The Economics of Transition: Developing and Reforming Emerging Economies 2019053717, 2019053718, 9780367210335, 9780367210342, 9780429264979

Citation preview

“The collapse of communism and the transition to capitalism caused major upheavals not only in the economies of Eastern Europe and the Soviet Union but also in eco­ nomics. This volume provides an invaluable service by surveying the literature on transition in a systematic way, mainly via meta-analysis. It brings authoritative clar­ ity to the controversies about transition and should be read by all economists inter­ ested in transition, development and policy making.” ─ Josef C. Brada, Professor Emeritus, Department of Economics, Arizona State University “With a talented team of contributors, Iwasaki has produced a masterful work that uses meta-analysis to draw on thousands of academic studies to evaluate central issues in transition economics—from privatization and growth to poverty and corrup­ tion. In addition, the chapters provide valuable syntheses of the main debates on these topics. Anyone interested in economic development and transition should read this book.” ─ Timothy Frye, Marshall Shulman Professor of Post-Soviet Politics, Department of Political Science, Columbia University “After the fall of the Berlin Wall in 1989, socialist countries embarked on a monumental transition from centrally planned to market-based economies. This pro­ cess has been one of the greatest transformations of our time but even now many elem­ ents of the process continue to generate heated debates. This important book provides a bird’s eye view on 30 years of research in transition economics. Academics, policy­ makers and students should find this book’s synthesis most informative.” ─ Yuriy Gorodnichenko, Professor, Department of Economics, University of California Berkeley “This book presents one of the most systematic and comprehensive surveys of tran­ sition experiences in the former socialist transition economies. By applying metaanalysis to the extensive literature, it lucidly summarizes what we have learnt from the grand scale of transformation in economic systems. As the most important achievement in surveying wide-ranging issues on transition, it is clearly a must-read for scholars and graduate students in this field.” ─ Byung-Yeon Kim, Professor, Department of Economics, Seoul National University “Thirty years after the fall of the Iron Curtain, the Economics of Transition provides a comprehensive, yet deep, assessment of its economic consequences. It masterfully blends essential theoretical discussions of reform and emerging markets with state-of­ the-art syntheses of all of the related research evidence. For scholars, policy makers and students, Economics of Transition reveals the important lessons of three decades of tran­ sition research like no other scholarly work.” ─ T. D. Stanley, Professor of Meta-Analysis, School of Business and Law, Deakin University

THE ECONOMICS OF TRANSITION

In the last three decades since the fall of the Berlin Wall, there has been a vast amount of study looking at transforming the planned economy to a market economy from both theoretical and empirical aspects. This book provides an overview and insight into transition economies in the recent decades and looks at key economics topics from the so-called “transition strategy debate” to environmental reform. The book also includes an analytical review and meta-analysis of the existing litera­ ture. By integrating theoretical discussions and synthesizing empirical findings in a systematic manner, this book may help to enlighten the debate on the timing, speed, and policy sequence of economic transition. The book will particularly appeal to researchers, policy makers, other practitioners, and under- and post-graduate students who are interested in transition economies in Eastern Europe, the former Soviet Union, Southeast Asia, and China. It aims to be read as an advanced reader. Ichiro Iwasaki is Professor at the Institute of Economic Research of Hitotsubashi University, Tokyo, Japan.

Routledge Advanced Texts in Economics and Finance

27. Regional Economics, Second Edition Roberta Capello 28. Game Theory and Exercises Gisèle Umbhauer 29. Innovation and Technology Business and Economics Approaches Nikos Vernardakis 30. Behavioral Economics, Third Edition Edward Cartwright 31. Applied Econometrics A Practical Guide Chung-ki Min 32. The Economics of Transition Developing and Reforming Emerging Economies Edited by Ichiro Iwasaki For more information about this series, please visit www.routledge.com/Routledge­ Advanced-Texts-in-Economics-and-Finance/book-series/SE0757

THE ECONOMICS OF

TRANSITION

Developing and Reforming Emerging

Economies

Edited by Ichiro Iwasaki

First published 2020 by Routledge 2 Park Square, Milton Park, Abingdon, Oxon OX14 4RN and by Routledge 52 Vanderbilt Avenue, New York, NY 10017 Routledge is an imprint of the Taylor & Francis Group, an informa business © 2020 selection and editorial matter, Ichiro Iwasaki; individual chapters, the contributors The right of Ichiro Iwasaki to be identified as the author of the editorial material, and of the authors for their individual chapters, has been asserted in accordance with sections 77 and 78 of the Copyright, Designs and Patents Act 1988. All rights reserved. No part of this book may be reprinted or reproduced or utilised in any form or by any electronic, mechanical, or other means, now known or hereafter invented, including photocopying and recording, or in any information storage or retrieval system, without permission in writing from the publishers. Trademark notice: Product or corporate names may be trademarks or registered trademarks, and are used only for identification and explanation without intent to infringe. British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library Library of Congress Cataloging-in-Publication Data Names: Iwasaki, Ichiro, editor.

Title: The economics of transition: developing and reforming emerging economies/

edited by Ichiro Iwasaki.

Description: Milton Park, Abingdon, Oxon; New York, NY: Routledge, 2020. |

Series: Routledge advanced texts in economics and finance | Includes

bibliographical references and index.

Identifiers: LCCN 2019053717 (print) | LCCN 2019053718 (ebook) | ISBN

9780367210335 (hardback) | ISBN 9780367210342 (paperback) | ISBN

9780429264979 (ebook)

Subjects: LCSH: Developing countries–Economic conditions–20th century. |

Developing countries–Economic conditions–21st century. | Economic development–

Developing countries–History–20th century. | Economic development–Developing

countries–History–21st century.

Classification: LCC HC59.7 .E3143 2020 (print) | LCC HC59.7 (ebook) | DDC

338.9009172/4–dc23

LC record available at https://lccn.loc.gov/2019053717

LC ebook record available at https://lccn.loc.gov/2019053718

ISBN: 978-0-367-21033-5 (hbk)

ISBN: 978-0-367-21034-2 (pbk)

ISBN: 978-0-429-26497-9 (ebk)

Typeset in Times New Roman

by Integra Software Services Pvt. Ltd.

Visit the eResources: www.routledge.com/9780367210342

Contents

List of figures List of tables List of contributors Preface Abbreviations Chapter 1 The economics of transition: Aim, methodology, and structure

ix xi xiv xv xvii 1

ICHIRO IWASAKI

Chapter 2 Transition strategy debate: Radicalism versus gradualism

25

ICHIRO IWASAKI AND TAKU SUZUKI

Chapter 3 Transformational recession and recovery: Determinants of the J-curved growth path

67

ICHIRO IWASAKI AND KAZUHIRO KUMO

Chapter 4 Economic transition and poverty: Changes in the determinants of poverty

119

KAZUHIRO KUMO

Chapter 5 Social confusion and corruption: Investigating the causes and effects of a breakdown of ethics 145 TAKU SUZUKI AND SATOSHI MIZOBATA

Chapter 6 Privatization, corporate ownership, and enterprise restructuring ICHIRO IWASAKI AND SATOSHI MIZOBATA

179

CONTENTS

Chapter 7 Human resource management in transition

239

NORIO HORIE AND KAZUHIRO KUMO

Chapter 8 The collapse of the COMECON system and trade in transition countries

263

AKIRA UEGAKI AND KAZUHIRO KUMO

Chapter 9 Foreign direct investment in transition economies: Its determinants and macroeconomic impacts

285

ICHIRO IWASAKI AND MASAHIRO TOKUNAGA

Chapter 10 Regime change and environmental reform

329

MASAHIRO TOKUNAGA

Index

374

Figures

1.1 1.2 1.3 1.4 1.5 1.6 2.1 2.2 2.3 2.4 3.1 3.2 3.3 3.4 3.5 3.6 4.1 4.2 4.3 4.4 4.5

4.6

Hierarchical structure of the literature survey Synthesizing results from multiple studies (image) Basic meta-analysis procedure Example of a funnel plot Example of a Galbraith plot Comparison of the ideal types of a socialist planned economy and a

capitalist market economy Frequency distribution of publication years of all searched literature and

the basic collection Breakdown of the basic collection by literature attribute Overall structure of the transition strategy debate Breakdown of the basic collection by debate attitude Output fall and recovery in CEE and FSU countries during 25 years of

transition Growth path of three clusters of transition economies Breakdown of collected estimates by growth-determining variable type Kernel density estimation of partial correlation coefficients and t values

by growth-determining variable type Funnel plot of estimates by growth-determining variable type Galbraith plot of estimates by growth-determining variable type The number of population with income below the poverty line Poverty headcount and Gini coefficients of income in Russia, 1980–2016 Poverty headcount and GNI per capita in Russia, 1980–2016 The number of papers by target country, by keyword searches using

“poverty” and country names of CEE and FSU The number of research article on poverty, targeting transition

economies and published in academic journals,

January 1989–December 2015 The number of poverty studies in general in academic journals, poverty studies on transition economies in journals, and the ratio of poverty

4

4

5

9

10

13

28

30

35

36

72

74

85

88

104

106

120

121

122

126

127

FIGURES

4.7

4.8 5.1 5.2 5.3 6.1 6.2 6.3 6.4 6.5 6.6 7.1 8.1 9.1 9.2 9.3 9.4 9.5 9.6 9.7 9.8 9.9 9.10 9.11 10.1 10.2 10.3

studies in transition economies to poverty studies in general, January 1989 – December 2015 The number of studies on transition economies in general in academic

journals, poverty studies on transition economies in journals, and the

ratio of poverty studies in transition economies to studies on transition

economies in general, January 1989–December 2015 Funnel plots for estimation results of various factors on the risk of

poverty/poverty ratio Change of Corruption Perceptions Index rankings Number of publications in the basic collection by year Breakdown of the basic collection by literature attribute Relationship between privatization policy and enterprise reform in

transition economies Breakdown of collected estimates by basic category of ownership variable Breakdown of collected estimates by aggregated category of ownership

variable Chronological order of partial correlation coefficients by aggregated

category of ownership variable Funnel plot of estimates by aggregated category of ownership variable Galbraith plot of estimates by aggregated category of ownership

variable Number of extracted basic references by publication year Funnel plots of estimates by aggregated category of factors affecting

trade volumes The dynamics of FDI into CEE and FSU countries in 1990-2017 FDI stock and per capita value in CEE and FSU countries in 2017 Relationship between FDI inflow, transition reform, and economic

growth in the CEE and FSU countries Distribution of partial correlation coefficients and t values of the

collected estimates of determinants of FDI Chronological order of partial correlation coefficients and t values of the

collected estimates of determinants of FDI Funnel plot of partial correlation coefficients of collected estimates of

determinants of FDI Galbraith plot of t values collected estimates of determinants of FDI Distribution of partial correlation coefficients and t values of the

collected estimates of macroeconomic impacts of FDI Chronological order of partial correlation coefficients and t values of the

collected estimates of macroeconomic impacts of FDI Funnel plot of partial correlation coefficients of collected estimates of

macroeconomic impacts of FDI Galbraith plot of t values collected estimates of macroeconomic impacts

of FDI Publication year and number of publications of the literature Basic characteristics of selected studies First year and last year of the analysis period of selected studies

128

129

138

148

155

156

180

193

194

197

223

224

242

281

286

287

289

296

296

307

308

312

312

319

320

338

339

344

Tables

1.1 Ranking of 13 leading economic journals and 16 representative journals

in transition economics in IDEAS list of economic journals, their com­ prehensive evaluation scores, and journal grades 2.1 Cross tabulation analysis of the relationship between the debate attitudes

and the literature attributes 2.2 Type and descriptive statistics of variables used in regression estimation

and correlation coefficients between independent variables and two

dependent variables 2.3 Estimation results of ordered probit model on the relationship between

the debate attitudes and the literature attributes 2.4 Estimation results of multinomial logit model on the relationship

between the debate attitudes and the literature attributes in the

gradualism literature 3.1 Length and depth of economic crisis, and recovery speed after the crisis

period in 28 CEE and FSU countries 3.2 Descriptive statistics of the partial correlation coefficients and the t

values of collected estimates and Shapiro–Wilk normality test by

growth-determining variable type 3.3 Synthesis of estimates by growth-determining variable type 3.4 Name, definition, and descriptive statistics of meta-independent

variables 3.5 Meta-regression analysis 3.6 Synthesis of estimates by subcategory of structural change and

transformation policy variable 3.7 Meta-regression analysis using estimates of structural change variable 3.8 Meta-regression analysis using estimates of transformation policy

variable 3.9 Univariate test of publication selection bias by growth-determining

variable type

12

43

49

52

55

68

87

89

90

92

98

99

101

105

TABLES

3.10 Meta-regression analysis of publication selection bias by growth-determining variable type 3.11 Summary of publication selection bias test 4.1 Poverty headcount in transition economies, 1993–2016 4.2 GDP per capita in transition economies, 1989–2010 4.3 Papers, the results of which would be utilized in meta-analysis: Explained variable—poverty risk/poverty ratio 4.4 Reporting style of empirical studies: Journal articles in the US in 1965 vs. Russian journal articles in Russia, 1992–2006 4.5 Meta-synthesis of estimates 4.6 Meta-regression analysis on publication biases and the existence of genuine effects of household size on poverty risks: Comparable with Figure 4.8a 4.7 Meta-regression analysis on publication biases and the existence of genuine effects of educational attainment on poverty risks: Comparable with Figure 4.8b 4.8 Meta-regression analysis on publication biases and the existence of genuine effects of rural residence on poverty risks: Comparable with Figure 4.8c 5.1 Result of vote counting on hypotheses by studies of corruption in transition countries 5.2 Results of examination of hypotheses 6.1 Privatization method and private sector size in transition economies 6.2 Characteristics of privatization methods in terms of mode of distribution of state properties and selection of their acquisitors 6.3 Synthesis of estimates 6.4 Name, definition, and descriptive statistics of meta-independent variables 6.5 Meta-regression analysis using the aggregated category of ownership variable: Base-line estimation 6.6 Meta-regression analysis using the basic category of ownership variable: Base-line estimation 6.7 Meta-regression analysis of the idiosyncrasy of CEE countries: Estimation using the aggregated category of ownership variable 6.8 Meta-regression analysis of the idiosyncrasy of voucher privatization countries: Estimation using the aggregated category of ownership variable 6.9 Meta-regression analysis of the idiosyncrasy of MEBO privatization countries: Estimation using the aggregated category of ownership variable 6.10 Meta-regression analysis of the idiosyncrasy of direct-sale privatization countries: Estimation using the aggregated category of ownership variable 6.11 Meta-regression analysis of the idiosyncrasy of slow-speed privatization countries: Estimation using the aggregated category of ownership 6.12 Meta-regression analysis of publication selection bias by aggregated category of ownership variable

108 111 123 130 131 133 135

139

140

141 160 171 182 189 195 198 202 206 209

212

215

218 220 226

TABLES

6.13 7.1 7.2 7.3 7.4 7.5 7.6 8.1 8.2 8.3 8.4 8.5 8.6 8.7 8.8 8.9 8.10 8.11 9.1 9.2

9.3 9.4 9.5 9.6

9.7 9.8 10.1 10.2 10.3 10.4

Summary of publication selection bias test Location of the institutional affiliations of the first authors Research target regions Research target industries Distribution of the determinants of the transition Descriptive statistics on the variables introduced into the analyses Estimation results Increase in the trade volume Regional structure of trade partners of CEE/FSU, changes by year, in

per cent Steps of literature selection Classification of literatures using econometric analysis Export volume of goods per capita Meta-synthesis of partial correlation coefficient and t values Meta-regression analysis of publication selection bias and genuine

effects by structural change variables Meta-regression analysis of publication selection bias and genuine

effects by structural reform variables Meta-regression analysis of publication selection bias and genuine

effects by GDP variables Meta-regression analysis of publication selection bias and genuine

effects by distance variables Meta-regression analysis of publication selection bias and genuine

effects by EU factors variables Synthesis of collected estimates of determinants of FDI Name, definition, and descriptive statistics of meta-independent

variables used in meta-regression analysis of heterogeneity among

studies of determinants of FDI Meta-regression analysis of heterogeneity among studies of

determinants of FDI Meta-regression analysis of publication selection in the studies on

determinants of FDI Synthesis of collected estimates of macroeconomic impacts of FDI Name, definition, and descriptive statistics of meta-independent

variables used in meta-regression analysis of heterogeneity among

studies of macroeconomic impacts of FDI Meta-regression analysis of heterogeneity among studies of

macroeconomic impacts of FDI Meta-regression analysis of publication selection in the studies on

macroeconomic impacts of FDI Research topics (finely classified) of selected studies and authors’

academic disciplines Cross table for four-point scale evaluations on environmental reform

and basic characteristics Descriptive statistics of dependent and independent variables for

ordinary probit regression analysis Estimation results of ordinary probit regression analysis

229

242

243

243

245

252

254

264

265

266

267

269

273

276

277

278

279

280

297

299

301

309

314

315

316

321

346

351

354

356

Contributors

Norio Horie is Professor at the Center for Far Eastern Studies of University of Toyama, Toyama, Japan. Ichiro Iwasaki is Professor at the Institute of Economic Research of Hitotsubashi University, Tokyo, Japan. Kazuhiro Kumo is Professor at the Institute of Economic Research of Hitotsubashi University, Tokyo, Japan. Satoshi Mizobata is Professor at the Institute of Economic Research of Kyoto University, Kyoto, Japan. Taku Suzuki is Professor at Faculty of Economics of Teikyo University, Tokyo, Japan. Masahiro Tokunaga is Professor at Faculty of Business and Commerce of Kansai University, Osaka, Japan. Akira Uegaki is Professor at Department of Economics of Seinan Gakuin University, Fukuoka, Japan.

Preface

The year 2019 marks the 30th anniversary of the fall of the Berlin Wall in 1989, a historic event symbolizing the end of the communist period and the Cold War. In the last three decades, researchers have produced a vast amount of research regarding the transformation process from the planned system to a market economy from both theoretical and empirical aspects. Today, the knowledge and evidence concerning the former socialist economies in Central and Eastern Europe, the former Soviet Union, and Asia, as well as the Chinese economy are abundant and have contributed greatly to the development of modern economics. This book, titled The Economics of Transi­ tion: Developing and Reforming Emerging Economies, attempts to show an overall perspective of the research activities in the field of transition economics that was built over the last 30 years, while focusing on the nine most important topics in this research area from the so-called “transition strategy debate” to environmental reform. The authors of the book accomplish this aim by conducting an analytical review and meta-analysis of the existing literature. In this sense, this book marks a sharp distinc­ tion from previous textbooks and readers regarding transition economics. The book is designed to serve as an advanced textbook for researchers, policy­ makers, other practitioners, and under- and post-graduate students who are interested in transition economies. Existing textbooks are mainly intended to describe socioeco­ nomic circumstances and issues in the former socialist countries and China. In con­ trast, this book has an essential goal of understanding academic research activities in and outcomes from the study of transition economies over the last three decades. Moreover, the accumulation of research in this field is remarkable, and we can now derive a wealth of knowledge about the transition process from the socialist planned system to a market-oriented economy from prior studies. Some important points of contention still have not been settled. By integrating theoretical discussions and syn­ thesizing empirical findings in a systematic manner, analytical review and meta­

PREFACE

analysis of the relevant literature provide a definite answer to a debate in question. In this book, the authors employ state-of-the-art techniques of the quantitative literature survey approach and clarify what the true attainment of academic research activities on the critical issues of transition economies around the world is. The authors believe that the above objectives have been achieved successfully. However, they would like to leave the final judgment to readers. This book is based on the research outcomes from the project called Toward Com­ parative Transition Economics: A Meta-Analysis of the 30-Year History of PostCommunist Transition to a Market Economy launched in April 2011. The project was financially supported by the Ministry of Education, Culture, Sports, Science and Tech­ nology of Japan (Grant Nos. 23243032, 23330089, 26245034, 15H01849, 17H04533, 19H01478). The authors also appreciate additional support from the Institute of Eco­ nomic Research of Hitotsubashi University and the Institute of Economic Research of Kyoto University through their Joint Usage and Research Center Programs in FY2015, 2016, and 2018 for organizing the research meetings and carrying out the editorial work and English proofreading of the articles contained in this book. Other support that the authors received individually are acknowledged in their chapters. The editor thanks all contributors to this book for their great efforts and tenacious cooperation in the years 2011 to 2019. The editor is also thankful to the staff of the Institute of Economic Research of Hitotsubashi University for their administration of the project; to Eriko Yoshida for her research assistance; and to Tammy Bicket, Dawn Brandon, Akira Ishida, and Mai Shibata for their editorial support in preparing the manuscripts. Furthermore, both the editor and contributors wish to express their deepest respect to the authors of the literature subject to the systematic reviews and meta-analyses conducted in this book. It is not an exaggeration to say that this book can exist only because of the effort of these researchers over the past 30 years. Last, but not least, the authors would all like to express thanks to Yong Ling Lam, Peter Hall, Mark Newby, and Samantha Phua of the editorial staff at Routledge for their kind support and careful coordination of this project.

Abbreviations

2SLS 3SLS AIC ANOVA BIC CE CEE CEO CIS COMECON CPI DF EBRD EC EKC ENGO ESOPs ETEs EU FAT FDI FE FSU GDP GDR GEE

two-stage least squares three-stage least squares Akaike’s information criterion analysis of variance Bayesian information criterion Central Europe Central and Eastern Europe chief executive officer Commonwealth of Independent States Council for Mutual Economic Assistance Corruption Perceptions Index degree of freedom European Bank for Reconstruction and Development European Community environmental Kuznets curve environmental non-governmental organization employee stock ownership plans European transition economies European Union funnel asymmetry test foreign direct investment fixed-effect(s) Former Soviet Union gross domestic product German Democratic Republic generalized estimating equation

ABBREVIATIONS

GLS GMM HRM IMF ISSP IV JEL LSDV LiTS MEBOs MNE MRA NGO OECD OLS PCC PEESE PET PHARE PSB RE RML ROA SD SE SEE SUR UK UNCTAD UNECE US USA US$ WLS

generalized least squares generalized method of moments human resource management International Monetary Fund International Social Survey Programme instrumental variable Journal of Economic Literature least square dummy variable Life in Transition Survey management and employee buyouts multi-national enterprise meta-regression analysis non-governmental organization Organization for Economic Cooperation and Development ordinary least squares partial correlation coefficient precision effect estimate with standard error precision effect test Poland and Hungary Action for Restructuring of the Economy publication selection bias random-effects restricted maximum likelihood return on assets standard deviation standard error South Eastern Europe seemingly unrelated regression United Kingdom United Nations Conference on Trade and Development United Nations Economic Commission for Europe United States United States of America US dollar weighted least squares

1

The economics of transition Aim, methodology, and structure Ichiro Iwasaki

1.1 AIM OF THE BOOK This book contains nine studies in the field of transition economics, focusing mostly on Central and Eastern Europe (CEE), the Russian Federation, Ukraine, and other countries in the former Soviet Union (FSU) but also covering China, which advocates a socialist market economy,1 and other former socialist states in Asia, when they are relevant to the research subjects in question. Although these nine works deal with dis­ tinct topics that may appear unrelated to one another, they all share a common goal. Namely, in this book, we focus our attention on an area of grave concern that has become evident during the great transformation of national economic systems from socialist planned economies to capitalist market economies since 1989 when the fall of the Berlin Wall signaled the end of the Cold War. Our aim is to clarify the overall picture of the research activities that have been undertaken in this area during the last 30 years and to elucidate an underlying theory. The aforementioned countries, which often include China, are collectively referred to as “the former socialist (or post-communist) transition economies.” There exist many textbooks and academic literature that deal extensively with these transition economies. In fact, in the 2010s alone, we can refer to Myant and Drahokoupil 2010, Turley and Luke 2010, Tridico 2011, Roland 2012, Åslund 2013, Hare and Turley 2013, Gevorkyan 2013, Åslund and Djankov 2014, Myant and Drahokoupil 2015, Ateljević and Trivić 2016, Douarin and Mickiewicz 2017, Havlik and Iwasaki 2017, and Gevorkyan 2018. These publications essentially focus on the following three points: (1) the social and economic issues encountered by transition economies, (2) the processes and consequences of structural reforms and economic policies, and (3) the economic activities of national authorities, politicians, bureaucrats, compan­ ies, farms, households, and individuals during the period of economic transition. In other words, they mainly explore the economic circumstances and conditions of the former socialist bloc and the countries comprising it or call attention to the discus­ sion and empirical analysis of specific economic issues. Consequently, none addresses the academic research activities themselves that have been undertaken to study transition economies.

C H A P T E R 1

THE ECONOMICS OF TRANSITION

The ultimate goal of this book is to find out precisely how researchers have argued about the subjects of heated debate that have been central to transition economics over the past three decades, what findings have been accumulated so far, and what conclu­ sions and common understanding have been drawn in the relevant research fields. In this aspect, this book is fundamentally different from most existing textbooks and other academic publications that deal with transition economies.

1.2 METHODOLOGY APPLIED IN THE BOOK

To achieve the aforementioned aim, this book elaborates on the development of major research in the field of transition economics and the achievements that have been made in the past 30 years. Consequently, a comprehensive review of precedent research works forms the foundation of the discussions that unfold in each chapter of the book. In this respect, our approach is not far from that of the majority of existing textbooks. However, this book extensively incorporates quantitative and stat­ istical methods of literature review, which clearly distinguish this volume from simi­ lar publications. As far as textbooks in the field of economics are concerned, the use of such a review methodology is probably unprecedented. It is natural, therefore, that many readers may have difficulty taking in what the quantitative and statistical methods in the literature review actually mean. Hence, let us begin by discussing the forms of literature review. Scientific papers that focus on the identification and review of previous studies in a specific research field are generally referred to as “survey articles.” Survey articles often play a critical role in determining the future direction of research. It is, there­ fore, not unusual for prominent scholars to publish survey articles in the research fields of their interest (e.g. Hahn and Matthews 1964; Amemiya 1984; Shleifer and Vishny 1997). Furthermore, successful survey articles can even boast a citation ratio comparable to, or even greater than, those of representative theoretical papers, experimental reports, case studies, and empirical analyses in the applicable field.2 It is, therefore, no wonder that there are researchers who earn their living by writing survey articles. The overwhelming majority of survey articles produced in the field of economics are so-called “narrative reviews.” A narrative review typically provides a summary of existing studies by stating, for example, that Studies A, B, and C make a certain argument based on certain theoretical considerations or empirical evidence. Survey articles that review empirical studies may regard a certain theory as convincing or not convincing based on the number of articles supporting the theory compared with those against it. Although narrative reviews are quite effective in grasping a general overview of the selected existing studies, they bear the risk of being unduly affected by the subjective views of the authors of the survey article or can suffer a “missed studies” problem, where research works that should be included in the review are neglected due to compelling circumstances, such as space limitations. It goes without

2

METHODOLOGY APPLIED IN THE BOOK

1.2

saying that such shortcomings of narrative reviews can pose a major obstacle for survey articles that aim to gain an accurate perspective of overall research activities. An approach to overcoming or substantially mitigating the shortcomings of narra­ tive review by making full use of quantitative and statistical methods is called a “systematic review.” “Meta-analysis” is a form of systematic review of literature that involves thoroughly conducted statistical/regression analysis of numerically expressed experiment results and empirical evidence. In this respect, Mullen argues that “a narrative review is ill-equipped to take into account the interrelations between significance level, sample size, and effect size, in the manner of a carefully con­ ducted meta-analysis” (1989, p. 7), and he points out that meta-analysis has advan­ tages over narrative review in terms of “precision,” among other things. He state that “objectivity” is another benefit of meta-analysis, explaining that “the rules and standards for including studies in the review process, abstracting results from them, weighing them in the final integration are never made explicit in a traditional narra­ tive review. In contrast, these rules and standards must be made explicit in any meta-analytic review” (ibid.). The third benefit of meta-analysis is “replicability,” and Mullen stresses the relative advantage of meta-analysis over narrative review in terms of replicability, stating that “the conclusions derived from a meta-analytic review of a given research domain are the same conclusions as which anyone would have arrived if they had included the same studies and followed the same rules for integrating study outcomes. On the other hand, because the rules and strategies of the traditional narrative review are not objective and public, the same research domain can be reviewed by two different narrative reviewers and give rise to two very different conclusions” (ibid., p. 8). Depending on the nature of the selected research questions, meta-analysis might not be necessary or applicable for all systematic reviews. According to Mullen (1989), in terms of “precision,” “objectivity,” and “replicability,” a systematic review that does not require the use of meta-analysis lies somewhere between a narrative review and metaanalysis. In other words, this type of literature survey is vulnerable to the influence of the subjective views of analysts, which could be introduced into the review during the process of extracting or classifying study results. This is precisely why this kind of survey article is inferior to meta-analysis in all three aspects of precision, objectivity, and replicability. At the same time, however, both non-meta-analytic systematic review and meta-analysis are not necessarily free of the elements of a narrative review. In sum, these three types of literature review methods constitute a hierarchical structure, as illus­ trated in Figure 1.1.3 As described above, systematic review and meta-analysis are superior to narrative reviews in terms of precision, objectivity, and replicability. The methodological dif­ ferences between them, however, are not limited to these three characteristics. Rather, systematic review and meta-analysis fundamentally differ from narrative review in that their ultimate goal is to “synthesize” evidence. In the context of a literature review, “synthesis” means “summarizing the results of independently conducted primary studies into statistics such as meta-synthesis values and metaregression coefficients.” Figure 1.2 gives an image of how results from multiple 3

C H A P T E R 1

C H A P T E R

THE ECONOMICS OF TRANSITION

Narrative review

Systematic review

1

Meta-analysis

Figure 1.1 Hierarchical structure of the literature survey

Figure 1.2 Synthesizing results from multiple studies (image)

studies are synthesized in literature reviews. Meta-analysis, in particular, “embodies a constellation of different statistical techniques” (Mullen 1989, p. 1) and is developed to achieve suc­ cessful synthesis of research evidence. The procedure, as well as the statis­ tical and quantitative methods involved in meta-analysis, is rigorous and highly advanced. Therefore, in the latter half of this section, we will focus on this topic. Note that some parts of the descriptions provided below are also applicable to system­ atic reviews that do not involve metaanalysis. In Figure 1.3, the basic procedure of meta-analysis of economic literature is diagrammed. This process is strictly applied to the series of chapters in this book that conduct meta-analyses. Each of these chapters provides a detailed description of Step 1 (establishing a research subject), Step 2 (searching for and collecting relevant studies for meta-analysis), and Step 3 (extracting and coding estimates from the selected studies) of meta-analysis in the context of their respective research themes. Therefore, the following subsections— 1.2.1, 1.2.2, and 1.2.3—provide details regarding Step 4 (meta-synthesis of extracted estimates), Step 5 (meta­ regression analysis of heterogeneity across studies), and Step 6 (testing for publication selection bias), respectively.

