.271, 0.640], only very slightly different from its counterpart in Reg. 1–1. The estimated value of ϕ is about –0.07, slightly higher than that in the previous regression and not significantly negative. Therefore, we fail to detect any ‘club convergence’ in TFP across the Chinese regions over the sample period. The estimated coefficients on the two zone dummies both have the expected sign, but neither is statistically significant. To check the robustness of the results above, we now run pooled cross section regressions, each time setting the time interval τ = 1, 3, and 6. The regression results are summarized in Table 4.2. Regressions on the left-hand side of Table 4.2 do not include the two zone dummies as explanatory variables while, as a comparison, regressions on the right-hand side of Table 4.2 control for the effects of the zone dummies. All regressions in Table 4.2 do not control for the effect of the province heterogeneity (for the time being), but they include (a proper number of) time dummy variables to take account of the time-varying intercept in Eq. (4.8). There are three major findings from the results in Table 4.2. First, for the regressions that include the zone dummies, the estimated coefficients on the zone dummies have the expected sign but are practically small and never statistically significant. Nor does the inclusion of the zone dummies alter the estimates of α and ϕ in any significant ways. The point and interval estimates of α and ϕ change only very slightly in response to the inclusion of the zone dummies. Second, the significantly negative values of ϕˆ throughout Table 4.2 suggest absolute divergence in TFP across the Chinese regions over the sample period. Third, the point estimates of α decreases with τ going from 1 to 3 to 6. However, these estimated values of α do not differ very much from those in Table 4.1. Next, we are going to see how regression results will change if we now control for the effects of the province heterogeneity. We include province dummy variables for the different provinces to take account of the latent component ci in Eq. (4.8).

42

4 Interregional Disparity in Productivity Growth

Table 4.2 Pooled cross sectional regressions τ = 1 (Obs: 261) Reg. 2–1

R2 = 0.9621

Reg. 2–4

R2 = 0.9624

Parameter

Estimate

95% conf. interval

Estimate

95% conf. interval

α

0.527

0.469

0.585

0.531

0.472

0.589

ϕ

−0.012

−0.017

−0.006

−0.009

−0.016

−0.002

Coef. on E

–

–

0.002

−0.005

0.009

Coef. on W

–

–

–

−0.002

−0.008

0.004

Reg. 2–2

R2 = 0.5494

Reg. 2–5

R2 = 0.5527

Parameter

Estimate

95% conf. interval

Estimate

95% conf. interval

α

0.436

0.370

0.501

ϕ

−0.037

−0.050

−0.023

Coef. on E

–

–

Coef. on W

–

–

Reg. 2–3

R2 = 0.5155

Parameter

Estimate

95% conf. interval

α

0.396

0.307

0.485

ϕ

−0.074

−0.102

−0.046

Coef. on E

–

–

–

Coef. on W

–

–

–

τ = 3 (Obs: 203)

0.440

0.374

0.506

−0.029

−0.048

−0.010

–

0.008

−0.011

0.026

–

−0.004

−0.019

0.011

Reg. 2–6

R2 = 0.5194

Estimate

95% conf. interval

τ = 6 (Obs: 116)

0.401

0.310

0.492

−0.062

−0.101

−0.024

0.011

−0.026

0.049

−0.007

−0.036

0.023

Regression results are summarized in Table 4.3. The estimated values of ϕ are now all significantly positive. This result suggests that once the province-specific effects are controlled for, the provinces show conditional convergence in TFP over the sample period. The estimated values of the output elasticity of capital, α, are somewhat higher than those in Table 4.2, and are closer to traditionally accepted values for the case of China and its regions, which are around 0.5.3

4.4 Including Human Capital In this section, we will incorporate human capital into our regression analysis. We use a simple approach to calculating human capital for the Chinese provinces.4 Our approach follows Hall and Jones (1999). In calculating per worker human capital 3 See,

for example, Zheng et al. (2009) and Brandt and Zhu (2010). mentioned earlier, in growth empirics calculating human capital has always been a weak spot fraught with measurement difficulties. It is beyond the scope of this chapter to design a more refined method of measuring human capital stocks for the Chinese provinces.

4 As

4.4 Including Human Capital

43

Table 4.3 Pooled cross sectional regressions with province dummy variables τ = 1 (Obs: 261)

Reg. 3–1

R2 = 0.9725

Parameter

Estimate

Std. Err.

95% conf. interval

α

0.553

0.031

0.491

0.614

ϕ

0.092

0.031

0.031

0.154

τ = 3 (Obs: 203)

Reg. 3–2

Parameter

Estimate

Std. Err.

95% conf. interval

α

0.467

0.034

0.401

0.534

ϕ

0.638

0.076

0.489

0.787

τ = 6 (Obs: 116)

Reg. 3–3

Parameter

Estimate

Std. Err.

95% conf. interval

α

0.491

0.043

0.405

0.576

ϕ

1.190

0.137

0.916

1.463

R2 = 0.8153

R2 = 0.9437

h i in a cross-country growth study, Hall and Jones (1999) have assumed that h i is related to educational attainment by h i = exp[μ(E i )]. E i denotes the average years of schooling attained by a worker in economy i. Therefore, the function μ(E) indicates the relative efficiency of one worker with E years of schooling compared with one with zero schooling (μ(0) = 0). The derivative μ (E) is the return to schooling estimated in a Mincerian wage regression (Mincer, 1974). In Hall and Jones (1999), μ(E) is assumed to be piecewise linear, with the rate of return being 13.4%, 10.1% and 6.8% respectively for schooling of the first four years, the second four years, and that beyond the eighth year. These rates of return are all based on Psacharopoulos (1994)’s survey of evidence from many countries on return-to-schooling estimates. The rate for the first four years, 13.4%, corresponds to the average return to an additional year of schooling in sub-Saharan Africa. The rate for the second four years, 10.1%, is the average return to an additional year of schooling worldwide, while that for schooling above the eighth year, 6.8%, is taken from the average return to an additional year in the OECD. In this chapter, our measure of per worker human capital of province i at time t, denoted h it , is constructed as h it = (1/L it∗ )



j

j

h j L it

(4.9)

 j where j L it = L it∗ (where j = a, b, c, d, e). L it∗ denotes province i’s population aged six and above at time t. We divide L it∗ into five groups by educational attainment: group a through group e. L ita denotes the total number of people aged six and above who have received zero schooling. L itb through L ite respectively denote the total number of people aged six and above who have received schooling up to the primary school level, the junior secondary school level, the senior secondary school level, and

44

4 Interregional Disparity in Productivity Growth

the university and higher level.5 h a through h e are per worker human capital in each of the five groups respectively. Therefore, the provincial per worker human capital h it j is now a weighted average of the h j ’s, with the weights being the (L it /L it∗ )’s. Data on j these (L it /L it∗ )’s for 29 Chinese provinces for each year during 2003–2012 are found in Chinese Statistical Yearbooks. Constructing h it thus boils down to determining the values of the h j ’s. Necessarily h a = 1 by construction, so that h i = 1 for a (fictitious) province that has only workers with zero schooling. We set h b = 2, h c = 2.6, h d = 3.2, and h e = 4.4 for all provinces in each year during 2003–2012. These assigned values of the h j ’s are calculated exactly according to the aforementioned piecewise linear rates of return to schooling based on Psacharopoulos (1994)’s survey, i.e. 13.4, 10.1 and 6.8% for schooling of the first four years, the second four years, and beyond the eighth year.6 Results of regressions parallel to those in Tables 4.1 and 4.2 but including human capital are summarized in Table 4.4. There are two important findings. First, the same as before, for the regressions that include the zone dummy variables, the estimated coefficients on the zone dummies have the expected sign but are practically small and never statistically significant. Nor does the inclusion of the zone dummies alter the estimates of the other parameters in any significant ways. The point and interval estimates of α, β and ϕ change only very slightly in response to the inclusion of the zone dummies. Second, the estimated values of β do not have the expected positive sign, nor are they significantly different from zero. Compared with results in Tables 4.1 and 4.2, the inclusion of human capital in the regressions only alters the estimates of α and ϕ very slightly. Next we run regressions parallel to those in Table 4.3. These are regressions that include the full set of province dummy variables as explanatory variables to control for the province-specific effects. Regression results are summarized in Table 4.5. Estimated values of α and ϕ deviate very slightly from their counterparts in Table 4.3. The parameter β is not precisely estimated: only the estimated value of β for τ = 6 (Reg. 5–3) has the expected positive sign and is significant. In this case, the estimate of β is quite sensitive to the setting of the time interval τ . The fact that the parameter β is not precisely estimated plus that the estimated values are sensitive to the setting of the time interval τ may be due to a poor measurement of the provincial human capital stocks, or possibly due to a lagged effect of human capital formation on output, or simply due to too much multicollinearity between the explanatory variables in our regressions. We will come back to the issue of the linkage between human capital and TFP growth in the next section.

5 This

five-group division is performed on the provincial population aged six and above because of unavailability of data on the distribution of educational attainment in the provincial employed population or working-age population. 6 Here, in calculating h e , we assume that a worker who has completed university or higher level of education has 17 years of schooling on average.

4.5 The Province Heterogeneity

45

Table 4.4 Cross sectional regressions with human capital τ = 1 (Obs: 261) Parameter

Reg. 4–1

R2 = 0.9623

Reg. 4–5

R2 = 0.9625

Estimate

95% conf. interval

Estimate

95% conf. interval

α

0.529

0.471

0.587

0.532

0.474

0.591

β

−0.028

−0.081

0.025

−0.028

−0.081

0.025

ϕ

−0.011

−0.017

−0.006

−0.009

−0.016

−0.002

Coef. on E

–

–

–

0.002

−0.005

0.009

Coef. on W

–

–

–

−0.002

−0.008

0.004

Reg. 4–2

R2 = 0.5723

Reg. 4–6

R2 = 0.5756

Estimate

95% conf. interval

Estimate

95% conf. interval

τ = 3 (Obs: 203) Parameter α

0.458

0.392

0.524

0.463

0.396

0.529

β

−0.150

−0.242

−0.058

−0.151

−0.243

−0.058

ϕ

−0.037

−0.050

−0.023

−0.029

−0.047

0.011

Coef. on E

–

–

–

0.009

−0.009

0.027

Coef. on W

–

–

–

−0.003

−0.018

0.012

Reg. 4–3

R2 = 0.5167

Reg. 4–7

R2 = 0.5206

Estimate

95% conf. interval

Estimate

95% conf. interval

τ = 6 (Obs: 116) Parameter α

0.402

0.310

0.494

0.408

0.314

0.502

β

−0.090

−0.441

0.260

−0.100

−0.478

0.277

ϕ

−0.073

−0.101

−0.045

−0.061

−0.100

−0.021

Coef. on E

–

–

–

0.014

−0.025

0.052

Coef. on W

–

–

–

−0.005

−0.035

0.026

Reg. 4–4

R2 = 0.5723

Estimate

95% conf. interval

τ = 9 (Obs: 29) Parameter α

0.456

0.272

β

−0.172

ϕ

−0.101

Coef. on E Coef. on W

Reg. 4–8

R2 = 0.5829

Estimate

95% conf. interval

0.640

0.466

0.272

0.661

−1.142

0.798

−0.215

−1.329

0.900

−0.183

−0.019

−0.073

−0.191

0.045

–

–

–

0.027

−0.086

0.140

–

–

–

−0.014

−0.102

0.074

4.5 The Province Heterogeneity We have shown that conditional on the province-specific effects as picked up by the province dummy variables in the regressions in Tables 4.3 and 4.5, the 29 Chinese provinces exhibit clear conditional convergence in TFP growth over the sample

46

4 Interregional Disparity in Productivity Growth

Table 4.5 Pooled cross sectional regressions with human capital and province dummies τ = 1 (Obs: 261)

Reg. 5–1

Parameter

Estimate

R2 = 0.9726 Std. Err.

95% conf. interval

α

0.555

0.031

0.493

0.616

β

−0.027

0.026

−0.077

0.024

ϕ

0.092

0.031

0.030

0.154

τ = 3 (Obs: 203) Parameter

R2

Reg. 5–2 Estimate

Std. Err.

= 0.8292

95% conf. interval

α

0.493

0.033

0.428

0.559

β

−0.167

0.045

−0.256

−0.077

ϕ

0.604

0.074

0.459

0.750

τ = 6 (Obs: 116) Parameter

R2 = 0.9463

Reg. 5–3 Estimate

Std. Err.

95% conf. interval

α

0.501

0.042

0.416

0.586

β

0.260

0.129

0.004

0.515

ϕ

1.26

0.131

0.994

1.517

period. Therefore, we are interested in seeing what factors underlie the province heterogeneity that may affect the rate of provincial TFP growth. Our analysis in the section is based on the estimated coefficients on the province dummies (i.e. the province intercepts) from the regressions in Table 4.3.7 We specifically focus on the regression under τ = 1 because this regression has the largest adjusted R-squared (not reported in the table) compared with other regressions under alternative values of τ . The (normalized) estimated province intercepts from this regression is listed in Table 4.6. We have normalized the intercept for Beijing to zero. We call these values of the province intercepts the ‘province effect indexes’. The difference between the highest value of these indexes (Shanghai) and the lowest value (Sichuan) is roughly 0.22. First we investigate how the province-specific effects are related to the geographical locations of the provinces. Regressing the province effect indexes on the two zone dummy variables E and W, we find that the estimated coefficient on E is significantly positive, being 0.049 with a 95% confidence interval of [0.015, 0.083], and the estimated coefficient on W has the expected negative sign but is not significantly different from zero. This result of the simple exercise roughly shows that the expected value of the province effect index will be higher by about 0.05 for a coastal province than an inland province. Regressing the province effect indexes on E alone produces 7 We

could as well use the estimated coefficients on the province dummies from the regressions in Table 4.5 because it has been shown that whether or not we have the human capital variables in the regression equation makes no significant difference as to the estimated values of the coefficients on the province dummies.

4.5 The Province Heterogeneity Table 4.6 Calculated province effect indexes

47 Province Beijing Tianjin Hebei Shanxi

Effect index 0 0.0562 0.0067 −0.0424

Province

Effect index

Henan

−0.0505

Hubei

−0.0108

Hunan

−0.0249

Guangdong

−0.0033

Inner Mongolia

0.0037

Guangxi

−0.0418

Liaoning

0.0936

Sichuan

−0.1100

Jilin

0.0224

Guizhou

−0.0733

Heilongjiang

0.0138

Yunnan

Shanghai

0.1096

Tibet

−0.0400

Jiangsu

0.0181

Shaanxi

−0.0195

Zhejiang

0.0097

Gansu

−0.0332

0.0046

Anhui

−0.0093

Qinghai

−0.0521

Fujian

0.0355

Ningxia

−0.0032

Jiangxi

−0.0346

Xinjiang

−0.0207

Shandong

−0.0042

an even clearer picture: the estimated coefficient on E is now 0.060 with a 95% confidence interval of [0.031, 0.088]. The R-squared of this latter regression is 0.41, showing that the zone dummy E alone explains over 40% of the sample variation in the province-specific effects. This result leads us to see that the eastern coastal provinces tend to have faster TFP growth over the sample period. Next, we are interested in seeing how the province-specific effects are related to the provinces’ human capital stocks. We thus run a regression of the province effect indexes on the provinces’ initial per worker human capital stocks (in logs) in 2003. The regression produces a significantly positive estimated coefficient on the latter, which is 0.178 with the 95% interval estimate being [0.063, 0.294]. The R-squared of this regression is 0.27, showing that the initial per worker human capital stock alone explains nearly 30% of the total sample variation in the province-specific effects. The sample correlation coefficient between the logs of the initial per worker human capital stocks and the province effect indexes is 0.52. Since the province-specific effects are supposed to capture ‘permanent’ or stable geographical, social, institutional, and policy differences across the Chinese provinces, our conjecture is that relatively higher values of the province effect indexes for the coastal provinces are due to more exposure of these provinces to international economic activities such as foreign direct investment and foreign trade, which in turn rely heavily on the favorable geographical locations of these coastal provinces and the preferential policies they receive. Demurger et al. (2002) have constructed a set of preferential policy indexes for the Chinese provinces to study the effect

48

4 Interregional Disparity in Productivity Growth

of open-door preferential policies on provincial economic performance.8 We run a regression of the province effect index on the preferential policy index and find that the estimated coefficient on the latter is significantly positive, which is 0.035 with a 95% confidence interval of [0.013, 0.057].9 The R-squared of this regression is 0.28, showing that the provincial preferential policy index explains nearly 30% of the sample variation in the province-specific effects. Not surprisingly, this result suggests that open-door preferential policies (coupled with favorable geographical locations) of the eastern coastal provinces are conducive to TFP growth by enhancing the exposure of these provinces to foreign direct investment and foreign trade, which in turn facilitate technology spillovers. In addition, according to our analysis above, we strongly suspect that the provincespecific factors embodied in the province effect indexes, i.e. the ‘permanent’ or stable geographical, social, institutional, and policy differences across the Chinese provinces, affect human capital accumulation in the provinces. This is in fact what the central idea of Hall and Jones (1999) is: differences in capital accumulation, productivity, and therefore output per worker are fundamentally related to differences in social infrastructure across economies. According to Hall and Jones (1999), social infrastructure refers to the institutions and government policies that determine the economic environment within which individuals accumulate skills, and firms accumulate capital and produce output. Therefore, as far as human capital is concerned, a social infrastructure favorable to high levels of output per worker should encourage educational attainment by ensuring that individuals capture the social returns to their education as private returns. A higher level of provincial educational attainment in turn facilitates provincial TFP growth by, for example, increasing the provincial absorptive capacity regarding foreign technology spillovers.10 Further studies on the relations between social infrastructure, human capital, and TFP growth for the Chinese provinces are on our future agenda.

8 Demurger

et al. (2002) stress that their construction of the index is restricted to purely open-door preferential policies and does not take into account other factors, such as the business environment. They also point out that disentangling geography and policy is not an easy task because preferential treatments are obviously related to geography. 9 See Table 11 of Demurger et al. (2002) for the values of the preferential policy indexes. The specific value for each province is calculated as the average of an evaluation of the province’s preferential policy environment on a 0–3 scale for each of the years over 1978–1998. Therefore, the preferential policy index of Demurger et al. (2002) to a large extent reflects the stable or ‘permanent’ provincial preferential policy environment for the sample period of our analysis 1996–2005. This feature of the preferential policy index is exactly what we desire. 10 Prescott (1998) believes that the resistance to the efficient use of currently operating technologies and to the adoption of new technologies is an important factor that hinders TFP growth. Prescott’s idea is compatible with the hypothesis that human capital is a crucial in determining the absorptive ability regarding foreign technology spillovers because better-educated workers have a comparative advantage in, and are thus less resistant to, the adoption of new technologies.

4.6 Concluding Remarks

49

4.6 Concluding Remarks In this chapter, we explore the characteristics of interregional TFP disparity in China and provide a related empirical analysis to enrich our understanding of the catch-up and convergence tendencies of the Chinese regions in terms of TFP. We build our baseline regression model on the basis of the Cobb–Douglas aggregate production function with Hicks-neutral TFP, where growth of per worker output is divided into growth of per worker capital accumulation and growth of TFP. Our regression analysis fails to detect any absolute convergence in TFP across the Chinese provinces over the sample period. Provinces that had higher levels of TFP initially tended to experience faster growth in TFP over the sample period. However, after controlling for the province heterogeneity, our regression results show that the Chinese provinces exhibited significant conditional convergence in TFP growth. This indicates that province-specific factors play an important role in determining provincial TFP growth. The important policy implication of our regression results is that economic policies conducive to faster TFP growth should thus be directed to the relevant factors underlying the province heterogeneity. We conjecture that relatively higher values of the province effect indexes for the coastal provinces are due to more exposure of these provinces to international economic activities such as foreign direct investment and foreign trade, which are dependent on the favorable geographical locations of these coastal provinces and the preferential policies they receive. Our empirical analysis accordingly suggests that open-door preferential policies coupled with favorable geographical locations of the eastern coastal provinces are conducive to TFP growth by enhancing the exposure of these provinces to foreign direct investment and foreign trade, which in turn facilitate technology spillovers. As a byproduct, our regressions have shown a large (direct) contribution of physical capital accumulation to output growth.11 The regressions have all produced values of the output elasticity of physical capital α that are consistent with its empirically accepted values. In addition, although owing to imprecise estimations we have failed to show a direct contribution to output growth of human capital as an accumulable production input, our empirical analysis does suggest that human capital accumulation is associated with TFP growth: a higher level of provincial per worker human capital stock promotes provincial TFP growth by increasing the provincial absorptive capacity of technology diffusion from technologically advanced countries. Further studies on the linkage between the economic environment and policy, openness, human capital, and TFP growth for the Chinese provinces are on our future agenda.

11 The

ultimate cause of capital accumulation, however, may be TFP growth.

50

4 Interregional Disparity in Productivity Growth

References Brandt, Loren, and Xiaodong Zhu. 2010. Accounting for China’s Growth. Working Papers tecipa394, Department of Economics, University of Toronto. Dacosta, Maria, and Wayne Carroll. 2001. Township and Village Enterprises, Openness and Regional Economic Growth in China. Post-Communist Economies 13 (2): 229–241. Demurger, Sylvie. 2001. Infrastructure Development and Economic Growth: An Explanation for Regional Disparities in China? Journal of Comparative Economics 29 (1): 95–117. Demurger, Sylvie, Jeffrey D. Sachs, Wing Thye Woo, Shuming Bao, Gene Chang, and Andrew Mellinger. 2002. Geography, Economic Policy, and Regional Development in China. Asian Economic Papers 1 (1): 146–197. Fleisher, Belton, Haizheng Li, and Min Qiang Zhao. 2010. Human Capital, Economic Growth, and Regional Inequality in China. Journal of Development Economics 92 (2): 215–231. Hall, Robert E., and Charles I. Jones. 1999. Why Do Some Countries Produce So Much More Output per Worker than Others? Quarterly Journal of Economics 114 (1): 83–116. Huang, Zheng-Ming, Y-Z. Zhang, Masaya Kotaki, and Seeram Ramakrishna. 2003. A Review on Polymer Nanofibers by Electrospinning and Their Applications in Nanocomposites. Composites Science and Technology 63 (15): 2223–2253. Jian, Tianlun, Jeffrey D. Sachs, and Andrew M. Warner. 1996. Trends in Regional Inequality in China. China Economic Review 7 (1): 1–21. Jiang, Yanqing. 2011. Economic Environment, Technology Diffusion, and Growth of Regional Total Factor Productivity in China. Journal of Chinese Economic and Business Studies 9 (2): 151–161. Kanbur, Ravi, and Xiaobo Zhang. 2005. Fifty Years of Regional Inequality in China: A Journey Through Central Planning, Reform, and Openness. Review of Development Economics 9 (1): 87–106. O’Neill, J. 2001. Building Better Global Economic BRICs. Global Economics Paper, No. 66, Goldman Sachs, November 30. Prescott, Edward C. 1998. Lawrence R. Klein Lecture 1997: Needed: A Theory of Total Factor Productivity. International Economic Review 39 (3): 525–551. Zhang, Xiaobo, and Kevin Zhang. 2003. How does Globalisation Affect Regional Inequality within a Developing Country? Evidence from China. Journal of Development Studies 39 (4): 47–67. Zhang, Bing, Bin-Bin Zhang, En-Wei Liang, Neil Gehrels, David N. Burrows, and Peter Mészáros. 2007. Making a Short Gamma-ray Burst from a Long One: Implications for the Nature of GRB 060614. The Astrophysical Journal Letters 655 (1): L25. Zheng, Jinghai, Angang Hu, and Arne Bigsten. 2009. Measuring Potential Output in a Rapidly Developing Economy: The Case of China in Comparison with the US and EU. Federal Reserve Bank of St. Louis Review, July/August 2009, 91 (4): 317–342. Zhu, S., M. Lai, and X. Fu. 2008. Spatial Characteristics and Dynamics of Provincial Total Factor Productivity in China. Journal of Chinese Economic and Business Studies 6 (2): 197–217.

Chapter 5

Trade Imbalance and Protectionist Policy on Imported Intermediate Inputs

5.1 Introduction Since March 2018, there has been an emergence of concern by US President Donald Trump about the size of the trade deficit in the US economy, especially given that China accounts for about three-fifths of US trade deficit. Impervious to economic logic, President Trump thinks that the US loses when it imports more than it exports. Trump’s logic is that if better economic policies eliminated this “trade deficit drag”, then US gross domestic product (GDP) would be higher and more American people would be employed. To solve the problem of “trade deficit”, Trump’s trade policy addresses his concern over US reliance on imports, by imposing tariffs on foreign imports, including imported intermediate inputs from China. In this chapter, we address President Trump’s concern over US trade deficit, and examine the effects of an increase in the price of imported intermediate goods on the trade balance. The remainder of this chapter proceeds as follows. In Sect. 5.2, we conduct literature review on growth models, including exogenous growth models, endogenous growth models, and non-scale growth models. In Sect. 5.3, we introduce our analytical framework, which is a non-scale growth model, with features of endogenous labor supply, leisure choice, and an imported intermediate input used in the domestic production. In Sect. 5.4, we derive the macroeconomic equilibrium for this model economy, followed by an analysis of the dynamic effects due to a change in the price of the imported intermediate input in Sect. 5.4. Section 5.5 concludes.

5.2 Theoretical Development In this section, we conduct literature review on economic growth models, including theoretical development on exogenous growth models (Sect. 5.2.1), endogenous growth models (Sect. 5.2.2), and non-scale growth models (Sect. 5.2.3), which we use as our analytical framework in Sect. 5.3. © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_5

51

52

5 Trade Imbalance and Protectionist Policy on Imported …

5.2.1 Exogenous Growth Models The exogenous growth literature starts with the Solow–Swan model (Solow 1956; Swan 1956). The motivation behind the Solow–Swan model is the questions on why our national income grows, and why some economies grow faster than others. So we must broaden our analysis to describe changes in the economy over time. By developing such a model, we make our analysis dynamic, which is more like a movie. The research contribution of the Solow (1956), Swan (1956) model is that they analyzed the role of saving and the role of population growth, introduced technological progress, and showed how saving, population growth, and technological progress affect the level of an economy’s output and its growth over time. There are several assumptions in the Solow (1956), Swan (1956) model. The production function is assumed to have constant returns to scale (CRS), for example, Cobb–Douglas production function. Each year, people save a fraction s of their income and consume a fraction (1-s). Population growth is assumed to be exogenous, ˙ with LL = n. Technological progress is also assumed to be exogenous. The evolution of capital k follows the process that the change in capital stock k equals investment s f (k) minus the break-even investment (δ + n + g)k, where δ is the rate of depreciation, n is the rate of population growth, and g is the rate of labor-augmenting technical change. k = s f (k) − (δ + n + g)k The economy converges on steady states in which the growth rate is (n + g). That is, the long-run growth rate depends on the rate of population growth n, and the rate of labor-augmenting technical change g. 





Y = A+L =n+g

The implication from the Solow (1956), Swan (1956) model is that policy does not matter. However, empirical evidence offers mixed support for theoretical implication from the Solow–Swan model. Its limitation is that it fails to explain how or why technological progress occurs.

5.2.2 Endogenous Growth Models 5.2.2.1

Romer (1990) Model 

In the Solow (1956), Swan (1956) model, Y = n + g. A natural question is: what is g? Romer (1990) answered this question on g, which started the literature on endogenous growth models. The research contribution of Romer (1990) paper is that

5.2 Theoretical Development

53

he developed a two-sector model of a closed economy, where new knowledge that is produced in one sector is used as an input in the production of the final output in the other sector. He formulated an explicit growth model with technological progress, which results from deliberate actions that are taken by private-sector agents who respond to market incentives. The key assumption in Romer (1990) model is that the production of technology is crucial to the growth process. More human capital in the research sector leads to a higher rate of production of new knowledge. Larger total stock of knowledge results in higher productivity of a researcher in the research sector. In the research sector, the production function for new technology is: A˙ t = δ H At At where δ is a productivity parameter. The implication from Romer (1990) model is that a doubling of the population devoted to research will double the growth rate in the economy. For example, 2 × H - > 2 × Y . This is the scale effect, that is, variations in the size or scale of the economy will affect the size of the equilibrium growth rate in the long run. 

5.2.2.2

Jones (1995b) Model

Following Romer (1990) paper on endogenous growth model with scale effect, Jones (1995a) conducted empirical study to examine whether there are “scale effects” in the data. He found that a simple linear trend fits per capita GDP (in logarithm) extremely well for the US data, and drew the implication that policy does not matter. This casual observation is confirmed rigorously by several empirical methods. Jones (1995a) also attempted to extend his analysis to growth rates for a sample of 14 additional advanced OECD economies, the evidence from which does not support the existence of scale effects. In particular, variations in the level of research and employment have exerted no influence on the long-run growth rates of the OECD economies. Thus, Jones (1995a) paper contradicts the predictions of Romer (1990) model. A potential explanation of this contradiction is that productivity may increase with population until some critical population level is reached, which indicates the possible existence of an optimal population level. So the equilibrium growth rate will also increase. However, congestion and other impediments to productivity will eventually set in, and therefore production and the equilibrium growth rate cannot continue increase indefinitely. Following Jones (1995a) paper, Jones (1995b) paper proposed a modified version of Romer (1990) model. This modified model better matches data and is consistent with empirical evidence. In particular, Jones (1995b) model altered a key assumption that is usually found in the literature on endogenous growth theory.

54

5 Trade Imbalance and Protectionist Policy on Imported …

In Romer (1990) model, the production function for new technology in the research sector is: A˙ = δ H A In Jones (1995b) model, the production function for new technology in the research sector is: A˙ = δ H λ Aφ The equilibrium growth rate in Romer (1990) model is: 



Y = A = δH

The equilibrium growth rate in Jones (1995b) model is: 

λH λn = Y =A= 1−φ 1−φ





Although economic growth in Jones (1995b) extended model is generated endogenously through R&D, the long-run growth rate depends only on parameters that are usually taken to be exogenous, including the rate of population growth. The implication from Jones (1995b) model is that policy does not matter.

5.2.3 “Non-scale” Growth Models 5.2.3.1

Closed-Economy “Non-scale” Growth Models

There were two main classes of new growth models: R&D based growth and investment-based growth models. In R&D based growth models, growth arises from technological innovation, as in Romer (1990). In investment-based models, growth originates from private investment, either in physical capital, or human capital. Endogenous growth models imply that the size of an economy may influences its long-run growth rate. Such implied scale effects run counter to the empirical evidence in OECD economies. Jones (1995b) developed a “semi-endogenous” model to address this concern. Eicher and Turnovsky (1999a) generalized Jones (1995b) model to “non-scale” models of economic growth. There are two research objectives in Eicher and Turnovsky (1999a) paper. Their first objective was to show how long-run growth can emerge as an equilibrium phenomenon that reflects the structural characteristics of the economy, for example, the endogenous accumulation of knowledge, tastes, the productivity of capital, and economic policy. Their second objective was to construct a model that still features

5.2 Theoretical Development

55

endogenous growth, but addresses the shortcomings of endogenous growth models, with respect to their empirical implication and theoretical restriction that are imposed on the underlying technologies. In Eicher and Turnovsky (1999a) paper, they developed a general two-sector model of economic growth with exogenous population growth and with accumulation of capital and technology. They parameterized the model’s scale by the population size, and they restricted neither the parameters nor the forms of the production function. They also characterized the non-scale balanced growth equilibria in the general two-sector economy. In Eicher and Turnovsky (1999a) model, the economy produces two goods: final output, and technological change (new knowledge). The final good is produced using three productive factors: the social stocks of technology, labor, and capital. The production function for the final good is: Y = F(A, θ N , φ K ) where Y denotes output of the final good, A denotes the stock of technology, N ˙ denotes population (labor force), which is assumed to grow at the steady rate NN = n, all at time t. K denotes the stock of physical capital. The parameter θ and the parameter φ are the fractions of labor and capital that are devoted to the production of the final good. In contrast to physical capital, technology is not only a public good, but it is also produced in an alternative sector. The production function for technology is: A˙ = J [A, (1 − θ )N , (1 − φ)K ] The issue of non-scale equilibria pertains to the long run. So Eicher and Turnovsky (1999a) focused their attention on characterizing the equilibrium balanced growth of the economy. They defined a balanced growth path as being one along which all real quantities grow at a constant rate, though not necessarily identical. Along the balanced growth path, the output/capital ratio KY is required to remain constant, with capital being accumulated from new final output. Together with the product market equilibrium condition (capital accumulation equation), this implies that the consumption/output ratio CY must remain constant as well. 







Y = K =C =c+n

The balanced growth rates of these real quantities can be obtained by taking the differentials of the two production functions. This leads to the following pair of linear equations in A and K . 







(1 − σ K )K − σ A A = σ N n

56

5 Trade Imbalance and Protectionist Policy on Imported … 



−η K K + (1 − η A )A = η N n where σ X = FXFX and η X = JXJ X , (x = A, N, K), denote the structural elasticities in the production sector and the knowledge sector, respectively. For n > 0, the previous two equations jointly determine the rates of growth of physical capital, output, consumption, and knowledge, as functions of the population growth rate and various production elasticities in the two sectors. 

A=



K =

[η N (1 − σ K ) + σ N η K ]n = βAn  [σ N (1 − η A ) + η N σ A ]n = βK n 

where  = (1 − η A )(1 − σ K ) − η K σ A . The key features of Eicher and Turnovsky (1999a) model is that technology and capital accumulation are still endogenous, but long-run growth rates are now independent of changes in policy and other scale variables. Instead, long-run growth rates are now determined by the exogenous labor growth rate in conjunction with production elasticities. The research contribution from Eicher and Turnovsky (1999a) paper is that they generalized Jones (1995b) model to a two-sector “non-scale” growth model. This general growth model is developed to examine conditions under which balanced growth is void of scale effects. It provides a unifying framework for considering a wide variety of growth model, as this “non-scale” growth model replicates wellknown exogenous, as well as endogenous, (non-) scale models. Replication 1: The Solow–Swan (1956) Model In the Eicher and Turnovsky (1999a) model, the linear relationship among the equilibrium growth rates of output, capital, knowledge, and labor is: 





Y = K = β K n = β A λn = λA

where λ is the relative sectoral growth rate. If production in both the output and R&D sectors exhibit constant-returns-to-scale (CRS) in all three factors, then σ A + σ K + σ N = 1 and η A + η K + η N = 1. Therefore, Y = K = A = n. This is the non-scale growth rate in Eicher and Turnovsky (1999a) model, which is identical to that in the Solow (1956), Swan (1956) model. 





Replication 2: Romer (1990) Model Romer (1990) is a R&D based model, which assumes that final output is generated by a production function that exhibits constant returns to scale (CRS) in knowledgeaugmented labor efficiency units AN, and physical capital K. It specifies the quantity

5.2 Theoretical Development

57

of new output as a linear function of the fraction of quality-adjusted labor employed in the technology sector. In terms of Eicher and Turnovsky (1999a)’s general framework, Romer (1990) model can be represented by: Y = (θ AN )σ K 1−σ A˙ = (1 − θ )AN where σ N = σ A = 1 − σ K = σ , and η N = η A = 1, η K = 0. If n = 0, population is stationary. In terms of Romer (1990) model, the stock of skilled labor is stationary. In order for the balanced growth path to exist, the equilibrium growth rate of the economy is: 







c = Y = K = A = (1 − θ )N

which equals the share of population engaged in research. If the size of the population N increases, the share of (1 − θ )N also increases. This is the scale effect as in Romer (1990) model. Replication 3: Jones (1995b) Model Jones (1995b) model is a non-scale version of Romer (1990) model. It retains the same production function for final output as in Romer (1990) model, and it modifies the production function in the knowledge sector, with η A < 1. In terms of Eicher and Turnovsky (1999a)’s notation, the production function in the knowledge sector in Jones (1995b) model can be written as: A˙ = Aη A [(1 − θ )N ]η N and η K = 0. The equilibrium growth rate can be written as: 



K −n = A =

ηn n 1 − ηA

as in Jones (1995b) model. To sum up, the Solow (1956), Swan (1956) model assumes that A is exogenously given. Romer (1990) and Jones (1995b) model assumes A = f(AN), but not as a function of K. Eicher and Turnovsky (1999a) is a hybrid model: A˙ = Aη A [(1 − θ )N ]η N [(1 − φ)K ]η K

58

5 Trade Imbalance and Protectionist Policy on Imported …

Different knife-edge conditions on Eicher and Turnovsky (1999a) model lead to different exogenous or endogenous growth models.