1.2.1 Meta-synthesis of extracted estimates In this book, we employ the partial correlation coefficient (PCC) and the t value to synthesize estimates extracted from studies subject to meta-analysis. The PCC is a measure of the association of a dependent variable and the independent variable in question when other variables are held constant.

4

METHODOLOGY APPLIED IN THE BOOK

1.2

The PCC is calculated in the following equation: tk rk ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ; k ¼ 1; 2; . . . ; K tk2 þ dfk

ð1:1Þ

1

where tk and dfk denote the t value and the degree of freedom of the k-th estimate, respectively, while K denotes the total number of collected estimates. The qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi ( ) ffi 1 - rk2 =dfk.4 As standard error (SE) of rk is given by Equation 1.1 indicates, the PCC is a unitless measure; hence, empirical results reported in different scales can be compared. To synthesize PCCs, we use the following method: hereinafter, the corresponding population and standard deviation of rk are labeled θk and sk, respectively. We assume that θ1 = θ2 = … = θK = θ, which implies that each study in a meta-analysis estimates the common underlying population effect, and that the estimates differ only by random sampling errors. An asymptotic­ ally efficient estimator of the unknown true population parameter θ is a weighted mean by the inverse variance of each estimate: R¼ R

XK

wr= k¼1 k k

XK k¼1

wk

ð1:2Þ

( )2 w r R r χ 2 ðK - 1Þ k k f k ¼1

ð1:3Þ

which has a chi-square distribution with N − 1 degrees of freedom. The null hypothesis is rejected if Qr exceeds the critical value. In this case, we assume that heterogeneity exists among the studies and adopt a meta random-effects model that incorporates a sampling variation due to an underlying population of effect sizes as well as a study-

Figure 1.3

XK

Basic meta-analysis procedure

of the synwhere wk ¼ 1=vk and vk ¼ s2k . The varianceP R is given by 1= K wk : thesized partial correlation R k ¼1 This is the meta fixed-effect model. Hereinafter, we denote estimates of the meta fixed-effect model using Rf . In order to utilize this method to synthesize PCCs, we need to confirm that the estimates are homogeneous. A homogeneity test uses the Q statistic: Qr ¼

C H A P T E R

5

C H A P T E R 1

THE ECONOMICS OF TRANSITION 2 level sampling error. If the deviation between estimates (is expressed ) as δθ , the uncondi­ 2 u tional variance of the k-th estimate is given as vk ¼ vk þ δθ . In the meta random­ effects model, the population θ is estimated by replacing the weight wk with the weight wuk ¼ 1=vuk in Equation 1.2.5 For the between-studies variance component, we use the method of moments estimator computed by the next equation using the value of the homogeneity test value Qr obtained from Equation 1.3:

^δ2 ¼ θ

K P k¼1

Qr - ðK - 1Þ PK PK u2 u wku k¼1 wk = k-1 wk

ð1:4Þ

Hereinafter, we denote estimates of the meta random-effects model as Rr . To combine t values, we use the next equation: Tw ¼

K X k¼1

rffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi XK wk t k = w2 r Nð0; 1Þ k¼1 k

ð1:5Þ

Here, wk is the weight assigned to the t value of the k-th estimate. As the weight wk in Equation 5, we use a 10-point scale to mirror the quality level of each relevant study ð1 : wk : 10Þ, as is described in Subsection 1.2.4. Moreover, we report not only the combined t value Tw weighted by the quality level of the study, but also the unweighted combined t value Tu . In addition, as a supplemental statistic for evaluating the reliability of the above-mentioned combined t value, we also report Rosenthal’s failsafe N (fsN) as computed by the next formula: PK fsN ðp ¼ 0:05Þ ¼

k¼1 tk 1:645

!2 -K

ð1:6Þ

Rosenthal’s failsafe N denotes the number of studies with an average effect size equal to zero, which needs to be added in order to bring the combined probability level of all studies to the standard significance level to determine the presence or absence of the effect. The larger the value of fsN in Equation 1.6, the more reliable is the estimation of the combined t value. 1.2.2 Meta-regression analysis of heterogeneity across studies Meta-regression analysis (MRA) is the “regression analysis of regression analyses” (Stanley and Jarrell 2005, p. 299). It is an effective tool for exploring how differences in study conditions and research quality explain the heterogeneity among the reported empirical results. In some chapters, we conduct MRA to determine the factors causing heterogeneity across studies, in addition to meta-synthesis of extracted estimates.

6

METHODOLOGY APPLIED IN THE BOOK

1.2

To this end, we estimate a meta-regression model: yk ¼ β0 þ

XN n¼1

βn xkn þ ek ; k ¼ 1; 2; � � � ; K

ð1:7Þ

C H A P T E R 1

where yk is either the PCC (i.e. rk) or the t value of the k-th estimate; xkn denotes a meta-independent variable that captures relevant characteristics of an empirical study and explains its systematic variation from other empirical results in the literature; βn denotes the meta-regression coefficient to be estimated; and ek is the meta-regression disturbance term. There is no clear consensus among meta-analysts about the “best” model for esti­ mating Equation 1.7. In fact, although it is conventional thinking that the method chosen must be either fixed-effect or random-effects model (Borenstein et al. 2009), Gonzalez-Mulé and Aguinis (2017) recommended using the mixed-effects model, except for specific instances where the fixed-effects model is appropriate. However, Stanley and Doucouliagos (2015, 2017) demonstrated that the weighted least squares (WLS) model outperforms the mixed-effects model, especially when there is selective reporting. To check the statistical robustness of coefficient βn in Equation 1.7, we therefore per­ form an MRA using the following five estimators: (1) the cluster-robust ordinary least squares (OLS) estimator, which clusters the collected estimates by study and computes robust standard errors; (2) the cluster-robust weighted least squares (WLS) estimator, which uses either the above-mentioned quality level of the study, the number of observa­ tions (N), the degree of freedom (df), or the inverse of the standard error (1/SE) as an analytical weight; (3) the multilevel mixed-effects estimator; (4) the cluster-robust unbalanced random-effects panel estimator; and (5) the cluster-robust fixed-effects estimator.6 1.2.3 Testing for publication selection bias Assessment of publication selection bias (PSB) is the most unique aspect of metaanalysis (Iwasaki 2020). Publication selection occurs when researchers, reviewers, and editors are inclined to publish research results that are consistent with the conventional view and/or statistically significant. Consequently, larger and more significant effects will be overrepresented in the research record. Stanley and Doucouliagos pointed out that “the real problem of publication selection is not its existence, but the large biases that it can impact upon any summary of empirical economic knowledge, when uncor­ rected” (2012, p. 52). Therefore, examining the likelihood of PSB and the presence of genuine empirical evidence beyond the bias is one of the most important missions of meta-analysis. In this book, we examine this issue by using the funnel plot and the Galbraith plot as well as by estimating a meta-regression model that is designed espe­ cially for this purpose. The funnel plot is a scatter plot that takes into account the effect size on the hori­ zontal axis and the precision of the estimate on the vertical axis. In the absence of

7

C H A P T E R 1

THE ECONOMICS OF TRANSITION

PSB, effect sizes reported by independent studies vary randomly and symmetrically around the true effect. Moreover, according to the statistical theory, the dispersion of effect sizes is negatively correlated with the precision of the estimate. Therefore, the shape of the plot must look like an inverted funnel. If the funnel plot is not bilat­ erally symmetrical but deflected to one side, then an arbitrary manipulation of the study area in question is suspected, in the sense that estimates in favor of a specific conclusion (i.e. estimates with an expected sign) are more frequently published. This is type I PSB. Figure 1.4 shows three examples of a funnel plot. Panels a and b of this figure display symmetrical distributions of estimates. In contrast, the distribution of estimates in Panel c is evidently skewed toward the positive side. From observing these funnel plots, we can conclude that the risk of type I publication selection is low in the study of the technology transfer effect of foreign direct investment (FDI) and the effect of ownership concentration on firm performance; however, this type of bias is highly likely to emerge in studying the effect of foreign ownership on firm performance. A Galbraith plot is a scatter plot with the precision of the estimate on the horizontal axis and the statistical significance on the vertical axis. We use this plot for testing another arbitrary manipulation, in the sense that estimates with higher statistical sig­ nificance are more frequently published, irrespective of their sign. This is classified as type II PSB. In general, the statistic, j(the k-th estimate – the true effect)=SEk j, should not exceed the critical value of ±1.96 by more than 5% of the total estimates. In other words, when a true effect does not exist and there is no publication selection, the reported t values should vary randomly around zero, and 95% of them should be within a range of ±1.96. A Galbraith plot tests whether the above-mentioned relation­ ship can be observed in the statistical significance of the collected estimates, and, thereby, identifies the presence of type II PSB. In Figure 1.5, three examples of a Galbraith plot are exhibited. In Panel a of this figure, we see that most estimates are distributed within the critical value of ±1.96, although it is unlikely that only 5% of the total estimates exceed this threshold. At the same time, Panels b and c indicate the high risk of type II PSB. Consequently, from these Galbraith plots, we can judge that studies of the effect of ownership con­ centration and foreign ownership on firm performance may have a strong tendency to selectively report statistically significant empirical results, while this kind of manipulation is weak in the study of the FDI technology transfer effect. Further, in addition to the above-mentioned two scatter plots, we report estimates of meta-regression models that have been developed to examine in a more rigorous manner the two types of PSB and the presence of the true effect. We can test for type I PSB by regressing the t value of the k-th estimate on the inverse of the standard error (1/SE) using the following equation: tk ¼ β0 þ β1 ð1=SEk Þ þ vk

8

ð1:8Þ

METHODOLOGY APPLIED IN THE BOOK 800

1.2

450

750 400

700 650

350 600

550

1

300

500

250

450

1/SE

1/SE

400 350 300

200

150

250

100

200

150

50

100

50

0

0

–50

–0.12 –0.10 –0.08 –0.06 –0.04 –0.02 0.00 0.02

0.04 0.06 0.08

0.10 0.12

Estimates (r)

(a) Technology transfer effect of foreign direct investment ( K = 625)a

C H A P T E R

–50 –0.7 –0.6 –0.5 –0.4 –0.3 –0.2 –0.1 0.0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

Estimates (r )

(b) Effect of ownership concentration on firm performance ( K = 1376)b

thereby testing the null hypothesis that the intercept term β0 is equal to zero.7 vk is the error term. When the intercept term β0 is statistically significantly different from zero in Equation 1.8, we can conclude that the distribution of the effect sizes is asymmetric. For this reason, this test is called the funnel asymmetry test (FAT). Meanwhile, type II PSB can be tested by estimating the next equation, where the left side of Equation 1.8 is replaced with the absolute t value: jtk j ¼ β0 þ β1 ð1=SEk Þ þ vk

ð1:9Þ

thereby testing the null hypothesis of β0 ¼ 0 in the same way as with the FAT. Figure 1.4 Even if there is a publication selection, Example of a funnel plot

a genuine effect may exist in the available Notes: The solid line in the figures represents the mean of

empirical evidence. Stanley and Doucou­ the estimates whose accuracy is in the top 10%.

a liagos (2012) proposed examining this From Iwasaki and Tokunaga (2016)

b From Iwasaki and Mizobata (2020)

possibility by testing the null hypothesis c From Figure 6.5 in Chapter 6 of this book

that the coefficient β1 is equal to zero in Equation 1.8. The rejection of the null hypothesis implies the presence of a genuine (i.e. statistically significant nonzero) effect. This is the precision effect test (PET). Moreover, they also stated that an estimate of the publication-selection-bias-adjusted effect size can be obtained by estimating the following equation, which has no intercept:

9

C H A P T E R

THE ECONOMICS OF TRANSITION

1

tk ¼ β0 SEk þ β1 ð1=SEk Þ þ vk

ð1:10Þ

thereby obtaining the coefficient β1. This means that if the null hypothesis of β1 ¼ 0 is rejected, then a nonzero effect actually exists in the literature, and the coefficient β1 can be regarded as its estimate. Stanley and Doucouliagos (2012) called this pro­ cedure “the precision effect estimate with standard error” (PEESE) approach.8 To test the robustness of the regression coefficient, we estimate Equations 8–10 using not only the OLS estimator, but also the clusterrobust OLS estimator, the multilevel mixed-effects estimator, and the unbal­ anced panel estimator,9 all of which treat Figure 1.5 possible heterogeneity among the studies. Example of a Galbraith plot To summarize, to test for PSB and the Notes: Solid lines indicate the thresholds of two-sided crit­ presence of a genuine empirical effect, ical values at the 5% significance level ±1.96.

a we take the following four steps: first,

From Iwasaki and Tokunaga (2016) b From Iwasaki and Mizobata (2020)

we test for type I PSB by estimating c From Figure 6.5 in Chapter 6 of this book Equation 1.8 to examine the FAT and type II PSB by estimating Equation 1.9. Second, regardless of the outcome of the PSB tests, we conduct the PET to test the existence of a genuine effect in the collected estimates beyond possible contamination from publication bias. Third, in cases where the null hypothesis of the PET is rejected in Equation 1.8, we obtain an estimate of β1 in Equation 1.10 using the PEESE approach. Finally, if β1 in Equation 1.10 is

10

STRUCTURE OF THE BOOK

1.3

statistically significantly different from zero, we report β1 as the estimate of the publi­ cation-selection-bias-adjusted effect size. In cases where the null hypothesis of PET is accepted, we judge that the literature in question fails to provide sufficient evidence to capture the genuine effect.10 1.2.4 Method for evaluating the quality level of a study Last, we describe the evaluation method used to determine the quality level of the studies subjected to meta-analysis in this book. For journal articles, we used the rankings of economics journals published as of November 1, 2012, by IDEAS—the largest bibliographical database dedicated to economics and available freely on the internet (http://ideas.repec.org)—as the most basic information source for our evaluation of quality level. IDEAS provides the world’s most comprehensive ranking of economics journals: as of November 2012, a total of 1,173 academic journals were ranked. We divided these 1,173 journals into ten clusters, using a cluster analysis based on overall evaluation scores.11 We then assigned each of these journal clusters a score (weight) from 1 (the lowest journal cluster) to 10 (the highest). Table 1.1 shows the ranking of 13 leading economic journals and 16 representa­ tive journals in transition economics in the aforementioned IDEAS list of economic journals, their comprehensive evaluation scores, and journal grades. In this book, we use the journal grade as a proxy for the quality level of a study subject to metaanalysis (i.e. wk in Equation 1.2). In addition, as mentioned in Subsection 1.2.2, the journal grade is also utilized to estimate Equation 1.7 by using the cluster-robust WLS estimator as an analytical weight. For academic journals not ranked by IDEAS, we referred to the Clarivate Analytics (formerly Thomson Reuters) Impact Factor and other journal rankings and identified the same level of IDEAS ranking-listed journals that correspond to these non-listed journals; we assigned each of them the same score as its counterparts. Meanwhile, for academic books and book chapters, we assigned a score of 1, in principle; however, if at least one of the following conditions was met, each of the relevant books or chapters uniformly received a score of 4, which is the median value of the scores assigned to the above-mentioned IDEAS ranking-listed econom­ ics journals: (1) the academic book or book chapter clearly states that it has gone through the peer review process; (2) its publisher is a leading academic publisher that has external evaluations carried out by experts; or (3) the research level of the study was evaluated by the authors as being obviously high.

1.3 STRUCTURE OF THE BOOK

In his masterpiece, Capital: A Critique of Political Economy, Karl Marx drew an ideal type of socialist planned system: “Based on the social ownership of the means of

11

C H A P T E R 1

C H A P T E R

THE ECONOMICS OF TRANSITION Table 1.1 Ranking of 13 leading economic journals and 16 representative journals in transition economics in IDEAS list of economic journals, their comprehensive evalu­ ation scores, and journal grades

1 Journal title Quarterly Journal of Economics American Economic Review Review of Economic Studies Economic Journal European Economic Review World Development International Economic Review Journal of Economic Surveys American Economic Journal: Applied Economics Scandinavian Journal of Economics IMF Economic Review Economica Canadian Journal of Economics Journal of Comparative Economics Economics of Transition Emerging Markets Review China Economic Review Economic Systems Economic Change and Restructuring Comparative Economic Studies Emerging Markets Finance & Trade Journal of Chinese Economic and Business Studies European Journal of Comparative Economics Post-Communist Economies China & World Economy Eastern European Economics Problems of Economic Transition Transition Studies Review International Journal of Economic Policy in Emerging Economies

Ranking in IDEAS list of economic journals

Comprehensive evaluation score Journal grade

1 4 10 16 28 33 38 49

1.56 3.57 11.02 20.51 28.93 33.35 42.49 55.02

10 10 10 10 10 10 9 9

58 70 85 93 97 129 138 162 169 230 362 397 419

64.16 79.03 89.38 95.10 99.71 129.98 137.84 160.99 164.32 216.02 338.54 370.99 393.71

9 9 9 9 9 8 8 7 7 7 5 5 5

438 443 449 457 483 626 663

416.86 421.53 425.82 430.31 456.52 590.06 625.18

5 5 5 5 4 4 3

895

837.22

2

Note: The journal grade is utilized as a proxy for the quality level of a study subject to meta-analysis in this book.

production, workers act as the main players in the management of the economy, engaging in production activities with the basic goal of satisfying the needs of the soci­ ety, establishing a national plan to adjust production and consumption, and contributing their abilities to the workforce and being compensated for it” (Nishimura 1995). As seen in Figure 1.6, the socialist planned economy does indeed fundamentally differ from the capitalist market economy in many aspects, from ownership of the means of production to income distribution. However, the concept of scientific socialism proposed by Marx did not account for a specific mechanism governing workers’ subordination to the social plan, which

12

STRUCTURE OF THE BOOK

1.3

C H A P T E R 1

Figure 1.6 Comparison of the ideal types of a socialist planned economy and a capitalist market economy

is essential to the realization of socialist principles. To address this issue, Soviet pol­ itical leaders, such as Vladimir Illich Lenin, created a mechanism that legally binds people to carry out the social plan established by the state. This is exactly the real socialist economy that we witnessed in the twentieth century. Researchers of socialism often referred to this type of economic mechanism as the “Soviet-type planned system,” which embodied the following three characteristics: (1) ownership of all means of production by the state, (2) a system to allocate pro­ duction assignments and goods produced to workers based on the administrative orders of the central planning authority, and (3) a massive top-down bureaucracy with the Council of Ministers at the top and state-owned enterprises at the bottom of the hierarchy. This type of economic system, where the central government has deci­ sion-making power over not only the overarching national economic issues but also the detailed aspects of company management, was called the “centralized system” (Nove 1986). This centralized system can be extremely effective when it comes to mobilizing resources to selectively enhance the growth of a specific sector in a structurally simple

13

C H A P T E R 1

THE ECONOMICS OF TRANSITION

and underdeveloped economy. In fact, the centralized system was very forceful in promoting modernization in Russia and other Soviet republics that lagged behind Western Europe and the United States economically in the early Soviet period. However, as economic growth was achieved, and, hence, the policy focus shifted from selective industrial development to comprehensive progress of the whole national economy, the centralized system not only lost superiority over the market system but also led the Soviet economy into deep stagnation during the latter half of the twentieth century. No doubt this long-term slump of the Soviet economy was mainly due to inefficient production systems, slow capital accumulation, and the lack of technological innovations. These problems were also commonly observed across socialist countries in CEE. From the moment socialism collapsed, it was obvious what had been wrong with the Soviet-type centralized system and what kind of rules needed to be introduced to replace the socialist planned economy with a capitalist market economy. However, although academic scholars in those days could suggest to policy-makers and citi­ zens which goal to pursue according to Figure 1.6, they failed to show them how to get there. This is exactly what troubled the former socialist countries and is the reason why economists created a new field of study called “transition economics” to understand the realities of their economic situation and to propose possible policy measures for them. The transition economies have been investigated from various aspects over the past three decades. Nevertheless, it was clear which reform measures were of par­ ticular importance for promoting transformation from the planned system to a market economy. As a consequence, many researchers tackled a number of specific issues aggressively and eventually produced several study areas that together consti­ tute the “core” of transition economics. Among these core areas, we identify a total of nine fields that have produced many studies, therefore enabling us to carry out a systematic review or meta-analysis based on the extant literature. The outcomes of this attempt comprise the main body of the book. Next, to show the overall outlook of the book, we provide a short description of the aim and content of each chapter below.12 Chapter 2 discusses the “transition strategy debate.” The transition strategy debate deals with policy arguments over what basic courses of action should be taken by CEE and FSU countries that renounced the socialist planned economy as well as by countries such as China and Cuba, which advocate radical reform of the economic structure while maintaining the one-party rule of the Communist Party, to establish a market-oriented system. This debate received considerable attention around the world during the early 1990s. Surprisingly, it is still ongoing, because this debate actually addresses a rather universal topic that is relevant not only to CEE and FSU countries and China but also to other developing economies and emerging markets in the world. This chapter presents an overview of the transition strategy debate by systematically reviewing 140 earlier works that contributed to the radicalism-versus-gradualism debate, which was the center of the transition strategy debate, and examines the relationship between debate attitudes and 14

STRUCTURE OF THE BOOK

1.3

literature attributes by means of statistical and econometrical methods. The results of the literature survey in this chapter indicate that radicalists maintain a monolithic debate attitude, as a whole, while that of gradualists is more diversi­ fied. In fact, the latter can be divided into slow-paced gradualism, step-by-step gradualism, and eclectic gradualism groups, whose respective presences are almost balanced. There is also a group of researchers who take a middle ground between radicalism and gradualism, keeping their distance from these two extreme view­ points but still participating in the radicalism-versus-gradualism debate. Further­ more, cross-tabulation analysis and regression estimation of a qualitative selection model provide deep insight regarding the relationship between the debate attitudes and the literature attributes in related studies. Chapter 3 touches on macroeconomic dynamics in CEE and FSU countries during the transition period. Immediately after the collapse of socialism, these countries plunged into a serious economic crisis called the “transformational recession” (Kornai 1994) and then slowly recovered from it. In other words, all of these coun­ tries, without exception, followed a J-curved growth path. There were, however, considerable differences among these countries in terms of the depth and length of the crisis and the speed of economic recovery. In relation to this interesting phe­ nomenon, we all acknowledge that not only education levels and human capital investment, which are emphasized in traditional growth theory, but even inputs such as capital and labor were not critical explanatory variables for economic growth rates during the crisis and the initial phase of recovery: those that were critical were unique to the former socialist transition economies and CEE/FSU regions, namely, (1) structural changes in the national economy, (2) the transform­ ation policy toward a market economy, (3) the legacy of socialism as an initial condition, (4) inflation, and (5) regional conflict. Unless the interactions among these five factors are identified, it is hardly possible to understand how the J-curved growth path occurred in the first place. This chapter approaches this issue by comparing these five factors by performing meta-analysis of their effect sizes and statistical significance. To this end, 3,279 estimates extracted from 123 previ­ ous studies were used. The meta-synthesis of the extracted estimates revealed that, while the growth-enhancing effects of structural change and transformation policy were small yet significant, inflation and regional conflict had a highly significant and strongly negative effect on output. In addition, the legacy of socialism might exacerbate the decline in production in the early stages of transition. The metaregression analysis that simultaneously controls for various research conditions and the assessment of publication selection bias provides supporting evidence for the results obtained from the meta-synthesis. Chapter 4 attempts to integrate the findings of studies of poverty in CEE/FSU countries. Research on the increase in poverty in the transitional economies affected by the collapse of socialism began soon after the beginning of the transition to capit­ alism. However, the nature of poverty in the former Soviet Union and Central and Eastern Europe differs, and two phases have been observed: a phase of increasing and stabilizing poverty in the 1990s and a phase of declining poverty in the 2000s. 15

C H A P T E R 1

C H A P T E R 1

THE ECONOMICS OF TRANSITION

Taking into account the possibility that the impact of household size, education level, and urban residence, which are factors employed in traditional poverty research, may differ depending on the year or the region, this chapter attempted a meta-analysis. The results generally supported the hypothesis. In the 1990s, there was no difference in the probability of falling into poverty between urban and rural populations. After 2000, however, urban residence became a significant factor in reducing the probability of falling into poverty. In addition, different factors affected poverty in the former Soviet Union and Central and Eastern Europe. This phenom­ enon is believed to indicate an important future direction for research in comparative transitional economics. Furthermore, the trend in poverty dynamics seen here can probably also be regarded as indicating steady progress in transition. Chapter 5 deals with corruption. The word “corruption” is an ambiguous, broadranging concept that encompasses a wide range of matters. It is sometimes defined as “behavior which deviates from the formal duties of a public role because of privateregarding wealth or status gains” (Nye 1967, p. 419). This definition includes bribery and nepotism but not acts offensive at a moral level, such as murder of the opposition. Other views of corruption, as extensions of this definition, include a “misuse of public office for private gain” (Treisman 2007, p. 360) or the “abuse of entrusted power for private gain” (Transparency International: www.transparency.org). The corruption issue associ­ ated with the transition process was adopted as an important research topic very early on, and such studies have focused mainly on macro and micro studies. Even after all this time, the debate over this issue does not seem to have been settled at all; rather, it seems to be expanding. By taking this research trend into consideration, this chapter carries out a systematic review of 558 previous studies to verify a total of 14 hypotheses that deal with the following three main aspects of the corruption issue: (1) factors causing the cor­ ruption issue, (2) factors influenced by the corruption issue, and (3) culture and values. The findings of this systematic review suggest not only that researchers mostly agree on the causes of corruption and the various factors that may be affected by the corruption issue but also that the overwhelming majority of researchers acknowledge the negative impact of corrupt practices from both social and economic perspectives, although there are a few exceptional cases where researchers could disagree. Nevertheless, almost all researchers are totally opposed to the so-called “greasing-the-wheels hypothesis” that emphasizes the positive aspects of corruption based on the understanding that, during the confusion of the transition period, corruption was one survival tactic that actually contrib­ uted to economic growth to some extent. As such, this chapter describes the details regarding the causes and effects of corruption and elucidates the researchers’ understand­ ing of these matters by summarizing and reviewing previous studies. Chapter 6 examines the privatization and restructuring of enterprises. As described at the beginning of this section, if state ownership of the means of production com­ poses the essence of the socialist planned economy, private ownership is the institu­ tional foundation of the capitalist market economy. This is precisely why the view that the transfer of state-owned enterprises to private companies by itself constitutes a systematic transformation is not necessarily misguided. In fact, among the many policies implemented to facilitate the transition to a market economy, enterprise 16

STRUCTURE OF THE BOOK

1.3

privatization received the greatest attention from researchers, with countless articles published on the topic. The relationship between ownership structure and the per­ formance of privatized companies is a major research topic in this field. However, initial conditions and actual measures implemented to privatize companies vary from country to country. Because many different kinds of individuals and organization ended up acquiring state-owned firms, the previous literature on this topic has dem­ onstrated mixed findings. To grasp the overall picture of this research field, this chapter performs a large meta-analysis of the relationship between post-privatization ownership and firm performance. The baseline estimation of a meta-regression model that employed a total of 2,894 estimates drawn from 121 previous studies indicated the superior impact of foreign ownership on firm performance as compared with state and domestic private entities. However, it did not go so far as to compre­ hensively verify the series of hypotheses concerning the interrelationship between different ownership types. The estimation of an extended meta-regression model that explicitly controls for the idiosyncrasies of transition economies and privatization policies strongly suggested that differences between countries in terms of location, privatization method, and policy implementation speed are causes of the opaqueness seen in the empirical results of the previous literature. The definite evidence of the harmfulness of voucher privatization for ex-post firm performance is one of the most noteworthy empirical findings obtained from the meta-analysis in this chapter. Chapter 7 focuses on the human resource management (HRM) employed by com­ panies in European transition economies. Under the socialist planned economy, enterprises did not have their own HRM. Transplanting Western-style HRM practices that are appropriate to the market economy and making changes to the way in which socialist personnel management was carried out during the socialist era presented new challenges for companies in transition economies. However, the introduction of new HRM practices and a departure from the old socialistic corporate culture was not something that could be achieved overnight. Rather, many enterprises ended up combining the former socialistic practices with new human resource management, which made HRM in European economies unique. This chapter focuses on the art­ icles that cover HRM in European transition economies in the context of transition to the market economy. It empirically verifies the relationship between the existing studies that deal with socialist institutional and cultural legacies and their literature attributes. A systematic review shows the following findings: (1) even after the pass­ ing of 25 years since the transition to a market economy, the socialist legacies in HRM in the European transition economies remain an important research topic; (2) studies of the traditional industries that used to have strategic importance under the socialist system tend to focus on the socialist legacies; and (3) although studies on human resource management divergence based on socialist cultural legacies tend to lose their significance in accordance with their deepening economic integration into the European Union (EU), the socialist institutional legacy could keep contributing to understanding the diversity of European HRM. Socialist legacies are still alive in the HRM of this region. The reader may realize these legacies will, therefore, con­ tinue to be a unique and important research topic for a long time to come. 17