5.2.3.2

Open-Economy “Non-scale” Growth Models

Eicher and Turnovsky (1999b) extended Eicher and Turnovsky (1999a) model, and opened the non-scale closed economy to international trade in goods and assets. They took into account factor mobility and its influence on economic growth, and they characterized the dynamic structure of a non-scale open economy, that is opened up to the rest of the world. The research methodology in Eicher and Turnovsky (1999b) model is that they first considered a conventional pure small open economy that faces a perfect world capital market, and assumed that capital accumulation is subject to convex adjustment costs. They then modified the model by introducing a capital market imperfection, in the form of an upward-sloping supply schedule of funds. In Eicher and Turnovsky (1999b) model, they restricted the production technology to only one single sector. There is a small open economy, which produces and consumes a single traded commodity. Each individual is endowed with a fixed quantity of labor L i . Labor is fully employed, such that total labor supply is equal to population N. The labor/population ratio grows exponentially at the steady rate N˙ = n N . Domestic output Yi of the traded commodity is determined by the private capital stock K i , labor supply L i , and the aggregate capital stock K = N K i . Agents also accumulate foreign bonds Bi . Bi pays a fixed rate of return ri , which is determined exogenously in the world bond market. The individual agent’s instantaneous budget constraint is:     h Ii + Ti B˙i = 1 − τ y Yi + [r (1 − τb ) − n]Bi − (1 − τc )Ci − Ii 1 + 2 Ki where τ y is the income tax rate from current production, τb is the income tax rate from bonds, τc is the consumption tax rate, and T is lump-sum transfer, which are revenues from all taxes. There are two departures in Eicher and Turnovsky (1999b) model. The first departure is non-constant returns to scale. The second departure is population growth. Aside from these two departures, their model is a standard endogenous growth model of a small open economy. Their two departures render the model within the class of non-scale growth models. Aggregate Dynamics To analyze the dynamics of the aggregate economy along the long-run stationary growth path, Eicher and Turnovsky (1999b) made the assumption that along such an equilibrium path, aggregate output and capital stock grow at the same constant rate, such that aggregate output-capital ratio remains unchanged.

5.2 Theoretical Development

59

The aggregate production is calculated by summing the individual production functions over the N agents: Y = α K η+σ N 1−σ = α K σ K N σ N where σ N = 1 − σ = share of labor in aggregate output, σ K = η + σ = share of capital in aggregate output. So σ K + σ N = 1 + η, which measures total returns to scale (TRS) of the social aggregate production function. By taking the percentage changes of the aggregate production function, and imposing the long-run condition of a constant output-capital ratio, they reached the long-run equilibrium growth rate of capital and output g as: g=

σN n>0 1 − σK

In a stable dynamic system, σ K < 1, which implies g > 0. Aggregate Dynamics—Scale Adjustment To analyze the transitional dynamics of the economy about the long-run stationary growth path, Eicher and Turnovsky (1999b) expressed the system in terms of the relative price of installed capital q. They defined the following stationary variables: c= k= b=

C σN

N 1−σ K K σN

N 1−σ K B σN

N 1−σ K

Aggregate Dynamics—Consumption The individual’s consumption grows at the constant rate: r (1 − τb ) − ρ − n C˙i = ψi = Ci 1−γ The aggregate consumption grows at the constant rate: r (1 − τb ) − ρ − γ n C˙ = =ψ C 1−γ The scale-adjusted per capita consumption grows at the constant rate:

60

5 Trade Imbalance and Protectionist Policy on Imported …

c˙ r (1 − τb ) − ρ − γ n σN =ψ−g = − c 1−γ 1 − σK With a perfect world capital market, the corresponding consumption growth rates are constant and independent of the production characteristics of the domestic economy. Those equilibrium growth rates vary inversely with the tax rate on foreign bond income, but are independent of all other tax rates. Aggregate Dynamics—K and PK PK The scale-adjusted capital-labor ratio grows at the constant rate: q −1 σN k˙ = − n =φ−g k h 1 − σK The law of motion for the relative price of the capital good is: q˙ = r (1 − τb )q −

(q − 1)2 − (1 − τ y )ασ k σ K −1 2h

k˙ equation and q˙ equation comprise a pair of equations in k and q that evolve independently of consumption. Aggregate Dynamics—Stationary Solution By setting k˙ = q˙ = 0, the steady-state price of installed capital is: ∼

q= 1 + h

σN n = 1 + hg 1 − σK ∼

The equilibrium scale-adjusted capital-labor ratio k is determined from the steadystate no-arbitrage condition: ∼

  ∼σ K −1 (q −1)2 ∼ = r (1 − τb ) q 1 − τ y ασ k + 2h Under the transversality condition, the long-run equilibrium must satisfy: r (1 − τb ) > g > 0 That is, the after-tax interest rate on foreign bonds must exceed the growth rate of domestic aggregate output.

5.2 Theoretical Development

61

Aggregate Dynamics—Foreign Debt Accumulation Eicher and Turnovsky (1999b) assumed that the domestic government maintains a continuously balanced budget, and they also assumed that all tax revenues are rebated back to the private sector. T = N Ti = τ y α K σ K N σ N + τb r B + τc C The nation’s current account balance is the aggregate net rate of accumulation of traded goods.   h I σK σN ˙ B = r B + αK N − C − I 1 + 2K Overall, the introduction of the capital market imperfect in Eicher and Turnovsky (1999b) model fundamentally changes both the dynamic structure and policy effectiveness in the long run. The evolution of all variables becomes inter-related, subject to transitional dynamics. Borrowing constraints thus yield an equilibrium in which the equilibrium growth rates of both output and consumption are determined by the production and population growth parameters alone, precisely as in a non-scale closed economy. A full non-scale equilibrium is thus obtained in this open economy model.

5.3 Analytical Framework In our analytical framework, we follow Schubert and Turnovsky (2007) to consider a one-sector “non-scale” open economy growth model. The domestic production includes one traded good Y, which can be consumed or invested or exported. There is an imported intermediate good that is used solely as an input in domestic production. Its relative price, expressed in terms of the traded final good, is p, which is assumed to remain constant over time. We relax Schubert and Turnovsky (2007)’s assumption of fixed labor supply, and introduce endogenous labor supply and leisure choice. This extension captures more realistic features. In our model economy, a representative agent obtains utility from his consumption and leisure choice, which is represented by an intertemporal iso-elastic utility function over an infinite time horizon. ∞ Ui = 0

where −∞ < γ < 1, θ > 0.

1  θ γ −βt Ci l e dt γ

(5.1)

62

5 Trade Imbalance and Protectionist Policy on Imported …

In the utility function, Ci is each individual’s consumption choice. Each representative agent is endowed with a unit of time that is divided between leisure choice l and labor choice (1−l). The parameter θ measures the degree of substitutability between 1 is the inter-temporal consumption and leisure choice in utility by each agent. 1−γ elasticity of substitution. The parameter β denotes the rate of time preference by the representative agent, which is assumed to constant over time. The production function is: 1−α−ξ

Yi = αL i

ξ

ξ

K iσ Z i K η = α(1 − l)1−σ −ξ K iσ Z i K η

where 0 < σ < 1, 0 < ξ < 1, 0 < σ + ξ < 1. In the production function, there are three private factors. The first factor is each individual’s labor supply L i (= 1 − l). The second factor is private capital K i . The their factor is imported intermediate good Z i . The parameter α is like a proxy for total factor productivity (TFP) in the domestic production of the traded good Yi . The fourth factor K is the economy-wide capital stock, with K = N K i . As in Romer (1986), the parameter η captures the spillover effect from the economy-wide capital stock. We also assume constant returns to scale (CRS) in the three private factors L i , K i , Z i , and total returns to scale (TRS) of degree (1 + η) in all four factors. Depending on the qualitative nature of the spillover effect, the total returns to scale (TRS) can be increasing, constant, or decreasing. The aggregate production function is obtained by summing individual production functions. Y = α(1 − l)σL N σ N K σ K Z σ Z

(5.2)

where σ L = σ N = 1 − σ − ξ , σ K = σ + η, an σ Z = ξ are the shares of labor, capital, and imported input in aggregate output, respectively. Hence, σ L + σ K + σ Z = 1 + η measures total returns to scale (TRS) of the aggregate production function. A representative agent accumulates physical capital K i . The investment in the accumulation of physical capital is associated with installation costs. We follow Hayashi (1982) and adopt a quadratic (convex) investment function with adjustment costs.   Ii2 h Ii = Ii 1 + (Ii , K i ) = Ii + h 2K i 2 Ki where the adjustment costs are proportional to the rate (rather than its level) of investment per unit of installed capital, KIii . The linear homogeneity of this function is necessary to sustain steady-state equilibrium growth. The aggregate investment function is obtained by summing individual investment functions.

5.3 Analytical Framework

63

  h I I2 = I 1+ (I, K ) = I + h 2K 2K

(5.3)

For simplicity, it is further assumed that the capital stock does not depreciate, such that the net rate of capital accumulation per agent, taking into account population growth, is given by: K˙ i = Ii − n K i ˙

where n = NN is the population growth rate. N is the population size, which equals total labor supply that is implied by full employment condition. At the aggregate level, the economy faces the constraint of physical capital accumulation. K˙ = I

(5.4)

Domestic agents have access to a perfect world capital market, which allows them to accumulate bonds from the rest of the world. These bonds are denominated in terms of the traded good, and pay a fixed interest rate r in the world market, yielding a net rate of return of (r−n) to individual agents. So a representative agent’s budget constraint in the short run, expressed in terms of the traded good, is: B˙i = Yi − Ci − p Z i − (Ii , K i )(r − n)Bi To the extent that the representative agent’s income from production Yi plus net interest income (r − n)Bi exceed his consumption Ci , raw material costs p Z i , and investment costs (Ii , K i ), he will accumulate his holdings of foreign (traded) bonds, that is, Bi > 0. Otherwise, Bi < 0, in which case the agent is a debtor. At the aggregate level, the economy accumulates net foreign bonds, B, that pay an exogenously given world interest rate, r , subject to the accumulation equation:   h I +rB B˙ = Y − C − p Z − I 1 + 2K

(5.5)

We consider the general equilibrium that is generated in a centrally planned economy, in which the social planner chooses C, l, Z , and I , together with K and B. We maximize the utility level of a representative agent as in Eq. (5.1), subject to the aggregate production function (Eq. 5.2), the capital accumulation Eq. (5.4), and the aggregate resource constraint (Eq. 5.5).  ∞    1 C θ γ −βt l e dt − q ∗ K˙ − I e−βt H= γ N 0

64

5 Trade Imbalance and Protectionist Policy on Imported …

   h I − r B e−βt − λ B˙ − Y + C + p Z + I 1 + 2K where Y = α(1 − l)σL N σ N K σ K Z σ Z . The first-order optimality condition with respect to consumption C is: N −γ C γ −1 l θγ = λ

(5.6)

The first-order optimality condition with respect to leisure l is: N −γ θC γ l θγ −1 = λσ L

Y 1−l

(5.7)

The first two optimality conditions give us the consumption-output ratio: C 1 σL = Y 1−l θ

(5.8)

The first-order optimality condition with respect to the imported input Z is: Z = σZ

Y p

(5.9)

The first-order optimality condition with respect to investment I is: 1+h

I =q K

(5.10)

In these first-order conditions, λ is the shadow value (marginal utility) of wealth in∗ the form of international traded bonds, q ∗ is the shadow value of capital, and q = qλ is the value of capital in terms of the (unitary) price of foreign bonds. Equation (5.6) is the usual static optimality condition on consumption, which requires that along an optimal path, the marginal utility of consumption has to equal the shadow value of wealth. Equation (5.7) determines the optimal leisure choice. Equation (5.9) determines the optimal choice of the level of imported input Z i . The production function belongs to the Cobb–Douglas type, which implies that the demand for the imported input is directly proportional to output, and inversely proportional to its relative price. Equation (5.10) asserts that the marginal cost of investment is equated to the marginal value of installed capital q. This gives us a Tobin’s Q type of investment. Next, the Euler condition with respect to foreign bond B is: β−

λ˙ =r λ

The Euler condition with respect to capital K is:

(5.11)

5.3 Analytical Framework

65

  q˙ h q −1 2 σK Y + + =r q K q 2q h

(5.12)

The Euler conditions are the dynamic efficiency conditions. Equation (5.11) is the dynamic efficiency condition on foreign bonds, with requires that the rate of return on consumption, denoted in terms of the traded good, is equal to the net growth interest rate. This implies a constant growth rate of the agent’s marginal utility, with constant rate of time preference β, constant world interest rate r , and constant population growth rate n. Equation (5.12) is the dynamic efficiency condition on capital stocks, which equates the rate of return on domestic capital to the rate of return on foreign (traded) bonds. The rate of return on domestic capital comprises three components. The first component is the marginal output per unit of installed capital, which is valued at the (shadow) price q. The second component is the capital gain, and the third component captures the benefit arising from the fact that higher capital stock leads to lower installation cost. Finally, the transversality conditions are: lim λi Bi e−βt = 0

t→∞

lim λi qi K i e−βt = 0

t→∞

These two transversality conditions are imposed in order to ensure that the agent’s inter-temporal budget constraint is met.

5.4 Macroeconomic Equilibrium 5.4.1 Balanced Growth Our objective is to analyze the dynamics of the domestic economy in response to an increase in the price of imported intermediate goods along a stationary growth. So we must determine the long-run growth rate along a balanced growth path for our model economy. At the steady state, q˙ = 0. Equation (5.12) implies that along a balanced growth path, KYii remains constant. Yi also remains constant at all points of time, which is implied by the optimality Ki condition for the imported input (Eq. 5.9). So along a balanced growth path, Yi , K i , and Z i all grow at a common constant rate. With identical agents, the aggregate quantities are defined by Y = N Yi , K = N K i , Z = N Z i , from which it follows that Y , K , and Z also grow at a common rate, that is, Y = K = Z , where “ ” denotes the growth rate in the variable. 







66

5 Trade Imbalance and Protectionist Policy on Imported …

Taking percentage change of the aggregate production function and imposing the long-run equilibrium relationship in the growth rate Y = K = Z gives us the steady-state growth rate: 



Y =g=



σN 1−σ −ξ n= n 1 − σK − σZ 1−σ −η−ξ



(5.13)

Therefore, the equilibrium growth rate depends on the production technology, which is summarized by the shares σ j in the aggregate production function. It also depends on the population growth rate n, which is assumed to be exogenous in the model. σ K + σ Z < 1 is assumed, so as to have a positive balanced growth. Under constant total returns to scale, the equilibrium growth rate equals the population growth rate, g = n. In the presence of spillover effect (from the economy-wide capital stock) as we have in this model, g can be bigger than or smaller than n, which depends on the qualitative form of this spillover effect η, that is, whether we have increasing or decreasing total returns to scale. This in turn also determines whether the introduction of an imported raw material aids or impedes the growth rate in the long run.

5.4.2 Consumption Dynamics The growth rate of consumption is determined by taking the time derivative of the optimality condition on consumption combined with the dynamic efficiency condition on foreign bonds.   1 l˙ C˙ = r − β − γ n + θβ C 1−γ l

(5.14)

Knife-Edge Conditions When the assumption of inelastic labor supply is relaxed and labor is endogenously supplied, a much stronger knife-edge condition is required. This is because the opti1 σL mality condition on the consumption-output ratio CY = 1−l (Eq. 5.8) must now be θ taken into account. The fraction of time that is allocated to work is constant at the steady state. This relationship thus implies that the steady-state consumption-output ratio must be constant, and as a result it imposes the equality of the long-run growth rates of consumption and output. At the steady state, l˙ = 0, so the consumption dynamics (Eq. 5.14) becomes: r − β − γn C˙ = =ψ C 1−γ By equating (5.13) and (5.14), we have:

(5.15)

5.4 Macroeconomic Equilibrium

ψ=

67

σn r − β − γn = n=g 1−γ 1 − σK − σZ

(5.16)

The condition (5.16) is the knife-edge condition for this growth model, which indicates that the implied growth rate of consumption is driven to equal the growth rate of capital, which is determined by the population growth rate in conjunction with the production elasticities in accordance with the “non-scale” growth model.

5.4.3 Capital Dynamics The dynamics of capital converge to a long-run steady state growth rate along a transitional growth path, with the stationary “scale-adjusted” capital stock to be: K

k= N

σN 1−σ K −σ Z

In equilibrium, the growth rate of the aggregate capital stock is: q −1 K˙ = K h

(5.17)

The growth rate of the stationary “scale-adjusted” capital stock is: k˙ q −1 σN = − n k h 1 − σK − σZ

(5.18)

5.4.4 Dynamic System The aggregate macroeconomic equilibrium is derived from the optimality conditions for individual agents. We begin by taking the time derivatives of the optimality condition for consumption (Eq. 5.6), the equilibrium consumption-output ratio (Eq. 5.8), and the production function (Eq. 5.2). This leads to the following relationships: −γ n + (γ − 1)

l˙ C˙ + θγ = β − r C l

Y˙ l˙ l˙ C˙ − = + C Y l 1−l

(5.19)

(5.20)

68

5 Trade Imbalance and Protectionist Policy on Imported …

Y˙ l˙ σ K K˙ σL σN n+ =− + Y 1 − σZ 1 − l 1 − σZ 1 − σZ K

(5.21)

The macroeconomic equilibrium can be expressed by the pair of differential equations in q and l. Y (q − 1)2 − σK 2h K    ˙l = 1 r − β − γ n − 1 − γ σ K q − 1 − σ N F(l) 1 − σZ h q˙ = rq −

(5.22) (5.23)

where F(l) =

  1 − γ (θ + 1) 1 − γ σL >0 + 1− l 1−l 1 − σZ

The dynamic system is Eqs. (5.22) + (55.23) + (5.18): σZ   1−σ σZ σ K +σ Z −1 1 Z σZ (q − 1)2 1−σ Z 1−σ Z − σK α q˙ = rq − k 1−σ Z (1 − l) 2h p    1 q −1 1 − γ r − β − γn σK − − σN n l˙ = F(l) 1−γ 1 − σZ h   σN q −1 − n k k˙ = h 1 − σK − σZ

where q and l are jump variables, and k is a sluggish variable. We apply the general knife-edge condition (Eq. 5.16), to get the macro dynamic equilibrium equations. σZ   1−σ σZ σ K +σ Z −1 1 Z σZ (q − 1)2 − σ K α 1−σ Z (1 − l) 1−σ Z k 1−σ Z 2h p    1 q −1 ˙l = 1 − γ g − σK − σN n F(l) 1 − σZ h   q −1 −g k k˙ = h

q˙ = rq −

In order for the domestic labor supply, capital stock, and output to converge to a balanced growth path with a constant growth rate, the stationary solution to the dynamic system above must have at least one real solution, when q˙ = l˙ = k˙ = 0.

5.4 Macroeconomic Equilibrium

69

∼ 2 q −1

 σ Z σ +σ −1 σZ 

∼ 1−σ Z σ Z 1−σ Z ∼ K1−σZZ 1− l q˙ = r q − =0 − σK α k 2h p 

∼ q −1 1 − γ 1 − σN n l˙ = =0 σK g− ∼ 1 − σZ h F( l )

∼  q −1 ∼ k˙ = −g k= 0 h ∼

1 1−σ Z

From k˙ = 0 equation, we get: ∼

q = 1 + gh

(5.24)

∼

So q is determined by the growth rate g. By substituting Eq. (5.24) into the q˙ = 0 and l˙ = 0 equations, we find that l and ∼

∼

k are jointly determined. So we can take l as given, and solve for k . (Likewise, we ∼ ∼ can also take k as given, and solve for l . σZ 1−σ Z  − σ +σ  σZ

∼− σ K +σ Z −1 σ Z K Z −1 Z g 2 h σ K +σ Z −1 − σ 1−σ − σ +σ1 −1 +σ −1 σ K K Z α K Z 1− l k = r (1 + gh) − 2 p

∼

˙ and l˙ around their steady states: Linearize q, ˙ k, ⎛

⎞

⎛

⎜ q˙ ⎟ ⎜ a11 a12 a13 ⎜ ˙⎟=⎝ ⎝k⎠ a21 a22 a23 a31 a32 a33 l˙

⎞ ⎛ ⎞ ∼ ⎜ q− q ⎟ ⎟ ⎟⎜ ⎠⎜ k− ∼ ⎟ k⎠ ⎝ ∼ l− l

where: a11 = r − g

a12

 σZ σZ  σ ∼ 1−σ Z σ Z 1−σ Z ∼ 1−σKZ −2 σ K (σ K + σ Z − 1) 1−σ1

=− α Z 1− l k 1 − σZ p a13 =

 σ Z σ +σ −1 σ L −1+σ Z  ∼ 1−σ Z σ Z 1−σ Z ∼ K1−σZZ σ K σ Z 1−σ1

α Z 1− l k 1 − σZ p ∼

a21 =

k h

70

5 Trade Imbalance and Protectionist Policy on Imported …

a22 = 0 a23 = 0 a31 = −

1−γ

1 1 σK 1 − σ h Z F( l ) ∼

a32 = 0 a33 =

1−γ ∼ 2





g−

F( l )

1 q −1 σK − σN n 1 − σZ h

 

  σL 1−γ 1 − γ (θ + 1) 1 − + l2 1 − σZ (1 − l)2

So we obtain: ⎞ ⎞⎛ ∼ q− q ⎟ ⎜ q˙ ⎟ ⎜ a11 a12 a13 ⎟⎜ ⎟⎜ ⎜ ˙⎟=⎜∼ ∼⎟ ⎜ ⎝ k ⎠ ⎝ k / h 0 0 ⎠⎝ k− k ⎟ ⎠ ∼ l˙ a31 0 a33 l− l ⎛

⎞

⎛

5.4.5 Non-scale Growth Recall from Sect. 5.4.3 on the capital dynamics, the dynamics of capital converge to a long-run steady-state growth rate along a transitional growth path. In this section, we also define the stationary “scale-adjusted” imported input and aggregate output. The imported input is:  1  σN σL σK σ Z 1−σ Z Z= α (1 − l) 1−σ Z K 1−σ Z N 1−σ Z p

(5.25)

The corresponding “scale-adjusted” per capita quantity of the imported input is: z=

Z σN

N 1−σ K −σ Z

By expressing the imported input in terms of its corresponding “scale-adjusted” per capita quantity, we get the direct relationship between the time path of capital and imported input.  1  σL σK σ Z 1−σ Z z= α (1 − l) 1−σ Z k 1−σ Z p

(5.26)

5.4 Macroeconomic Equilibrium

71

The aggregate output is: 1



Y = α 1−σ Z

σZ p

σZ  1−σ

Z

σN

σK

K 1−σ Z N 1−σ Z

(5.27)

The corresponding “scale-adjusted” per capita quantity of the aggregate output is: Y

y=

N

σN 1−σ K −σ Z

By expressing the aggregate output in terms of its corresponding “scale-adjusted” per capita quantity, we obtain the direct relationship between the time path of capital and aggregate output. y=α

1 1−σ Z

(1 − l)

σL 1−σ Z



σZ p

σZ  1−σ

Z

σK

k 1−σ Z

(5.28)

5.4.6 Accumulation of Foreign Assets In the equilibrium, differential growth rates of consumption and domestic output can be sustained. This has implication on its net asset position. The individual consumer’s flow budget constraint is: B˙i = Yi − Ci − p Z i − (Ii , K i ) + (r − n)Bi By aggregating all the individual consumers’ flow budget constraints, we get the aggregate net rate of accumulation of traded bonds by the private sector, which is also the current account balance in this economy.   h I B˙ = r B + Y − C − p Z − I 1 + 2K Its corresponding “scaled-adjusted” form is: b˙ = (r − g)b + (1 − σ Z )α 1−σ Z 1



σZ p

σZ  1−σ

Z

σK

k 1−σ Z −

q2 − 1 k−c 2h

If the transversality conditions hold, then the linearized solution to the “scaleadjusted” per capita stock of bonds, starting from the initial stock of bonds b0 is given by:

72

5 Trade Imbalance and Protectionist Policy on Imported …

b(t) = −

L M c(0) (ψ−g)t − e e μ1 t + r − g r − g − μ1 r −ψ

The inter-temporal budget constraint is:  c(0) = (r − ψ) b0 +

L M + r − g r − g − μ1



where M and L are constants. The inter-temporal budget constraint shows the present value of the resources available for initial consumption after the investment needs have been met along the transition path.

5.5 Shock Analysis In this section, we analyze the impact of a US tariff on imported intermediate inputs, which is represented by an increase in the price of the imported intermediate input p in our model economy.

5.5.1 Steady-State Changes First, we consider the effect of an unanticipated permanent increase in the price of the imported intermediate input (dp > 0) at time t = 0. With perfect foresight, agents’ expectation of the steady-state response determines the dynamic evolution of the economy. So we start our analysis with the study of the long-run steady-state effect of an increase in the price of the imported intermediate input. At the steady state, ∼

q = 1 + gh ∼

This implies that the long-run market price of installed capital q does not change, because the steady-state growth rate g is independent of the imported intermediate input price in the economy. We rewrite the “scale-adjusted” stationary quantities here: ∼

k=



g2 h r (1 + gh) − 2



1−σ Z σ K +σ Z −1

1−σ

σK

1 − σ +σ Z −1 − σ +σ K Z α K Z −1



σZ σZ    ∼ − σ K +σ Z −1 σ Z − σ K +σ Z −1

1− l

 1  σL σK ∼ 1−σ Z ∼ 1−σ Z σ Z 1−σ Z

1− l z= α k p

p

5.5 Shock Analysis

73

y=α

1 1−σ Z

 σZ σL  σ

∼ 1−σ Z σ Z 1−σ Z ∼ 1−σKZ 1− l k p

From these “scale-adjusted” stationary quantities, we can measure the long-run impact of the price of the imported intermediate input on the “scale-adjusted” capital stock, the demand for the imported intermediate input, and the aggregate output. ⎛

∼

dk ∼

=−

k

⎞

σZ ⎝ dp + d l ⎠ < 0 ∼ 1 − σK − σZ p 1− l

  ∼ dp d l σK σZ 1 σ + − L ∼ ∼ 0} Turning Points in Business Cycles Measurement of business cycles provides us with a benchmark against which macroeconomic theories and policy assessment can be applied. This process requires an operational definition of a business cycle. In the literature, there are several different notions of business cycles, among which classical cycles and growth cycles are the most common ones. A classical cycle is identified by finding the turning points in the level of the variable, and a growth cycle is identified as the turning points in the level of the variable less a permanent component. In addition, an acceleration cycle is identified by locating the turning points in the growth rate of the variable, which could be either a quarterly growth rate, or an annual growth rate. (a) Classical Cycle From a historical perspective, one major concern of the NBER’s Business Cycle Dating Committee is the classical cycles in the US economy. A classical cycle refers to the recurrence of expansions and contractions in the aggregate level of economic activity. In a classical cycle, the peaks measure the dates from which economic activity suffers a sustained decline; and the troughs measure the dates from which economic activity ends its decline and begins a sustained increase. According to the

80

6 When the US Sneezes, Will China Catch a Cold?

NBER, an expansion starts at the trough of a business cycle and ends at the peak of the cycle, while a contraction (recession) starts at the peak of a business cycle and ends at the trough of the cycle. In other words, an expansion is a period between the trough and the peak of a business cycle with increasing economic activity spread across an economy, and a recession is a period between the peak and the trough of a business cycle with declining economic activity across an economy, which usually lasts for a minimum of two consecutive quarters. As per the definition of a classical cycle, China has not experienced any economic recessions, which is due to the fact that the level of China’s GDP does not fall, even during economic downturns, and therefore, the notion of a classical cycle is not appropriate to describe the business cycles for the Chinese economy. (b) Growth Cycle Burns and Mitchell (1946) simply referred to the time series yt as the aggregate level of economic activity. Mintz (1969, 1972) had difficulty locating turning points in the aggregate level of economic activity in surging economies such as West Germany at that time. This led Mintz to first extract a permanent component pt from yt , and then examine the turning points in z t = yt − pt . It is the cycle in z t , rather than yt , that is then examined, with the requisite indicators being derived from observations on z t . A growth cycle reflects the fluctuation in the rate of economic growth, which takes into account the long-run trend rate of growth in the economy (Boehm and Moore 1984). In a growth cycle, the peaks measure the points at which economic growth moves from above the trend rate to below the trend rate, and the troughs measure the points at which growth moves from below the trend rate to above the trend rate for a sustained period of time. In our analysis on the growth cycle, we use the logarithm of the aggregate level of economic activity, that is, the real GDP level, ln(G D P t ); and we use the turning points in ln(G D P t − a − bt ), where (a + bt ) is the permanent component in the real GDP data, to measure the growth cycle. Binary random variables are used to summarize the expansion and contraction phases of the growth cycle. We let SCGN ,t be the binary variable that represents the growth cycle for China, and let SUG S,t be the binary variable that represents the growth cycle for the US. SCGN ,t and SUG S,t take the value of unity if the country is in a growth cycle expansion at time t, and they take the value of zero if the economy is in a growth cycle recession at time t. In Table 6.1 and Chart 6.1, we report our finding of the turning points in the growth cycles of China and the US. The source for China’s quarterly GDP is CEIC database and China’s National Bureau of Statistics (NBS). The source for the US quarterly GDP data is Federal Reserve Economic Data (FRED), and the data series is real gross domestic product, billions of chained 2012 dollars, with seasonal adjustment. We choose the common start date of 1992Q1, and the end date of 2019Q4, which is before the outbreak of the COVID-19 pandemic. From Table 6.1, there are two main observations. The first observation is that in terms of the number of growth cycles, China has experienced much fewer growth cycles than the US, with its earliest trough occurred in 2008Q1 when the Global

6.2 Measurement of Business Cycles

81

Table 6.1 Turning points in the growth cycles–China and the US China Trough

US Peak

Trough

Peak

1993Q1

1993Q4

1995Q1

1995Q3

2000Q3

2002Q1

2002Q3

2003Q2

2008Q1

2009Q2

2006Q2

2009Q4

2012Q1

2019Q2

2010Q4

2011Q4

2012Q2

2014Q2

2015Q3

2017Q3

2018Q4

Chart 6.1: Turning points in the growth cycles–China and the US

Financial Crisis (GFC) happened. The second observation is that in terms of the duration of growth cycles, the US growth cycle is usually short in duration, ranging from two quarters to three years and a half. By comparison, China has only experienced two growth cycles, with the first one during the GFC, and a very long growth cycle expansion from 2012Q1 to 2019Q2. (c) Acceleration Cycle An acceleration cycle is identified by locating the turning points in the growth rate of the variable. In an acceleration cycle, the peaks measure the dates at which the rate of economic growth begins to slow down, whilst the troughs measure the dates at which the rate of economic growth begins to increase. In our analysis on the acceleration cycle, we use the turning points in ln(G D P t ) − ln(G D P t−4 ), where the subscript 4 represents four quarters, which is one year. We let SCAN ,t be the binary variable that represents the acceleration cycle for China, and let SUA S,t be the binary variable that represents the acceleration cycle for the US. SCAN ,t and SUA S,t take the value of one if the country is in an acceleration cycle expansion at time t, and they take the value of zero if the economy is in an acceleration cycle recession at time t.

82

6 When the US Sneezes, Will China Catch a Cold?

Table 6.2 Turning points in the acceleration cycles–China and the US China

US

Trough

Peak

Trough 1992Q1

1993Q4

1994Q3

1998Q3

1994Q4

1996Q1

1999Q1

1999Q4

1997Q1

1997Q3

2001Q2

2002Q1

1999Q1

1999Q3

2004Q3

2005Q4

2000Q3

2002Q1

2007Q4

2009Q3

2002Q4

2003Q2

2011Q3

2012Q4

2004Q1

2007Q2

2014Q1

2016Q1

2007Q4

2009Q3

2010Q4

2011Q4

2012Q2

2013Q3

2015Q2

2016Q3

2018Q2

Peak

2018Q3

In Table 6.2, we present our finding of the turning points in the acceleration cycles of China and the US. There are two main observations. First, compared with growth cycles, there are many more acceleration cycles for both China and the US, which also have shorter duration. This indicates that acceleration cycles occur more frequently than growth cycles, which is confirmed in Chart 6.2. Second, in terms of the number of acceleration cycles, China has experienced fewer acceleration cycles than the US, but both China and the US experienced contractions in the early 1990s.

Chart 6.2: Turning points in the acceleration cycles–China and the US. Source CEIC, FRED.

6.3 Synchronization of Business Cycles

83

6.3 Synchronization of Business Cycles 6.3.1 Analytical Framework In Sect. 6.2, we date the business cycles and find the turning points in the growth cycles and acceleration cycles for China and the US. In this section, we use the binary states from the cycle dating and estimate the parameters that are related to the measurement of cycle synchronization between China and the US. The mathematical notations and computer codes are based on and adapted from Hamilton (1994), and Harding and Pagan (2006). In what follows, we use generalized methods of moments (GMM), because we would like to seek a model that is free in measuring synchronization, and to obtain estimates that are robust to serial correlation and heteroscedasticity. Economic theory only provides us with information about the moments, but no information is provided about the distribution from which the shocks are drawn. Unless one is willing to go beyond the information provided by economic theory, it is not possible to use maximum likelihood (ML) to estimate these models. Hence, we use GMM in our analysis. For the two binary series, the correlation between them is: j ρC N ,U S

= Corr



j j SC N ,t , SU S,t



j

=

j

Cov(SC N ,t , SU S,t ) j

j

V ar (SC N ,t )V ar (SU S,t )

where j = growth cycle, acceleration cycle. The covariance between the two binary series is:        j j j j j j j j j j Cov SC N ,t , SU S,t = E SC N ,t − μC N SU S,t − μU S = E SC N ,t SU S,t − μC N μU S The variance in each binary series is:   j j j V ar Si,t = μi (1 − μi ) where i = China, US.