C H A P T E R 1

C H A P T E R 1

THE ECONOMICS OF TRANSITION

Chapter 8 discusses how the transformation of the economic system has changed trade activities in the former socialist bock that were once supported by two systems: the state monopoly over trade and the Council for Mutual Economic Assistance (COMECON). This chapter examines this issue by using meta-analysis to synthesize the empirical results from the previous studies that investigated the determinants of trade volume. States lost their control over trade after the collapse of the socialist system. This would have increased the impact of more common determinants of trade volume, including the size of the market and transportation costs. In view of this, many transition economists began publishing estimates derived from the so-called “gravity model,” which uses distance between two trading countries and economic scale as the most basic independent variables. At the same time, they conducted empirical analyses to test the hypothesis that the progress of systematic transformation can by itself remove the barriers between two countries and contribute to increasing trade volume. Several findings were obtained from the meta-analysis of the verification results published by these previous studies. First, in line with a theoretical hypothesis, both the distance between two trading companies and the gross domestic product (GDP) as a surrogate for economic scale, which were assumed to be the basic determinants of trade volume in the gravity model, had statistically significant impacts on trade volume, even in tran­ sition economies. Second, the hypothesis that the transformation factor (or the structural change variable) works in the direction of increasing the trade volume (in other words, the trade volume increases as transformation progresses) was proved. However, the ana­ lysis failed to capture the true effect of the factors expressed as the “structural reform variable” and the “EU factor variable.” Further details are provided in the chapter. Chapter 9 discusses foreign direct investment. Direct capital flows from abroad are expected not only to increase the financial sources of investment for former socialist countries but also to create the momentum to fundamentally reform their national economies, which were considerably more inefficient than those of Western countries. In the early transition period, however, due to the deep skepticism of for­ eign investors and firms regarding the perspective of the former socialist bloc on the serious economic crisis, foreign investment in this region generally fell far short of expectations throughout the 1990s. The situation surrounding foreign investment changed substantially after the turn of the century. The remarkable progress toward a market economy that resulted in the belief that a return to the old regime would never occur in CEE and FSU countries, a redefinition of the transition economies as emerging markets against the background of a dramatic business recovery, and the psychological effects on foreign investors and multinational enterprises stemming from the accelerating globalization of the world economy all contributed to the rapid growth of FDI into the region. In view of these facts, many researchers investigated the determinants and economic impacts of FDI in transition economies intensively during the last three decades. As a result, there now exists a wealth of empirical literature on these topics that enables us to perform a meta-analysis. Taking advan­ tage of this opportunity, the early part of this chapter examines how transitionspecific factors affect FDI in CEE and FSU countries. The latter part explores how large the impact of FDI on macroeconomic growth in the region is. The results of 18

CONCLUDING REMARKS

1.4

meta-analysis revealed that empirical results reported in previous studies present the close relationship between the progress in transition to a market economy and FDI and a positive effect of FDI on macroeconomic growth in the literature as a whole. This sug­ gests that, in transition economies, the success of transformation toward a marketoriented system and foreign capital flow has created a kind of virtuous cycle. Finally, Chapter 10 focuses on environmental reform. Important political concerns for CEE countries after the revolutions of 1989 include not only the transition to market economies (economic reform) and the promotion of democratic political sys­ tems (political reform) but also the implementation of environmental policies neces­ sary to solve environmental issues (environmental reform). Historically speaking, it has been pointed out that the decades of centralized planned economic systems and iron-fisted dictatorship in these countries not only left both the national economy and the civil society in shambles but also left devastating scars on the natural environment. One of the wishes of the people who led the revolutions of 1989 that resulted in suc­ cessful regime transformation was to end serious industrial pollution so as to restore nature and the living environment to their healthy states. In addition, the accession to the EU proposed after this CEE region-wide change required that candidate countries observe the environmental laws and regulations of the EU, which aim to achieve eco­ nomic growth while preserving the environment: that is, sustainable development. This meant that they had to fundamentally revise their own laws and regulations to ensure consistency with those of the EU. All candidate countries are required to accept and implement the acquis communautaire, which comprises the whole body of EU law, including the treaties, regulations, directives and decisions, and judgments of the Euro­ pean Court of Justice. Membership requirements related to environmental issues, called environmental acquis, necessitated the revision of approximately 450 provisions of the legal system, which imposed tremendous costs and burdens on the candidate countries. Naturally, more than a few candidate countries harbored resentment of the EU for demanding their acceptance of the environmental acquis. Similar issues persist even today. In particular, Southeastern Europe and the Western Balkan region, which are still recovering from the ravages of a civil war that broke out in the former Yugo­ slavia (Bosnian War), are currently undergoing a three-part reconstruction under the supervision of the EU: reconstruction of politics, the economy, and the environment. In view of these facts, and based on a systematic review of previous studies that investigated a number of issues concerning regime transformation and environmental reform in CEE countries, this chapter summarizes various views on the relationship between regime transformation and environmental reform, systematically verifying the factors of the literature that contribute to conflicting views regarding this relationship.

1.4 CONCLUDING REMARKS

In this chapter, the Introduction, we described the book’s aim, methodology, and structure. Our wish is that, by reading through the nine chapters contained in this

19

C H A P T E R 1

C H A P T E R 1

THE ECONOMICS OF TRANSITION

book, readers will be able to comprehend the overall picture of the research activity of the past 30 years in the field of the transition economics. Needless to say, the important research subjects of transition economics are not limited to those covered by this book. From this viewpoint, we strongly encourage readers to read the text­ books and other academic literature mentioned in Section 1.1. We also recommend that readers look at our additional analytical surveys and meta-analyses of the transi­ tion literature, which supplement the arguments in this book.13 It is our sincere hope that this volume will be of great benefit to those wishing to gain a deeper under­ standing of transition economics along with other materials.

NOTES 1 Socialist market economy refers to the policy set by the Communist Party of China, which aims to introduce market principles into the economy while politic­ ally maintaining one-party rule. It was first proposed by Deng Xiaoping during the 14th National Congress of the Communist Party of China, held in the fall of 1992; in the following year of 1993, it was incorporated into the Constitution of the People’s Republic of China as the basic principle governing China’s eco­ nomic policy. 2 For instance, the number of citations of Shleifer and Vishny’s survey article (1997) exceeded 5,000, as of September 2019. That is equivalent to the number of citations of Robert M. Solow’s famous article titled “A Contribution to the Theory of Economic Growth,” published in 1956 (Solow 1956). 3 Some meta-analysts consider that systematic reviews are practically the same as meta-analysis, and they tend to refuse to acknowledge the above-mentioned differences between the two. In economic papers published in international journals, literature reviews that use quantitative and statistical methods but do not use meta-analysis are often referred to as “analytical surveys” or “quantita­ tive reviews”: they are distinguished from typical narrative reviews (e.g. Glo­ wienka 2015; Cornelson and Siow 2016; Hummel and Maedche 2019). Based on what we saw and heard at several workshops and research conferences in the past, many meta-analysts seem to agree that there is a group of “intermedi­ ary” literature reviews that lie in between narrative reviews and meta-analyses, which could be generally referred to as “systematic reviews.” The classification provided in this section reflects the views of such academic meetings and researchers. 4 A benefit of the PCC is that it makes comparing and synthesizing collected esti­ mates easier concerning independent variables of which the definitions or units differ. However, a flaw of the PCC is that its distribution is not normal when the coefficient is close ( to −1 (or1þr+1 )) (Stanley and Doucouliagos 2012, p. 25). Fisher’s is the most well-known solution to this problem. z-transformation z ¼ 12 ln 1-r As in overall economic studies, the PCC of each estimate used for our meta­

20

NOTES

analysis is rarely observed to be close to the upper or lower limit; thus, we use the PCC as calculated in Equation 1 throughout the book. 5 This means that the meta fixed-effect model is a special case based on the assumption that δ2θ ¼ 0. 6 In addition to MRA using these orthodox estimators, some meta-analysts employ several types of model-averaging approaches, including frequentist model aver­ aging and Bayesian model averaging, to tackle the issue of model uncertainty. For instance, see Ahtiainen and Vanhatalo 2012, Babecky and Havranek 2014, and Polák 2019. 7 Equation 1.8 is an alternative to the following meta-regression model that takes the effect size as the dependent variable and the standard error as the independ­ ent variable: Effect sizek ¼ β0 SEk þ β1 þ εk

ð1:8bÞ

More specifically, Equation 1.8 is obtained by dividing both sides of the equation above by the standard error. The error term εk in Equation 1.8b does not often satisfy the assumption of being i.i.d. (independent and identically distributed). In contrast, the error term in Equation 1.8, vk ¼ εk =SEk , is normally distributed: thus, it can be estimated by OLS. Type I PSB can also be detected by estimating Equation ( 1.8b) using the WLS estimator with the inverse of the squared standard error 1=SEk2 as the analytical weight and, thereby, testing the null hypothesis of β0 = 0 (Stanley 2008; Stanley and Doucouliagos 2012, pp. 60–61). 8 We can see that the coefficient β1 in Equation 1.10 may become the estimate of the publication-bias-adjusted effect size in light of the fact that the following equation is obtained when both sides of Equation 1.10 are multiplied by the standard error: Effect sizek ¼ β0 SEk2 þ β1 þ εk

9

10

11 12 13

ð1:10bÞ

When directly estimating Equation 1.10b, the WLS method, with 1=SEk2 as the analytical weight, is used (Stanley and Doucouliagos 2012, pp. 65–67). To estimate Equations 1.8 and 1.9, we use the cluster-robust random-effects esti­ mator or the cluster-robust fixed-effects estimator. With regard to Equation 1.10, which does not have an intercept term, we report the random-effects model as estimated by the maximum likelihood method. As mentioned above, we basically follow the FAT–PET–PEESE approach advo­ cated by Stanley and Doucouliagos (2012, pp. 78–79) as the test procedures for publication selection. However, we also test for type II PSB using Equation 1.9 as our first step, as this kind of bias is very likely to be present in the literature of transition economies, as indicated in Iwasaki and Tokunaga (2014, 2016). For more details regarding these scores, see https://ideas.repec.org/t/ranking.html. The description of each chapter was provided by the respective authors themselves. They include Iwasaki and Tokunaga 2016, on FDI spillover and technology transfer; Iwasaki and Kočenda 2017, on the impacts of voucher privatization on

21

C H A P T E R 1

C H A P T E R

THE ECONOMICS OF TRANSITION

enterprise restructuring in the Czech Republic; Iwasaki et al. 2018, on postprivatization ownership and firm performance in Russia; Mizobata and Horie 2019, on the path dependency of economic transition; Iwasaki and Uegaki 2019, on the disinflation effect of central bank independence; Iwasaki et al. 2019, on corporate ownership and managerial turnover; and Iwasaki and Mizobata 2020, on ownership concentration and firm performance.

1

REFERENCES Ahtiainen, Heini, and Jarno Vanhatalo (2012) The value of reducing eutrophication in Euro­ pean marine areas: A Bayesian meta-analysis. Ecological Economics, 83, pp. 1–10. Amemiya, Takeshi (1984) Tobit models: A survey. Journal of Econometrics, 24(1–2), pp. 3–61. Åslund, Anders (2013) How Capitalism was Built: The Transformation of Central and Eastern Europe, Russia, and Central Asia. Second edition, Cambridge University Press: New York. Åslund, Anders, and Simeon Djankov (2014) (eds.) The Great Rebirth: Lessons from the Vic­ tory of Capitalism over Communism. Peterson Institute for International Economics: Washington DC. Ateljević, Jovo, and Jelena Trivić (2016) Economic Development and Entrepreneurship in Transition Economies: Issues, Obstacles and Perspectives. Springer: Switzerland. Babecky, Jan, and Tomas Havranek (2014) Structural reforms and growth in transition: A meta-analysis. Economics of Transition, 22(1), pp. 13–42. Borenstein, Michael, Larry V. Hedges, Julian P. T. Higgins, and Hannah R. Rothstein (2009) Introduction to Meta-Analysis. Wiley: Chichester. Cornelson, Kirsten, and Aloysius Siow (2016) A quantitative review of marriage markets: How inequality is remaking the American family by Carbone and Cahn. Journal of Eco­ nomic Literature, 54(1), pp. 193–207. Douarin, Elodie, and Tomasz Mickiewicz (2017) Economics of Institutional Change: Central and Eastern Europe Revisited. Third edition, Palgrave Macmillan: Basingstoke. Gevorkyan, Aleksandr V. (2013) Innovative Fiscal Policy and Economic Development in Transition Economies. Routledge: Abingdon. Gevorkyan, Aleksandr V. (2018) Transition Economies: Transformation, Development, and Society in Eastern Europe and the Former Soviet Union. Routledge: Abingdon. Glowienka, Jens (2015) The refusal of the subsistence level for nonresident taxpayers: An analytical survey of the relevant legal framework. Journal of Business Economics, 85(1), pp. 85–106. Gonzalez-Mulé, Erik, and Herman Aguinis (2017) Advancing theory by assessing boundary conditions with metaregression: A critical review and best-practice recommendations. Journal of Management, 44(6), pp. 2246–2273. Hahn, F. H., and R. C. O. Matthews (1964) The theory of economic growth: A survey. Eco­ nomic Journal, 74(296), pp. 779–902. Hare, Paul, and Gerard Turley (2013) (eds.) Handbook of the Economics and Political Econ­ omy of Transition. Routledge: Abingdon and New York.

22

REFERENCES Havlik, Peter, and Ichiro Iwasaki (2017) (eds.) Economics of European Crises and Emerging Markets. Palgrave Macmillan: Singapore. Hummel, Dennis, and Alexander Maedche (2019) How effective is nudging? A quantitative review on the effect sizes and limits of empirical nudging studies. Journal of Behavioral and Experimental Economics, 80, pp. 47–58. Iwasaki, Ichiro (2020) Meta-analysis of emerging markets and economies: An introductory note for the special issue. Emerging Markets Finance & Trade, 56(1), pp. 1–9. Iwasaki, Ichiro, and Evzen Koč enda (2017) Are some owners better than others in Czech privatized firms? Even meta-analysis can’t make us perfectly sure. Economic Systems, 41(1), pp. 537–568. Iwasaki, Ichiro, and Satoshi Mizobata (2020) Ownership concentration and firm perform­ ance in European emerging economies: A meta-analysis. Emerging Markets Finance & Trade, 56(1), pp. 32–67. Iwasaki, Ichiro, and Masahiro Tokunaga (2014) Macroeconomic impacts of FDI in transition economies: A meta-analysis. World Development, 61, pp. 53–69. Iwasaki, Ichiro, and Masahiro Tokunaga (2016) Technology transfer and spillovers from FDI in transition economies: A meta-analysis. Journal of Comparative Economics, 44(4), pp. 1086–1114. Iwasaki, Ichiro, and Akira Uegaki (2019) The disinflation effect of central bank independ­ ence: A comparative meta-analysis between transition economies and the rest of the world. In Julien Chevallier et al. (eds.), International Financial Markets. Volume 1, Routle­ dge: Abingdon, pp. 227–287. Iwasaki, Ichiro, Xinxin Ma, and Satoshi Mizobata (2019) Corporate ownership and manager­ ial turnover in China and Eastern Europe: A comparative meta-analysis. CEI Working Paper No. 2019-1, Center for Economic Institutions, Institute of Economic Research of Hitotsubashi University: Kunitachi, Tokyo. Iwasaki, Ichiro, Satoshi Mizobata, and Alexander Muravyev (2018) Ownership dynamics and firm performance in an emerging economy: A meta-analysis of the Russian literature. Post-Communist Economies, 30(3), 2018, pp. 290–333. Kornai, János (1994) Transformational recession: The main causes. Journal of Comparative Economics, 19(1), pp. 39–63. Mizobata, Satoshi, and Norio Horie (2019) Path-dependency of economic transition: An analytical review. KIER Discussion Paper No. 1014, Institute of Economic Research, Kyoto University: Kyoto. Mullen, Brian (1989) Advanced BASIC Meta-analysis. Lawrence Erlbaum Associates: Hillsdale. Myant, Martin, and Jan Drahokoupil (2010) Transition Economies: Political Economy in Russia, Eastern Europe, and Central Asia. J. Wiley: Hoboken, NJ. Myant, Martin, and Jan Drahokoupil (2015) Transition Economies after 2008: Responses to the Crisis in Russia and Eastern Europe. Routledge: London and New York. Nishimura, Yoshiaki (1995) The ideology, institutions, current status, and reforms of the Soviet-type planned economy. In Hiroshi Kimura et al. (eds.), The Soviet Study. KyoikuSha: Tokyo, pp. 113–188 (Japanese). Nove, Alec (1986) The Soviet Economic System. Allen & Unwin: Boston. Nye, Joseph S. (1967) Corruption and political development: A cost-benefit analysis. Ameri­ can Political Science Review, 61(2), pp. 417–427.

23

C H A P T E R 1

C H A P T E R 1

THE ECONOMICS OF TRANSITION Polák, Petr (2019) The Euro’s trade effect: A meta-analysis. Journal of Economic Surveys, 33(1), pp. 101–124. Roland, Gérard (ed.) (2012) Economies in Transition: The Long-Run View. Palgrave Macmil­ lan: Basingstoke. Shleifer, Andrei, and Robert W. Vishny (1997) A survey of corporate governance. Journal of Finance, 52(2), pp. 737–783. Solow, Robert M. (1956) A contribution to the theory of economic growth. Quarterly Journal of Economics, 70(1), pp. 65–94. Stanley, T. D. (2008) Meta-regression methods for detecting and estimating empirical effects in the presence of publication selection. Oxford Bulletin of Economics and Statis­ tics, 70(1), pp. 103–127. Stanley, T. D., and Hristos Doucouliagos (2012) Meta-regression Analysis in Economics and Business. Routledge: London and New York. Stanley, T. D., and Hristos Doucouliagos (2015) Neither fixed nor random: Weighted least squares meta-analysis. Statistics in Medicine, 34(13), pp. 2116–2127. Stanley, T. D., and Hristos Doucouliagos (2017) Neither fixed nor random: Weighted least squares meta-regression analysis. Research Synthesis Methods, 8(1), pp. 19–42. Stanley, T. D., and Stephen B. Jarrell (2005) Meta-regression analysis: A quantitative method of literature surveys. Journal of Economic Surveys, 19(3), pp. 299–308. Treisman, Daniel (2007) The causes of corruption: A cross-national study. In Erik Berglof and Gérard Roland (eds.), The Economics of Transition: The Fifth Nobel Symposium in Economics. Palgrave Macmillan: New York, pp. 251–271. Tridico, Pasquale (2011) Institutions, Human Development and Economic Growth in Transi­ tion Economies. Palgrave Macmillan: Basingstoke. Turley, Gerard, and Peter J. Luke (2010) Transition Economics: Two Decades On. Routle­ dge: London and New York.

24

2

Transition strategy debate Radicalism versus gradualism Ichiro Iwasaki and Taku Suzuki

2.1 INTRODUCTION If someone becomes excited when he or she hears the term “transition strategy debate,” that person is highly likely to have been engaged in policy practices or research activities related to the countries of Central and Eastern Europe (CEE) and the former Soviet Union (FSU) for at least the past quarter century. These countries renounced a socialist planned economy between the late 1980s and the early 1990s. The debate about what kind of reform track they should adopt toward the creation of a capitalist market economy, which is a more efficient economic system for pursuing further social welfare and economic growth, was so fierce at that time that policy­ makers and researchers who were involved in this debate have strong impressions that still return vividly from time to time. While enjoying great relief and freedom in the wake of the Cold War that had con­ strained the world for as long as 70 years, both the former socialist countries and the rest of the world immediately had to take their next step toward creating a new economic order. The transition strategy debate was about deciding the road map, and, hence, the arguments on this subject involved many leading economists and spread beyond the boundaries of the academic world. A dozen years later, “transition economics” was estab­ lished as a major study area of modern economics, and the pioneering transition strategy debate has been one of the most important subjects in this research field.1 The transition strategy debate has developed as an argument between two conflict­ ing reform philosophies, radicalism and gradualism. Here, “radicalism” denotes a policy philosophy that demands prompt and parallel implementations of the reform packages advocated by the Washington Consensus.2 It is also called “shock therapy” or the “big-bang approach,” reflecting the content of its relevant policy recommenda­ tions. “Gradualism” is a collective term for views antithetical to radicalism, and the reform measures recommended by its advocates are extremely varied. Gradualists, however, show a certain congruity in their debate attitudes toward a transition strat­ egy, approving a milder policy implementation process in terms of speed of reform and/or emphasizing theoretical and practical needs to promote structural reforms in a reasonable policy sequence as compared to the radicalists.

C H A P T E R 2

TRANSITION STRATEGY DEBATE

As we will describe later, some researchers point out that, among transition econ­ omies, some nations have followed a unique reform track that cannot be categorized either as radicalism or as gradualism; some skeptics even question the raison d’être of the transition strategy debate itself. However, they are in a minority, and it is indisputable that the overwhelming majority of people who have participated in the debate so far have stated their opinions, focusing on the validity and relevance of the two contrasting reform philosophies. More than a quarter of a century has now passed since the fall of the Berlin Wall, an event symbolic of the demise of the Communist Bloc: the fever of that time has already become a memory of the past. However, despite Vladimir Popov’s declaring “the end of the debate” (2000a), academic discussion regarding transition strategies has continued, and there is no sign of agreement. It has been pointed out that one of the main reasons for the ongoing irreconcilable debate, in terms of ex-post economic performance, is that there are no obvious, definitive dif­ ferences between countries that have promoted an economic transition that follows radicalism-based policy guidance and countries that have carried out gradualism-based structural reforms. Notwithstanding, from the viewpoint of historical path dependency, the choice of a transition strategy still has significant influence on various levels and aspects of their national economies. Therefore, it is still not the time to sum up the transition strategy debate. Never­ theless, it is possible to provide an overall picture of the debate based on studies accumulated during the past 30 years and to examine the relationship between debate attitudes and literature attributes, such as authorship, research contents, and publication media. These efforts are meaningful research tasks for the future devel­ opment of the debate. Therefore, in this chapter, we will attempt to perform these tasks through an analytical survey of 140 earlier studies that have contributed to the international debate on radicalism versus gradualism. In addition, we will try to unveil some issues to be tackled within the framework of the transition strategy debate by paying attention to the above-mentioned minority’s views. The results of our literature survey indicate that radicalists maintain a monolithic debate attitude, as a whole, while that of gradualists is more diversified. In fact, the latter can be divided into slow-paced gradualism, step-by-step gradualism, and eclectic gradualism groups, whose respective presences are almost balanced. Moreover, we also found that there is another group of researchers who stay within the radicalism­ versus-gradualism debate framework while at the same time staying at arm’s length from both the radicalists and the gradualists. In addition, our cross tabulation analysis and regression estimation of the qualitative selection models provide interesting find­ ings regarding the relationship between debate attitudes and literature attributes in related studies. The remainder of this chapter is organized as follows: the next section will discuss the methodology of the literature search and an outline of the studies subject to our analytical survey. Section 2.3 will look at the overall structure of the transition strat­ egy debate. Section 2.4 will review representative research works in the radicalism­ versus-gradualism debate. Section 2.5 will examine the relationship between the 26

METHODOLOGY OF THE LITERATURE SEARCH

2.2

debate attitudes and the literature attributes by means of statistical and econometrical methods. Finally, Section 2.6 will discuss a future agenda for paving the way toward a deeper debate beyond the traditional dichotomy, referring to heterodox views.

C H A P T E R 2

2.2 METHODOLOGY OF THE LITERATURE SEARCH AND OUTLINE OF THE STUDIES SUBJECT TO ANALYTICAL SURVEY As the first step in identifying relevant studies that argue for radicalism or gradualism as transition strategies from a planned system to a market economy during the period from 1989 to 2018, we searched EconLit, a representative electronic database of eco­ nomic literature.3 We conducted this search using combinations of two keywords or terms including one of the following: “big bang,” “gradualism,” “radicalism,” “shock therapy,” and “Washington Consensus,” which are inseparable from the transition strat­ egy debate, as well as “inflation,” “institution,” “liberalization,” “stabilization,” and “social costs,” which also have deep connections with the debate from the viewpoint of transition policy. We used another keyword or phrase from among the following: “transition economies,” “Central Europe,” “Eastern Europe,” “former Soviet Union,” “China,” or the names of each CEE and FSU country. Then, judging from the paper titles and abstracts, we excluded studies that were irrelevant to our issues and interests in this chapter. As a result of this procedure, we found slightly more than 300 studies. In addition, we also collected studies (centering on books) that are widely regarded as having an important influence on the transition strategy debate, although they were not picked up in our mechanical search, and as many non-overlapping related research works as possible that were cited in the approximately 300 papers above. In this way, we collected a total of 378 studies. Next, we further narrowed our focus to study works that can be subjected to our analytical survey by carefully reading the 378 studies one by one. As a result, we ultimately selected a total of 140 studies from Svejnar 1989 and Lipton and Sachs 1990, both pioneering works regarding transition strategies, to the latest publica­ tions, including Jiang et al. 2016, Mikeladze and Gelashvili 2016, and Hartwell 2017.4 Hereafter, we refer to these 140 studies as the “basic collection.” Figure 2.1 shows the frequency distribution of the publication year of the 378 papers searched and the basic collection. As this figure shows, both are remarkably similar in terms of the composition by publication year. In fact, the correlation coefficient of the number of studies by publication year amounts to 0.819. Figure 2.1 also indicates that the debate on transition strategy had become substantially active immediately after the breakdown of the Soviet Union in late 1991 and that it also had gathered remarkable momentum in 1994, five years after the fall of the Berlin Wall; in 1996, five years after the downfall of the Soviet Union; in 2000, the end of the century; and in the two years between 2009 and 2010, which marked the 20th anniversary of the demise of the Communist Bloc. This suggests that the transition

27

28

Figure 2.1

Frequency distribution of publication years of all searched literature and the basic collection

C H A P T E R

TRANSITION STRATEGY DEBATE

2

METHODOLOGY OF THE LITERATURE SEARCH

2.2

strategy debate was strongly inspired by the exit of the Soviet Union from world history and that transition researchers have continually revived their interest in this issue at each historical milestone. Figure 2.2 shows the outline of the basic collection in terms of authorship, research content, and publication media. According to Panel a of this figure, the 140 studies in the basic collection were written by 202 authors on a gross basis, of which an overwhelmingly majority (166 authors) belonged to universities or other academic research institutions. Meanwhile, 16 authors, accounting for 7.9% of the total number, have been involved in the transition strategy debate, worked at the IMF or the World Bank, both of which had strong and direct influences on policy decisions in the transition countries. Following them, the third largest group is com­ prised of 11 staff members at think tanks. Geographically, 152 authors (75.2% of the total) are based in North America or Western Europe, while only 29 authors (14.4% of the total) belong to institutions located in the CEE and FSU states. This fact proves that the international debate on transition strategies has been led by observers outside of the transition countries rather than by researchers in the very countries that have been carrying out the reforms. In addition to affiliated institutions and their locations, we also investigated each author’s research experience, gender, and assertiveness toward academic society, which may be related to their debate attitude.5 As Panel a of Figure 2.2 shows, the literature composition, in terms of the median of the author’s first publication year,6 reveals that papers written by the generation that started their research careers in the era when social­ ism existed in the CEE and FSU region and papers presented by the post-socialism gen­ eration almost perfectly counterbalance each other, with the rate being 69:71 in this order. The largest bloc is comprised of authors who made their debut in the 1990s, pro­ ducing 35.7% (50 studies) of the basic collection. Moreover, studies written at least partly by a female researcher accounted for 12.1% (17 studies) of the total, while those written at least partly by either a world-famous or a very influential economist in the field of transition economics accounted for 20.0% (28 studies) of the total. Panel b of Figure 2.2 shows the literature’s composition on the basis of their research content. As shown in this panel, most of the literature that shapes the transition strategy debate neither focused on any particular region or country as the subject of their study nor limited their debate to any particular policy areas. The same panel also shows that only 11.4% (16 studies) of the basic collection were published as research outcomes of academic projects. We can further confirm that literature proving their own assertions about ideal or desirable modes of transition strategy by means of mathematical economic models accounts for 16.4% (23 studies), and literature justifying their own assertions quantitatively by either conducting econometric analyses or using quantitative data (mostly official statistics) accounts for 34.3% (48 studies). The attributes of the basic collection, in terms of publication media, are shown in Panel c of Figure 2.2. This panel reveals that most studies in the basic collection were published as economic journal articles. In fact, 126 studies (90.0%) of the entire collection are journal articles, and 110 studies (78.6%) have been published in 29

C H A P T E R 2

C H A P T E R

TRANSITION STRATEGY DEBATE

2

Figure 2.2 Breakdown of the basic collection by literature attributea

Notes:

a

Numbers in parentheses are those of the relevant literature.

b

For more details on the evaluation method, see Chapter 1.