6.3.1.1

Moment Conditions

We have the following moment conditions in the system.   j j E Si,t − μi = 0

84

6 When the US Sneezes, Will China Catch a Cold?

where i = China, US; j = growth cycle, acceleration cycle. ⎡

⎤ j j j j (SC N ,t − μC N )(SU S,t − μU S ) j E⎣  − ρC N ,U S ⎦ = 0 j j j j μC N (1 − μC N )μU S (1 − μU S ) where j = growth cycle, acceleration cycle.  j j j Let θ = (μC N , μU S , ρC N ,U S ) be a vector of parameters for the population means j j (μC N , μU S ) in the Chinese and US binary states, and the population correlation j j (ρC N ,U S ) between these two binary states. Let S be a Tx2 matrix with elements SC N j and SU S in the two columns, respectively. Then we can write the moment conditions as follows: ⎛ ⎜ m t (θ, St ) = ⎝

j

j

j

j

⎞

SC N ,t −μC N

√

SU S,t −μU S j j j j SC N ,t SU S,t −μC N μU S j

j

j

j

μC N (1−μC N )μU S (1−μU S )

−

j ρC N ,U S

⎟ ⎠

where m t is a Tx3 matrix, and T   1  T {S} g θ, m t (θ, St ) t=1 = T t=1

In the m t matrix, we have 3 moment conditions, and we also have 3 parameters  j j j in the system, θ = (μC N , μU S , ρC N ,U S ), so it is a just identified system. 6.3.1.2

Moment Estimation

Since we have an exactly identified model, we can use the method of moments. The estimators can be solved analytically via the following equations. j

μi = j ρC N ,U S









j

j

1 T

= 

T 1  j S T t=1 i,t

T

j j t=1 (SC N ,t SU S,t

j

j

j

j

− μC N μU S ) j

j

μC N (1 − μC N )μU S (1 − μU S )



j

Let θ = (μC N , μU S , ρC N ,U S ) be the vector of parameters for the sample means in the Chinese and US binary states, and the sample correlation between these two binary states. The next step is to obtain the S matrix.

6.3 Synchronization of Business Cycles

85

The Newey-West (1987) estimate of the S matrix is5 : 



ST = 0,T +

q 

(1 −

v=1

v  )(v,T + v,T ) q +1 



where q is the window width. 

v,T =

T   1  [m t θ , St ][m t (θ, St−v )] T t=v+1 





where θ is an initial consistent estimator of θ .

6.3.2 Single Synchronization Tests In this section, we test for the synchronization of business cycles between China j j j and the US. ρC N ,U S = Corr (SC N ,t , SU S,t ). To test the correlation of business cycles between China and the US, we have the null hypothesis (H0 ) and the alternative hypothesis (H1 ) as follows: j

H0 : ρC N ,U S = 0 j

H1 : ρC N ,U S = 0 where j = growth cycle, acceleration cycle. 



Following Sect. 6.3.1, we let θ =





j j (μC N , μU S , 0)

be the restricted parameter 

j

vector for the (strict) non-synchronization (SNS) case, where ρC N ,U S = 0. Under the null hypothesis H0 , the test statistic is: W = −

√

−1 √



T T g(θ, {S}t=1 ) S T

T T g(θ, {S}t=1 )

Under the alternative hypothesis H1 , the model is exactly identified, and hence −

−

m (θ ) = 0. The test statistic is a J test, the form of which is: J =m (θ )ST−1 m (θ ) . The J test is distributed χ12 asymptotically. For both types of business cycles, the test statistic is for the null hypothesis of non-synchronization. As there is only one restriction here, the test statistic has a χ12 distribution at 1 degree of freedom.

5 See

Hamilton (1994: p. 414), equations [14.1.19] and [14.1.20].

86 Table 6.3 Single synchronization tests–China and the US, (1992Q1-2019Q4)

6 When the US Sneezes, Will China Catch a Cold? Growth cycle

Acceleration cycle

μC N

0.6964

0.5179

μU S

0.5268

0.4286

ρ C N,U S

0.2688

0.1857

Test statistic

4.4379

2.0289

p-value

0.0637

0.1543

The Results: In Table 6.3, we present the results from the single synchronization tests in the growth cycles and acceleration cycles between China and the US, for the sample time period from 1992Q1 to 2019Q4, which is before the outbreak of the coronavirus. For the growth cycles, we find evidence of positive correlation with a coefficient of 0.27, a χ12 test statistic of 3.4, and a p-value of 0.06 at the 10% significance level. Hence, we reject the null hypothesis of non-synchronization, accept the alternative hypothesis and conclude that the growth cycles are synchronized between China and the US. Recall the definition of a growth cycle from Sect. 6.2. In a growth cycle, the peaks measure the points at which economic growth moves from above trend rate to below trend rate, while the troughs measure the points at which growth moves from below trend rate to above trend rate for a sustained period of time. So the positive correlation of their growth cycles between China and the US implies that when the US economy grows above its trend rate, the Chinese economy will also grow above its own trend growth rate. For the acceleration cycles, we find that their contemporaneous cycle correlation is 0.19 with a χ12 test statistic of 2.0 and a p-value of 0.15, which is not significant. Thus, we fail to reject the null hypothesis and conclude that the acceleration cycles are non-synchronized between China and the US.

6.3.3 Single Synchronization Tests—Extension In Sect. 6.3.2, we conduct single synchronization tests in the contemporaneous binary states of the business cycles, and find evidence of growth cycle synchronization between China and the US. In this section, we conduct correlation tests in the leads and lags of the binary states for the US, through which we would like to investigate further the relationship in the business cycles between China and the US. In the extension, we fix the binary states for China, but lag and lead the binary states for the US by 1, 2, 3, 4, 5, 6, 7, 8 periods, respectively. As an illustration, if we lag the US binary states by 8 periods, and if we find evidence of synchronization, then this means the US business cycle leads that of China by 2 years. Conversely, if we lead the US binary states by 8 periods, and if we find evidence of synchronization, then it means the US cycle lags that of China by 2 years.

6.3 Synchronization of Business Cycles

87

The correlation between the binary states becomes: S, j

j

j

ρC N ,U S = Corr (SC N ,t , SU S,t+l ) where l = –8, –7, –6, –5, –4, –3, –2, –1, 0, 1, 2, 3, 4, 5, 6, 7, 8. The Results: In Table 6.4 and Chart 6.3, we provide a summary of the results from the crosstime correlation tests between the binary states in the growth cycles and acceleration cycles for China and the US, for the sample time period from 1992Q1 to 2019Q4. For the growth cycles, we find strong evidence of positive correlation between China and the US, from the lags of 3 quarters to the leads of 8 quarters. The interpretation of this finding is that the US growth cycle leads that of China by 3 quarters, and China’s growth cycle leads that of the US by 8 quarters. So there is interdependence and synchronization between China and the US in their growth cycles, with spillover effects on each other. If one economy grows above its trend rate, then the other economy will also grow above its own trend rate. For the acceleration cycles, we find evidence of negative correlation between China and the US at the lags of 7 to 8 quarters, and at the leads of 5 to 7 quarters. In an acceleration cycle, the peaks measure the dates at which the rate of economic Table 6.4 Cross-time correlation tests–China and the US, (1992Q1–2019Q4) Growth cycle Correlation

Acceleration cycle p-value

Correlation

p-value

– 8

0.08830845

0.547861442

– 0.417693169

0.004359366

– 7

0.072362723

0.630452263

– 0.288905647

0.022720696

– 6

0.086539496

0.558373424

– 0.124281609

0.324469145

– 5

0.140843798

0.31313083

– 0.038420927

0.752660013

– 4

0.193223737

0.163577994

0.008304548

0.939608

– 3

0.256263692

0.075357762

0.016989687

0.899129749

– 2

0.299931722

0.048350697

0.06208416

0.644190047

– 1

0.303857887

0.044127607

0.142674524

0.274754811

0

0.26877586

0.063716701

0.185695338

0.154333571

1

0.237819465

0.107721399

0.085549073

0.495277512

2

0.246017934

0.104056812

– 0.052668352

0.702343201

3

0.294233716

0.042747668

– 0.119026416

0.389857421

4

0.343345708

0.01742395

– 0.149071198

0.175262109

5

0.39340221

0.008396904

– 0.254657556

0.013926339

6

0.389948681

0.008499649

– 0.324037319

0.01142625

7

0.386387056

0.013213564

– 0.261682248

0.059152465

8

0.300476479

0.038402066

– 0.120717495

0.375118574

88

6 When the US Sneezes, Will China Catch a Cold?

Chart 6.3 Cross-time correlation tests–China and the US, (1992Q1–2019Q4)

growth begins to slow down, whilst the troughs measure the dates at which the rate of economic growth begins to increase. The interpretation of the negative correlation in their acceleration cycles is that if the rate of economic growth in the US economy starts to slow down, the Chinese economy will be resilient. So when the US sneezes, China will not catch a cold.

6.3.4 Joint Synchronization Tests In this section, we conduct joint non-synchronization tests, with the null hypothesis that both the growth cycle and the acceleration cycle are jointly non-synchronized, and the alternative hypothesis that these two business cycles are jointly synchronized. H0 : ρCGN ,U S = ρCAN ,U S = 0 In this joint non-synchronization test, the moment conditions become: ⎛

SCGN ,t −μCG N SUG S,t −μUG S A SC N ,t − μCA N SUA S,t − μUA S G G SC N ,t SU S,t −μCG N μUG S

⎞

⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ ⎜ ⎟ m t (θ, St ) = ⎜ ⎟ ⎜ G ⎟ ⎜√ − ρ C N ,U S ⎟ ⎜ μCG N (1−μCG N )μUG S (1−μUG S ) ⎠ ⎝ A A A A S −μ μ S √ AC N ,t U S,tA CA N U S A − ρCAN ,U S μC N (1−μC N )μU S (1−μU S )

6.3 Synchronization of Business Cycles Table 6.5 Joint Synchronization Tests–China and the US, (1992Q1–2019Q4)

89 Growth cycle

μCG N G μU S G ρC N ,U S

0.6964

Test statistic

4.5977

p-value

0.2037

0.5268 0.2688

Acceleration cycle μCA N μUA S ρCAN ,U S

0.5179 0.4286 0.1857



where θ = (μCG N , μUG S , μCA N , μUA S , ρCGN ,U S , ρCAN ,U S ), and S = (SCGN ∼ SUG S ∼ SCAN ∼ SUA S ). In Table 6.5 the results for the joint non-synchronization tests are summarized. We find that the test statistics is 4.6, which is distributed Chi square with 3 degrees of freedom χ32 , and it has a p-value of 0.20. So we fail to the reject the null hypothesis and conclude that their growth cycles and acceleration cycles are not jointly synchronized for China and the US. This is consistent with what find in the single synchronization tests (as reported in Table 6.3), in which we find that their growth cycles are synchronized but their acceleration cycles are non-synchronized.

6.4 Conclusion In this chapter, we examine the synchronization of business cycles between China and the US. To that end, the concept of business cycles is explored and the comparison among them is made with various definitions that exist in the literature. We then measure their business cycles and apply the turning point approach to date the business cycles for China and the US. With the binary states from cycle dating, we conduct single synchronization tests (with an extension) and joint synchronization tests under a GMM framework. From our synchronization tests, we find strong evidence of positive correlation in the growth cycles between China and the US, with spillover effects on each other. In addition, we find evidence of negative correlation in the acceleration cycles between China and the US across time, which implies that when the rate of US economic growth begins to slow down, the Chinese economy will be resilient and will not follow the US into a contraction. So when the US sneezes, China will not catch a cold.

90

6 When the US Sneezes, Will China Catch a Cold?

References Boehm, Ernst A., and Geoffrey H. Moore 1984. New Economic Indicators for Australia, 1949–84. Australian Economic Review, 4th Quarter, pp. 34–56. Bry, C., and C. Boschan. 1971. Cyclical Analysis of Time Series: Selected Procedures and Computer Programs. Cambridge MA: National Bureau of Economic Research (NBER). Burns, A.F., and W.C. Mitchell. 1946. Measuring Business Cycles. Cambridge MA: National Bureau of Economic Research (NBER). Hamilton, James D. 1994. Time Series Analysis. Princeton, NJ: Princeton University Press. Harding, D., and A. Pagan. 2002. Dissecting the Cycle: A Methodological Investigation. Journal of Monetary Economics 49: 365–381. Harding, D., and A. Pagan. 2006. Synchronization of Cycles. Journal of Econometrics 132: 59–79. IMF. 2002. Recessions and Recoveries. International Monetary Fund (IMF), Washington DC, April: IMF World Economic Outlook. Mintz, I. 1969. Dating Postwar Business Cycles: Methods and Their Application to Western Germany, 1950–1967”, NBER Occasional Paper, No. 107, National Bureau of Economic Research (NBER), Cambridge MA. Mintz, I. 1972. “Dating American Growth Cycles”, in Zarnowitz, V. (ed.), The Business Cycle Today, National Bureau of Economic Research (NBER), Cambridge MA. Newey, Whitney K., and Kenneth D. West. 1987. A Simple Positive Semi- Definite, Heteroskedasticity and Autocorrelation Consistent Covariance Matrix. Econometrica 55: 703–708.

Chapter 7

Growth Cycles in the BRICS

7.1 Introduction Almost two decades have passed since the acronym ‘BRIC’ was first coined by Jim O’Neill (2001) of Goldman Sachs back in the year 2001. The four BRIC countries— Brazil, Russia, India and China, were identified as four major emerging market economies. At end-2000, GDP in US$ on a PPP basis in BRIC was about 23.3% of world GDP. O’Neill (2001) predicted that the weight of the BRICs and especially China in world GDP would grow, raising important issues about the global economic impact of the BRICs. The cooperation within the BRIC countries was formalized in the year 2009, with a full-scale diplomatic meeting held in Yekaterinburg, Russia, on June 16, 2009. The BRIC grouping’s first formal summit had the focus on improving the global economic situation and reforming financial institutions. There was further discussion on how the four BRIC countries could better cooperate in the future. In 2010, South Africa made an effort to join the BRIC grouping. The group was renamed BRICS, with the ‘S’ standing for South Africa, to reflect the group’s expanded membership. The five BRICS countries—Brazil, Russia, India, China, and South Africa are major economic forces in their respective regions, and are known for their significant influence on regional affairs. In 2019, these five BRICS nations had a combined GDP on a PPP basis of around US$47 trillion, with China’s growth rate being 6.1%, followed by India 4.2%, Russia 1.3%, Brazil 1.1%, and South Africa 0.2%1 (see Chart 7.1). All the five BRICS nations are also members of the G20.2 The growth and development of the BRICS are of importance for the rest of the world.

1 IMF

(2020), “World Economic Outlook”, International Monetary Fund (IMF), Washington DC, April. 2 See https://g20.org/en/about/Pages/Participants.aspx. © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_7

91

92

7 Growth Cycles in the BRICS

Chart 7.1 BRICS—GDP Growth (The size of the bubble measures the BRICS’ GDP on a PPP basis), 2019. Data Source IMF-World Economic Outlook (WEO) database, April 2020

The BRICS Forum was formed in 2011, which is an independent international organization that encourages commercial, political, and cultural cooperation between the BRICS nations. The five BRICS countries have met annually at formal summits, with member countries taking turns to host, the latest of which was the 11th BRICS summit hosted by Brazil in November 2019. One basis of bilateral relations among the BRICS is mutual benefit. With their economic linkages strengthening, there are also spillover effects from one economy to another. The correlation of growth cycles among the BRICS determines whether these economies have scope for macro cooperation. Macro interdependence among BRICS nations since the formalization of the BRICS is our research interest in this paper. The rest of this paper proceeds as follows. In Sect. 7.2, we present our analytical framework, with measurement of growth cycles in Sect. 7.2.1, and correlation of growth cycles in Sect. 7.2.2. An extension is also conducted to further investigate the cross-time correlation of growth cycles among the BRICS. Section 7.3 concludes and indicates policy implication.

7.2 Analytical Framework To address our research question on the macro interdependence among the BRICS, we follow the analytical framework as presented in Chapter 6 to examine the correlation of growth cycles among BRICS countries.

7.2 Analytical Framework

93

7.2.1 Measurement of Growth Cycles For the quarterly GDP data, we collect China’s quarterly GDP data from China’s National Bureau of Statistics (NBS) and CEIC, with a start date of 1992Q1. Our data source for the other four BRICS countries is from US Federal Reserve Economic Data (FRED). The data series are “Gross Domestic Product by Expenditure in Constant Prices: Total Gross Domestic Product, Chained 2000 National Currency Units, Quarterly, Seasonally Adjusted”. The start date for Brazil is 1996Q1, the start date for Russia is 2003Q1, the start date for India is 1996Q2, and the start date for South Africa is 2010Q1. In Chart 7.2, we present our finding of turning points in the growth cycles for the BRICS. From the common start date of 2010Q1, Brazil has experienced a growth cycle contraction since 2013Q3. Russia experienced a growth cycle contraction in 2010Q3 for two quarters, a growth cycle contraction in 2015Q4 for two quarters, and another growth cycle contraction in 2018Q3 for four quarters, which was the period of the Global Financial Crisis (GFC). India had a growth cycle contraction in 2011Q2

Chart 7.2 BRICS—Turning Points in Growth Cycles. Data Sources China’s National Bureau of Statistics (NBS), CEIC, FRED

94

7 Growth Cycles in the BRICS

and bounced back in 2014Q2, it had a short growth cycle contraction in 2016Q2 for two quarters, and now it has been in a growth cycle contraction again since 2018Q3. In the case of China, it was in a growth cycle contraction from 2012Q1 to 2019Q1. South Africa experienced a growth cycle contraction in 2014Q1 for two quarters, it had another growth cycle contraction in 2015Q2 and bounced back in 2017Q2, and now it has been in a growth cycle contraction since 2018Q1 (Chart 7.3).