30

METHODOLOGY OF THE LITERATURE SEARCH 100%

Unpublished working papers (4)

Regional study (6)

2014–2018 (5) Grades 9–10 (21)

Int’l relations (6) 90%

Politics (4) Sociology (9)

2.2

2009–2013 (25)

Business (5)

80%

2 Grades 6–8 (30)

70%

2004–2008 (29)

60%

50%

Journal articles (126)

Grades 3–5 (36) 1999–2003 (35) Economics (110)

40%

30% 1994–1998 (28)

20%

Grades 1–2 (53)

10% 1989–1993 (18)

Book chapters (6) 0%

Academic books (4) Type of publication media

Specialized fields of publication media

C H A P T E R

Publication year

b

Quality level

(c) Publication media attributes

Figure 2.2 contd.

journals specializing in economics.7 Moreover, looking at the publication year of the 140 studies in five-year intervals, we found that, by and large, these studies are evenly distributed over the entire period, although there were slightly fewer studies published between 1989 and 1993 and slightly more studies published between 1999 and 2003. In addition to these attributes, we set ten grading criteria for evaluating the quality level of the publication media. More specifically, in the case of journal articles, the criteria are based on journal rankings and impact factor, while in the case of academic books and book chapters, the criteria are based on the presence of a peer-review system and literature information, such as the publishers. According to our evaluation, the quality level of the publication media has a negative correlation with the number of studies. Nevertheless, the number of studies published either in top-ranking journals (the 9th to 10th grade) or in the next-highest journals (the 6th to 8th grade) is not small. Actually, such studies account for 36.4% (51 studies) of the basic collection.8 This fact may reflect how the transition strategy debate, which has involved even a great number of world-famous researchers, was regarded as a critically important research subject for the entire economics society, which has been undergoing a big paradigm shift inspired by the end of socialism.

31

C H A P T E R 2

TRANSITION STRATEGY DEBATE

2.3 OVERALL STRUCTURE OF THE TRANSITION STRATEGY DEBATE In general, the opposing axis of academic debates among different schools or research groups with different stances gradually becomes more apparent as their arguments deepen or as time goes on. This also holds true for the quarter-century­ old transition strategy debate. In fact, the initial debate, in which the pros and cons of radicalism and gradualism were discussed to determine which was the more appropriate transition strategy for former socialist countries, was not necessarily focused: it even seemed confused. However, when we look at the flow of the debate during the past 30 years from the present position, we can see clear boundaries between the debate attitudes in the relevant literature, although that is, of course, the benefit of hindsight. In our view, the most fundamental criteria are the following three perspectives: speed of reform, policy sequence, and institutions. By checking each of 140 studies in the basic collection against these three criteria, we identified authors’ willingness to accept or reject radicalism and their grounds for objection; subsequently, we created a bold classification of the preceding studies.9 In this sec­ tion, we present the overall structure of the transition strategy debate emerging from this classification work. With regard to advocates of radicalism, including Lipton and Sachs (1990), Bal­ cerowicz (1994), and others, their debate attitude is remarkably consistent regarding speed of reform and policy sequence. Indeed, the radicalists share an idea that bigbang and speedy implementations of policy packages are indispensable for establish­ ing a market economy in the former socialist economies. The underlying logic of this idea is that economic transitions must be carried forward as quickly as possible, and, consequently, a single round of expeditious execution of necessary reform measures is essential. In terms of the reasoning about why the transition to a market economy should be achieved speedily, radicalists underline the following three points: (1) a strong demand from the international community, calling for the deter­ rence of backsliding into the Cold War situation; (2) a survival strategy for reformers who face pro-communist opposing forces; and (3) the necessity of cultivating a middle class that will proactively support democracy and a market economy. In other words, the radicalists tend to stress political reasoning to justify their debate attitude toward a transition strategy (Åslund 2007; Turley and Luke 2010). Despite the consistency of its policy recommendations, however, the radicalists can be divided into two research groups. One is the “universal radicalism” group, represented by Murphy et al. (1992), de Melo et al. (1996), and Åslund (2007, 2009), which maintains that the best option for the CEE and FSU countries should be radicalism, irrespective of differences in the degree of perfection of the planned system and other historical preconditions. The second is the “conditional radicalism” group, including Klaus (1993) and others, which affirms that implementing a transition strategy based on radicalism is a better option than gradualism as long as a series of initial conditions antecedent to reforms, such as the government’s adequate policy capability and the citizens’ sufficient understanding and tolerance of

32

OVERALL STRUCTURE OF THE TRANSITION STRATEGY DEBATE

2.3

capitalism, is minimally met.10 We note, however, that the latter group is clearly a minority, as compared to the former. Meanwhile, gradualism advocates have a unified voice against the radicalists in criti­ cizing radicalism’s “speed-before-quality,” “haphazard,” and “unrealistic” approach. Moreover, gradualists contend that radicalism is highly likely to be associated with socially intolerable negative side effects. When it comes to basic reasons for justifying the gradualist approach, however, gradualists have a more varied rationale in compari­ son with their radicalist counterparts. Nevertheless, it is possible to classify the gradual­ ists into several research groups, based upon the speed of reform and the policy sequence emphasized in their debate attitude. In this way, we come up with the first research group among the gradualists, which can be called the “slow-paced gradualism” group, including Etzioni (1992), Murrell (1992a, 1992b), Blanchard and Kremer (1997), and King (2002). This group asserts that the tran­ sition to a market economy should be carried forward over time so that any social pitfalls can be avoided, in light of the necessity of effectively controlling the side effects of struc­ tural reforms, such as political and social unrest, transformational recession (Kornai 1994), unfair distribution of wealth, and increases in unemployment and poverty, as well as in light of the lawmaking and administrative capabilities of governments in transition economies, their security-enforcement power, and the limited capacity of citizens to adapt to large-scale social changes. In this regard, we point out that the slow-paced gradualism group does not necessarily express a strong objection to radicalism, with regard to an allout and simultaneous undertaking of reform measures for economic transition. In contrast to the debate attitude of the slow-paced gradualism group, some researchers particularly emphasize the importance of policy sequence in order to suc­ cessfully carry out structural reforms that might drastically change a given economic system, while at the same time avoiding excessive social confusion. We call them the “step-by-step gradualism” group. Van Brabant (1993, 1994a, 1994b), Lian and Wei (1998), and Calcagno et al. (2006) represent this group. In addition to these step-by-step transition strategy advocates, the gradualists also embrace the “institutional gradualism” group, consisting of Hecht (1994), Liew (1995), Popov (2000a, 2000b, 2007, 2009, 2012), and many others, who stress that the establishment of institutions that constitute the foundation of the market economy and democracy, such as property rights and the rule of law, should be the top priority in order to advance the transformation from the planned system to a market economy. They also argue that the upgrading of institutions is a basic precondition for carrying out marketization policies, including freeing up of prices and privatization of enterprises, and thus, it should precede these policies. There­ fore, the basic standpoint of the institutional gradualism group is not essentially different from that of the step-by-step gradualism group in the sense that both groups emphasize policy sequence for carrying out economic transition.11 For this reason, we include the institutional gradualism group in the step-by-step gradualism group in the broad sense. Moreover, the gradualists comprise another group of researchers who regard the assertions of both the slow-paced and the step-by-step gradualism groups as equally important justification for denouncing radicalism. Quite a few famous scholars, such as Dewatripont and Roland (1992a, 1992b, 1995), Aghion and Blanchard (1994), 33

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE

North (1994), Stiglitz (1999), and Arrow (2000), belong to this group. We call them the “eclectic gradualism” group. Further, there is another research group, including McMillan and Naughton (1992), Islam (1993), Fan (1994), and Papapanagos and Sanfey (2003), that takes the side of neither the radicalists nor the gradualists. The essence of their argument is that radical­ ism and gradualism are not intrinsically opposed to each other, but rather are mutually alternative options; therefore, neither of the two can always be superior to the other theoretically and practically. Based on this notion, they further maintain that, in the real world, policy-makers may well choose either gradualism or radicalism as their basic transition strategy on a case-by-case basis, depending on the country’s actual conditions: in some cases, a mixture of both or switching between the two at different stages would be even possible. Their debate attitude can be described as “neutralism” because they remain within the framework of the radicalism-versus-gradualism debate while, at the same time, keeping at arm’s length from both the radicalists and the grad­ ualists. Neutralists have something in common with the conditional radicalism group; however, they should be clearly distinguished in the sense that their debate stance is more thoroughly neutral than is that of conditional radicalism. As mentioned in the Introduction, not all researchers who have discussed transition strat­ egy stay within the radicalism-versus-gradualism framework. For instance, Pomfret (2000), Herrmann-Pillath (2006), and some other researchers have claimed that some tran­ sition countries have carried out a “third” reform track, which cannot be classified as either radicalism or gradualism. Moreover, researchers such as Hoen (1996, 2010), Swaan and Lissowska (1996), and Liodakis (2001) have serious doubts about the significance of the transition strategy debate itself. These research groups can be respectively called “third­ way thinkers” or “transcendentalists.” These two groups almost complete our categoriza­ tion of researchers who have been deeply involved in the transition strategy debate. As for the heterodox research groups that keep a certain distance from the orthodox transition strategy debate, we will cover them in the concluding section of this chapter. Summing up the preceding discussion, we have illustrated in Figure 2.3 the over­ all structure of the transition strategy debate during the 30 years after the demise of the Communist Bloc in the CEE and FSU region. In accordance with this figure, we classified the 140 studies of the basic collection based on their respective debate atti­ tudes. Figure 2.4 summarizes the results. As shown in Panel a of this figure, the gradualists leave both the radicalists and the neutralists far behind in terms of the number of their publications. In fact, 97 research works are classified as gradualismadvocating literature, accounting for 69.3% of the entire basic collection. In this sense, gradualism is the majority view.12 Meanwhile, 30 studies belong to the radicalists, accounting for 21.4% of the total. As Panel b of Figure 2.4 shows, 27 of these 30 studies were written by researchers who firmly believe in the universality of radicalism. This fact reflects the monolithic nature of the radicalists. The remaining 13 studies are products of researchers who expressed a neutral position in the debate; however, the neutralists do not achieve even half the number of studies as the radicalists. In this way, the conflict between the radicalists and the gradualists is obvious. 34

OVERVIEW OF THE RADICALISM-VERSUS-GRADUALISM DEBATE

2.4

C H A P T E R 2

Figure 2.3 Overall structure of the transition strategy debate

Panel c of Figure 2.4 exhibits the subclassifications of the gradualists. According to this panel, the slow-paced gradualism group published 43 out of the 97 studies, or 44.3% of the total studies created by the gradualists. Meanwhile, the step-by-step gradualism group and the eclectic gradualism group published 33 studies (34.0% of the total) and 21 studies (21.6%), respectively, suggesting that the power balance is almost even among these three research groups. Moreover, 18 studies belong to the institutional gradualism group, which places the most emphasis on the importance of institution building, accounting for more than half (54.5%) of the total studies from the step-by-step gradualism group. This demonstrates that the debate attitude of the gradualists is varied, and no particu­ lar view overwhelms the others. This is in clear contrast to the radicalists, who demon­ strate a consistent view about their policy recommendation for economic transition.

2.4 OVERVIEW OF THE RADICALISM-VERSUS­ GRADUALISM DEBATE In this section, we will outline the main arguments in the radicalism-versus­ gradualism debate. First, we review representative research works of the radicalists. Then we will deal with those of the gradualists and then the neutralists.

35

C H A P T E R

TRANSITION STRATEGY DEBATE

2

Figure 2.4 Breakdown of the basic collection by debate attitude Note: Numbers in parentheses are those of the relevant literature.

2.4.1 Radicalism As described in Section 2.3, research works presented by the universal radicalism group, which strongly advocates the superiority of radicalism over other options, account for most of the studies produced by radicalists. Accordingly, we first touch on the debate attitude of this research group and then move on to the conditional radicalism group. UNIVERSAL RADICALISM

Many early studies of the universal radicalism group emphasized the speed of reform, taking reform experiences within socialism in the CEE countries into account: their 36

OVERVIEW OF THE RADICALISM-VERSUS-GRADUALISM DEBATE

2.4

momentum had a great influence on the subsequent research trends. For example, just after the fall of the Berlin Wall, Lipton and Sachs (1990) initiated the debate on transi­ tion strategy and took a leading role among Western radicalists, arguing that speed itself is the most important reform element in economic transition. Murphy et al. (1992) also advocated for the speedy and all-out implementation of reform measures, arguing that partial reforms (gradualism-based transition strategy) would easily destroy an existing economic system; however, they can hardly accomplish an efficient market and, as a result, they might cause a substantial decrease in production. At almost the same time, Brada (1993) stated that rapid reforms would benefit post-communist countries and that the claim of gradualists is not only inappropriate, in light of his­ tory and reality in the CEE countries, but also inconsistent with democracy. More­ over, Balcerowicz and Gelb (1995) discussed how radicalism-based liberalization policies and macroeconomic stabilization measures would involve the lowest risk and would not prevent reforms and production, even in the medium term. Thereafter, the universal radicalism group has shown a stronger tendency to sup­ port and justify the rapid big-bang approach in a more proactive manner, referring to the actual transition processes in the CEE and FSU countries and their ex-post eco­ nomic performances. For example, Lavigne (2000) endorsed radicalism as a proper reform track based on her retrospection on reform achievements during the 1990s. Anders Åslund (2007), who acknowledges that he is a promoter of radicalism, refuted criticism from gradualists by claiming that failures in transition countries, especially economic turmoil in Russia, were caused not by the radicalism-based reforms themselves but by the lukewarm implementation of necessary reform meas­ ures by the transition governments. Two years later, he also asserted that although Russia might have failed in politics, it had achieved a certain success in its economic transition, suggesting that the source of success in the Russian structural reforms should be attributed to shock therapy (Åslund 2009). As reported in Section 2.2, the number of studies that performed empirical analysis is very limited, and most of them belong to the universal radicalism group. Sachs 1996 is one of pioneering works in this research field. Using a cross-sectional dataset of 25 transition economies in 1995, Sachs found a positive correlation between pro­ gress in liberalization and economic growth and, consequently, reached a conclusion in favor of radicalism. Moreover, de Melo et al. (1996, 2001) gave countenance to radicalism by verifying a positive correlation between the speed of liberalization and economic growth using macroeconomic data of 28 transition economies for the first half of the 1990s. Selowsky and Martin (1997) and Berg et al. (1999) also demon­ strated their own support for radicalism, presenting empirical evidence similar to that of Sachs (1996) and de Melo et al. (1996, 2001). CONDITIONAL RADICALISM

Some radicalists express hesitations regarding the rapid and all-out execution of reform measures to advance the economic transition to a market economy, although they are extremely minor. For example, Klaus (1993) wrote that radicalism can be 37

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE

constantly recommended only if institutions and other environments are consistent with policies. Balcerowicz (1994) also stated that, as long as minimum conditions for carrying out radicalism-based transition strategies are satisfied, radicalism should precede all alternative approaches in order to ensure the irreversibility of reforms.13 The conditional radicalism group keeps their debate attitude essentially the same as that of the universal radicalism group in the sense that they consider radicalism to be a reform philosophy superior to gradualism. Nevertheless, it is also true that there is an obvious enthusiasm gap between the groups. 2.4.2 Gradualism As discussed in Section 2.3, gradualists are divided into the following three research groups in accordance with their respective logic lines for criticizing radicalism: the slow-paced gradualism group, the step-by-step gradualism group, including the insti­ tutional gradualism group, and the eclectic gradualism group. In the following sub­ sections, we will look at the debate content of these three groups in turn. SLOW-PACED GRADUALISM

Arguments of the slow-paced gradualism group mirror those of the universal radical­ ism group. This means that, in the early stage of the debate, this group tried to jus­ tify gradualism based on economic theories, and, as economic transition progressed in the former Communist Bloc, it began to support gradualism, referring to the real­ ity of structural reforms and/or socioeconomic situations in the CEE and FSU coun­ tries or China. Peter Murrell (1992a, 1992b) manifested a view representative of slow-paced gradualism at the very beginning of the debate. For example, in Murrell 1992a, he pointed out that drastic reforms of state-owned enterprises have a great risk of inter­ fering with private entrepreneurship; thus, he advocated for slow-paced implementa­ tion of enterprise reforms. Meanwhile, in Murrell 1992b, he argued that most successful reform experiments in CEE are rooted in small-scale accumulated institu­ tional changes over time; therefore, he strongly recommended slow-paced structural reforms. In the same period, Etzioni (1992) also maintained that social changes are intrinsically incremental and, thus, impossible to accelerate by any means; accord­ ingly, he argued that the transition strategy itself should also be gradual. During the 1990s, numerous papers that share the debate attitude of Etzioni (1992) were pub­ lished, such as Bhagwati 1994, Gel′vanovskii 1994, Dehejia 1996, and many others. Afterward, as is the case with the universal radicalism group, the slow-paced grad­ ualism group has increasingly leaned on ex-post reform achievements or economic performances in transition countries as supporting facts for gradualism. Above all, the remarkable successes of China and the tragic failures of Russia have frequently been cited by this group to enforce their assertions. A good example is Blanchard and Kremer 1997. In this well-known paper, they argued that, in the case of China, in addition to the low degree of industrialization, maintenance of the Communist 38

OVERVIEW OF THE RADICALISM-VERSUS-GRADUALISM DEBATE

2.4

Party’s political power, and the decentralized development system, its gradual and slow-paced economic reforms were significantly effective in avoiding what they called “disorganization.” Comparing Russia with Poland, where the privatization policy had been postponed until promising investors appeared, King (2002) expressed criticism of radicalism by stating that, in Russia, fast and sloppy privatiza­ tion of enterprises caused serious damage to company management, which in turn deepened the economic crisis in the country. Many of the studies that expressed debate attitudes similar to that of Blanchard and Kremer (1997) and King (2002) appeared in the 2000s, including Rosefielde 2001, Molchanov 2005, Zweynert 2006, and Rozzelle and Swinnen 2009. As demonstrated above, discussions by the slow-paced gradualism group were remarkably active. However, the number of studies that empirically verified their arguments is extremely limited. Sušjan and Redek 2008 is a valuable study from this viewpoint. Using panel data of 23 transition economies for the period 1995 to 2002, they found that transition-specific uncertainty caused by radicalism-based high-speed reforms is highly likely to disturb economic growth in these countries. Thus, their empirical results are regarded to serve as good reasons for the slow-paced transition strategy. STEP-BY-STEP AND INSTITUTIONAL GRADUALISM

Together with discussions related to slow-paced gradualism, those regarding step-by­ step gradualism were also very active from the beginning of the debate. Jozef Van Brabant provided a series of pioneering studies to support the step-wise transition strategy in the early 1990s (1993, 1994a, 1994b). In these works, he consistently stressed the importance of policy sequence, although he did not completely rule out rapid reforms. Actually, he wrote that the stabilization policy is the very first one to be implemented among marketization measures, and trade liberalization and disposal of non-performing loans should be dealt with in a careful manner based on a reasonable policy sequence. Lian and Wei (1998) also argued that partial reforms are more desirable than the big-bang approach; hence, gradualism, which represents a series of partial reforms, is a superior transition strategy to radicalism. Following these works, several country studies that give positive evaluation to the step-by-step approach were published one after the other. For instance, Calcagno et al. (2006), who investigated the steel industry in Romania, pointed out that the massive restruc­ turing of state-owned companies that had been implemented almost in parallel with enterprise privatization subsequently caused serious problems in this country. Minniti and Polutnik (2007) evaluated the Slovenian step-by-step reforms based on their study of the country’s currency conversion process and anti-inflation policies. Meanwhile, Hecht presented a groundbreaking study (1994) advocating institu­ tional gradualism. In this paper, he stated that structural reforms in the former social­ ist countries should focus more on the establishment of property rights and the rule of law; then, from this viewpoint, he sharply criticized radicalism, which downplays the role of institutions. Dealing with Chinese experiences, Liew (1995) also 39

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE

emphasized how ensuring sound institutions, including a strong central state, is essential for promoting reforms. Furthermore, Noman (1999) stressed how strong institutions are more important than anything else to avoid or mitigate the adverse effects of economic liberalization, including income disparity and capital flight. In addition, Coyne and Boettke (2006) pointed out that keeping consistency among dif­ ferent institutions is more essential than adjusting the speed of reform for successful economic transition. As demonstrated in their works, researchers who belong to the institutional gradualism group presented a series of discussions that placed emphasis on the priority of institution building over other reform measures. It was Vladimir Popov who intensely and tirelessly advocated the institutionalist approach through his numerous studies. As one of the leading figures in the transi­ tion strategy debate, he repeatedly claimed the importance of institutions in the sys­ temic transformation process toward a market economy. In fact, in Popov 2000b, he pointed out that 90% of the failures in transition economies are attributable to grave blunders in maintaining strong institutions, not to the liberalization policies them­ selves. Nine years later, he also stated that the creation and maintenance of powerful institutions is the most essential part for the sake of economic growth in transition countries (Popov, 2009). At the same time, he presented supporting evidence for the institutionalist approach in several empirical studies (Popov 2000a, 2007, 2012). As mentioned in Section 2.2, the international debate has been and still is led mainly by American and Western European researchers. Among them, Vladimir Popov is play­ ing a great role as one of the strong promoters of institutional gradualism from the Russian viewpoint.14 ECLECTIC GRADUALISM

More than a few researchers place emphasis on both speed of reform and policy sequence as reasons for gradualism. From this point of view, Dewatripont and Roland’s studies (1992a, 1992b, 1995) are representative of studies from the early stage of the debate. In the two papers published in 1992, they asserted that gradual reforms with a careful and step-by-step approach would be the most appropriate strategy, taking the fiscal cost of structural reforms into account. Three years later, they also stated that slow-paced and step-by-step reform measures are not only easier to implement but also get returns from investment faster than a rapid and allout marketization, on the condition that the relevant policy sequence must get public support. At almost the same time, Aghion and Blanchard (1994) made policy recom­ mendations that take account of both speed of reform and policy sequence for arran­ ging the transition strategy in the proper way, arguing that it is difficult and undesirable to accelerate economic transition and that creating employment should be the first priority among various reform measures. The eclectic gradualism group is characterized by the outstanding presence of Nobel laureates. Indeed, the slow-paced and step-by-step transition strategies are strongly encouraged by Douglass C. North, a prominent scholar in the new institu­ tional economics, Joseph E. Stiglitz, who paved the way for the economics of 40

OVERVIEW OF THE RADICALISM-VERSUS-GRADUALISM DEBATE

2.4

asymmetric information, and Kenneth J. Arrow, a leading figure in social choice theory. More specifically, North (1994) advocated the slow-paced and step-by-step reform approaches from his insights into the role of informal institutions (norms) in economic systems and the long-term nature of institutional changes. From the viewpoint of bounded rationality, Stiglitz (1999) severely criticized the unreality of the neoclassical model that many radicalists rely upon, while Arrow (2000) dir­ ected attention to the difficulty of citizens in the former socialist countries to per­ ceive and understand newly introduced price and market mechanisms. The involvement of these world-famous economists in the transition strategy debate greatly influenced both the direction of the debate and the power balance among research groups. Together with literature by the aforementioned Nobel Prize winners, a series of research works by John Marangos also made noteworthy contributions to the eclectic gradualism group (2003, 2004a, 2004b, 2006). In these papers, he consistently touched on the importance of the speed of reform, and his emphasis on policy sequence has also been increasing in proportion to the development of his research. In recent years, he not only advocated gradualism but also argued that former social­ ist countries should aim for a post-Keynesian and market socialism economic system. 2.4.3 Neutralism Finally, we will briefly describe the main perspectives of the neutralists. As with the gradualists, the naturalists also published more than a few papers based on the reform achievements and/or economic performances of China. McMillan and Naugh­ ton 1992 is one such study that delivered a unique policy perspective, which depends neither on radicalism nor gradualism. More specifically, describing the many problems of China’s reform and open-door policies, although these reform measures were successful in various aspects, they concluded that China’s experiences neither deny the radicalism-based transition strategy nor justify the gradualist approach. Moreover, Islam (1993) showed his neutral stance, that radicalism and gradualism are both feasible options, stating that speed of reform should be con­ sidered in light of different time spans for policy effects to emerge in each economic field: this is true for policy sequence, also. Neutralists also published several works that emphasize the essential nature of strategic flexibility for successful promotion of the economic transition. For example, according to Fan (1994), both radicalism and gradualism can serve as the more appropriate solution, subject to given conditions in a transition economy. Further­ more, some studies proposed that transition countries should switch their reform modes flexibly with passing time. A typical example is Papapanagos and Sanfey 2003. In this paper, developing a unique mathematical model, they asserted that labor reallocation is better implemented in a gradual manner in the early stage of transition, but a rapid restructuring of the labor market is more desirable in the longrun; consequently, mode switching is required for successful transition. Despite their 41

C H A P T E R 2

C H A P T E R

TRANSITION STRATEGY DEBATE

minor presence, neutralists provide a bridge between radicalists and gradualists and, by playing such a role, present meaningful opinions toward a constructive conver­ gence in the transition strategy debate.

2

2.5 RELATIONSHIP BETWEEN DEBATE ATTITUDES AND LITERATURE ATTRIBUTES Through the previous discussions, we have revealed the overall picture of the litera­ ture attributes and the debate attitudes of the basic collection. As a next step, we will examine the relationship between these two elements by means of statistical and econometric methods. First, we will conduct a cross tabulation analysis to test the independence of the literature attributes from the debate attitudes. Then, we will esti­ mate qualitative choice models to regress the debate attitudes towards a series of the literature attributes simultaneously. Table 2.1 shows the cross tables. In reference to the overall structure of the transi­ tion strategy debate described in Section 2.3, the columns of this table feature not only the three main categories of debate attitudes—radicalism, neutralism, and grad­ ualism—but also five subcategories, comprised of the two radicalist groups and the three gradualist groups. Meanwhile, the table rows contain all 14 kinds of literature attributes mentioned in Section 2.2. Unlike Figure 2.2, however, this table provides a more-detailed breakdown of subject regions and subject policy areas. We test the interdependence of the debate attitudes and the literature attributes based on the three main categories (Test I) and the five subcategories plus neutralism (Test II), separately. Table 2.1 reports the results of the chi-square test of independence as well as Cramér’s coefficient of association.15 According to the test results, the null hypothesis that the debate attitudes and a given literature attribute are independent is rejected at the 10% or less significance level, both in the cases of Test I and II in relation to the six attributes including: (1) author affiliation, (2) location, (3) median value of first publication year, (4) intensity of empirical examination, (5) type of publication media, and (6) quality level. In addition, in the case of subject policy areas and specialized fields of publication media, the null hypothesis is rejected in Test II. These results indicate that the differ­ ences in debate attitudes are statistically associated with many of the literature attri­ butes. Nevertheless, according to Cramér’s coefficient of association, there is no remarkable difference in the degree of correlation among these literature attributes. Next, we estimate qualitative choice models to examine whether each literature attribute is correlated with the debate attitudes when the other attributes are simul­ taneously controlled. As explained in Section 2.3, the debate attitude of the step-by­ step gradualism group, which stresses the importance of policy sequence in its opposition to radicalism, theoretically, provides much clearer reason to support grad­ ualism in comparison with that of the slow-paced gradualism group. Therefore, together with the relationship between the degree of the radicalism stance and

42

1

2

48

45

0 0

3

2 1 0

6 1 0

(c) Median value of authors’ first publication year Until the 1960s 0 0 1970s 7 7

Total

0

3

1

0 0 0

2

Conditional radicalism (1b)

38

45

6 14 1

6 14 1

48

23

Universal radicalism (1a)

25

(b) Location of affiliated institutions North America or 38 Western Europe CEE 8 FSU 2 Asia and Oceania 0

Total

(a) Affiliated institutions Universities or academic research institutions Think tanks IMF or World Bank Other international institutions Others

Total of radicalism (1)

Radicalism

1 5

23

6 1 1

15

23

0

1 1 1

20

Neutralism (2)

4 22

131

7 8 17

99

131

1

4 1 4

121

1 8

57

3 3 8

43

57

1

3 0 0

53

3 4

27

0 0 3

24

27

0

0 1 0

26

0 10

47

4 5 6

32

47

0

1 0 4

42

SlowTotal of paced Eclectic Step-bygradual- gradual- gradual- step gradism (3) ism (3a) ism (3b) ualism (3c)

Gradualism

5 34

202

21 11 18

152

202

3

11 16 6

166

Total

0.214

18.432***

0.360

52.422***

Upper: Chisquare test of independence Lower: Cramér’s coefficient of association c

Test I a

Table 2.1 Cross tabulation analysis of the relationship between the debate attitudes and the literature attributes

(Continued )

0.256

39.849***

0.324

84.776***

Upper: Chisquare test of independ­ ence Lower: Cramér’s coefficient of association c

Test II b

RELATIONSHIP BETWEEN DEBATE ATTITUDES 2.5

43

C H A P T E R 2

44

(f) Focus on particular regions or countries Without any par24 21 ticular subject regions FSU bloc 0 0 China 0 0 0 3

0 0

13

8

3

27

30

3 10

13

3 10

13

2 4 1

Neutralism (2)

3

0 3

(e) Involvement of world-famous economists d Yes 7 7 No 23 20

Total

3

27

Total

30

0 3

3

27

6 21

30

1 2 0

Conditional radicalism (1b)

12 5 3

Universal radicalism (1a)

(d) Involvement of female researchers Yes 6 No 24

Total

13 7 3

Total of radicalism (1)

Radicalism

2 6

69

97

18 79

97

8 89

97

15 39 17

1 3

29

43

6 37

43

5 38

43

7 20 7

1 2

15

21

8 13

21

1 20

21

5 7 2

0 1

25

33

4 29

33

2 31

33

3 12 8

Slow- Total of paced Eclectic Step-bygradual- gradual- gradual- step gradism (3) ism (3a) ism (3b) ualism (3c)

Gradualism

2 9

101

140

28 112

140

0 17 123

140

30 50 21

Total

0.054

0.412

0.181

4.573

0.230

0.239

7.980

0.218

6.670

0.236

31.068*

Upper: Chisquare test of independ­ ence Lower: Cramér’s coefficient of association c

Upper: Chisquare test of independence Lower: Cramér’s coefficient of association c

14.812*

Test II b

Test I a

2

1980s 1990s 2000s and after

Table 2.1 contd.

C H A P T E R

TRANSITION STRATEGY DEBATE

4 0

4 0

27

13

3

Total

27

10

2

30

3

13

3 10

13

1

3

1 2

3

1 0

0

12 0

13

1 0

0 0 1

0 0

(i) Employment of mathematical economic models Studies using 4 3 mathematical eco­ nomic models Others 26 24

30

Total

27

0 0

0 1

0 27

30

0

1

(h) Outcomes from academic projects Project outcomes 1 Others 29

Total

2 1

3

0 0

0 0 0

0 0

24 1

27

0 2 0

0 2 0

30

0 0

0 0

(g) Focus on particular policy areas Policies in general 26 Economic 2 liberalization Macroeconomic 1 stabilization Privatization 0 Enterprise reform 1 and corporate restructuring

Total

Cuba Czech Republic or Czechoslovakia Hungary Poland Certain countries in Southeastern Europe Russia Uzbekistan

97

81

16

97

12 85

97

4 5

6

82 0

97

7 3

1 0 6

1 2

43

34

9

43

6 37

43

2 1

3

37 0

43

4 3

1 0 1

0 1

21

15

6

21

3 18

21

0 1

1

19 0

21

2 0

0 0 0

0 1

33

32

1

33

3 30

33

2 3

2

26 0

33

1 0

0 0 5

1 0

140

117

23

140

16 124

140

5 6

7

120 2

140

12 3

1 2 7

1 2

0.067

0.628

0.164

3.770

0.199

11.071

0.282

22.313

(Continued )

0.251

8.804

0.228

7.267

0.238

31.729**

0.264

48.738

RELATIONSHIP BETWEEN DEBATE ATTITUDES 2.5

45

C H A P T E R 2

Total of radicalism (1)

46

3

2 0

27

(l) Specialized fields of publication media Economics 23 21 Business 1 1

30

Total

0 0 3 0

1 1 21 4

(k) Type of publication media Academic books 1 Book chapters 1 Journal articles 24 Unpublished work4 ing papers

3

3

11

27

0

5

30

Total

0

Conditional radicalism (1b)

11

Universal radicalism (1a)

Radicalism

12 0

13

0 1 12 0

13

10

2

1

Neutralism (2)

75 4

97

3 4 90 0

97

68

24

5

34 0

43

0 2 41 0

43

29

11

3

17 0

21

1 2 18 0

21

14

7

0

24 4

33

2 0 31 0

33

25

6

2

SlowTotal of paced Eclectic Step-bygradual- gradual- gradual- step gradism (3) ism (3a) ism (3b) ualism (3c)

Gradualism

110 5

140

4 6 126 4

140

92

31

17

Total

0.238

15.927**

0.281

0.239

23.946*

0.327

29.882***

Upper: Chisquare test of independ­ ence Lower: Cramér’s coefficient of association c

Upper: Chisquare test of independence Lower: Cramér’s coefficient of association c

22.131***

Test II b

Test I a

2

(j) Intensity of empirical examination Econometric 11 studies Studies using 5 quantitative data Studies without 14 quantitative analysis

Table 2.1 contd.

C H A P T E R

TRANSITION STRATEGY DEBATE

13 3 5 9

30

(n) Quality level e Grades 1–2 Grades 3–5 Grades 6–8 Grades 9–10

Total

27

12 3 4 8

27

3 7 6 3 8

3

1 0 1 1

3

2 1 0 0 0

3

0

0 1 0

13

3 2 4 4

13

3 4 2 2 2

13

0

0 0 1

97

37 31 21 8

97

10 16 27 24 20

97

6

6 2 4

43

10 16 15 2

43

6 6 13 10 8

43

3

2 2 2

21

8 4 4 5

21

3 3 8 4 3

21

3

1 0 0

33

19 11 2 1

33

1 7 6 10 9

33

0

3 0 2

140

53 36 30 21

140

18 28 35 29 30

140

6

9 4 6

0.242

16.334**

0.171

8.229

0.164

7.528

0.294

36.348***

0.205

23.442

0.225

35.517*

a

Notes:

Columns (1), (2), and (3) are subjects of the test

b Columns (1a), (1b), (2), (3a), (3b), and (3c) are subjects of the test.

c ***, **, and * denote statistical significance at the 1% level, 5% level, and 10% level, respectively.

d Involvement of world-famous economists denotes literature that includes at least one of the following researchers among the authors (Each author’s affiliated institution and position that we could

confirm while writing this article are indicated in parentheses): Anders Åslund (Senior Fellow, Peterson Institute), Kenneth J. Arrow (Emeritus Professor, Stanford University), Jagdish Bhagwati (Professor,

Columbia University), Olivier Blanchard (Professor, Massachusetts Institute of Technology; Economic Counsellor, IMF), Martha de Melo (Former Chief Economist, World Bank), Mathian Dewatripont

(Extraordinary Professor, Université libre de Bruxelles; Director, National Bank of Belgium), Stanley Fischer (Vice-Chairman, US Federal Reserve System), Alan Gelb (Senior Fellow, Center for Global

Development), Marie Lavigne (Senior Fellow, Institute of Mathematical Sciences and Applied Economics), John McMillan (Professor, Stanford University), Peter Murrell (Mancur Olson Professor, Univer­ sity of Maryland), Douglass C. North (Spencer T. Olin Professor, Washington University; Bartlett Burnap Senior Fellow, Hoover Institution, Stanford University), Vladimir Popov (Adjunct Research Profes­ sor, Carleton University at Ottawa; Inter-regional Advisor, United Nations), Gérard Roland (E. Morris Cox Professor, University of California, Berkeley), Jeffrey Sachs (Director of the Earth Institute,

Columbia University), Andrei Shleifer (Professor, Harvard University), Joseph E. Stiglitz (Professor, Columbia University).

e For more details on the evaluation method, see Chapter 1.

30

5 8 6 3 8

27

0

0

30

3 1 1

3 2 1

Total

(m) Publication year 1989–1993 1994–1998 1999–2003 2004–2008 2009–2018

Total

Sociology Politics International relations Regional study

RELATIONSHIP BETWEEN DEBATE ATTITUDES 2.5

47

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE

literature attributes, it is valuable to examine the association between the degree to which policy sequence is stressed and the literature attributes of the studies produced by gradualists. Therefore, we use the following two types of dependent variables for our regression estimation: One is a four-point ordered variable that gives a value of 0 to the literature of gradualism, 1 to that of neutralism, 2 to that of conditional rad­ icalism, and 3 to that of universal radicalism. This variable is used as a proxy for the degree of radicalism. The other is a three-point ordered variable that identifies the studies from the slow-paced gradualism group by 0, those from the eclectic grad­ ualism group by 1, and those from the step-by-step gradualism group by 2. This variable serves as a proxy for the degree to which policy sequence is stressed in the gradualists’ research work. As for the independent variable, we employ a total of 36 variables, which consist of 10 types of authorship attributes, 16 types of research content attributes, and 10 types of publication media attributes, which correspond with the row of cross tables in Table 2.1. Table 2.2 reports the type of variables used in our regression estimation, their descriptive statistics and correlation coefficients between each independent variable, and the two dependent variables. As shown in this table, both the degree of radicalism and the degree to which policy sequence is stressed are closely associated with the seven types of variables. However, combinations of these significantly correlated independent variables are completely different between the two. Table 2.3 reports estimation results of ordered probit models. Models 1 and 2 take the degree of the radicalism stance and the degree to which policy sequence is stressed as the dependent variable, respectively. We used the Huber–White sandwich estimator for computing robust standard errors. In this table, we report the most reli­ able models in terms of Akaike’s information criterion (AIC) and Bayesian informa­ tion criterion (BIC).16 The estimation results of Model 1 suggest the following relationship between the degree of the radicalism stance and the literature attributes in the basic collection: with regard to the authorship attributes, authors who belong to think tanks, the IMF, or the World Bank have a stronger tendency to support radicalism, as compared to authors who work for universities or for academic research institutions. Similarly, as compared with authors based in Asia and Oceania, those based in CEE tend to attach a higher value to the radicalist approach. Moreover, Model 1 also indicates that authors whose first publication year is more recent are less likely to support rad­ icalism, as is also true when an eminent economist is among the authors.17 Concerning the research content attributes, compared to studies that discuss the transition strategy in general or without any particular subject regions, studies that explicitly deal with the FSU countries, Cuba, the Czech Republic or Czechoslovakia, Hungary, and Uzbekistan tend to express a more negative debate attitude against rad­ icalism. In contrast, studies that discuss a transition strategy based on experiences or cases in China or Poland demonstrate a stronger support for radicalism.18 Further­ more, as compared with general policy studies, studies that argue for a transition strategy in line with the economic liberalization policy are more likely to express 48

Research content attributes

C C C

Proportion of location of affiliated institutions North America or Western Europe CEE FSU

Focus on particular regions or countries FSU bloc China

Other authorship attributes Median value of authors’ first publication year e Proportion of female researchers Involvement of world-famous economists f

C C C C

Proportion of affiliated institutions Think tanks IMF or World Bank Other international institutions Others

Authorship attributes

0.020 0.066

0.080 0.191

C D

D D

1988.234

0.741 0.092 0.061

0.076 0.039 0.039 0.013

0.714 0.897

Mean

C

O O

Dependent variables Degree of the radicalism stance c Degree to which policy sequence is stressed d

Debate attitude

Variable typea

Variable name

Variable group

0.140 0.249

0.243 0.394

11.293

0.430 0.284 0.238

0.262 0.187 0.187 0.099

1.189 0.884

S.D.

0 0

0 0

1990

1 0 0

0 0 0 0

0 1

Median

0 0 0

0 0 0 0

0 0

Min

1 1

1 1

0 0

0 0

2008.5 1951

1 1 1

1 1 1 1

3 2

Max

Descriptive statistics

−0.073 −0.084

0.085 0.060

−0.080

−0.018 0.153* −0.016

0.232*** 0.303*** −0.014 0.122

— —

Correlation coefficient with the degree of the radicalism stance b

(Continued )

−0.066 −0.067

−0.092 −0.004

−0.050

−0.119 0.159 0.144

−0.094 0.012 0.260*** −0.104

— —

Correlation coefficient with the degree to which policy sequence is stressed b

Table 2.2 Type and descriptive statistics of variables used in regression estimation and correlation coefficients between independent variables and two dependent variables

RELATIONSHIP BETWEEN DEBATE ATTITUDES 2.5

49

C H A P T E R 2

50

Publication media attributes

Type of publication media Academic book Book chapter Unpublished working paper

Focus on particular policy areas Economic liberalization Macroeconomic stabilization Privatization Enterprise reform and corporate restructuring Other research content attributes Outcomes from academic projects Employment of mathematical eco­ nomic model Intensity of empirical examination g

Cuba Czech Republic or Czechoslovakia Hungary Poland Certain countries in Southeastern Europe Russia Uzbekistan

Variable name

0.480

O

0.026 0.039 0.026

0.118 0.151

D D

D D D

0.013 0.046 0.039 0.033

0.092 0.026

D D D D D D

0.007 0.020 0.007 0.013 0.046

Mean

D D D D D

Variable typea

0.161 0.195 0.161

0.700

0.324 0.360

0.114 0.210 0.195 0.179

0.290 0.161

0.081 0.140 0.081 0.114 0.210

S.D.

0 0 0

0

0 0

0 0 0 0

0 0

0 0 0 0 0

Median

1 1 1

2

1 1

1 1 1 1

1 1

1 1 1 1 1

Max

Descriptive statistics

0 0 0

0

0 0

0 0 0 0

0 0

0 0 0 0 0

Min

0.005 −0.009 0.331***

0.156 −0.094 —

−0.071

−0.063 −0.201**

−0.129 −0.040 0.332***

— −0.019 0.024 0.133 0.181** −0.055 −0.084 −0.116

−0.103 −0.182*

0.128 −0.066 −0.104 — 0.225**

−0.051 −0.073 −0.051 0.232*** −0.111 0.095 −0.089

Correlation coefficient with the degree to which policy sequence is stressed b

Correlation coefficient with the degree of the radicalism stance b

2

Variable group

Table 2.2 contd.

C H A P T E R

TRANSITION STRATEGY DEBATE

C O

Other publication media attributes Publication year Quality level h 2002.026 4.033

0.039 0.059 0.026 0.039 0.039 6.749 3.304

0.195 0.237 0.161 0.195 0.195 2002 4

0 0 0 0 0 2017 9

1 1 1 1 1 1989 0

0 0 0 0 0

−0.053 0.087

−0.019 0.063 0.078 −0.009 −0.128

Notes: a C: Continuous variable; D: Dummy variable; O: Ordered variable b ***, **, and * denote statistical significance at the 1% level, 5% level, and 10% level, respectively c Ordered variable that gives a value of 0 to gradualism, 1 to neutralism, 2 to conditional radicalism, and 3 to universal radicalism d Ordered variable that gives a value of 0 to slow-paced gradualism, 1 to eclectic gradualism, and 2 to step-by-step gradualism e Based on the first publication year of each author registered in the ProQuest database f Dummy variable that gives a value of 1 to literature that includes at least one of the world-famous researchers among the authors (for details, see note d in Table 2.1) g Ordered variable that gives a value of 0 to studies without quantitative analysis, 1 to studies using quantitative data, and 2 to econometric studies h For more details on the evaluation method, see Chapter 1.

D D D D D

Specialized fields of publication media Business Sociology Politics International relations Regional study

0.094 −0.209**

0.260*** 0.079 −0.148 0.024 −0.116

RELATIONSHIP BETWEEN DEBATE ATTITUDES 2.5

51

C H A P T E R 2

C H A P T E R

TRANSITION STRATEGY DEBATE Table 2.3 Estimation results of ordered probit model on the relationship between the debate attitudes and the literature attributesa Model

[1]

[2]

Sample literature

Basic collection

Gradualism literature

Independent variables (Default category)/Dependent variable

Degree of the radicalism stance

Degree to which policy sequence is stressed

2

Proportion of affiliated institutions (Universities or academic research institutions) Think tanks 1.393** IMF or World Bank 5.235*** Other international institutions 0.324 Others 0.257 Proportion of location of affiliated institutions (Asia and Oceania) North America or Western Europe 1.339 CEE 2.132** FSU 1.259 Other authorship attributes Median value of authors’ first publication year Proportion of female researchers Involvement of world-famous economists

−0.935 5.880*** 5.752*** −10.758*** −0.016 1.378* 0.939

−0.034**

−0.005

0.461 −1.265**

−2.081 0.796

Focus on particular regions or countries (Without any particular subject regions or countries) FSU bloc −9.901*** −5.306*** China 1.493** −0.331 Cuba −6.441*** 8.567*** Czech Republic or Czechoslovakia −3.153*** −0.416 Hungary −6.557*** −5.119*** Poland 10.551*** — Certain countries in Southeastern −0.992 2.515** Europe Russia 0.164 −0.589 Uzbekistan −5.904*** −5.271*** Focus on particular policy areas (Policies in general) Economic liberalization 2.447*** Macroeconomic stabilization −7.623*** Privatization −0.464 Enterprise reform and corporate −6.567*** restructuring Other research content attributes Outcomes from academic projects Employment of mathematical economic model Intensity of empirical examination Type of publication media (Journal article) Academic book Book chapter Unpublished working paper

−0.594 0.228

— −0.101 −2.358** 0.505

−0.308 −0.134

1.057***

−0.431

1.634* 1.037 13.476***

1.103* −1.076 —

(Continued )

52

RELATIONSHIP BETWEEN DEBATE ATTITUDES

2.5

Table 2.3 contd. Model

[1]

[2]

Sample literature

Basic collection

Gradualism literature

Independent variables (Default category)/Dependent variable

Degree of the radicalism stance

Degree to which policy sequence is stressed

Specialized fields of publication media (Economics) Business −3.721*** Sociology 1.544** Politics 2.156*** International relations 0.122 Regional study −6.042***

6.730*** −0.176 −6.022*** 2.260*** 0.205

Other publication media attributes Publication year Quality level

0.031 −0.078

N Log pseudolikelihood Pseudo R2 Akaike’s information criterion (AIC) Bayesian information criterion (BIC) Wald test (χ2) a

0.038 0.220*** 140 −71.702 0.415 221.405 336.129 3605.08***

C H A P T E R 2

97 −70.390 0.315 208.781 296.321 3296.68***

Notes:

a Null hypothesis: All coefficients are zero.

For more details on definitions and descriptive statistics of the variables used for estimation, see Table 2.2.

Robust standard error is computed using the Huber–White sandwich estimator for hypothesis testing.

***, **, and * denote statistical significance at the 1% level, 5% level, and 10% level, respectively.

support for radicalism, while studies that handle issues related to macroeconomic stabilization or enterprise reform and corporate restructuring tend to stress a negative stance toward radicalism. It is also proved that empirical examination is more fre­ quently employed to endorse radicalism. With respect to the publication media attributes, the estimation results of Model 1 imply that support for radicalism is more likely to be manifested in academic books and unpublished working papers than in journal articles. In addition, compared with economics-related media, media that specialize in sociology or politics have a stronger tendency to carry views in favor of radicalism, while media devoted to business administration, and regional study are more likely to publish papers that distance themselves from radicalism. Moreover, it is also suggested from the estimates of Model 1 that publication media of a higher quality level tend to provide a platform to describe pro-radicalism discussions if other conditions remain unchanged. To move on to the estimation results of Model 2, we point out the following rela­ tionship between the degree to which policy sequence is stressed and the literature attributes of the gradualism literature. First, compared with authors who belong to universities or academic research institutions, staff members of international organ­ izations who advocate gradualism pay more attention to the importance of policy

53

C H A P T E R 2

TRANSITION STRATEGY DEBATE

sequence than to problems caused by hasty reforms. Second, research works that study the FSU states, Hungary, and Uzbekistan tend to advocate transition strategies based on slow-paced gradualism rather than on step-by-step gradualism. In contrast, studies that focus on Cuba or certain countries in Southeastern Europe are more likely to justify gradualism from the standpoint of step-by-step gradualism. Third, compared with economics-related media, media that specialize in business adminis­ tration and international relations more aggressively feature discussions that empha­ size policy sequence, while politics-related media have a stronger inclination to feature opinions that stress the time allocation for promoting reforms. To obtain deeper insights into the relationship between the degree to which policy sequence is stressed and the literature attributes within gradualism-advocating litera­ ture, we also estimated a multinomial logit choice model that sets the eclectic grad­ ualism group as its base category. Table 2.4 shows the results. According to this table, we can make additional remarks regarding the aforementioned observations earned from the ordered probit regression of Model 2. The first point is that thinktank staff members are more likely to construct arguments that rely on slow-paced gradualism. Second, authors based in the CEE and FSU countries have a strong inclination to oppose radicalism from the viewpoint of policy sequence more than the time allocation for reforms. Third, female researchers who support gradualism tend to participate in the transition strategy debate avoiding step-by-step gradualism. Fourth, eminent economists have a tendency to express opinions not following slowpaced gradualism. Fifth, compared with studies that discuss policies in general, papers that deal with a concrete policy measure put less emphasis on the policy sequence. Finally, media specializing in business administration and international relations are more likely to publish studies that promote step-by-step gradualism than slow-pace gradualism, while media related to sociology as well as regional study show the opposite relationship. Furthermore, media of politics has a strong preference to publish studies based on slow-pace gradualism. To summarize, there is a close relationship between the debate attitudes and the literature attributes in the basic collection, and the findings reported in this section are helpful for understanding the background of the transition strategy debate and its path to date.

2.6 BEYOND THE DICHOTOMY: CONCLUDING REMARKS The discussion of transition strategies still continues even now, more than a quarter of a century after the collapse of communism in the CEE and FSU region. Through an analytical survey of 140 related studies, we have presented an overall picture of the radicalism-versus-gradualism debate and examined the relationship between debate attitudes and literature attributes in the preceding studies. We found that radicalists maintain their monolithic debate attitude on the speed of reform and policy sequence of the transition strategy, while the gradualists’ debate attitude is

54

BEYOND THE DICHOTOMY: CONCLUDING REMARKS

2.6

Table 2.4 Estimation results of multinomial logit model on the relationship between the debate attitudes and the literature attributes in the gradualism literature Independent variables (Default category)/ Dependent variables (Base category: Eclectic gradualism)

Slow-paced gradualism

Proportion of affiliated institutions (Universities or academic research institutions) Think tanks 23.561*** IMF or World Bank −48.803*** Other international institutions −2.551 Others 48.886*** Proportion of locations of affiliated institutions (Asia and Oceania) North America or Western Europe 0.648 CEE 21.394*** FSU 22.977*** Other authorship attributes Median value of authors’ first publication year Proportion of female researchers Involvement of world-famous economists

0.047 −0.292 −2.046*

Step-by-step gradualism

2

1.226 −1.823 35.968*** 3.019 0.753 45.867*** 173.724*** 0.040 −412.801*** −1.778

Focus on particular regions or countries (Without any particular subject regions or countries) FSU bloc 22.671*** −1.767 China −1.257 0.122 Cuba −26.213*** 265.307*** Czech Republic or Czechoslovakia −1.600 −24.417*** Hungary 11.961*** −24.633*** Poland — — Certain countries in Southeastern Europe 22.335*** 416.637*** Russia −1.622 −174.882*** Uzbekistan 25.374*** 3.101 Focus on particular policy areas (Policies in general) Economic liberalization Macroeconomic stabilization Privatization Enterprise reform and corporate restructuring

— −0.823 24.706*** −0.786

Other research content attributes Outcomes from academic projects Employment of mathematical economic model Intensity of empirical examination

0.659 −0.666 0.338

0.724 −1.859 0.263

−24.036*** 0.645 —

0.061 −22.160*** —

0.455 1.413 26.606*** −3.283** −0.807

452.669*** −14.935*** 3.815 217.624*** −23.271***

Type of publication media (Journal article) Academic book Book chapter Unpublished working paper Specialized fields of publication media (Economics) Business Sociology Politics International relations Regional study Other publication media attributes Publication year Quality level Const.

−0.106 0.027 109.294

C H A P T E R

— −170.050*** −224.966*** −17.472***

−0.127 0.031 173.330

(Continued ) 55

C H A P T E R 2

TRANSITION STRATEGY DEBATE Table 2.4 contd. Independent variables (Default category)/ Dependent variables (Base category: Eclectic gradualism) N Log pseudolikelihood Pseudo R2 Akaike’s information criterion (AIC) Bayesian information criterion (BIC) Wald test (χ2) a

Slow-paced gradualism

Step-by-step gradualism 97 −49.288 0.520 218.577 373.059

7933.91***

38000.00***

Notes:

a Null hypothesis: All coefficients are zero.

For more details on definitions and descriptive statistics of the variables used for estimation, see Table 2.2.

Robust standard error is computed using the Huber–White sandwich estimator for hypothesis testing.

***, **, and * denote statistical significance at the 1% level, 5% level, and 10% level, respectively.

more multifaceted. In fact, gradualists can be divided into a slow-paced gradualism group, a step-by-step gradualism group, and an eclectic gradualism group: the pres­ ence of these three groups is almost balanced. In addition, the debate content of the step-by-step gradualism group is more multi-layered than that of the other two groups, as it is comprised of an institutional gradualism group, which stresses that building institutions should take top priority over any other reform measures. More­ over, we also found that there is another group of researchers that stays within the framework of the radicalism-versus-gradualism debate while, at the same time, remains at arm’s length from both the radicalists and the gradualists. However, these neutralists do not have much of a presence, and, hence, the confrontation between radicalists and gradualists is remarkably vivid in the transition strategy debate. Furthermore, the cross tabulation analysis and the regression estimation of qualitative choice models conducted in the previous section revealed interesting findings for deeper understanding of the transition strategy debate: that is, authorship attributes, including affiliated institutions, their locations, research experiences, and gender, as well as influ­ ence on the academic world are closely related both to the degree of the radicalism stance and the degree to which policy sequence is stressed among gradualists. It also becomes clear that studies that discuss a desirable mode of transition strategy in a specific country or a policy area tend to express much clearer debate attitudes, as compared with general policy discussions. In addition, it is also proved that empirical examination is more fre­ quently carried out to back up radicalism. In other words, the author profiles, research subjects, and methodologies are a major source of the diversified arguments regarding transition strategies during the last quarter century. Furthermore, we found that the types of publication media, their specialized study fields, and their quality level are significantly correlated with the likelihood of specific debate attitudes being published. These results imply that a sort of publication selection bias may exist in this research area. The radicalism-versus-gradualism debate has been developed through the production of a great number of research works characterized by the preceeding findings. In this

56

BEYOND THE DICHOTOMY: CONCLUDING REMARKS

2.6

way, it has not only served as a bellwether of academic argument on transition strategy, but it has also played a significant role in the creation of a new research field called “transition economics.” In recent years, a quarter of a century after the end of the Cold War, some researchers have even declared “the end of economic transition” (Sonin 2013). Under these circumstances, it is difficult to deny that selection of the reform track is no longer a strategically important matter for the majority of CEE and FSU countries, in reality. On a global scale, however, some countries still maintain a strict socialist regime, and, in the near future, these countries might face political and eco­ nomic issues similar to those with which the CEE and FSU countries struggled. In add­ ition, radicalism and gradualism are never confined to the former socialist economies, but rather both policy philosophies can constitute an important platform for discussing structural reforms and other economic policies in developing countries. In some cases, they may even be useful for developed countries. In this sense, we strongly believe that the transition strategy debate, which has developed mainly through the study of the former socialist economies in the CEE and FSU regions and China, should be further deepened and systematized toward an upgrade to a more general policy theory. In our view, one urgent task that researchers must tackle to this end is to expand and enrich the empirical studies. As Panel c of Figure 2.2 shows, econometric stud­ ies account for only 12.1% of the entire basic collection, or 17 of the 140 studies. This means that the transition strategy debate has thus far advanced without having sufficient empirical examination. Lack of empirical evidence allows for discretion and arbitrariness by researchers and ultimately keeps the debate from converging. It is possible that the transition strategy debate has actually been trapped in this bottle­ neck. We understand that this kind of policy debate is difficult to fit into empirical analysis, due to its nature and scope. Nevertheless, we maintain that there is still much room for improvement of this aspect.19 Another task that may significantly contribute to the further development of the transi­ tion strategy debate is the deconstruction of the traditional dichotomy. We agree that this conventional debate format is useful for both the clarification of controversial issues and theoretical considerations. However, there are more than a few cases in which the under­ standing of the reality is excessively simplified and/or trivialized because of enthusiasm to interpret every insight obtained from observations of transition economies within this framework. This adverse effect seems to be getting more pronounced as knowledge and data on the process of economic transition accumulate in larger quantities. One possible breakthrough solution to this problem is provided by the third-way thinkers mentioned in Section 2.3. The dichotomy of radicalism versus gradualism implicitly presumes that all transition economies are aiming to establish a capitalist market economy as their ultimate goal. Only if this precondition is met are we allowed to classify all observable transition economies into three categories—radical reformers, gradual reformers, or intermediate reformers between the two—and then compare the country groups with each other. However, the third-way thinkers are trying to overcome the serious contradictions caused by an unreasonable attempt to discuss all reform results and economic performances in transition countries within this traditional framework by explicitly identifying some former socialist countries 57

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE

(and China, in some researchers’ views) as another country group, which does not necessarily intend to introduce the capitalist market system. A representative scholar among the third-way thinkers is Richard Pomfret, who clearly distinguished Uzbekistan from gradualism-based reforming countries. In Pomfret 2000, he stated that this Central Asian country adopted a development model that is less consistent with the ideal gradualism. Nevertheless, the Uzbek economy was relatively stable throughout the early stage of transition. The key to such a better performance was its unique economic and industrial policies rather than the modest reform speed as stressed by many other researchers. Zettelmeyer (1999) also dealt with the uniqueness of Uzbekistan’s way of transition. He argued that Uzbekistan might have rebuilt its economy by a series of measures that did not match the gradualists’ policy recommendations, indicating that the economic crisis that followed the collapse of the Soviet Union was relatively mild in Uzbekistan, due to the implementation of industrial policy that imposed a strong grip on produc­ tion, in addition to some favorable preconditions including the country’s underdevel­ opment and rich energy resources. Moreover, Herrmann-Pillath (2006) claimed that China’s reform agenda does not set transition to a capitalist market economy as the final goal, but rather its contents change flexibly and opportunistically, depending on the circumstances of the moment. Thus, he concluded that the Chinese way should be distinguished from the standard gradualism model. We ourselves also discussed the economic performance and corruption in the FSU countries from a viewpoint similar to that of Pomfret (2000) and Zettelmeyer (1999) (Iwasaki 2004; Iwasaki and Suzuki 2007). Our arguments start from the fact that the FSU countries can be classified into two types: the first consists of states that adopted a decentralization strategy and, according to this strategy, made efforts to restore their economic systems by devolving the economic power held by the central government under socialism to domestic firms; the second consists of states that fol­ lowed a recentralization strategy to fill the institutional vacuum that was brought about immediately after the breakdown of the Soviet Union by centralizing control over domestic firms in the hands of newly born independent governments and by restructuring industrial organizations to accommodate this change. It is obvious that the decentralization strategy has a high affinity to the debate on radicalism versus gradualism, while the recentralization strategy cannot be handled within this traditional framework due to its heterogeneous nature. Belarus, Turkmeni­ stan, and Uzbekistan have consistently pursued this recentralization strategy through­ out the entire course of their transitions. These three countries are definitely different from the Baltic states, which have carried forward their decentralization strategy in a thorough manner, and the other FSU states including Russia, which share the same policy objective as the Baltic states, although they are lagging behind in terms of the separation between the state and enterprises.20 Therefore, it is highly effective to classify these three countries as a third country group, which belongs neither to the radicalism-adopting countries nor to the gradualism-following countries, to avoid any misleading conclusions from handling their reform experiences in the framework of the orthodox dichotomy. In this sense, we and Myant and Drahokoupil (2010), 58

NOTES

who also clearly distinguish states that adopted the recentralization strategy from other transition countries from the same standpoint as ours, are third-way thinkers. In order to promote the deconstruction of the transition strategy debate, we also need to listen to the opinions of the transcendentalists, who raise a question about the raison d’être of the debate itself. For example, Hoen (1996) criticized the attempt to divide the former socialist countries into two categories, such as the Czech Republic and Poland as radical reformers and Hungary as a gradual reformer, as totally unrealistic, given the fact that both radicalism and gradualism are actually blended, depending on the policy areas in each transition country. From the same point of view, Louzek (2009) argued that it is inappropriate to classify transition economies as radicalism-based or gradual­ ism-based reformers in reference to their respective privatization policies. Meanwhile, Liodakis (2001) raised a question about the essence of the transition strategy debate from an angle substantially different from that of Hoen (1996) and Louzek (2009). He claimed that the radicalism-versus-gradualism debate is built on the premise of a transition from a socialist planned economy. According to him, however, economic transition in the CEE countries started with state capitalism; hence, the debate has missed the point in the first place. Taking additional steps for­ ward, Leijonhufvud and Rühl (1997) expressed the severe view that the transition strategy debate is no longer of importance after a certain degree of advancement in marketization; thus, it makes no sense to continue the debate at all. We believe that the transition strategy debate will further develop into a study area with richer content and insight by going through the process of responding to bitter criticism from the transcendentalists as well as constructive suggestions from the third-way thinkers. We expect great progress in the future.