7.2.2 Correlation of Growth Cycles In Sect. 7.2.1, we find the turning points in growth cycles for the BRICS. In this section, we use the binary states from the cycle dating, and estimate the parameters that are related to the measurement of growth cycle correlation among the BRICs (Figs. 7.1, 7.2, 7.3, 7.4, 7.5, 7.6 and 7.7). In Table 7.1, we summarize the results of pair-wise growth cycle correlation for the four BRIC countries (Brazil, Russia, India, China), for the time period from 2003Q1 to 2019Q4. The country pairs are Brazil and Russia, Brazil and India, Brazil and China, Russia and India, Russia and China, India and China. We find that the growth cycle correlation between Brazil and China has a coefficient of 0.59 at the 1% significance level, and the cycle correlation between India and China has a coefficient of 0.45 at the 5% significance level. For these two country pairs, we reject the null hypothesis of non-correlation, accept the alternative hypothesis of correlation and conclude that the growth cycles are correlated between Brazil and China, and between India and China. In the growth cycle, the peaks measure the points at which growth moves from above trend rate to below trend rate, while the troughs measure the points at which growth moves from below trend rate to above trend rate for a sustained period of time. In Table 7.1, the interpretation of the significant positive correlation is that when the Chinese economy grows above its trend growth rate, Brazil and India would also grow above their trend rates. The intuition behind the finding on Brazil and China is that as China is Brazil’s largest export destination (26.5%, Fig. 7.8), the growth cycle of Brazil depends highly on that of China. The intuition behind the finding on India and China is that their economic structure is complementary to each other, and hence there are two-way trades between them (Figs. 7.13 and 7.14), which in turn leads to the positive growth cycle correlation between India and China (Figs. 7.9, 7.10, 7.11, 7.12 and 7.15). In Table 7.2, the results of pair-wise growth cycle correlation for the five BRICS countries are summarized, for the time period from 2010Q1 to 2019Q4. The country pairs are Brazil and Russia, Brazil and India, Brazil and China, Brazil and South

7.2 Analytical Framework

Chart 7.3 BRICS—Growth Cycle Correlation

95

96

7 Growth Cycles in the BRICS

Fig. 7.1 Brazil’s export—by product, 2018 (total: $242bn)

Fig. 7.2 Russia’s export—by product, 2018 (total: $427bn)

Fig. 7.3 India’s export—by product, 2018 (total: $326bn)

Africa, Russia and India, Russia and China, Russia and South Africa, India and China, India and South Africa, China and South Africa. For the sample period from 2010 to 2019, we find the growth cycle correlation between Brazil and South Africa to be significant with a coefficient of 0.66 at the 1% significance level. So for the

7.2 Analytical Framework

97

Fig. 7.4 South Africa’s export—by product, 2018 (total: $115bn) Source OEC

Fig. 7.5 Brazil’s export—by destination, 2018 (total: $242bn)

Fig. 7.6 Russia’s export—by destination, 2018 (total: $427bn)

country pair of Brazil and South Africa, we reject the null and accept the alternative hypothesis that their growth cycles are correlated. As we live in a dynamic world, we next move from contemporaneous to cross-time cycle correlation among the BRICS.

98

7 Growth Cycles in the BRICS

Fig. 7.7 India’s export—by destination, 2018 (total: $326bn)

Table 7.1 BRIC—Growth Cycle Correlation, 2003Q1–2019Q4 Correlation p-value

Correlation p-value

Brazil, Russia

Brazil, India

Brazil, China

−0.2224

0.1869

0.5923

0.1902

0.3603

0.0128

Russia, India

Russia, China

−0.0962

−0.1846

0.6218

0.3679 India, China

Correlation p-value

Fig. 7.8 South Africa’s export—by destination, 2018 (total: $115bn)

0.4508 0.0320

7.2 Analytical Framework

Fig. 7.9 Brazil’s exports to China, 2018 (total: $64.3bn)

Fig. 7.10 Russia’s exports to China, 2018 (total: $55.2bn)

Fig. 7.11 India’s exports to China, 2018 (total: $16.6bn)

99

100

7 Growth Cycles in the BRICS

Fig. 7.12 China’s exports to India, 2018 (total: $75.5bn) Source OEC

Fig. 7.13 South Africa’s exports to China, 2018 (total: $18.1bn)

Fig. 7.14 Exporters—crude & refined petroleum, 2018 (total: $1.83tn)

7.3 Extension

101

Fig. 7.15 Importers—crude & refined petroleum, 2018 (total: $1.83tn). Source OEC

Table 7.2 BRICS—Growth Cycle Correlation, 2010Q1–2019Q4 Brazil, Russia

Brazil, India

Brazil, China

Brazil, South Africa

Correlation

0.1048

−0.1370

0.4871

0.6637

p-value

0.5934

0.6136

0.1139

0.0081

Russia, India

Russia, China

Russia, South Africa

Correlation

−0.0503

−0.1120

0.3015

p-value

0.7825

0.4966

0.1332

India, China

India, South Africa

Correlation

0.0056

0.1111

p-value

0.9788

0.6787 China, South Africa

Correlation

0.2195

p-value

0.3469

7.3 Extension Building from the contemporaneous correlation tests, we conduct correlation tests in the leads and lags of the binary states, in order to further examine the relationship between the growth cycles among the BRICS across time. To summarize the findings of our cross-time correlation tests, we report the results of pair-wise growth cycle correlation with leads and lags of up to 8 quarters for the four BRIC countries from 2003Q1 to 2019Q4 in Table 7.3, and for the five BRICS countries form 2010Q1 to 2019Q4 in 7. Table 7.4.3 For the country pair of Brazil and Russia, their growth cycles are negatively correlated at the leads of 4 to 8 quarters, for the period of 2003–2019; and their growth cycle are positively correlated at the lags of 2 to 8 quarters, for the period of 2010–2019. The interpretation for the period of 2010–2019 is that Russia’s growth 3 The

result for the country-pair of Brazil and China is not available as the matrix is not positive definite.

102

7 Growth Cycles in the BRICS

Table 7.3 BRIC—Cycle Correlation—Lead/Lag by 8 quarters, 2003Q1–2019Q4 Brazil, Russia Correlation

Brazil, India p-value

Correlation

Brazil, China p-value

Correlation

p-value

−8

0.076828052 0.638313863

−7

0.03307466

−6

0.060938922 0.697858342 −0.030160929 0.880153635 0.255423625 0.185128972

−5

0.08642212

0.006951992 0.971958968 0.17681305

0.347759879

0.836450112 −0.012280488 0.950521707 0.216583154 0.251869379 0.625727758 −0.046829291 0.819984151 0.293439166 0.140314582

−4 −0.022792311 0.886107275 −0.062406593 0.771597045 0.330719289 0.109138344 −3 −0.127390998 0.444578775 −0.076998051 0.734057406 0.367340977 0.087154337 −2 −0.202776776 0.223668609 −0.028342376 0.896177721 0.403370811 0.07130929 −1 −0.275687252 0.124279944

0.080664045 0.692053707 0.499120902 0.027651629

0

−0.222384975 0.190165777

0.186936049 0.360316735 0.592348878 0.012843151

1

−0.150033879 0.410986467

0.231581745 0.245128436 0.679883607 0.004305434

2

−0.139046932 0.464979178

0.277693204 0.165400212 0.77142344

3

−0.188926711 0.306355667

0.300099451 0.137848216 0.830652817 0.000831176

4

−0.306708251 0.069154153

0.258472551 0.194035211 0.791749806 0.001258976

5

−0.429166254 0.014310259

0.215913467 0.304490198 0.75254943

6

−0.487560341 0.007448209

0.240788889 0.249162531 0.777844468 0.002989085

7

−0.50318907

0.007230022

0.267678204 0.206452547 0.738946696 0.003342868

8

−0.51970115

0.007161259

0.296844459 0.18020857

Russia, India Correlation

Russia, China p-value

Correlation

0.00142675

0.00312238

0.699722426 0.004718982 India, China

p-value

Correlation

p-value

−8 −0.213200716 0.180354233 −0.309173471 0.117816953

0.233779739 0.270130266

−7 −0.341164194 0.042380349 −0.432218151 0.022710548

0.200549222 0.330787479

−6 −0.397326668 0.018595275 −0.552419285 0.006736187

0.170098573 0.433432626

−5 −0.340206909 0.050428761 −0.630528289 0.004111461

0.142089855 0.550560449

−4 −0.218575039 0.210641085 −0.572715214 0.004705954

0.246751291 0.25746542

−3 −0.23203299

0.347604514 0.0948988

0.198913858 −0.517322818 0.009925213

−2 −0.180537944 0.348475191 −0.401055592 0.03812328

0.38256478

−1 −0.106088008 0.594091715 −0.260208665 0.199852362

0.416955601 0.028175788

0.041606375

0

−0.096189478 0.62182441

−0.184637236 0.367921367

0.450834817 0.032040522

1

−0.147779178 0.416659327 −0.197944449 0.327387007

0.383357659 0.046821343

2

−0.200695654 0.276384658 −0.275009549 0.149735127

0.313913971 0.098963262

3

−0.321138211 0.062688909 −0.354478049 0.068382565

0.24232965

4

−0.357370095 0.024931861 −0.316640692 0.105348738

0.131562186 0.525288007

5

−0.327050998 0.041378396 −0.211432084 0.272085824

0.018330889 0.934335561

6

−0.296167247 0.057303847 −0.170399544 0.350228646 −0.032530002 0.882110253

7

−0.115853659 0.469970661 −0.128253617 0.468670516 −0.084417364 0.706740686

8

−0.001283697 0.994178316 −0.08484178

0.631858821 −0.13743154

0.226230355

0.564923556

7.3 Extension

103

Table 7.4 BRICS—Cycle Correlation—Lead/Lag by 8 quarters, 2010Q1–2019Q4 Brazil, Russia Correlation

Brazil, India p-value

Correlation

Brazil, South Africa p-value

Correlation

p-value

−8

0.423659273 0.017166666 −0.08818078

0.737253753 0.619047619 0.033535884

−7

0.404651319 0.015620185 −0.190126438

0.458030317 0.694049177 0.026972657

−6

0.387298335 0.014215107 −0.289354105

0.233142201 0.647951595 0.029244731

−5

0.371390676 0.012938493 −0.38624364

0.086042735 0.606788036 0.036571296

−4

0.356753034 0.011778698 −0.402597403

0.058240524 0.569802882 0.047097042

−3

0.343237617 0.010725013 −0.417209321

0.11019213

−2

0.330718914 0.009767538 −0.319853947

0.237108351 0.724568837 0.009714077

−1

0.210714286 0.172709542 −0.226643897

0.405606073 0.692820323 0.008841765

0

0.104828484 0.593393413 −0.136963567

0.613607936 0.663746518 0.008052847

1

0.089802651 0.652313304 −0.109108945

0.682662174 0.621581561 0.008823165

2

0.073088168 0.718480091 −0.077579111

0.767951705 0.57936546

3

0.01224133

4

−0.05547002

5

−0.133440128 0.525025156

0.048581924

0.832654757 0.452267017 0.012783496

6

−0.227866358 0.32807171

0.106229573

0.607006905 0.409673245 0.014080068

7

−0.261528431 0.26621871

0.175932888

0.316207518 0.366899693 0.015487413

8

−0.302613766 0.197883262

0.262265264

0.070080727 0.323875138 0.016942045

0.952057236 −0.041566866 0.786776616

Russia, India Correlation −8

2.75412E−17 1

0.120246518 0.391003273

Correlation

0.009668661

0.871662925 0.537086156 0.010596464

Russia, China p-value

0.647884323 0.016373069

0.494727445 0.011627174

Russia, South Africa p-value

Correlation

p-value

0.121566135 0.015734941 −0.238095238 0.110749037

−7 −0.080614352 0.747950563 −0.146551724 0.460449657 −0.406838102 0.015193314 −6 −0.088399139 0.716617229 −0.363887061 0.269060161 −0.387298335 0.013826151 −5

0.144841365 0.256779943 −0.464238345 0.050650002 −0.093352006 0.509795889

−4

0.203858877 0.189458361 −0.533600097 0.024493419

0.149071198 0.452873389

−3 −0.022792115 0.915217132 −0.584434803 0.04353554

0.220155384 0.266832294

−2 −0.082572282 0.712971545 −0.471472246 0.045242892

0.285773803 0.159217666

−1 −0.065670202 0.741408474 −0.283392761 0.112755591

0.293948176 0.130826649

−0.050251891 0.782492082 −0.111978502 0.496640679

0.301511345 0.133215194

0 1

0.030929479 0.846549414

0.121629434 0.381337865

2

0.121716124 0.50564081

0.24122532

3

−0.011894652 0.937481734

0.231068515 0.010654306

4

−0.149071198 0.327914359

0.239045722 0.011702051 −0.050964719 0.75999414

5

−0.29036179

0.062686194

0.247593784 0.012855807 −0.192757291 0.151490502

6

−0.281864775 0.064477804

0.256776296 0.014126728 −0.018993429 0.870153187

7

0.043033148 0.724863568

0.009575075

0.266666667 0.015527209

0.397317807 0.051165563 0.5077524

0.009765573

0.234107082 0.093057042

0.333333333 0.145308851 (continued)

104

7 Growth Cycles in the BRICS

Table 7.4 (continued) Russia, India Correlation 8

Russia, China p-value

0.221916762 0.174932874 India, China Correlation

p-value

Correlation

Russia, South Africa p-value

0.277350098 0.017071023

Correlation 0.36705849

p-value 0.05129805

India, South Africa

China, South Africa

Correlation

Correlation

p-value

p-value

−8

0.283654314 0.01632974

0.015873016 0.902640267

0.654653671 0.016977419

−7

0.152564288 0.183380122 −0.022222222 0.863500899

0.619677335 0.015416959

−6

0.058722022 0.745185966 −0.055555556 0.739514804

0.588348405 0.014013015

−5 −0.013001274 0.962892553 −0.08496732

0.693245508

0.560112034 0.012748746

−4

0.070186241 0.805711366 −0.222222222 0.380748079

0.534522484 0.011607637

−3

0.145554894 0.610717848 −0.242690058 0.360430433

0.511217187 0.010575017

−2

0.091345367 0.695579178 −0.155555556 0.564203077

0.404529485 0.04355731

−1

0.045305024 0.830980832

0.031746032 0.902718325

0.30807416

0

0.005627131 0.978764162

0.111111111 0.678685118

0.219458127 0.346936276

1

−0.042510209 0.834293399 −0.061497326 0.791851689

0.160931504 0.490366672

2

−0.098975948 0.62602088

−0.136363636 0.491368018

0.098975948 0.678666727

3

−0.166248168 0.397236319 −0.215151515 0.297927303

0.032526816 0.897301716

4

−0.289343829 0.090644866 −0.181818182 0.323851461 −0.039992322 0.881631593

5

−0.41843061

0.012791731 −0.143356643 0.39468936

6

−0.43643578

0.014039271 −0.052148852 0.713520513 −0.040888768 0.827335711

7

−0.45607017

0.015407085 −0.087705802 0.598800721

8

−0.477566933 0.016906522 −0.12869493

0.593808349

0.157810983

−0.121033648 0.6663034 0.065653216 0.535944973 0.216845437 0.016264883

cycle leads the growth cycle of Brazil by 8 quarters, and this leading effect is significant for 7 quarters before it fades away over time. By product, Russia’s largest export is petroleum (49.5%, Fig. 7.5), and petroleum is also a major export from Brazil (12.11%, Fig. 7.4). As of 2018, Russia is the world’s largest exporter of crude petroleum and refined petroleum (11.6%, Fig. 7.17), and hence its growth cycle reflects the demand for petroleum products in the world market, which is a leading indicator for the growth cycles of other petroleum exporters such as Brazil (1.6%, Fig. 7.17). But for the period of 2003–2009, Russia’s growth cycle lags that of Brazil, with a negative correlation by 4 to 8 quarters. The interpretation of this finding is that if the Russian economy grows below its trend rate, then the Brazilian economy would be resilient and would still grow above its own trend rate after 4 to 8 quarters. The intuition behind this result is that both Brazil and Russia are commodity exporters, and their growth cycles depend less on each other, but more on the demand from the world market, especially from their largest export destination—China (26.5% for Brazil, Fig. 7.8; 12.9% for Russia, Fig. 7.9), which is also the world’s largest importer of petroleum products (12.8%, Fig. 7.18).

7.3 Extension

105

For the country pair of Brazil and India in 2010–2019, their growth cycles are negatively correlated at the lags of 4 to 5 quarters, and positively correlated at the lead of 8 quarters. The interpretation of the negative correlation is that if the Indian economy grows below its trend rate, the Brazilian economy would be resilient and grow above its trend rate after 4 to 5 quarters. The intuition behind this result is that by export destination, Brazil’s largest trading partner is China (26.5%, Fig. 7.8) and India’s largest trading partner is the United States (16%, Fig. 7.10), so Brazil and India have their growth cycles less dependent on each other, but more on their largest export destination—China and the US, respectively. On the other hand, Brazil’s growth cycle leads that of India by 8 quarters. The intuition behind this finding is that Brazil’s growth cycle depends highly on the growth cycle of China, which is a leading indicator for the demand in the world market, including that for India (Fig. 7.16). For the country pair of Brazil and China in 2003–2019, we find evidence of positive contemporaneous positive correlation of 0.59 (Table 7.2), and cross-time positive correlation from the lags of 3 quarters to the leads of 8 quarters (Table 7.3) between their growth cycles. The interpretation of this finding is that China’s growth cycle leads the growth cycle of Brazil by 3 quarters, and this effect lasts for 11 quarters. By export destination, China is Brazil’s largest trading partner (26.5%, Fig. 7.8). By product, Brazil’s largest export to China is soybeans (42.5%, Fig. 7.12). As agriculture commodities like soybeans are seasonal, and it also takes time to order and ship these products, there are lags and persistence in the growth cycle correlation between Brazil and China. For the country pair of Brazil and South Africa in 2010–2019, we find evidence of contemporaneous positive correlation (Table 7.2), and cross-time positive correlation at all lags and leads (Table 7.4) between their growth cycles. The intuition behind this finding is that both Brazil and South Africa have China as their largest export destination with 26.5% for Brazil (Fig. 7.8) and 15.7% for South Africa (Fig. 7.11), and hence their growth cycles are highly correlated with that of China. This is confirmed by our finding of positive correlation in growth cycles between Brazil and China (Tables 7.2 and 7.3), and evidence that China’s growth cycle leads that of South Africa by 8 quarters (Table 7.4). So it is the common dependence on China that drives the positive correlation of growth cycles between Brazil and South Africa. For the country pair of Russia and India, their growth cycles are negatively correlated at the lags of 5 to 7 quarters and at the leads of 3 to 6 quarters for the period of 2003–2019; and the negative correlation is significant at the leads of 5 to 6 quarters for the period of 2010–2019. The interpretation of the negative correlation is that if the Russian economy grows below its trend rate, the Indian economy would be resilient and grow above its trend rate. By product, Russia’s largest export is petroleum (49.5%, Fig. 7.5), and India’s largest export is refined petroleum (12.7%, Fig. 7.6), so to some extent there is competition effect between their major exports. By export destination, Russia’s largest trading partner is China (12.9%, Fig. 7.9) and India’s largest trading partner is the United States (16%, Fig. 7.10), and hence Russia and India have their growth cycles less dependent on each other, but more dependent on their largest trading partners—China and the US, respectively.

106

7 Growth Cycles in the BRICS

For the country pair of Russia and China, we find evidence of negative correlation at the lags of 2 to 7 quarters and at the leads of 3 to 4 quarter in 2003–2019, and also negative correlation at the lags of 2 to 5 quarters in 2010–2019 between their growth cycles. During the recent decade of 2010–2019, evidence of positive correlation is found at the leads of 2 to 8 quarters, and also at the lag of 8 quarters, that is, China’s growth cycle leads the growth cycle of Russia by 8 quarters. This positive correlation in their growth cycles reflects the fact that by export destination, China has become Russia’s largest trading partner (12.9%, Fig. 7.9), with petroleum (70%, Fig. 7.13) being the largest export product from Russia to China. It is due to the strong economic growth in China and the huge demand from the Chinese market that drive the positive correlation in the growth cycles between Russia and China in the more recent decade of 2010–2019. For the country pair of Russia and South Africa in 2010–2019, their growth cycles are negatively correlated at the lags of 6 to 7 quarters, and positively correlated at the leads of 1 to 3 quarters, and 8 quarters. By product, Russia’s largest export is petroleum (49.5%, Fig. 7.5), and South Africa’s largest export is gold (15.5%, Fig. 7.7), so Russia and South Africa are not in competition with each other, in terms of their major exports. By export destination, both Russia and South Africa have China as their largest trading partner (12.9% for Russia, Fig. 7.9; 15.7% for South Africa, Fig. 7.11). Therefore, the growth cycles of Russia and South Africa depend highly on the demand of their major exports and also on the growth cycle of China. For the country pair of India and China, we find evidence of positive cross-time correlation from lags of 3 quarters to leads of 2 quarters in 2003–2019 between their growth cycles. This is consistent with our finding of contemporaneous correlation in Table 7.2, which is due to the complementary export structure of India and China. For the period of 2010–2019, we find that China’s growth cycle leads that of India by 8 quarters, and this leading effect fades away over time, with a negative correlation after 12 quarters. By comparing the cross-time correlation between the time period of 2003–2019 and 2010–2019, we find that the trend is similar, but the magnitude of correlation is much smaller in the more recent decade of 2010–2019, which partly reflects the weak export dependence between India and China. For the country pair of China and South Africa in 2010–2019, we find significant evidence of positive cross-time correlation at the lags of 1 to 8 quarters, and also at lead of 8 quarters between their growth cycles. The interpretation of this finding is that China’s growth cycle leads the growth cycle of South Africa by 8 quarters, and this leading effect is significant for 8 quarters. By export destination, China is the largest trading partner of South Africa (15.7%, Fig. 7.11). So it is very intuitive that the growth cycle of South Africa is highly correlated with that of China, and the growth cycle of China is a leading indicator for South Africa.

7.4 Conclusion In this chapter, we examine the correlation of growth cycles among the BRICS countries. We first conduct contemporaneous correlation tests among the original four BRIC countries for the time period from 2003Q1 to 2019Q4, and then we

7.4 Conclusion

107

conduct cross-time correlation tests at the lags and leads of 8 quarters among the five BRICS countries for the time period from 2010Q1 to 2019Q4. From our empirical study, there are several main results. First, China is the largest export destination for Brazil (26.5%, Fig. 7.8), Russia (12.9%, Fig. 7.9), and South Africa (15.7%, Fig. 7.11), and hence China’s growth cycle is highly correlated with those of Brazil (0.59 in 2003–2019, Table 7.2), Russia (0.24, China leading by 8 quarters in 2010– 2019, Table 7.4), and South Africa (0.65, China leading by 8 quarters in 2010–2019, Table 7.4). Second, among the BRICS countries, Brazil, Russia, and South Africa are major commodity exporters in the world market. By product, Brazil’s largest export is soybean (13.7%, Fig. 7.4), and its largest export to China is soybean (42.5%, Fig. 7.12). For Russia, its largest export is petroleum (49.5%, Fig. 7.5), and that is also Russia’s largest export to the Chinese market (70%, Fig. 7.13). South Africa’s largest export is gold (15.5%, Fig. 7.7), which is its largest export to China (36.9%, Fig. 7.16). Therefore, the growth cycles in Brazil, Russia, and South Africa depend highly on the growth in the Chinese economy and the domestic demand from the Chinese market. Third, Brazil, Russia, and South Africa have common dependence on the growth cycle of China, which in turns drives the correlation of growth cycles among these three economies over time—Brazil and South Africa (0.66, in 2010–2019, Table 7.4), Russia and South Africa (0.39, Russia leading by 1 quarter in 2010–2019, Table 7.4), Brazil and Russia (0.42, Russia leading by 8 quarters in 2010–2019, Table 7.4). The last result is due to the fact that Russia is the largest exporter of petroleum products in the world market (11.6%, Fig. 7.17), its growth cycle reflects the demand from China, which is Russia’s largest importer of petroleum products. So China’s growth cycle affects Russia’s growth cycle of its major export—petroleum, which in turn drives the growth cycles of other petroleum exporters such as Brazil (1.6%, Fig. 7.17). It has been almost two decades since the acronym BRIC was first coined, and has been over a decade since the cooperation within the BRICS was formalized. The economic linkages have been strengthening among these countries, with spillover effects from one economy to another. The correlation of growth cycles in the BRICS determines the degree of their macro interdependence. As the BRICS are major economic forces in their respective regions, the findings from our research study have policy implication for future cooperation among the BRICS.

Reference O’Neill, J. 2001. Building Better Global Economic BRICs. Global Economics Paper, No. 66, Goldman Sachs, November 30.

Chapter 8

China’s Growth Targeting and Policy Implication for LDCs

8.1 Introduction The current wave of globalization has encouraged economic growth in the world economy and affected all sides of international economic involvement. A new “conventional wisdom” on globalization is that trade and financial openness do not lead to higher economic growth by themselves, in the absence of institutional reforms. So globalization needs to be complemented by institutional reforms, in both developed and developing countries, so as to fully reap its potential benefits. Over the past few decades, developing and emerging market economies have also introduced major policy reforms and some have experienced significant economic growth. However, different policy patterns have led to different growth path for developing countries. Some economies are faced with substantial difficulties, while others have experienced rapid economic growth. It is a useful lesson for policy makers in least developed countries (LDCs) to learn from the successful stories of other developing countries; and it is also important for academic economists to investigate the successful cases of developing countries, abstract from the reality and formulate an analytical framework, from which policy implication can be drawn for LDCs on how to achieve positive and sustainable economic growth. In this paper, we study the case of China with a focus on its economic growth targeting, which can be thought of as an implicit adoption of a nominal GDP (NGDP) targeting rule and a special form of a rule-based monetary policy regime. To that end, we start with a review of the literature on monetary economics and a discussion on rule-based monetary policy settings in Sect. 8.2. In Sect. 8.3, we develop an analytical framework to examine the case of economic growth targeting in China. In our model economy, key features for developing countries include the degree of commitment by the central bank, semi-dependence between the monetary authority and the fiscal authority, and the risk of sovereign default. From our analytical results, we find that under a GDP growth targeting regime, when the central bank is more committed to its growth target, the actual growth rate becomes close to the target level, the inflation rate increases moderately for those countries with flatter supply curves (which holds © Springer Nature Singapore Pte Ltd. 2020 Y. Jiang and Y. Dai, Globalization and Sustainable Growth in China, https://doi.org/10.1007/978-981-15-9825-8_8

109

110

8 China’s Growth Targeting and Policy Implication for LDCs

true for most developing countries), and the social welfare improves. Moreover, higher growth rate is associated with lower tax levied by the fiscal authority, and it is inversely related to the government’s default probability. Section 8.4 concludes and indicates policy implications for developing countries.

8.2 Background: Rule-Based Monetary Policy All central banks are responsible for managing their monetary policy. Monetary policy is broadly defined into two types—discretionary monetary policy, and rulebased monetary policy. In the literature on monetary economics, there is the “inflationary bias” result with the influential papers by Kydland and Prescott (1977), and Barro and Gordon (1983a) predicting that the rate of inflation will be biased toward a higher level under a discretionary monetary policy than under a rule-based monetary policy regime. The reason for this counter-intuitive “inflationary bias” result is due to the time-inconsistency problem, in which a central bank with discretion will have the incentive to raise the inflation ex post, and the public knows this ex ante. To eliminate this “inflationary bias”, it is well established that a credible commitment to a nominal targeting rule is required in a rule-based policy setting. In a rule-based monetary policy regime, central banks (monetary authorities) must set parameters to formulate their monetary policy. It has become a common practice for central banks in many countries around the world, including both developed economies and developing economies, to gravitate toward an approach that is focused on the strong commitment to a binding rule for a nominal anchor. The reasons, for having a simple and publicly announced nominal variable as the monetary target, are transparency and the anchoring of expectation. As long as the public believes the central bank’s promise to target a certain variable—whether it is the money supply, the exchange rate, or the inflation level, this monetary target is established as the nominal anchor, based upon which household and business form their expectation and make their economic decision accordingly. To conduct rule-based monetary policy, the choice of a nominal anchor by central banks is like a “fashion business”. From a historical point of view, “money supply targeting” gained prominence in the 1970s, which was adopted as the main monetary policy regime to bring down inflation by central banks in major advanced economies, including the United Kingdom and the United States. The monetarism school of economic thought maintains that the total amount of money in an economy, that is, the money supply, is the major determinant of GDP in the short run and the price level over the long run. Monetarists believe that the policy objective of central banks is to achieve low and stable inflation, which is best met by targeting the growth rate of money supply. However, the adoption of “money supply targeting” as the main monetary policy regime was ended due to the velocity shocks in the 1980s. Whilst advanced economies adopted “money supply targeting”, the exchange rate was the favored nominal anchor that was adopted to stabilize inflation by central banks in many developing economies in the late 1980s and early 1990s. In an

8.2 Background: Rule-Based Monetary Policy

111

Table 8.1 “Inflation Targeting” countries and years of adoption Country

Year

Country

Year

Country

Year

New Zealand

1990

Hungary

2001

Ghana

2007

Canada

1991

Mexico

2001

Uruguay

2007

United Kingdom

1992

Iceland

2001

Albania

2009

Australia

1993

South Korea

2001

Georgia

2009 2011

Sweden

1993

Norway

2001

Paraguay

Czech Republic

1997

Peru

2002

Uganda

2011

Israel

1997

Philippines

2002

Dominican Republic

2012

Poland

1998

Guatemala

2005

Japan

2013

Brazil

1999

Indonesia

2005

Moldova

2013

Chile

1999

Romania

2005

India

2015

Colombia

1999

Serbia

2006

Kazakhstan

2015

South Africa

2000

Turkey

2006

Russia

2015

Thailand

2000

Armenia

2006

Source IMF AREAER database

“exchange rate targeting” policy regime, the central bank targets the level of the exchange rate of an “anchor” currency (for example, the US dollar), which has low and stable inflation. However, the vulnerability of “exchange rate targeting” policy regime was exposed during the currency crises in the 1990s. After the currency crises of 1994–2001, “inflation targeting” has become the preferred nominal anchor in place of “exchange rate targeting”. According to Mishkin (1999), and Mishkin and Savastano (2001), ‘inflation targeting’ involves the public announcement of a well-defined numerical target for inflation - a small range or a point target, which shows a strong commitment of the central bank to price stability as its prime monetary policy objective, and a high degree of accountability and transparency in its strategy and implementation. In 1990, New Zealand was the first country to adopt inflation targeting. Since then, many monetary authorities around the world have either implicitly or explicitly adopted inflation targeting. The IMF (2019)’s Annual Report on Exchange Arrangements and Exchange Restrictions1 indicates that in 2015, 38 countries were found to be directly targeting inflation, and 26 of these countries could be classified as developing economies2 (see Table 8.1).

1 IMF (2019), “Annual Report on Exchange Arrangements and Exchange Restrictions 2018”, Mone-

tary and Capital Markets (MCM) Department, International Monetary Fund (IMF), Washington DC, April 16. 2 They are: Albania (2009), Armenia (2006), Brazil (1999), Chile (1999), Colombia (1999), Dominican Republic (2012), Georgia (2009), Ghana (2007), Guatemala (2005), Hungary (2001), India (2015), Indonesia (2005), Korea (2001), Mexico (2001), Moldova (2013), Paraguay (2011), Peru (2002), Philippines (2002), Poland (1998), Romania (2005), Russia (2015), Serbia (2006), South Africa (2000), Thailand (2000), Turkey (2006), Uganda (2011), and Uruguay (2007).

112

8 China’s Growth Targeting and Policy Implication for LDCs

In theory, “inflation targeting” should function well, in terms of achieving low and stable inflation, and anchoring expectation by the public. Empirically, there are mixed findings of the effect of inflation targeting, with the effectiveness of inflation targeting largely debated (Walsh 2009). On the one hand, some empirical studies suggest that inflation targeting helps to enhance macroeconomic performance of adopted economies, by keeping a check on both inflation and inflation expectation while managing long-term interest rates. On the other hand, Mishkin and SchmidtHebbel (2007) find no evidence that inflation targeting improves economic performance. Samarina et al. (2014) argue that the impact of inflation targeting on inflation performance depends on the stage of economic advancement that a country is at, and they find that inflation targeting has no effect for advanced economies, but a significant negative impact for emerging and developing economies. As the year 2020 gathers pace, there has been an outbreak of COVID-19 pandemic, which is an infectious disease that is caused by a newly discovered coronavirus. With the pandemic spreading around the world, it induced a sudden and violent shock across the economy. Many governments have implemented massive fiscal stimulus programs, some fiscal authorities have borrowed heavily to finance schemes that support firms and workers, and some governments have financed their fiscal deficits by printing the extra money that they need to cover their public expenditure. Against the backdrop of COVID-19 pandemic and the monetization of fiscal deficits, some economists start to talk about the return of inflation, and the limitation of “inflation targeting”. In such an economic environment, another danger will be that policy makers withdraw stimulus too soon, as central banks that adopt inflation targeting may worry about overshooting their inflation target. Recently (late August 2020), Jerome Powell, the chairman of the US Federal Reserve, announced changes to the Fed’s monetary framework and made its biggest monetary policy shift in decades, from “inflation targeting” to “average inflation targeting” (AIT).3 In the face of persistently low inflation, the Fed’s existing inflation target of 2% will henceforth be an average, and the central bank may pursue efforts to push inflation above the target. Whether and how monetary policy will change in practice is still not clear. But the macroeconomic developments that led the Fed to revise its strategy would definitely also influence other central banks to adjust their own monetary policies. The Fed’s policy change could start a trend and initiate a global shift in central bank practice. For developing and emerging market economies, their monetary authorities might choose to stay put or follow suit, or seek an alternative nominal anchor in conducting their monetary policies. In addition to money supply targeting, exchange rate targeting, and inflation targeting, another potential nominal anchor is “nominal GDP (NGDP) targeting”. Originally proposed in Meade (1978) and Tobin (1980), the idea of directing monetary policy toward targeting nominal GDP attracted interest in the 1980s and the 1990s. See, for example, Bean (1983), West (1986), Frankel (1990), Feldstein and 3 See,

for example, https://www.economist.com/finance-and-economics/2020/08/27/the-fedmakes-its-biggest-inflation-policy-change-in-decades, https://www.economist.com/finance-andeconomics/2020/09/05/will-the-feds-policy-shift-start-a-trend.

8.2 Background: Rule-Based Monetary Policy

113

Stock (1994), Hall and Mankiw (1994), among others. Since the Global Finance Crisis (GFC), there has been a revival of research proposals that central banks should consider targeting the nominal GDP level, including Woodford (2012), Sumner (2014), Frankel (2014), Armenter (2017), Bai, Kirsanova and Leith (2017), Bhandari and Frankel (2017), Dai and Xu (2019). When faced with macroeconomic shocks, an “inflation targeting” regime places the entire burden on output adjustment, while a NGDP targeting rule, as argued in Bhandari and Frankel (2017), has the advantage of absorbing shocks through both the price and output adjustments. Dai and Xu (2019) show another advantage of a NGDP targeting rule is that it has the inherent flexibility as a NGDP target can be divided between a price goal and an output goal. According to the IMF’s AREAER database for its “Annual Report on Exchange Arrangements and Exchange Restrictions”, the People’s Bank of China (PBoC) adopts a monetary policy framework of monetary aggregate target.4 Moreover, the Law of the People’s Republic of China (PRC) on the PBoC stipulates that “the aim of monetary policies must … promote economic growth”. So the nominal anchor of “monetary aggregate”, combined with China’s emphasis on GDP growth target, resembles that of a NGDP targeting rule. In the next section, we examine China’s case of economic growth targeting, which can be thought of as an implicit adoption of a NGDP targeting rule and a special form of a rule-based monetary policy regime.