ACKNOWLEDGMENTS This chapter is a substantially extended and updated version of Iwasaki and Suzuki 2016. We thank Robert M. Buckley, Maren Duvendack, Donald George, Yasushi Nakamura, and Tom D. Stanley for their helpful comments and suggestions on the earlier version of this paper.

NOTES 1 In fact, Myant and Drahokoupil 2010, Turley and Luke 2010, and Åslund 2013, which represent recent basic textbooks of transition economics, devote many pages to discussing the transition strategy debate in their respective introductory chapters. 2 The Washington Consensus refers to a set of economic policy prescriptions for­ mulated by a group of policy-makers and researchers from Washington-based international institutions or administrative organizations such as the International

59

C H A P T E R 2

C H A P T E R

TRANSITION STRATEGY DEBATE

2

3 4 5

6

7

8

9

10

11

12

60

Monetary Fund (IMF), the World Bank, and the US Treasury Department in the course of a series of economic crises that hit developing countries in the 1980s. According to Williamson (1990), the consensus basically consists of the following ten policy agendas: reduction of fiscal deficits, public expenditure priorities, tax reform, liberalization of interest rates, flexible foreign exchange rates, trade liberalization, promotion of foreign direct investment, privatization of state enterprises and other public assets, deregulation, and reinforcement of property rights. See Rodrik 2006, Lütz and Kranke 2014, for valuable com­ ments and critiques on applying the Washington Consensus to CEE and FSU countries. The final literature search was conducted in March 2019.

For another 135 selected studies, see Iwasaki and Suzuki 2015, app. A.

The attention to gender arises from our experimental intuition that there may be

a difference between genders in the degree of relative priority given to social stability as a criterion for policy judgments. We obtained information concerning the author’s first publication year from the ProQuest database (http://www.proquest.com), which extensively covers researchers worldwide. Taking co-authored papers into account, we used the median value as the proxy for the research experience. These high numbers are more likely to have resulted from our methodology of using EconLit as a prime means for the literature search and, hence, the above observation does not necessarily mean that economic journals are the main battlefield for the transition strategy debate. According to the journal ranking that we used to evaluate of the quality level of the publication media, the proportion of 6th to 10th grade journals accounts for 26.0% (304 out of all 1,171 ranked journals). Based on the evaluation criteria described below, we performed this classification work in as objective a manner as possible. However, it is hard to say that arbi­ trariness has been completely eliminated. In addition, the classification result does not necessarily correspond to the individual belief and/or stance of each author in the basic collection at the time of his or her writing. Although their conclusion or policy recommendation is different from that of “conditional radicalism,” more than a few researchers emphasize the effect of the initial conditions on the transition process. For the latest studies, see BenYishay and Grosjean 2014 and Grosfeld and Zhravskaya 2015. It is noteworthy that the argument of the institutional gradualism group has something in common with a series of articles by Daron Acemoglu and others that insists upon the importance of institutions as a crucial determinant of the rise and fall of the state (Acemoglu et al. 2008, 2011; Acemoglu and Robinson 2012). Without duplication, the number of all authors in the basic collection is 163. Among them are 159 authors whose articles are all classified in a single category: these include 37 radicalists (23.3%), 19 neutralists (11.9%), and 103 gradualists (64.8%). Therefore, we can also confirm the predominance of gradualists based on the number of authors in each category.

NOTES

13 There are some cases, such as Balcerowicz 1994 and Balcerowicz and Gelb 1995, where the same author belongs to different research groups at the same time, reflecting the presence of co-authors or different timing of publications, etc., although such cases are extremely limited. In this regard, however, we have not found any single case where the same author belongs to two groups in which the debate attitudes are extremely different from each other. 14 In this regard, however, he is mainly based at the Institute of European, Russian, and Eurasian Studies at Carleton University in Ottawa as well as the United Nations. 15 It is also called Cramér’s V. The value of this coefficient ranges from 0 to 1. If it is closer to 1, the association is evaluated to be stronger. 16 For a robustness check, we also performed an estimation of ordered logit models and found that the results are not much different from those in Table 2.3. 17 Though it has not yet been proven, it is possible that researchers of the postsocialism generation recognized that the situation during the early stage of transi­ tion in the countries required a shock-therapy strategy as a more critical phase than did those of the socialism generation. We also conjecture that eminent economists may have a tendency to avoid going so far as to advocate a shocktherapy strategy, due to concern for their reputations. 18 This result, indicating a close relationship between debate attitudes and focus on particular regions or countries, strongly suggests the possibility that both radical­ ists and gradualists purposefully choose a specific country or region that provides favorable evidence for their arguments. However, this result also suggests that China, which is regarded as a typical country that embraces gradualism, tends to be cited more often than other transition economies to support the radicalismbased transition strategy. As indicated in cross table f in Table 2.1, this seem­ ingly strange result comes from the fact that three studies on China express a neutral debate attitude. 19 Reevaluation and meta-analysis of empirical evidence reported in previous stud­ ied that examined the impact of structural reforms on economic performance in transition economies are also valuable from this viewpoint. See Babecký and Campos 2011 and Babecky and Havranek 2014. 20 As pointed out in Iwasaki 2004, in some cases, such as Azerbaijan and Tajiki­ stan, the government turned their reform strategies from recentralization to decentralization in the course of the economic transition. In Iwasaki 2004 and Iwasaki and Suzuki 2007, we called Belarus, Turkmenistan, and Uzbekistan, all of which adopted and carried out the recentralization strategy, “order states,” in light of their top-down administration system in the relationship between the government and domestic companies. Meanwhile, the Baltic states, which estab­ lished the principle of bringing bankrupt enterprises to justice based on the rule of law, are called “punish states.” Russia and the other remaining former Soviet Republics, in which governments frequently take actions to rescue poorperforming companies due to the incomplete separation of the state-business relationship, are called “rescue states.” 61

C H A P T E R 2

C H A P T E R

TRANSITION STRATEGY DEBATE

2

Acemoglu, Daron, Davide Cantoni, Simon Johnson, and James A. Robinson (2011) The consequences of radical reform: The French revolution. American Economic Review, 101 (7), pp. 3286–3307.

REFERENCES

Acemoglu, Daron, and James A. Robinson (2012) Why Nations Fail: The Origins of Power, Prosperity and Poverty. Crown Publishers: New York.

Acemoglu, Daron, Simon Johnson, James A. Robinson, and Pierre Yared (2008) Income and democracy. American Economic Review, 98(3), pp. 808–842. Aghion, Philippe and Olivier J. Blanchard (1994) On the speed of transition in Central Europe. In Stanley Fischer and Julio J. Rotemberg (eds.), NBER Macroeco­ nomics Annual 1994. Volume 9, MIT Press: Cambridge, MA and London, pp. 283–320. Arrow, Kenneth J. (2000) Economic transition: Speed and scope. Journal of Institutional and Theoretical Economics, 156(1), pp. 9–18. Åslund, Anders (2007) How Capitalism was Built: The Transformation of Central and Eastern Europe, Russia, and Central Asia. Cambridge University Press: New York. Åslund, Anders (2009) Why market reform succeeded and democracy failed in Russia. Social Research, 76(1), pp. 1–28. Åslund, Anders (2013) How Capitalism was Built: The Transformation of Central and Eastern Europe, Russia, and Central Asia. Second edition, Cambridge University Press: New York. Babecký, Jan, and Nauro F. Campos (2011), Does reform work? An econometric survey of the reform–growth puzzle. Journal of Comparative Economics, 39(2), pp. 140–158. Babecky, Jan, and Tomas Havranek (2014), Structural reforms and growth in transition: A meta-analysis. Economics of Transition, 22(1), pp. 13–42. Balcerowicz, Leszek (1994) Common fallacies in the debate on the transition to a market economy. Economic Policy, 9(19) (Supplement), pp. 16–50. Balcerowicz, Leszek, and Alan Gelb (1995) Macropolicies in transition to a market economy: A three-year perspective. In Michael Bruno and Boris Pleskovič (eds.), Proceedings of the World Bank Annual Conference on Development Economics 1994. World Bank: Washington DC, pp. 21–44. BenYishay, Ariel, and Pauline Grosjean (2014) Initial endowments and economic reform in 27 post-socialist countries, Journal of Comparative Economics, 42(3), pp. 892–906. Berg, Andrew, Eduardo Borensztein, Ratna Sahay, and Jeromin Zettelmeyer (1999) The evolution of output in transition economies: Explaining the differences. Working Paper No.WP/99/73, IMF: Wasington DC. Bhagwati, Jagdish (1994) Shock treatments. New Republic, 210(13), pp. 39–43. Blanchard, Oliver, and Michael Kremer (1997) Disorganization. Quarterly Journal of Eco­ nomics, 112(4), pp. 1091–1126. Brada, Josef C. (1993) The transformation from communism to capitalism: How far? How fast? Post-Soviet Affairs, 9(2), pp. 87–110. Calcagno, Peter T., Frank Hefner, and Marius Dan (2006) Restructuring before privatization —putting the cart before the horse: A case study of the steel industry in Romania. Quar­ terly Journal of Austrian Economics, 9(1), pp. 27–45. Coyne, Christopher J., and Peter J. Boettke (2006) The role of the economist in economic development. Quarterly Journal of Austrian Economics, 9(2), pp. 47–68.

62

REFERENCES de Melo, Martha, Cevdet Denizer, and Alan Gelb (1996) Patterns of transition from plan to market. World Bank Economic Review, 10(3), pp. 397–424. de Melo, Martha, Cevdet Denizer, Alan Gelb, and Stoyan Tenev (2001) Circumstance and choice: The role of initial conditions and policies in transition economies. World Bank Economic Review, 15(1), pp. 1–31. Dehejia, Vivek H. (1996) Shock therapy vs. gradualism: A neoclassical perspective. Eastern Economic Journal, 22(4), pp. 425–431. Dewatripont, Mathias, and Gérard Roland (1992a) Economic reform and dynamic political constraints. Review of Economic Studies, 59(4), pp. 703–730. Dewatripont, Mathias, and Gérard Roland (1992b) The virtues of gradualism and legitimacy in the transition to a market economy. Economic Journal, 102(411), pp. 291–300. Dewatripont, Mathias, and Gérard Roland (1995) The design of reform packages under uncertainty. American Economic Review, 85(5), pp. 1207–1223. Etzioni, Amitai (1992) How is Russia bearing up? Challenge, 35(3), pp. 40–43. Fan, Gang (1994) Incremental changes and dual-track transition: Understanding the case of China. Economic Policy, 9(19) (Supplement), pp. 99–122. Gel′vanovskii, Mikhail (1994) Russia on the path to a normal economic system. Problems of Economic Transition, 37(3), pp. 47–61. Grosfeld, Irena, and Ekaterina Zhuravskaya (2015) Cultural vs. economic legacies of empires: Evidence from the partition of Poland. Journal of Comparative Economics, 43 (1), pp. 55–75. Hartwell, Christopher A. (2017) If you’re going through hell, keep going: Nonlinear effects of financial liberalization in transition economies. Emerging Markets Finance & Trade, 53(2), pp. 250–275. Hecht, James L. (1994) Shocked Russians, chagrined economists. Orbis, 38(3), pp. 499–504. Herrmann-Pillath, Carsten (2006) Cultural species and institutional change in China. Journal of Economic Issues, 40(3), pp. 539–574. Hoen, Herman W. (1996) Shock versus gradualism in the Central Europe reconsidered. Comparative Economic Studies, 38(1), pp. 1–20. Hoen, Herman W. (2010) Transition strategies in Central Asia: Is there such a thing as “shock-versus-gradualism”? Economic and Environmental Studies, 10(2), pp. 229–245. Islam, Shafiqul (1993) Russia’s rough road to capitalism. Foreign Affairs, 72(2), pp. 57–66. Iwasaki, Ichiro (2004) Evolution of the government-business relationship and economic per­ formance in the former Soviet states: Order state, rescue state, punish state. Economics of Planning, 36(3), pp. 223–257. Iwasaki, Ichiro, and Taku Suzuki (2007) Transition strategy, corporate exploitation, and state capture: An empirical analysis of the former Soviet states. Communist and PostCommunist Studies, 40(4), pp. 393–422. Iwasaki, Ichiro, and Taku Suzuki (2015) Radicalism versus gradualism: An analytical survey of the transition strategy debate. RRC Working Paper No. 45, Russian Research Center, Institute of Economic Research, Hitotsubashi University: Kunitachi, Tokyo. Iwasaki, Ichiro, and Taku Suzuki (2016) Radicalism versus gradualism: An analytical survey of the transition strategy debate. Journal of Economic Surveys, 30(4), pp. 807–834. Jiang, Chenchao, Di Shang, and Shixiong Cao (2016) Managing institutional evolution to increase the likelihood of success: Examples of guidance from Chinese history. Social Indicators Research, 127(3), pp. 1157–1167.

63

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE King, Lawrence (2002) Postcommunist divergence: A comparative analysis of the transition to capitalism in Poland and Russia. Studies in Comparative International Development, 37(3), pp. 3–34. Klaus, Vaclav (1993) Interplay of political and economic reform measures. Vital Speeches of the Day, 60(5), pp. 130–132. Kornai, János (1994) Transformational recession: The main causes. Journal of Comparative Economics, 19(1), pp. 39–63. Lavigne, Marie (2000) The economics of the transition process: What have we learned? Problems of Post-Communism, 47(4), pp. 16–23. Leijonhufvud, Axel, and Christof Rühl (1997) Russian dilemma. American Economic Review, 87(2), pp. 344–348. Lian, Peng, and Shang-Jin Wei (1998) To shock or not to shock? Economics and political economy of large-scale reforms. Economics and Politics, 10(2), pp. 161–183. Liew, Leong H. (1995) Gradualism in China’s economic reform and the role for a strong cen­ tral state. Journal of Economic Issues, 29(3), pp. 883–895. Liodakis, George (2001) The political economy of the transition and the transformation crisis in Eastern Europe. East-West Journal of Economics and Business, 4(2), pp. 45–64. Lipton, David, and Jeffrey D. Sachs (1990) Creating a market economy in Eastern Europe: The case of Poland. Brookings Papers on Economic Activity, 21(1), pp. 75–148. Louzek, Marek (2009) The Czech privatisation after 20 years. Post-Communist Economies, 21(3), pp. 345–359. Lütz, Susanne, and Matthias Kranke (2014) The European rescue of the Washington con­ sensus? EU and IMF lending to Central and Eastern European countries. Review of Inter­ national Political Economy, 21(2), pp. 310–338. McMillan, John, and Barry Naughton (1992) How to reform a planned economy: Lessons from China. Oxford Review of Economic Policy, 8(1), pp. 130–143. Marangos, John (2003) Was shock therapy really a shock? Journal of Economic Issues, 37 (4), pp. 943–966. Marangos, John (2004a) A post-Keynesian approach to the transition process. Eastern Economic Journal, 30(3), pp. 441–465. Marangos, John (2004b) Alternative Economic Models of Transition. Ashgate: Burlington, VA. Marangos, John (2006) Was market socialism a feasible alternative for transition econ­ omies? International Journal of Political Economy, 35(3), pp. 64–88. Mikeladze, Giorgi, and Simon Gelashvili (2016) Gradualistic strategy of transition to market economy. Theoretical and Applied Economics, 23(4), pp. 237–242. Minniti, Maria, and Lidija Polutnik (2007) Currency conversion and the role of expectations: The case of Slovenia. Economic and Business Review, 9(1), pp. 5–21. Molchanov, Mikhail A. (2005) Russia and globalization. Perspectives on Global Develop­ ment and Technology, 4(3/4), pp. 397–429. Murphy, Kevin M., Andrei Shleifer, and Robert W. Vishny (1992) The transition to a market economy: Pitfalls of partial reform. Quarterly Journal of Economics, 107(3), pp. 889–906. Murrell, Peter (1992a) Evolutionary and radical approaches to economic reform. Economics of Planning, 25(1), pp. 79–95. Murrell, Peter (1992b) Evolution in economics and in the economic reform of the centrally planned economies. In Cristopher Clague and Gordon C. Rausser (eds.), The Emergence of Market Economies in Eastern Europe. Blackwell: Oxford, pp. 35–53.

64

REFERENCES Myant, Martin, and Jan Drahokoupil (2010) Transition Economies: Political Economy in Russia, Eastern Europe, and Central Asia. J. Wiley: Hoboken, NJ. Noman, Omar (1999) In the former eastern bloc, a hazardous transition. UNESCO Courier, 52(3), pp. 30–31. North, Douglass C. (1994) Economic performance through time. American Economic Review, 84(3), pp. 359–368. Papapanagos, Harry, and Peter Sanfey (2003) Emigration and the optimal speed of transition. Review of International Economics, 11(3), pp. 541–554. Pomfret, Richard (2000) The Uzbek model of economic development, 1991–1999. Econom­ ics of Transition, 8(3), pp. 733–748. Popov, Vladimir (2000a) Shock therapy versus gradualism: The end of the debate (explain­ ing the magnitude of transformational recession). Comparative Economic Studies, 42(1), pp. 1–57. Popov, Vladimir (2000b) Shock therapy versus gradualism: Ten years down the road [book review of “From Shock to Therapy. The Political Economy of Postsocialist Transformation,” by Grzegorz W. Kolodko]. Comparative Economic Studies, 42(3), pp. 121–125. Popov, Vladimir (2007) Shock therapy versus gradualism reconsidered: Lessons from transition economies after 15 years of reforms. Comparative Economic Studies, 49(1), pp. 1–31. Popov, Vladimir (2009) Lessons from the transition economies: Putting the success stories of the postcommunist world into a broader perspective. UNU-WIDER Research Paper No. RP2009/15, World Institute for Development Economics Research, United Nations University: Helsinki. Popov, Vladimir (2012) Russia: Austerity and deficit reduction in historical and comparative perspective. Cambridge Journal of Economics, 36(1), pp. 313–334. Rodrik, Dani (2006) Goodbye Washington consensus, hello Washington confusion? A review of the World Bank’s economic growth in the 1990s: Learning from a decade of reform. Journal of Economic Literature, 44(4), pp. 973–987. Rosefielde, Steven (2001) Premature deaths: Russia’s radical economic transition in Soviet perspective. Europe-Asia Studies, 53(8), pp. 1159–1176. Rozelle, Scott, and Johan F. M. Swinnen (2009) Why did the communist party reform in China, but not in the Soviet Union? The political economy of agricultural transition. China Economic Review, 20(2), pp. 275–287. Sachs, Jeffrey. D. (1996) The transition at mid decade. American Economic Review, 86(2), pp. 128–133. Selowsky, Marcelo, and Ricardo Martin (1997) Policy performance and output growth in the transition economies. American Economic Review, 87(2), pp. 349–353. Sonin, Konstantin (2013) The end of economic transition: Lessons for future reformers and students of reform. Economics of Transition, 21(1), pp. 1–10. Stiglitz, Joseph E. (1999) Whither reform? Ten years of the transition. Annual Bank Confer­ ence on Development Economics, April 1999, World Bank: Washington DC. Sušjan, Andrej, and Tjaša Redek (2008) Uncertainty and growth in transition economies. Review of Social Economy, 66(2), pp. 209–234. Svejnar, Jan (1989) A framework for the economic transformation of Czechoslovakia. PlanE­ con Report, 5(52), pp. 1–18.

65

C H A P T E R 2

C H A P T E R 2

TRANSITION STRATEGY DEBATE Swaan, Wim, and Maria Lissowska (1996) Capabilities, routines, and East European economic reform: Hungary and Poland before and after the 1989 revolutions. Journal of Economic Issues, 30(4), pp. 1031–1056. Turley, Gerard, and Peter J. Luke (2010) Transition Economics: Two Decades On. Routledge: London and New York. Van Brabant, Jozef M. (1993) Lessons from the wholesale transformations in the East. Com­ parative Economic Studies, 35(4), pp. 73–102. Van Brabant, Jozef M. (1994a) Alternative trade regimes and the economics of transition. Russian and East European Finance and Trade, 30(1), pp. 32–52. Van Brabant, Jozef M. (1994b) Bad debts and balance sheets in transforming Eastern Europe. Russian and East European Finance and Trade, 30(2), pp. 5–33. Williamson, John (ed.) (1990) Latin American Adjustment: How Much has Happened. Institute for International Economics: Washington DC. Zettelmeyer, Jeromin (1999) The Uzbek growth puzzle. IMF Staff Papers, 46(3), pp. 274–292. Zweynert, Joachim (2006) Shared mental models, catch-up development and economic policy-making: The case of Germany after World War II and its significance for contempor­ ary Russia. Eastern Economic Journal, 32(3), pp. 457–478.

66

3

Transformational recession and recovery

Determinants of the J-curved growth path Ichiro Iwasaki and Kazuhiro Kumo

3.1 INTRODUCTION When they initially abandoned socialism with the aim of restructuring and revitalizing their national economies with market principles, the countries of Central and Eastern Europe (CEE) and the former Soviet Union (FSU) were hit by a catastrophic collapse in economic activity. Even in the least affected countries, the size of this collapse ranged from 13% to 20% of gross domestic product (GDP) in the final stages of socialism, while in those countries most seriously affected, it amounted to declines of between 64% and 87%. Moreover, this dramatic drop in production continued for as long as between six and eight years in some countries (Table 3.1). That the transition to a market economy would probably trigger social disorder and economic decline was to some extent predicted by policy-makers and researchers both inside and outside the former socialist countries. At the end of the Cold War, however, hardly anyone could have foreseen that the CEE and FSU countries would experience such a serious and pro­ tracted output fall. Furthermore, the recovery process dashed the hopes of large numbers of people. This was because even the most developed CEE countries took between five and eight years to restore output to the levels at the end of socialism. Not only that, but even in the 25th year of the transition from socialism, some countries have yet to fully recover from the economic damage they suffered during the crisis.1 A sharp contraction in production at the start of the transition and the relatively grad­ ual recovery that followed was a situation common to all the CEE and FSU countries. In other words, without exception, the former socialist transition economies have fol­ lowed the so-called “J-curved” growth path to date (Brada and King 1992). Because the J-curve was observed not only in the countries with extensive transitions but also in non-reforming states like Belarus, Uzbekistan, and Turkmenistan, it is obvious that this phenomenon cannot be ascribed to economic policies alone. There must be a bigger pic­ ture beyond them. At the same time, as Table 3.1 shows, there were marked differences among CEE and FSU countries in the length and depth of the decline in output as well

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY Table 3.1 Length and depth of economic crisis, and recovery speed after the crisis period in 28 CEE and FSU countries

3 Region/Subregion/ Country 28 CEE and FSU countries (average) Central Europe (CE) and the Baltic countries (average) Croatia Czech Republic Estonia Hungary Latvia Lithuania Poland Slovak Republic Slovenia South Eastern Europe (SEE) (average) Albania Bosnia and Herzegovina Bulgaria FYR Macedonia Montenegro Romania Serbia FSU excluding the Baltic countries (average) Armenia Azerbaijan Belarus Georgia Kazakhstan Kyrgyz Republic Moldova Russia Tajikistan Turkmenistan Ukraine Uzbekistan

Average continuous years of the economic crisis in the beginning period of transition

Average output decline during the crisis period (end of socialism = 100) a

Average real GDP growth rate during first 10 years of recovery after the crisis

4.0

60.2

5.7

3.3

73.1

4.7

4 3 3 4 4 3 2 4 3 4.0

59.5 86.9 77.0 81.9 56.2 59.5 82.2 75.3 79.7 54.6

4.3 2.0 6.7 3.7 6.8 5.6 4.7 4.3 4.1 5.3

3 4

60.1 13.5

7.0 18.7

4 6 4 3 4 4.5

73.3 70.9 48.7 74.9 40.6 53.8

1.5 2.2 3.1 1.5 2.8 6.8

2 4 4 3 4 4 5 5 5 6 8 4

53.1 42.2 66.1 36.5 69.0 55.0 44.9 62.8 34.1 54.1 44.8 82.5

7.5 10.5 6.9 5.9 6.5 4.7 3.4 5.2 7.3 14.6 4.7 4.4

Multiple comparison of 3 subregions b ANOVA (F) 2.50 4.66* Bartlett test (χ2) Kruskal–Wallis test (χ2) 4.73*

4.40** 3.10 4.73*

0.86 12.45*** 5.90*

(Continued )

68

INTRODUCTION

3.1

Table 3.1 contd.

Region/Subregion/ Country

Average continuous years of the economic crisis in the beginning period of transition

Univariate comparison of country groups c (a) Structural change d Countries in which 3.3††† share of private sector in GDP is 75% or more Countries in which 4.6 share of private sector in GDP is less than 75% (b) Transformation policy e Countries of which 3.4††† average EBRD reform score is 3.5 or more Countries of which 4.6 average EBRD reform score is less than 3.5 (c) Socialist legacy (initial conditions) CEE countries 3.7 FSU including the 4.3 Baltic countries (d) Inflation f Countries with 4.4†† higher inflation in the first 5 years of transition Countries with 3.4 lower inflation in the first 5 years of transition (e) Regional conflict g Countries with a 4.1 regional conflict(s) in 1990s Countries without 3.9 regional conflict in 1990s

Average output decline during the crisis period (end of socialism = 100) a

Average real GDP growth rate during first 10 years of recovery after the crisis

65.2†

5.3

56.4

6.0

69.1†††

4.3

51.3

7.1

65.2† 55.8

4.6 6.7

55.0†††

6.9

71.0

4.3

44.4†††

6.6

69.0

5.3

Notes a Reference year for CEE countries is 1989, for FSU countries 1991. b ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. c One-sided t test. †††, ††, and † denote statistical significance at the 1%, 5%, and 10% levels, respectively. d EBRD estimation in 2010 e In 2010. Czech Republic is included to the upper country group. f Due to data limitations, Bosnia and Herzegovina, FYR Macedonia, Montenegro, and Serbia are excluded from the univari­ ate comparison. g Countries that experienced a regional conflict(s) in 1990s include the following 10 countries: Armenia, Azerbaijan, Bosnia and Herzegovina, Croatia, Georgia, FYR Macedonia, Moldova, Montenegro, Serbia, and Tajikistan. Source: EBRD website (http://www.ebrd.com)

69

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

as the speed of growth during the recovery phase. Faced with such a situation that was profoundly interesting from an academic point of view, researchers have not only offered various theoretical considerations of this unique phenomenon but also performed a variety of empirical analyses to specify its background factors. Currently, much atten­ tion is being paid to this accumulated research, and it is by no means an overstatement to say that it has become one of the most important research fields in the economics of transition. The results of this aggressive research have led us to share a common understanding of the determinants of macroeconomic growth in the CEE and FSU countries. Among others, not only education levels and human capital investment, which are emphasized in traditional growth theory, but even inputs such as capital and labor were not critical explanatory variables for economic growth rates during the crisis and the initial phase of recovery. Rather, the only interpretation is that various unique factors pertaining to the CEE/FSU zone and the former socialist transition economies were quite important in determining macroeconomic performance during these periods. Specifically, these unique factors are (1) structural changes in the national economy, (2) the transformation policy toward a market economy, (3) the legacy of socialism as an initial condition, (4) inflation, and (5) regional conflict. In fact, many previous studies have empirically verified that the first two factors serve to enhance economic growth while the last three factors tend to cause a downturn. Nevertheless, more than a few studies have produced results that contradict the above policy implications. Thus it cannot be said that the transition economy growth debate has reached a final conclusion. Furthermore, no comparison has been conducted on the effect size and statistical significance of the above five determinants of growth, so the question of why the CEE and FSU countries have followed not a U-shaped or V-shaped growth path but a J-shaped trajectory has not been answered by previous research. Elucidating this issue will therefore fill a big gap in the study of transition economies. Based on the above perception of the issues, in this chapter, we will attempt to shed light on the mechanism that generated the J-curved growth path in transition economies by performing a meta-analysis to compare effect size and statistical significance of structural change, transformation policy, the socialist legacy, infla­ tion, and regional conflict. The meta-synthesis, which comprised 3,279 estimates extracted from 123 earlier studies, revealed that the growth-enhancing effects of structural change and transformation policy were small, yet statistically significant, while inflation and regional conflict demonstrated a highly significant and strongly negative effect on output. The effect size and statistical significance of the social­ ist legacy were similar to those of structural change and transformation policy, so it is likely that this factor also contributed to the decline in production in the early stages of transition. The meta-regression analysis that simultaneously con­ trolled for various research conditions and the assessment of publication selection bias (PSB) provided supporting evidence for the policy implications obtained from the meta-synthesis. Based on these results, we conjecture that while the interaction between the above five factors led to a J-curved growth path in all of the CEE and FSU countries, differences among the countries in terms of historical 70

OUTPUT FALL AND RECOVERY IN TRANSITION ECONOMIES

3.2

preconditions, political circumstances, and reform efforts resulted in large differ­ ences in their growth trajectories. The earliest meta-analyses on the factors of macroeconomic growth in former social­ ist transition countries were performed by Babecký and Campos (2011) and Babecky and Havranek (2014). In this chapter, we will use the advantages of later research to supplement these two early studies in three ways. First, whereas the two early studies constituted meta-analyses focused on economic reform, this chapter will verify the growth-enhancing effect of transformation policy in a broader sense. Second, as stated above, because this study deals simultaneously with five determinants of growth that differ in nature, the effect size and statistical significance of transformation policy are in clear contrast to the other four factors. Third, by involving an extensive examination of related studies covering almost every piece of literature targeted by these previous metastudies, this chapter provides a wider picture of the research on transition economies.2 The remainder of this chapter is structured as follows: Section 3.2 gives an overview of the process of output decline and recovery in CEE and FSU countries during the past quarter century. Section 3.3 considers the factors we ought to focus on to understand the mechanism of the emergence of the J-curved growth path through a comprehensive review of previous studies. Section 3.4 describes the procedure of literature selection and an overview of the studies selected for meta-analysis. Section 3.5 conducts a metasynthesis and meta-regression analysis, and Section 3.6 verifies the presence and degree of PSB. Section 3.7 summarizes the major findings and draws conclusions.

3.2 OUTPUT FALL AND RECOVERY IN TRANSITION ECONOMIES: LOOKING BACK ON THE PAST QUARTER CENTURY In this section, we will employ time-series data for real GDP growth rates to identify the characteristics of 28 CEE and FSU countries during the past quarter century. As stated in the Introduction, in the several years immediately following the start of the systemic transformation from the planned system to a market economy, these former socialist countries were hit with a severe economic crisis that was described as a “transformational recession” (Kornai 1994). Later, the negative growth seen at the time of the crisis was replaced by a period of slow growth, with the absolute value of positive growth rates during this period not being as high as that of negative growth rates recorded during the crisis period. In that sense, every country followed a J-shaped growth path. However, conspicuous differences between countries emerged in the length of the economic crisis, the rate of decline in output during the crisis period, and the speed of recovery from the crisis. Figure 3.1 plots the path of output fall and recovery for all 28 CEE and FSU countries as well as for the Central European and Baltic region, the Southeast European (SEE) region, and the FSU region excluding the Baltic countries. The figure puts the starting points (t0) for real GDP at 1989 for the CEE countries and 1991 for the FSU states. A look at the

71

C H A P T E R 3

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

3

Figure 3.1 Output fall and recovery in CEE and FSU countries during 25 years of transition Note: The real GDP level at the end of the socialist regime (t0) is set at 100. The reference year for CEE countries is 1989, for FSU countries 1991. Source: EBRD website (http://www.ebrd.com)

overall trend for the 28 CEE and FSU countries reveals that their economic down­ turn continued until the fifth year following the beginning of system transformation. However, it took 11 years from the bottom for this drop in output to recover to the level at the end of socialism. In other words, compared with the output shrinkage during the transformational reces­ sion, the subsequent recovery took more than twice as long. However, a comparison by region reveals large differences in the patterns seen. In fact, the degree of decline in output during the economic crisis in the SEE and FSU regions was much larger than that in the Central European and Baltic region. Furthermore, the phase of recovery from the crisis took more time in the SEE and FSU regions than in the Central European and Baltic region. Actually, it took until 2004 for output in the FSU region to recover to the level in 1991, while it was not until 2007 that output in the SEE region returned to the 1989 level. These results contrasted sharply with the speed of economic recovery in the Central European and Baltic region. 72

OUTPUT FALL AND RECOVERY IN TRANSITION ECONOMIES

3.2

In Table 3.1, based on analysis of variance (ANOVA) and/or Kruskal–Wallis tests, we confirm that the number of years the output fall lasted, the rate of decline in output during the crisis, and the average real growth rate in the first decade of recovery all exhibited significant differences between the three regions. These results also show that in the SEE and FSU regions the crisis triggered at the initial stage of transition was much more severe than that in the Central European and Baltic region. According to country-level data, in the SEE region, three countries of the former Yugoslavia, namely, Serbia, Bosnia–Herzegovina, and Montenegro, and in the FSU region, five countries, Azerbaijan, Ukraine, Georgia, Tajikistan, and Moldova, lost more than 50% of their output, as compared with that at the end of socialism, during the crisis. In the Central European and Baltic region, however, not one single country experienced a production decline on a par with those seen in the eight SEE and FSU countries. However, from the viewpoint of the robustness of economic growth during the recovery phase, the SEE and FSU regions were not necessarily much inferior to the Central European and Baltic region. Conversely, economic growth in the FSU region was actually superior to that in the Central European and Baltic region. Nevertheless, trends like this, obtained by making comparisons among regions, are not observed for the most part at the level of the countries that comprise each region. Rather, in the background that led to this situation was the fact that several countries in the SEE and FSU regions experienced rapid economic growth as soon as the recovery period began. Furthermore, while not as clear-cut as economic growth rates in the recovery period, there were fairly large differences among countries in the SEE and FSU regions in the number of years the crisis lasted and the rate of decline in output during the crisis. As the above has shown, intra-regional differences seen in the growth paths during the transition period were actually more obvious than inter-regional differ­ ences. Therefore, we performed a nonhierarchical cluster analysis, employing a k-means algorithm using index data, with 100 as the figure for the end of social­ ism, in an attempt to perform comparisons for CEE and the FSU from a different perspective than regional differences. The number of clusters was designated as three. According to the results of the cluster analysis, the three countries from the former Yugoslavia as well as the four FSU countries, Ukraine, Georgia, Tajikistan, and Moldova, that experienced the most severe drop in output and were also the slowest to recover constitute a single cluster. This can be regarded as the group of transition economies that exhibited the worst macroeconomic performance during the past quarter century. However, a separate cluster comprises a group of coun­ tries characterized by the best macroeconomic performance. These are six countries in the Central Europe and Baltic region, Estonia, Slovakia, Slovenia, the Czech Republic, Hungary, and Poland, as well as five FSU countries in which economic administration was relatively good: Armenia, Uzbekistan, Kazakhstan, Turkmenistan, and Belarus.3 The other cluster, which comprises the ten remaining countries, includ­ ing Romania and Russia, can be positioned as neutral, lying somewhere between these two groups of countries. 73

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

Figure 3.2 plots the growth paths of the above three transition-country clusters. The group of nations that exhibited the worst macroeconomic performance during the past quarter century is Cluster 1, whereas the group of best performing economies is Cluster 3. The middle group is Cluster 2. Like the growth curves for each region shown in Figure 3.1, J-shaped growth curves that go beyond the differences among clusters are also reproduced in Figure 3.2. Furthermore, it is clear that the differences between groups of countries in the shapes of the growth curves are more distinct than in Figure 3.1. This fact strongly suggests that the processes of economic crisis and recovery in the CEE and FSU countries are highly likely to have been affected by differ­ ences in the three transition-country clusters and more so than by regional differences. As determinants of macroeconomic growth in transition countries that cannot simply be eliminated by regional differences, what sort of factors should we be focusing on? Providing an answer to this question is the task of the next section.

Figure 3.2 Growth path of three clusters of transition economies Note: The real GDP level at the end of the socialist regime (t0) is set at 100. The reference year for CEE countries is 1989,

for FSU countries 1991.

Source: EBRD website (http://www.ebrd.com)

74

THE DEBATE ON ECONOMIC CRISIS AND GROWTH

3.3

3.3 THE DEBATE ON ECONOMIC CRISIS AND GROWTH IN TRANSITION ECONOMIES From the beginning of the transitional processes, various debates have occurred among policy-makers and researchers concerning crisis and recovery in the CEE and FSU economies. Kornai’s “transformational recession” concept (1994), which focuses on the characteristics of socialist planned economies, and Blanchard and Kremer’s “disorganization hypothesis” (1997) had a big impact on the debate at the time by offering in-depth understandings of the economic crisis in the former socialist coun­ tries, which had become more serious than expected. Furthermore, a series of studies that appeared later, such as those by Heybey and Murrell (1999), de Melo et al. (2001), Havrylyshyn and Wolf (2001), and Falcetti et al. (2002), contended that the serious adverse impact of hyperinflation, which was triggered by the monetary over­ hang (excess liquidity) accumulated by the so-called economies of shortage during the socialist era, as well as the negative legacy of socialism as a historical initial condition and specific regional problems typified by civil wars, ethnic conflicts, and so on, had a major and negative effect on the CEE and FSU economies. Meanwhile, a series of studies by Mitrović and Ivančev (2010), Apolte (2011), and Peev and Mueller (2012), which focused on the recovery process in the transitional economies, also endeavored to empirically verify the growth-enhancing effect not only of economic reforms but also of various other policy measures that could affect national economic activity. In the end, the consensus reached by researchers on transition economies was that neither the long-term economic growth factors such as education level and human capital investment focused on by Mankiw et al. (1992) nor capital and labor inputs, which are essential for standard economic growth models, were important factors for determining output levels and growth rates in the CEE and FSU countries as meas­ ured in terms of GDP or GDP per citizen (or per worker). Rather, from the perspec­ tive of the depth of the economic crisis and the speed of the subsequent recovery, what led to the striking differences between these countries were the following five factors: (1) structural changes in the national economy, (2) a transformation policy aimed at establishing a market economy, (3) the legacy of socialism as an initial condition, (4) inflation, and (5) regional conflict (Havrylyshyn 2001; Campos and Coricelli 2002; Iwasaki 2004). Below, we will examine each of the five factors in turn, explore what sort of debate concerning them has existed in previous research, and use a range of vari­ ables to consider in what way the previous research attempted to empirically identify their impact on economic growth. At the end of this section, we will discuss why meta-analysis is required for the elucidation of the mechanism by which the J-shaped growth path appeared. 3.3.1 Structural changes in the national economy Socialist countries, which viewed the quantity of labor input and intermediate goods invested as the source of value, gave priority to material production and favored

75

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

heavy industry. They did not develop their financial sectors. Furthermore, to facili­ tate planning and management, these countries centralized the locations of produc­ tion facilities and constructed enormous factories (so-called gigantism). Moreover, the social system, which included educational and research institutions, was designed and developed to support this physical-good-focused production system. In light of such characteristics of the socialist economic system, researchers have employed a variety of indicators as variables for identifying structural changes in national economies that have occurred in conjunction with system transformation. These include the private-sector share of GDP, which indicates the degree of change in the composition of a production system previously dominated by staterun companies; the degree of trade openness, which reflects the extent of freedom and diversity in external economic activity, something that had been effectively monopolized by national governments under the COMECON structure; and the degree of penetration of bank lending and depth of financing, which indicate the development of the financial system, which had been given only an extremely limited role in planned economies. Since 2000, as state-owned businesses have been privatized, a great deal of progress has been made in the verification of the impact of the expansion of the private sector on economic growth. Fischer and Sahay (2001) used panel data covering 25 CEE and FSU countries during the period 1990 to 1998. Their study was one of the first to identify a positive correlation between private-sector GDP share and real economic growth rates. A study by Próchniak (2011), which was released approximately a decade later, constituted a quantitative analysis of the CEE countries newly joined to the EU and reproduced empirical results similar to those of Fischer and Sahay (2001). However, a number of studies, such as those by Bennett et al. (2004, 2007) and Sukiassyan (2007), while indicating that the private-sector GDP share is positively correlated with the rate of economic growth, have reported estimation results suggesting that this correlation is statis­ tically insignificant. Thus there are big differences among the results of empirical assessments. A well-known example of a study that verified the cause-and-effect relationship between trade openness and the economic growth rate was that performed by Cernat and Vranceanu (2002). They studied ten CEE countries during the 1990s and demon­ strated that a close positive correlation has been established between these two vari­ ables. Capolupo and Celi (2005), meanwhile, reported that the higher the degree of trade openness in a transitional country, the higher the rate of economic growth tends to be. Other examples of previous research that has verified the relationship between trade openness and GDP growth rate are studies by Nath (2009) and Josifi­ dis et al. (2012), and both of these studies identified a significant and positive correl­ ation between the two variables. Of course, studies with opposite findings, such as those by Campos and Kinoshita (2002) and Neyapti and Dincer (2005), also exist, but the number of such studies is relatively small. Studies on the relationship between the development of the financial sector and economic growth during the transition period have also been published one after 76

THE DEBATE ON ECONOMIC CRISIS AND GROWTH

3.3

another, starting with one by Halushka (1997). Recent empirical research in this area includes studies by Akimov et al. (2009), Gaffeo and Garalova (2014), and Cojocaru et al. (2016). These studies all confirmed that the degree of bank loan penetration and financial depth have a significant and positive impact on economic growth. Kornai (1994), who put forward the notion of a “transformational recession,” argued that the lack of financial system development was one of the factors behind that recession and that to overcome this problem, the establishment of private-sector commercial banks would be important. Furthermore, the results obtained by Akimov et al. (2009) provide empirical backing for that claim. Nevertheless, research in this area includes a lot of studies, such as those by Djalilov and Piesse (2011) and Dudian and Popa (2013), that have provided empirical evidence that the impact of financial-sector development on enhancing economic growth is either neutral or negative, and these empirical findings are more confusing than those concerning the effects of expansion of the private sector, trade activity, and so forth. In light of the above discussion, let us return to Table 3.1. Column a of this table demonstrates the impact of structural change in the economic system on the growth paths followed by the CEE and FSU countries. Here, based on the percentage share of the private sector in total GDP, a frequently used indicator as a proxy for struc­ tural change, the 28 transitional countries were divided into two groups: an upper group with a private-sector GDP share equal to or above 75% and a lower group with a private-sector GDP share below 75%. Then, from the perspective of the length of the economic crisis, the rate of decline in output during the crisis period, and the average rate of real economic growth during the first decade of the recovery phase, we examined whether there is a statistically significant difference between these two groups of countries. According to the results, in the case of the group of countries that have made a high degree of progress with structural change, the length of the economic crisis was, compared with the lower group, 1.3 years shorter on average, with a one-tailed 1% significance level. Furthermore, the rate of decline in output during the crisis period was also 8.8% lower at a 10% level. In other words, in countries with a relatively high degree of structural change, the results of the analysis shown in Column a of Table 3.1 are, in the sense that they hint at the possibility that the damage inflicted by the economic crisis was relatively minor, in line with the arguments put forward in previous research: that is, that structural change served to suppress the economic crisis. However, a significant difference in economic growth rates between the two groups of countries was not detected, so clear support for the contention that there was a positive correlation between struc­ tural change and economic growth was not obtained using this analytical framework. 3.3.2 Transformation policy toward a market economy Some of the most fundamental research questions in the economics of transition are what effect policies to promote system transformation had on market economies, how those policies were designed and implemented, and what sort of effects they had. Opinions concerning the answers to these questions in previous research, even 77

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

when limited to the relationship between system transformation policy and macro­ economic growth, are extremely varied. Actually, in the debate on the growth of transitional economies, the effect of transformation policy on promoting growth has been the topic of most interest to researchers in this field, and, because of that, it is no exaggeration to say that this issue has led to extremely vigorous theoretical and empirical investigations. In fact, as will be mentioned later, most of the previous studies covered by the meta-analysis performed in this chapter are focused on the empirical analysis of transformation policy. Among transformation policies, economic reform, which comprises liberalization, pricing/competition policies, corporate reform, privatization policy, financial/trade policy, and so on, have always been of the most interest to researchers since the work performed by Åslund et al. (1996) and, most recently, de Rocha (2015). There are far too many studies to mention in this chapter that have used either (1) the “Transition Indicators” of the European Bank for Reconstruction and Development (EBRD), which classify the extent of progress in economic reform in CEE and FSU countries by assigning each country one of five grades, or (2) the US Heritage Foundation’s “Index of Economic Freedom,” which gives each of the world’s countries an overall score based on its degree of economic freedom, or (3) variables that are the result of adjustments made to the “Transition Indicators” or the “Index of Economic Freedom” to verify the impact of economic reform on growth. An overview of previous research alone gives the impression that reform measures served, on the whole, to enhance eco­ nomic growth. At the same time, however, researchers’ choices about which countries/ territories to cover, the period of observation, and the types of reform measures to look at can dramatically alter the empirical results, and this is characteristic of this research field. Following economic reform, the transition policy given the next most attention by researchers has been democratization, the pillars of which include the introduction of a parliamentary system or a multi-party system. Fidrmuc (2001, 2003), Heckel­ man (2010), Apolte (2011), and Peev and Mueller (2012) produced research find­ ings by tackling the relationship between democratization and economic growth head on. On the whole, these previous studies, which have employed “democracy indicators” resulting from investigations and calculations performed by such organizations as Freedom House in the USA and the World Bank to carefully examine the interrelationship between the two phenomena, negate the minority view that democratization operates as a direct driver of economic growth. In fact, even Fidrmuc (2003), who acknowledges the economic-growth-enhancing effect of democracy, expresses a modest view, stating that democratization has, by increasing economic freedom, had a positive, yet indirect, impact on economic growth during the transition period. The estimation results of other previous stud­ ies which have employed democratization indicators as a control variable were mixed, and the overall trend in empirical findings has been less clear than that of economic reform. The scope of transformation policy considered by researchers extends beyond eco­ nomic reform and democratization. The rule of law and judicial reform are also 78

THE DEBATE ON ECONOMIC CRISIS AND GROWTH

3.3

important components. Most of the studies that have focused on these areas of reform have employed figures from third-party organizations, most notably Free­ dom House, to verify the relationship between economic growth in each country and the rule of law, the degree of the establishment of property rights, and the independence/fairness of judicial institutions, and they have identified a positive correlation between them (Grogan and Moers 2001; Godoy and Stiglitz 2006; Popov 2007; Eicher and Schreiber 2010). Other studies, albeit a much smaller number, have investigated the impact of the institutional nature of civic society, administrative reform, political reform/stability, and civic rights and the maturity of civic society on the economic crisis and the process of recovery, and these stud­ ies have found that there is a positive impact (Beck and Laeven 2006; Eicher and Schreiber 2010; Heckelman 2010). If a bold conclusion is to be drawn from the above, it would be that previous studies that have empirically examined the relationship between transformation policy and economic recovery have produced broadly similar findings, namely that measures that contribute to the transformation to a market economy either assist with economic growth or, at the very least, do not impede it, even though more than a few studies have produced contrary findings. Column b of Table 3.1 employs EBRD transition indicators to verify the relationship between the degree of progress in transformation policy and growth paths. Actually, similar to the case of structural change discussed earlier, the findings obtained support the aforementioned view that transformation policy served to suppress the crisis. However, previous studies have suggested the possibility that the relationship between transformation policy and economic growth is not a linear one. What needs to be pointed out here is that the claims of Fischer et al. (1996a, 1996b) and de Melo et al. (1997)—namely that liberalization and stabilization are two sides of the same coin but that if society does not stabilize, it will be difficult to achieve liberal­ ization and that therefore, stabilization should be given priority—are worthy of atten­ tion from this point of view. The view that “better policy,” as advocated by Selowsky and Martin (1997), may intensify an economic crisis at least in the initial phase, leaving aside a long-term effect, also cannot be ignored. An important issue in the transitional economic growth debate that is inseparable from and relates to the view of de Melo et al. (1997), who suggest that a non-linear relationship exists between policy and growth, is the conflicting views of radicalists and gradualists concerning the nature of the speed and policy sequence of economic transition (see Chapter 2 in this book). From the 1990s to the early 2000s, the groups clashed, with opposing claims concerning the relationship between the speed of transformation policy (and economic reform in particular) and economic growth. In fact, while a series of studies, such as that by Roland and Verdier (1999), argue that a gradualist policy implementation effectively softened the drop in output that occurred in the initial phase of the transition, numerous others, such as that by Wyplosz (2000), claim that a policy of rapid liberalization led to a rapid escape from the transformational recession and a subsequent swift economic recovery. With the aim of bringing this debate to a conclusion, a group of researchers in the field of 79

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

the economics of transition, which included Heybey and Murrell (1999), attempted to perform a unique empirical analysis by employing speed of reform as a variable for transformation policy in addition to the level of reform achieved, which had been used as such a variable in the past. Whether the differences between these two variables would lead to differences in empirical findings is an extremely interesting issue. 3.3.3 The legacy of socialism In the context of the transitional economic growth debate, a vigorous discussion also occurred concerning the impact that factors such as the number of years spent under socialism and the thoroughness of the planned economic structure had on the growth path during the transition phase. In this “socialist legacy” debate, the view existed that historical initial conditions had what could be described as a decisive impact on current and future economic activity. This view was jointly espoused by researchers in the fields of both institutional economics and evolutionary economics in the form of the concept of “path-dependency.” The important point is that the longer the period spent under socialism and the deeper its impact on industrial activity or citi­ zens’ livelihoods, the more difficult it was to transform the system to a capitalist market economy, and this also served to suppress economic growth during the transi­ tion period. From this viewpoint, Rosati (1994) expressed the opinion that the vestiges of the former system exacerbated the economic crisis during the initial phase of transition, while Stuart and Panayotopoulos (1999) expressed the view that early-stage macroeco­ nomic imbalances were directly connected to the depth and length of the transform­ ational recession. Furthermore, Polanec (2004) stated that the negative legacy of the socialist era, which took the form of distortions in the market structure, had an extremely adverse impact on productivity during the period of transformational reces­ sion. This shows that more than a few researchers regard the crisis that occurred during the initial phase of transition as being related to the economic situation at the tail end of socialism. Historical initial conditions are also a key determinant not only of the economic crisis at the beginning of transition but also of the subsequent process of systemic transformation. For this reason, Denizer (1997) argued that the FSU countries, espe­ cially those in Central Asia, had more issues to resolve than did the CEE countries. Kolodko (2001) also pointed out that the longer the socialist legacy took to be elim­ inated, the longer it took for economic growth to return. Writing from a similar per­ spective, Selowsky and Martin (1997) stated that the problems faced by the FSU countries, such as distortions in industrial location, an industrial structure geared excessively toward meeting military demand, problems with private property rights, and lack of the rule of law, were more widespread and serious than those faced by the CEE countries, and thus it took the former countries longer to reallocate resources. They therefore offered an approach that added depth to the arguments put forward by such researchers as Denizer (1997) and Kolodko (2001). The view that 80

THE DEBATE ON ECONOMIC CRISIS AND GROWTH

3.3

the historical initial situation functioned to significantly restrict macroeconomic per­ formance not only during the economic crisis but also during the recovery phase has been inherited by such researchers as Redek and Sušjan (2005) and Hodgson (2006). Quite a few studies have used the number of years under socialism as an indicator of the weight of the socialist legacy, and one of the first to do so was Wolf (1999), which identified a significant negative correlation between the planned-economy period and the real economic growth rate. Since then, however, numerous studies have produced empirical results indicating that the correlation between the number of years under socialism and the economic growth rate has been weakening recently. A typical example of such a study is that of Falcetti et al. (2002). The fact that his­ torical initial conditions recede over time has been empirically proven repeatedly in previous studies such as those by Iwasaki (2004), Cerović and Nojković (2009), and Mitrović and Ivančev (2010), who provide empirical background for the view, shared by most researchers of transitional economics, that while historical initial conditions are important, they are not necessarily insurmountable. Previous studies include several that have employed dummy variables for the former Soviet zone, the members of the Commonwealth of Independent States (CIS), and so on, to measure the negative impact of the socialist legacy on economic growth in FSU countries as compared with CEE countries. The majority of studies, however, have employed, in addition to the aforementioned number of years under socialism, comprehensive indicators of initial conditions such as those developed by the EBRD (1999) and de Melo et al. (2001) or the degree of industrialization, output levels, and so on, at the end of socialism to more rigorously verify the growth-suppressing effect of the socialist legacy. As stated above, in the FSU countries, the negative legacy of the socialist planned economy probably hindered economic growth more seriously than in the CEE countries. In Column c of Table 3.1, we compare the CEE countries with the FSU countries, which include the Baltic states, in order to verify the relationship between the legacy of socialism and the growth path. In doing so, we find that in the FSU countries, the rate of decline in output during the economic crisis was, on average, a statistically significant 9.4% greater than in the CEE countries. This hints at the economic-growth-suppressing effect of the socialist legacy. 3.3.4 Inflation It is a well-known fact that the CEE and FSU countries experienced high inflation throughout the transition period. In fact, according to the EBRD, in 1992 Russia was hit by a more than 1,500% increase in consumer prices. Furthermore, Ukraine and Armenia recorded inflation rates of almost 5,000% in 1993 and 1994, respectively. Annual price rises of over 500% were seen in most CEE countries (Iwasaki and Uegaki 2019). Kornai (1994) emphasized the particular importance of curbing infla­ tion and pointed out that if the same policies implemented under socialism had been maintained, the budget constraints would not have hardened. This would have wors­ ened inflation, which in turn would have suppressed investment and impeded 81

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

economic growth. Wyplosz (2000) and Radulescu and Barlow (2002) also stated that a high rate of inflation and a favorable macroeconomic structure are incompatible. Moreover, de Melo et al. (1997) presented the interesting empirical finding that allow­ ing prices to be determined freely serves to reduce prices more than maintaining price controls does. They also develop the argument that liberalization and inflation control can be pursued simultaneously and that this can also promote economic growth. Several studies have aimed at verifying the relationship between high inflation and economic downturn during the transition period, and almost all of them have employed the consumer price index compared with the previous year or natural loga­ rithms of that as an independent variable. One of these studies, by Brenton et al. (1997), reports that countries that succeeded in curtailing inflation were able to keep the decline in output during the crisis at a lower level. Meanwhile, Fischer et al. (1996a, 1996b) stated that countries that moved quickly to deal with inflation were seen to experience a swift recovery in production. In addition, Loungani and Sheets (1997) provided analytical results showing that countries that quickly succeeded in enhancing central bank independence also succeeded in keeping their inflation rates at low levels, and this served to improve their macroeconomic conditions. Further­ more, Gillman and Harris (2010) verified that the inflation rate exerts an extremely powerful, stable, and negative impact on economic growth but also that if the infla­ tion rate drops, its marginal effect is reduced. The results of analysis presented in Column d of Table 3.1 show that for the group of countries with a higher inflation rate during the first five years of transi­ tion, the length of the economic crisis was statistically significantly longer than that in the lower group and that the decline in output during the crisis period was more severe for the former. This finding backs up the aforementioned arguments presented in previous research concerning the economic-growth-suppressing effect of the hyperinflation that occurred in the immediate aftermath of the collapse of the socialist regime. 3.3.5 Regional conflict At the end of socialism and during the period after its collapse, regional conflicts broke out in various CEE and FSU countries. The one that is most deeply ingrained in our memories is probably the series of military conflicts that occurred in the former Yugoslavia. In this region, one war after another occurred during the 1990s: in the Croatian War, Croatia, which was striving for independence, clashed with the Federal Republic of Yugoslavia, which was trying to block these moves; the Bosnian War followed serious ethnic clashes concerning the independence of Bosnia–Herzegovina; and the Kosovo War broke out in the form of violent resistance by Albanians against Serbia. Conflicts also occurred frequently in the FSU region. First, armed clashes broke out between Armenia and Azerbaijan over the Nagorno-Karabakh region in 1988, during the Gorbachev administration in the Soviet Union. The war finally ended in 1994, but both countries had been left in ruins. In Tajikistan, a civil war between different ethnic groups lasted from 1992 to 1997, producing more than a million 82

THE DEBATE ON ECONOMIC CRISIS AND GROWTH

3.3

refugees. In 1992, a civil war also broke out in Moldova, after which some regions became independent of the central government. In Georgia, meanwhile, a civil war in 1991 resulted in the emergence of a number of semi-autonomous regions. The country also experienced armed clashes with Russia in 2008. Furthermore, the con­ flict in Ukraine, which broke out in 2014, has attracted international attention and continues to the date of publishing this volume. The economic impact that these regional conflicts wrought on the countries affected does not need to be emphasized. The impact was enormous. In the previous section, we reported that seven countries, Ukraine, Georgia, Tajikistan, and Moldova in the FSU and Serbia, Montenegro, and Bosnia–Herzegovina in the former Yugoslavia, formed a cluster of countries that had experienced the largest declines in production during the economic crisis and whose economies had also been slow to recover after the crisis. It is by no means coincidental that this group contains many of the countries that experienced the conflicts described above. With the aim of measuring the impact of regional conflict on economic growth, researchers in the field of the economics of transition have expressed the occurrence of a regional conflict as a dummy variable and, for the most part, obtained statistically significant estimation results.4 Most previous studies have treated this regional-conflict dummy variable as a control variable. However, a few studies have positioned regional conflict itself as one of the focal points of their empirical analysis. Moers (1999), for instance, pointed out that regional conflict had a more powerful impact on economic growth than systemic reform during the first half of the 1990s in 21 CEE and FSU countries. Furthermore, Hodgson (2006), who suspects that differences in ethnic com­ position were closely related to macroeconomic performance following the collapse of socialism, has produced empirical results that suggest a strong connection between ethnic conflict during the 1990s and economic downturns. Our univariate comparative analysis supports the claims made by Moers (1999) and Hodgson (2006). In fact, Column e of Table 3.1 clearly illustrates the destructive effect of conflict, as it shows that the rate of decline in output during the crisis period in the countries that experienced regional conflicts during the 1990s was 24.6%, which is significant for a one-tailed test, higher than in the other transitional countries. 3.3.6 The need for a meta-analysis Regarding the five factors discussed above, it is reasonable to say that the opinions of researchers concerning the direction of their impact on economic growth in the CEE and FSU countries are mostly aligned. In other words, the structural changes in the national economy and the implementation of a transformation policy had a positive, or at least a non-negative, impact on economic growth in these countries. In contrast, the legacy of socialism, inflation, and regional conflict had a negative effect on output. As stated above, the various results of the univariate comparative analysis performed for the attributes of each group of countries reported in Table 3.1 also support such a view in terms of the length of the economic crisis and the degree of decline in output during the crisis. 83

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

However, the previous literature does not provide a clear explanation of why the transition economies in the CEE and FSU regions followed a J-curved growth path without exception. None of the studies has explicitly discussed and empirically examined the relative differences in the degree of impact of each factor on economic growth. Only by quantitatively comparing and considering the impact of these five factors in terms of effect size and statistical significance we can answer the question of why the crisis and process of recovery in transition economies followed a J-curved path rather than a U-curved or V-curved one. Employing a meta-analysis of the empirical results of previous research to com­ pare effect size and statistical significance is a highly effective means of achieving this objective. Furthermore, meta-analysis allows us to tackle such questions as whether a widely accepted view is sufficiently valid when the existence of counterevidence is explicitly taken into account and whether the true effect size can be spe­ cified in research on transition economies as a whole. This is why, in this chapter, we attempt to perform a unique and large-scale meta-analysis based on the empirical evidence of the previous literature.