8.3 Analytical Framework Textbook treatment of monetary policies typically does not distinguish developing countries from developed countries. The literature on monetary policy for developing countries hardly existed until the publication of Agenor and Montiel (1996, 1999). There are important features about developing countries that differentiate them from developed countries, including less developed institution, lower central bank credibility. See, for example, Fraga et al. (2003). This suggests that the optimal design of monetary policy by central banks should not be the same in a systematical way between developing countries and developed countries. As such, we need different macroeconomic models, or at least some variants of existing models, for developing and emerging market economies. In this section, we aim to contribute to the literature on monetary policy for developing countries. To that end, we examine China’s case of economic growth targeting, and attempt to formulate an analytical framework that abstracts from China’s experience, based upon which useful lessons and importantly policy implication can be drawn for LDCs. As a developing country, China has set an annual economic growth target5 , which has driven the nation’s economic policies for decades. Gross domestic product (GDP) 4 See IMF’s AREAER—China country report, Section III.E.2, (position as of September 30, 2019). 5 Due

to the uncertainties associated with the COVID-19 pandemic, the Chinese government does not explicitly set a GDP growth target for the year 2020, which is the first time ever for China in

114

8 China’s Growth Targeting and Policy Implication for LDCs

100 80 60 40 20

20 0

0 20 01 20 02 20 03 20 04 20 05 20 06 20 07 20 08 20 09 20 10 20 11 20 12 20 13 20 14 20 15 20 16 20 17 20 18 20 19

0 -20 -40

Consumption

Investment

Net Exports

-60

Source: National Bureau of Statistics, China. Fig. 8.1 Decomposition of China’s growth (in percentage). Source National Bureau of Statistics, China.

is often considered to be one of the best measures of how well the economy is performing. The national income accounts divide GDP into four broad categories of spending: consumption, investment, government purchases, and net exports, Y = C + I + G + NX. There are two kinds of demand, domestic demand (C, I, G) and external demand (NX). On the one hand, the PBoC can directly affect the domestic demand via the interest rate channel. On the other hand, monetary policy can only exert influence over the external demand via the exchange rate channel. Interest rates and exchange rates are connected by the “interest rate parity” condition. So we omit the exchange rate but focus on the interest rate in our analysis. This modeling strategy is also justified6 by the fact China’s growth has been driven mainly by domestic demand, rather than external demand over the past two decades (see Fig. 8.1). In our model economy, the central bank uses the nominal interest rate as its monetary policy instrument to achieve its policy objective. We follow Barro and Gordon (1983a) to assume that there is a direct connection between the nominal interest rate and the inflation rate, and henceforth we follow Lohmann (1992) and 30 years. However, its budget projections suggest that the Chinese government is implicitly aiming for a GDP growth rate of 5.4% in 2020. 6 On July 31, 2020, Chinese leaders concluded at a meeting of the Political Bureau of the Communist Party of China (CPC) that in the post COVID-19 era China faces a protracted war with long- and medium-term challenges and risks, and the leaders stated that China will pursue a ‘dual circulation’ strategy that is centered on ‘internal circulation’ and is supplemented by ‘external circulation’. For ‘internal circulation’ to be the central driver of China’s economic growth, there must be enough domestic demand in the Chinese economy. This further justifies our modeling strategy of focusing on the domestic demand and the interest rate channel.

8.3 Analytical Framework

115

Walsh (1995) to treat price or the inflation rate as the choice variable by the central bank. We consider a central bank that establishes a target for the economic activity within the country for a given period of time. The central bank chooses its own policy objective function, with the policy goal of inflation stabilization and growth rate targeting. The policy objective function by the central bank is assumed to take the general form as: Q = (π − π ∗ )2 + ω(g − g ∗ )2 , where the variable g is the actual GDP growth rate. The parameter g ∗ is the target GDP growth rate. The variable π is the actual inflation rate, and the parameter π ∗ is the optimal inflation rate. The parameter ω > 0 and it measures the relative importance that is assigned to the policy goal of growth rate targeting as opposed to inflation stabilization by the central bank. One key feature about developing countries, which distinguishes them from developed countries, is the degree of central bank commitment. This also depends on the degree of central bank independence from the central government. To model that, we introduce the parameter δ, with δ ∈ (0, 1], as a proxy for the degree of commitment to economic growth targeting by the central bank. The higher the parameter δ, the more committed the central bank is to growth targeting, and the more credible the central bankers are, as perceived by the general public. In our model setup, we assume that the central bank aims to minimize a quadratic loss function which takes the form as follows: Q = (π − π ∗ )2 + ω(g − δg ∗ )2 To model the supply curve, we combine Okun (1962)’s Law and the Phillips curve. Okun’s Law is an empirical observation, or a “stylized fact”, which links the unemployment rate to the GDP growth rate relative to the trend growth rate or ¯ where the variable μ is the actual “natural” rate of growth, μ − μn = −β(g − g), unemployment rate, and the parameter μn is the “natural” rate of unemployment. The variable g is the actual GDP growth rate, and the parameter g¯ is the trend growth rate or the “natural” rate of growth, which is assumed to be known or can be estimated by the central bank. We assume it to equal to the target growth rate g ∗ as in the policy objective function by the central bank. The expectations-augmented Phillips curve is: π = π e − γ (μ − μn ), where the variable π is the actual inflation rate, and the parameter π e is the expected inflation rate, which we assume to be equal to the optimal inflation rate π ∗ , as in the central bank’s policy objective function. Combining Okun’s Law and the expectations-augmented Phillips curve, we obtain our supply equation, which shows that the economy faces a trade-off between inflation and output growth.7 This is the constraint in the optimization problem by the central bank.   g = g∗ + α π − π ∗ 7 −β(g

− g ∗ ) = μ − μn = − γ1 (π − π e ), g − g ∗ =

1 βγ

(π − π e ).

116

8 China’s Growth Targeting and Policy Implication for LDCs

The Lagrangian of the optimization problem becomes:    L = −[(π − π ∗ )2 + ω g − δg ∗ )2 ]+λ[g − g ∗ − α π − π ∗ The first-order conditions from the optimization problem give us the reaction 1 λ, where λ functions of inflation and output growth, π = π ∗ − α2 λ, and g = δg ∗ + 2ω is the Lagrangian multiplier, which is also known as the “shadow price” of relaxing the constraint in the optimization problem. Solving these two reaction functions along with the constraint, we obtain the Nash equilibrium inflation and growth rates as functions of the parameter δ.8 π = π∗ + g=

αω(δ − 1) ∗ g 1 + α2 ω

1 + α 2 ωδ ∗ g 1 + α2 ω

Next, we examine the effect of central bank commitment on inflation and output growth, by taking the partial derivatives of the Nash equilibrium with respect to the commitment parameter δ. αω ∂π = g∗ > 0 ∂δ 1 + α2 ω α2 ω ∂g = g∗ > 0 ∂δ 1 + α2 ω Two observations can be made. From the first result, the optimality of the degree of central bank’s commitment depends on the slope of the supply curve. The effect of commitment falls into two ranges. On the one hand, for flatter supply curves with α > 1, an increase in δ will cause a trivial increase in π . On the other hand, for steeper supply curves with α ∈ (0, 1), an increase in δ will cause a large increase in π . That is, a more committed central bank (with higher δ) could lead to an overheated economy with higher inflation. By taking into account the commitment of the central bank (as proxied by δ), we find that growth targeting may be optimal only for those countries with flatter supply curves (α > 1), which happens to be true for most developing countries, as they are usually price takers in the world market. From the second result, we observe that an increase in the commitment parameter δ will cause an increase in output growth. That is, the more committed the central 8 From

the Nash equilibrium solution for inflation, we can see that with the commitment parameter δ∈(0,1], the actual inflation rate will be less than the optimal inflation rate, π ≤ π ∗ , and there is a “downward” inflation bias here. This is in contrast to the “upward” inflation bias as in Barro and Gordon (1983a, b), which is a “puzzle” in the literature on monetary economics. So by introducing the degree of central bank commitment, the curious case of “upward” inflation bias disappears in our model. See Dai and Xu (2019) for more on “price bias” in a similar monetary policy setting.

8.3 Analytical Framework

117

bank is to its policy goal of growth targeting, the more likely it will reach its growth target. Lastly, the level of social welfare can be estimated by substituting the Nash equilibrium inflation and growth rates into the policy objective function Q = (π − π ∗ )2 + ω(g − δg ∗ )2 , and taking the negative sign of the quadratic loss func2 tion, W = − ω(1−δ) (g ∗ )2 . The effect of central bank’s commitment on the social 1+α 2 ω welfare can be examined by taking the partial derivative of the welfare with respect to the commitment parameter δ, which we find to be positive. So a more committed central bank will not only cause the growth rate to be close to its target level, but also ∗ 2 > 0. improves social welfare in the economy, ∂∂δW = 2ω(1−δ) 1+α 2 ω (g ) To summarize, we have: Under a GDP growth targeting regime, when the central bank is more committed to its growth target, (1) the actual growth rate becomes close to the target level; (2) the actual inflation rate increases moderately for those countries with flatter supply curves; and (3) the social welfare improves with a more committed central bank in this economy. Model Extension 1: with Fiscal Authority Another key feature about developing countries, which differs them from developed countries, is that their monetary authority and fiscal authority are not completely independent from each other. In other words, there is semi-dependence between the monetary authority and the fiscal authority in developing countries. In this extension, we introduce the fiscal authority in an extended model setup, with the loss function as: Q 2 = ω1 (π − π ∗ )2 + ω2 (g − δg ∗ )2 + ω3 (P G − P G)2 where the parameter P G is a non-negative target provision of public goods that are provided by the fiscal authority; the weight parameters ω1 , ω2 , and ω3 are the weights on the inflation objective, the growth objective, and the objective of minimizing the deviation of public goods provision in the loss function. These weights reflect the relative preference of each policy goal by the government. Following Alesina and Tabellini (1987), the government budget constraint, expressed in real terms, is: PG = τ + π where τ is the tax rate levied by the fiscal authority. This is the second constraint in the optimization problem by the central bank. The Lagrangian of the optimization problem becomes:   L2 = −[ω1 (π − π ∗ )2 + ω2 (g − δg ∗ )2 + ω3 P G − P G)2    + λ1 g − g ∗ − α π − π ∗ + λ2 (P G − τ − π )

118

8 China’s Growth Targeting and Policy Implication for LDCs

The Nash equilibrium solutions for inflation and output growth can be derived from the first-order conditions associated with the optimization problem. π=

  αω2 (δ − 1) ω3 ω1 + α 2 ω2 τ − PG π∗ + g∗ − 2 2 2 ω1 + α ω2 + ω3 ω1 + α ω2 + ω3 ω1 + α ω2 + ω3 g=

  ω1 + α 2 ω2 δ + ω3 ∗ αω3 τ + π∗ − PG g − 2 2 ω1 + α ω2 + ω3 ω1 + α ω2 + ω3

The effect of government tax rate τ on inflation and output growth can be examined by taking the partial derivatives of the Nash equilibrium. ω3 ∂π =− g ↑

Tax rate levied by the fiscal authority

Fiscal authority

p

p ↑=> g ↑

Probability of no sovereign default

  is that with no sovereign default, the return we get is 1 + r g ; and with default, the return we get is 0. The probability-weighted return under the scenario of no   default and default by the government should be equal to the risk-free rate of return 1 + r f . 1+r Solving for r g , we have: r g = p f − 1. The Lagrangian of the optimization problem becomes:   L3 = −[ω1 (π − π ∗ )2 + ω2 (g − δg ∗ )2 + ω3 P G − P G)2        + λ1 g − g ∗ − α π − π ∗ + λ 2 P G − τ − π − 1 + r g d Solving this new optimization problem, we derive the effect of sovereign default > 0 and ∂∂gp > 0. The interpretation is that risk on inflation and output growth as: ∂π ∂p with a lower default probability by the government, the actual inflation and output growth will be higher. The intuition behind this result is that if the government is perceived by the general public to have a lower chance of default, then it will be able to issue more debt d for public spending, which in turn will boost economic growth and put upward pressure on the price level. To summarize, we have: Under a GDP growth targeting regime with semidependence between the monetary authority and fiscal authority and with sovereign default risk, if we have more commitment from the central bank, lower tax rate levied by the fiscal authority, and more credibility from the fiscal authority, then it would be more likely that the GDP growth target would be achieved and sustainable economic growth would be maintained in this economy (Table 8.2).

8.4 Conclusion and Policy Implication Whilst textbook discussion of monetary policies usually comes in the context of developed countries, we contribute to the literature on monetary policy for developing countries in this paper. To formalize the practice of China’s unconventional economic growth targeting in an analytical framework, there are three contributions from our paper. First, we take into account the degree of central bank commitment, which is one key feature that distinguishes developing countries from developed countries. Second, we model the semi-dependence between the monetary authority and the fiscal authority, which is another key feature about developing countries, as their monetary authority is often not completely independent from the central government. Third,

120

8 China’s Growth Targeting and Policy Implication for LDCs

our modeling of sovereign default risk is a feature of some developing countries, especially least developed countries (LDCs). From our analytical results, we find that under a GDP growth targeting regime, when the central bank is more committed to its growth target, the actual growth rate becomes close to the target level; the inflation rate increases moderately for those countries with flatter supply curves. So the optimality of the degree of commitment by the central bank depends on the slope of supply curve, which hold for most developing countries, as they are usually “small open” economies that take prices as given in the world market. The social welfare improves with a more committed central bank in the economy. Our first model extension considers the case of semidependence between the monetary authority and fiscal authority, from which we find that higher growth rate is associated with lower tax levied by the fiscal authority. Our second extension considers the case with sovereign default risk, from which we find economic growth is inversely related to the government’s default probability. Lessons could be drawn from our model results for LDCs as they usually have less committed central banks, semi-dependence between their monetary authorities and fiscal authorities, and the risk of sovereign default. When they evaluate their alternative policy options, the analytical results from our paper can be taken into consideration. A growth target is more likely to be achieved with more commitment from the central bank, some coordination between the monetary authority and fiscal authority, a lower probability of sovereign default and more credibility from the central government. So if LDCs also consider the policy option of growth targeting as China does, then it is critical for them to improve and strengthen the governance, accountability, and institutional capacities of their central banks and fiscal authorities, which are the necessary requirements for them to achieve positive and sustainable economic growth.

References Agenor, P. R., and P. Montiel. 1996. Development Macroeconomics, 1st edition, Princeton University Press. Agenor, P. R., and P. Montiel. 1999. Development Macroeconomics, 2nd edition, Princeton University Press. Alesina, A., and G. Tabellini. 1987. Rules and Discretion with Non-Coordinated Monetary and Fiscal Policies. Economic Inquiry 25 (4): 619–630. Armenter, R. 2017. The Perils of Nominal Targets. The Review of Economic Studies 85 (1): 50–86. Bai, Y., T. Kirsanova, and C. Leith. 2017. Nominal Targeting in an Economy with Government Debt. European Economic Review 94: 103–125. Barro, R., and D.B. Gordon. 1983a. A Positive Theory of Monetary Policy in a Natural Rate Model. Journal of Political Economy 91 (4): 589–610. Barro, R., and D.B. Gordon. 1983b. Rules, Discretion and Reputation in a Model of Monetary Policy. Journal of Monetary Economics 12 (1): 101–121. Bean, C. R. 1983. Targeting Nominal Income: An Appraisal. The Economic Journal 93(372): 806–819, December.

References

121

Bhandari, P., and J.A. Frankel. 2017. Nominal GDP Targeting for Developing Countries. Research in Economics 71 (3): 491–506. Dai, Y., and B. Xu. 2019. Nominal GDP Targeting and Central Bank Conservativeness. Bulletin of Economic Research 71(3): 483–490, July. Feldstein, M., and J. H. Stock. 1994. The Use of a Monetary Aggregate to Target Nominal GDP. In Monetary Policy, ed. N.G. Mankiw, 7–69. The University of Chicago Press. Fraga, A., I. Goldfajn, and A. Minella. 2003. Inflation Targeting in Emerging Market Economies. In NBER Macro Annual 2003, ed. Kenneth Rogoff and Mark Gertler. Cambridge MA: MIT Press. Frankel, J. A. 1990. International Nominal Targeting (INT): A Proposal for Monetary Policy Coordination in the 1990s. The World Economy 13(2): 263–273, June. Frankel, J.A. 2014. Nominal GDP Targeting for Middle-Income Countries. Central Bank Review 14 (3): 1–14. Hall, E. R., and N. G. Mankiw. 1994. Nominal Income Targeting. In Monetary Policy, ed. N.G. Mankiw, 71–94. The University of Chicago Press. Kydland, F.E., and E.C. Prescott. 1977. Rules Rather than Discretion: The Inconsistency of Optimal Plans. Journal of Political Economy 85 (3): 473–491. Lohmann, S. 1992. Optimal Commitment in Monetary Policy: Credibility vs. Flexibility. American Economic Review 82 (1): 273–286. Meade, J. 1978. The Meaning of Internal Balance. The Economic Journal 88(351): 423–435, September. Mishkin, F. S. 1999. International Experiences with Different Monetary Policy Regimes. NBER Working Paper, No. 6965, National Bureau of Economic Research (NBER), Cambridge MA. Mishkin, F.S., and M.A. Savastano. 2001. Monetary Policy Strategies for Latin America. Journal of Development Economics 66 (2): 415–444. Mishkin, F. S., and K. Schmidt-Hebbel. 2007. Does Inflation Targeting Make a Difference. NBER Working Paper, No. 12876, National Bureau of Economic Research (NBER), Cambridge MA. Okun, A. M. 1962. Potential GNP: Its Measurement and Significance. In Proceedings of the Business and Economic Statistics Section of the American Statistical Association, Washington DC. Samarina, A., M. Terpstra, and J. De Haan. 2014. Inflation Targeting and Inflation Performance: A Comparative Analysis. Applied Economics 46(1): 41–56, January. Sumner, S. 2014. Nominal GDP Targeting: A Simple Rule to Improve Fed Performance. Cato Journal 34 (2): 315–337. Tobin, J. 1980. Stabilization Policy Ten Years After. Brookings Papers on Economic Activity, 1: 19–72. Walsh, C.E. 1995. Optimal Contracts for Central Bankers. American Economic Review 85 (1): 150–167. Walsh, C.E. 2009. Inflation Targeting: What Have We Learned? International Finance 12 (2): 195–233. West, K. D. 1986. Targeting Nominal Income: A Note. The Economic Journal 96(384): 1077–1083, December. Woodford, M. 2012. Methods of Policy Accommodation at the Interest-Rate Lower Bound. Paper presented at the Federal Reserve Bank–Kansas City Symposium on ‘The Changing Policy Landscape’, Jackson Hole, Wyoming, August.