3.4 PROCEDURE OF LITERATURE SELECTION AND OVERVIEW OF SELECTED STUDIES FOR META-ANALYSIS Taking into account the issues and research objectives described above, in this sec­ tion, we will describe the procedure of literature selection and give an overview of the studies selected for meta-analysis. As the first step toward identifying relevant research that involved empirical analysis of the determinants of output decline and growth during the transition period in the CEE and FSU countries, we used EconLit and Web of Science to search for literature pub­ lished between 1989 and 2016. When using these electronic databases, we employed as search terms combinations of one of “growth,” “decline,” “output,” “performance,” “gross domestic product,” and “GDP” and one of “transition economies,” “Central Europe,” “Eastern Europe,” “former Soviet Union,” and the name of a CEE or FSU country. This generated close to 3,500 hits. Then, judging from each title, abstract, and other related information, we narrowed the list and obtained more than 250 studies. In the second step, we closely examined the contents of these research works one by one and limited our literature list to those containing estimates that could be sub­ jected to meta-analysis in this chapter. As a result, we selected a total of 123 studies.5 These selected studies had been published continuously during the period between 1996 and 2016, but the years 2004 and 2009 saw the most publications, with ten in each of those years. The next most productive years were 2005 and 2006, with nine papers published in each, followed by 2001 and 2003, with eight in each. By decade, 20 papers were published in the 1990s (16.3%), 73 in the 2000s (59.3%), and 30 (24.4%) in the 2010s.

84

META-ANALYSIS

3.5

All of these 123 previous works are multinational studies, covering seven or more countries (mean: 22.3; median: 25); 120 of the studies deal with new EU member states as target countries. Studies covering non-EU CEE countries numbered 104, and those covering FSU countries, excluding the Baltic states, 109. Moreover, 18 stud­ ies included research on former socialist countries outside the CEE and FSU or other emer­ ging economies. These 123 studies cover the 33 years from 1979 to 2011 as a whole. The average period covered by each study is 9.9 years (median: 9). Eighty-one studies employ GDP as the base index of economic growth variable (i.e., dependent variable) in their empirical analysis. The number of studies using per-capita GDP and per-worker GDP is 40 and 50, respectively. Meanwhile, the number of studies using structural change, transformation policy, the socialist legacy, inflation, and regional con­ flict as growth-determining variables (i.e., independent variables) are 34, 96, 38, 68, and 36, respectively.6 From these 123 studies, we extracted a total of 3,279 estimates (mean: 26.7 per study; median: 16). Figure 3.3 gives a breakdown of the collected estimates by each growth-determining-variable type. As this figure shows, 1,702 extracted estimates (52%) are empirical results of the growth-enhancing 3.3 effect of transformation Figure Breakdown of collected estimates by growth-determining vari­ policy, reflecting the high able type level of interest in this aspect Note: Values following the category name denote the number of estimates and among researchers. Evidence the share in total collected estimates, respectively. regarding the effect of infla­ tion on growth takes the second largest share with 696 estimates. Those regarding the impacts of structural change, the socialist legacy, and regional conflict accounted for 8% to 10% of all the collected estimates.

3.5 META-ANALYSIS In this section, we will perform the meta-analysis employing the aforementioned 3,279 collected estimates according to the methodology described in Chapter 1. 85

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

First, we will examine the distribution of the estimates and, using a traditional metasynthesis method, perform a comparison of the effect size and statistical significance of the five growth-determining variables in question. Next, we will perform a metaregression analysis to verify whether the results of the meta-synthesis are supported even after simultaneously controlling for various research conditions. Furthermore, we will attempt a meta-analysis focusing solely on the structural change and trans­ formation policy, which are of great interest in the field of the economics of transition. 3.5.1 Meta-synthesis In Table 3.2, the descriptive statistics of the PCC and the t value and the results of the Shapiro–Wilk normality test are reported for each type of growth-determining variable. Figure 3.4 presents each kernel density estimation. From these materials, we can confirm that while the distribution of the collected estimates is not distrib­ uted normally for every variable type, more of the estimates relating to the growth effects of structural change and transformation policy are distributed on the posi­ tive side, and, in contrast, those of the socialist legacy, inflation, and regional con­ flict are clearly biased toward the negative side. In other words, most previous studies produced findings implying that while the former two factors demonstrated a growth-enhancing effect, the latter three factors served as triggers for negative growth. Table 3.3 reports the results of meta-synthesis. In this table, the PCCs are synthe­ sized using a meta fixed-effect model and a meta random-effects model. The t values are combined with and without a 10-point scale of the research quality level as a weight. We also report the median t value and Rosenthal’s fail-safe N (fsN). The latter serves as a supplemental statistic for evaluating the reliability of the com­ bined t value. As reported in Column a of Table 3.3, the test of homogeneity rejected the null hypothesis at a 1% significance level for all five variable types. Therefore, we adopt the coefficient of the random-effects model as a reference value of the synthesized effect size. The results indicate that the synthesized effect sizes for structural change and transformation policy both take positive values and are statistically significant. Meanwhile, those for the socialist legacy, inflation, and regional conflict are significant and negative. Furthermore, regarding the combined t values shown in Column b of the same table, even when differences in the quality level among the studies are taken into account, the overall statistical significance of the collected estimates is of an adequate level for all variable types. The sufficiently large fail-safe N also supports the results of a combination of t values. According to Doucouliagos (2011), concerning assessment of the PCC in eco­ nomic research,7 the impact on economic growth of structural change, transform­ ation policy, and the socialist legacy is regarded as small. However, the effect of regional conflict and inflation on output can be assessed as medium.8 In other words, regional conflict and inflation hindered economic growth with effect sizes 86

META-ANALYSIS

3.5

Table 3.2 Descriptive statistics of the partial correlation coefficients and the t values of collected estimates and Shapiro–Wilk normality test by growth-determining variable type (a) PCC

C H A P T E R 3

Number of collected estimates (K) Structural change

Transformation policy

Socialist legacy Inflation Regional conflict

Mean

Median S.D.

Max.

Min.

Kurtosis

Skewness

Shapiro–Wilk normality test (W)

280

0.087

0.099

0.213

0.681

−0.873

5.582

−0.859

0.944***

1702

0.104

0.113

0.281

0.891

−0.878

2.935

−0.058

0.998**

285 696 316

−0.095 −0.291 −0.209

−0.123 −0.295 −0.254

0.291 0.258 0.344

0.827 0.695 0.914

−0.853 −0.911 −0.878

3.801 3.060 2.637

0.580 −0.120 0.580

0.973***

0.988***

0.947***

(b) t value Number of collected estimates (K) Mean Structural change

Transformation policy

Socialist legacy Inflation Regional conflict

Median S.D.

Max.

Min.

Shapiro–Wilk Kurtosis Skewness normality test (W)

280

1.041

1.190

1.967

6.420

−8.597

5.601

−0.703

0.957***

1702

0.996

1.090

3.063 16.730

−8.000

5.024

0.574

0.966***

285 696 316

−1.029 −3.744 −2.378

−1.550 −3.000 −2.500

2.896 6.620 −7.300 3.654 4.635 −16.400 3.012 10.800 −15.570

3.013 4.043 4.308

0.508 −0.996 −0.052

0.974***

0.935***

0.968***

Note *** Null hypothesis of normal distribution is rejected at the 1% level ** at the 5% level.

that surpass those of structural change and transformation policy by a large margin, and the legacy of socialism also contributed to output decline with an effect size similar to those of structural change and transformation policy. This finding provides a clear answer to the question of why, during the initial years of transition, the CEE and FSU countries experienced a destructive drop in output. Fur­ thermore, the results, which indicate that the growth-enhancing effects of structural change and transformation policy are not as strong as they were assumed to be in early transition period, could provide evidence that the recovery process following the economic crisis was not V-shaped. For this reason, a comparative analysis involving a meta-synthesis with five types of growth-determining variables provides an unequivocal quantitative explanation of emergence of the J-curved growth path in tran­ sition economies.

87

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

3

Figure 3.4 Kernel density estimation of partial correlation coefficients and t values by growth-determining variable type Note: Vertical axis is Kernel density. Horizontal axis is variable value.

3.5.2 Meta-regression analysis It is difficult to say that the traditional meta-synthesis method can effectively con­ trol for possible heterogeneity between studies. Therefore, in this subsection, we perform a meta-regression analysis (MRA) to examine whether the results of metasynthesis described in the previous subsection can be reproduced after controlling for various research conditions that may have affected the empirical results in the previous literature. We introduce the PCC or the t value into the left-hand side of the meta-regression equation, while on its right-hand side, we adopt a series of meta-independent vari­ ables designed to capture not only the differences in growth-determining variables, target countries, the estimation period, and the base index of economic growth vari­ able that we mentioned in Section 3.4 but also the differences in data type, estimator, benchmark index of economic growth variable, degree of freedom, and quality of the study. The names, definitions, and descriptive statistics of these meta-independent vari­ ables are listed in Table 3.4. Table 3.5 provides the estimation results using all 3,279 collected estimates. Panel a of this table shows the estimation results of the meta-regression model with the PCC on the left-hand side, while Panel b gives those taking the t value as the dependent variable. Hereinafter, we will interpret the regression results under the assumption that the meta-independent variables that are statistically significant and have the same sign in at least five of eight models constitute statistically robust estimation results.

88

280 1702 285 696 316

Growth-determining variable type

Structural change Transformation policy Socialist legacy Inflation Regional conflict

0.097*** 0.077*** −0.080*** −0.315*** −0.267***

Fixed-effect model a

b

a

Notes

Null hypothesis: The synthesized effect size is zero.

Null hypothesis: Effect sizes are homogeneous.

*** denotes statistical significance at the 1% level.

Number of estimates (K) 0.090*** 0.096*** −0.091*** −0.295*** −0.232***

Random-effects model a 996.107*** 16000.000*** 2391.953*** 7455.361*** 2067.309***

Test of homogeneity b

(a) Synthesis of PCCs

Table 3.3 Synthesis of estimates by growth-determining variable type

17.417*** 41.071*** −17.367*** −98.773*** −42.267***

Unweighted combination

3.543*** 6.769*** −2.903*** −20.229*** −6.527***

Weighted combination

1.190 1.090 −1.550 −3.000 −2.500

Median of t values

(b) Combination of t values

31109 1059248 31482 2508612 208308

Failsafe N (fsN)

META-ANALYSIS 3.5

89

C H A P T E R 3

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

3

Variable name

Table 3.4 Name, definition, and descriptive statistics of meta-independent variables Descriptive statistics Definition

Transformation policy 1 = if growth-determining variable used for estimation belongs to the category of transformation policy, 0 = otherwise Socialist legacy 1 = if growth-determining variable used for estimation belongs to the category of socialist legacy, 0 = otherwise Inflation 1 = if growth-determining variable used for estimation belongs to the category of inflation, 0 = otherwise Regional conflict 1 = if growth-determining variable used for estimation belongs to the category of regional conflict, 0 = otherwise Trade openness a 1 = if trade openness is used as a proxy for structural change, 0 = otherwise Bank credit to private 1 = if bank credit to the private sector is used as a proxy for structural change, 0 = otherwise sector a Market capitalization a 1 = if market capitalization is used as a proxy for struc­ tural change, 0 = otherwise Development of 1 = if development of financial sector represents trans­ formation policy, 0 = otherwise financial sector a Comprehensive eco­ 1 = if comprehensive economic reform represents trans­ nomic reform b formation policy, 0 = otherwise 1 = if liberalization represents transformation policy, 0 = Liberalization b otherwise Price and competi­ 1 = if price and competition reform represents transform­ tion reform b ation policy, 0 = otherwise 1 = if enterprise reform represents transformation policy, Enterprise reform b 0 = otherwise Privatization b 1 = if privatization represents transformation policy, 0 = otherwise 1 = if financial reform represents transformation policy, 0 Financial reform b = otherwise Trade reform b 1 = if trade reform represents transformation policy, 0 = otherwise Institutional quality b 1 = if institutional quality represents transformation policy, 0 = otherwise Property rights 1 = if property rights reform represents transformation policy, 0 = otherwise reform b Government 1 = if government reform represents transformation reform b policy, 0 = otherwise Political reform/ 1 = if political reform/stability represents transformation policy, 0 = otherwise stability b Democratization b 1 = if democratization represents transformation policy, 0 = otherwise Rule of law/legal 1 = if rule of law/legal reform represents transformation policy, 0 = otherwise reform b Civil rights/society b 1 = if civil rights/society represents transformation policy, 0 = otherwise

Mean

Median

S.D.

0.519

1

0.500

0.087

0

0.282

0.212

0

0.409

0.096

0

0.295

0.354

0

0.479

0.143

0

0.351

0.266

0

0.442

0.152

0

0.359

0.266

0

0.442

0.152

0

0.359

0.067

0

0.250

0.028

0

0.166

0.068

0

0.251

0.036

0

0.186

0.039

0

0.193

0.014

0

0.118

0.009

0

0.097

0.018

0

0.132

0.018

0

0.134

0.088

0

0.283

0.055

0

0.227

0.023

0

0.150

(Continued ) 90

META-ANALYSIS

3.5

Table 3.4 contd. Descriptive statistics Variable name

Definition

Other transformation policies b

1 = if a policy other than comprehensive structural reform and the above policies/reforms represents trans­ formation policy, 0 = otherwise 1 = if reform speed is adopted as the benchmark index of the transformation policy variable, 0 = otherwise Proportion of non-EU CEE countries in target countries c

Reform speed Proportion of other CEE countries Proportion of FSU countries Proportion of nonCEE and FSU countries First year of estimation Length of estimation Cross-section data GLS FE RE SUR GMM Other estimators IV/2SLS/3SLS GDP per capita GDP per worker Growth level Lagged variable With an interaction term(s) √Degree of freedom Quality level

Proportion of FSU countries in target countries, exclud­ ing the Baltic countries Proportion of non-CEE and FSU countries in target countries First year of estimation period Years of estimation period 1 = if cross-section data is employed for empirical ana­ lysis, 0 = otherwise 1 = if generalized least squares estimator is used for esti­ mation, 0 = otherwise 1 = if fixed-effect panel estimator is used for estimation, 0 = otherwise 1 = if random-effects panel estimator is used for estima­ tion, 0 = otherwise 1 = if seemingly unrelated regression estimator is used for estimation, 0 = otherwise 1 = if generalized method of moments estimator is used for estimation, 0 = otherwise 1 = if an estimator other than OLS and the above estim­ ators is used for estimation, 0 = otherwise 1 = if instrumental variable method or 2SLS or 3SLS is used for estimation, 0 = otherwise 1 = if GDP per capita is used as the base index of eco­ nomic growth variable, 0 = otherwise 1 = if GDP per worker is used as the base index of eco­ nomic growth variable, 0 = otherwise 1 = if growth level is used as the benchmark index of economic growth variable, 0 = otherwise 1 = if a lagged growth-determining variable is used for estimation, 0 = otherwise 1 = if estimation is carried out with an interaction term(s) of growth-determining variable, 0 = otherwise Root of degree of freedom of the estimated model Ten-point scale of the quality level of the study d

Mean

Median

S.D.

0.022

0

0.148

0.159

0

0.366

0.119

0.12

0.081

0.435

0.48

0.194

0.011

0

0.029

1991.534

1990

3.210

10.756 0.177

10 0

4.391 0.382

0.049

0

0.215

0.402

0

0.490

0.026

0

0.158

0.000

0

0.017

0.110

0

0.313

0.002

0

0.039

0.131

0

0.338

0.306

0

0.461

0.018

0

0.132

0.031

0

0.174

0.169

0

0.374

0.036

0

0.187

10.935 4.992

11.662 4.617 5 2.918

Notes

a Descriptive statistics are computed using the estimates of structural change variable only.

b Descriptive statistics are computed using the estimates of transformation policy variable only.

c Including Albania, Bosnia and Herzegovina, Croatia, Kosovo, FYR Macedonia, Montenegro, and Serbia.

d See Chapter 1 for more details.

91

C H A P T E R 3

92

[1]

Meta-independent variable (default)/model [2]

Cluster-robust WLS [Quality level]

0.0166*** 0.0023

0.0420

Estimation period First year of estimation Length of estimation

Data type (panel data) Cross-section data 0.0628

0.0178*** 0.0032

Composition of target countries (CEE EU countries) Proportion of other CEE −0.0647 0.1363 countries Proportion of FSU −0.0763 −0.0268 countries Proportion of non-CEE −0.4829 −0.5147 and FSU countries

Growth-determining variable type (structural change) Transformation policy 0.0105 0.0016 −0.1443*** Socialist legacy −0.1433*** Inflation −0.3664*** −0.3522*** *** Regional conflict −0.3004 −0.3087***

Clusterrobust OLS

0.0630

0.0690

0.0093*** 0.0027

−0.0264

−0.0317

0.0098*** 0.0034

−0.0996*

−0.1065**

−0.0968*

0.0533

0.0139*** 0.0037

−0.2710

−0.0114

0.0047 −0.1255*** −0.3747*** −0.3097***

[5]

0.0513

−0.0194 −0.1199*** −0.3345*** −0.3024***

[4]

0.0467

−0.0203 −0.1198*** −0.3328*** −0.3018***

[3]

ClusterClusterClusterrobust WLS robust WLS robust WLS [N] [df] [1/SE]

−0.0062

0.0349*** 0.0147***

−0.4661

−0.0552

−0.1753***

0.0335 −0.1117** −0.3457*** −0.2682***

[6]

−0.0092

0.0357*** 0.0153***

−0.4829

−0.0546

−0.1744***

0.0333 −0.1113** −0.3457*** −0.2681***

[7]a

Multilevel Cluster-robust mixed-effects random-effects RML panel GLS

−0.1352

0.0508*** 0.0297***

−0.7500***

−0.0582

−0.1600***

0.0239 −0.1111** −0.3510*** −0.2744***

[8]b

Cluster-robust fixed-effects panel LSDV

3

Estimator (analytical weight in parentheses)

(a) Dependent variable — PCC

Table 3.5 Meta-regression analysis

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

−0.1053*** −0.0285 −0.0343 −0.2706*** −0.0342 −0.1132*** 0.0717***

K R2

3279 0.328

(9.629)

Degree of freedom and research quality √Degree of freedom −0.0008 Quality level 0.0086* *** −33.0555 Intercept 3279 0.315

(9.930)

−0.0003 − −35.4172***

3279 0.346

(5.630)

−0.0009 0.0035 −19.4525***

0.0844** −0.0893**

Other characteristics of growth-determining variable 0.1020* Lagged variable 0.1139** With an interaction term(s) −0.1048* −0.0906

0.0321 −0.0164

−0.0887*** −0.0265 −0.0430* −0.3177*** −0.0117 −0.0937*** 0.0734***

−0.0082

0.0301 −0.0564

−0.1287*** −0.0441 −0.1010** −0.3226*** −0.0463 −0.1113*** 0.0781***

Benchmark index of economic growth variable (growth rate) Growth level −0.0251 0.0182

Base index of economic growth variable (GDP) GDP per capita 0.0343 GDP per worker −0.0623

Estimator (OLS) GLS FE RE SUR GMM Other estimators IV/2SLS/3SLS

3279 0.355

(5.498)

− 0.0039 −18.3827***

0.0843** −0.0909**

0.0151

0.0356 −0.0194

−0.0872*** −0.0249 −0.0266 −0.3135*** −0.0113 −0.0937*** 0.0742***

3279 0.362

(7.333)

−0.0009 0.0062 −27.6109***

0.0967** −0.1005*

−0.0032

0.0345 −0.0594

−0.1012*** −0.0286 −0.0253 −0.2827*** −0.0225 −0.0954*** 0.0768***

3279 −

(15.962)

0.0069 0.0075 −69.6136***

0.1613*** −0.0242

−0.0932

−0.0188 −0.0903

−0.0666*** −0.0091 −0.0342** −0.0305 −0.0127 −0.0357*** 0.0232

3279 0.252

(15.913)

0.0070 0.0077 −71.2976***

0.1617*** −0.0237

−0.0953

−0.0201 −0.0945

−0.0650*** −0.0082 −0.0332** −0.0213 −0.0118 −0.0350*** 0.0220

(Continued )

3279 0.137

(11.682)

0.0077 dropped −101.5778***

0.1646*** −0.0214

−0.1848***

0.1407*** −0.0124*

−0.0324* 0.0142 −0.0102 0.0992*** 0.0114 −0.0390*** 0.0041

META-ANALYSIS 3.5

93

C H A P T E R 3

94

[9]

[10]

Cluster-robust WLS [Quality level]

−1.2178*** −0.1931 −0.1740 −0.2935 0.0049

0.1861

Data type (panel data) Cross-section data

Estimator (OLS) GLS FE RE SUR GMM

0.1266*** 0.0542

Estimation period First year of estimation Length of estimation

−1.4180*** −0.2113 −0.7506** −0.7266 0.0160

0.4201

0.1189*** 0.0426

Composition of target countries (CEE EU countries) Proportion of other CEE −0.4909 1.2194 countries −1.1009* Proportion of FSU −1.5191** countries Proportion of non-CEE −0.0396 −1.4241 and FSU countries

Growth-determining variable type (structural change) Transformation policy 0.1679 0.0658 −1.3238*** Socialist legacy −1.2831*** −4.3779*** Inflation −4.4848*** *** Regional conflict −3.0447 −3.0993***

Meta-independent vari­ able (default)/model

Clusterrobust OLS

2.3465

2.6031

−1.1382*** −0.2985 −0.3919 0.2895 0.0218

0.1531

−1.1570*** −0.3166 −0.3748 0.4913 −0.0098

0.4297

0.1036*** 0.0418

−1.8398***

−1.5311**

0.1013** 0.0630

−0.0873

−0.2334 −1.5826*** −4.8957*** −4.2383***

[12]

0.3116

−0.2190 −1.5540*** −4.8242*** −4.1402***

[11]

−1.2340*** −0.2621 −0.1634 0.1519 0.0067

0.2161

0.1262*** 0.0730*

0.8076

−1.8189***

−0.2465

0.1502 −1.3688*** −5.1052*** −3.6988***

[13]

ClusterClusterClusterrobust WLS robust WLS robust WLS [N] [df] [1/SE]

−0.7776** −0.0083 −0.3821** 0.2136 0.0925

0.2049

0.1666*** 0.0794**

1.3817

−0.9968**

−1.2600***

0.5978 −0.9670** −4.2021*** −2.6281***

[14]

−0.7315** −0.0026 −0.3824** 0.2661 0.0949

0.1816

0.1735*** 0.0832**

1.2651

−0.9479**

−1.2521***

0.6049 −0.9516** −4.1976*** −2.6114***

[15] c

Multilevel Cluster-robust mixed-effects random-effects RML panel GLS

−0.3236 0.0906 −0.3318** 0.8561*** 0.1651

−0.7550

0.2940*** 0.1728***

−1.7133

−0.5196

−0.9512***

0.6108 −0.8454 −4.1893*** −2.5394***

[16] d

Cluster-robust fixed-effects panel LSDV

3

Estimator (analytical weight in parentheses)

(b) Dependent variable — t value

Table 3.5 contd.

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

−1.4001*** 0.6220***

3279 0.345

3279 0.327

−0.0173 − −236.1619*** 3279 0.344

−0.0446 0.0410 −200.6579**

1.3645** −1.5517**

−0.0918

0.5643 0.2255

−1.1753*** 0.9217***

3279 0.348

3279 0.368

− −0.0375 0.0400 0.0483 −205.4785*** −250.5097***

1.2389** −1.4975**

0.1357

0.6449 0.3171

−1.2901*** 1.0953***

3279 −

0.0279 0.0326 −332.3560***

2.0560*** −0.7148

−0.2468

0.2224 0.1016

−1.0418*** 0.2481

3279 0.321

0.0300 0.0337 −346.0784***

2.0800*** −0.6734

−0.2680

0.1972 0.0443

−1.0140*** 0.2206

3279 0.258

0.0477 dropped −587.5625***

2.2327*** −0.4439

−0.5929

0.7854*** −1.1830***

−0.8646*** −0.0328

a

Notes Breusch–Pagan test: χ2=651.50, p=0.000 b Hausman test: χ2=102.40, p=0.000 c Breusch–Pagan test: χ2=615.09, p=0.000 d Hausman test: χ2=58.77, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 3.4 for definition and descriptive statistics of meta-independent variables.

K R2

Degree of freedom and research quality √Degree of freedom −0.0336 Quality level 0.0592 *** −251.4594 Intercept

1.2390** −1.4950**

Other characteristics of growth-determining variable 1.1483** Lagged variable 1.4378*** ** With an interaction term(s) −1.3960 −1.2543**

0.6085 0.3338

−1.2484*** 1.0521***

0.0574

0.4477 0.3641

−1.4118*** 0.5991***

Benchmark index of economic growth variable (growth rate) Growth level −0.0491 0.2152

Base index of economic growth variable (GDP) GDP per capita 0.4274 GDP per worker 0.1767

Other estimators IV/2SLS/3SLS

META-ANALYSIS 3.5

95

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

As Table 3.5 shows, if structural change is taken as the default category, then regardless of differences in the dependent variables, all of the meta-independent variables that specify the estimates of the socialist legacy, inflation, and regional conflict are robustly estimated to be negative. This is in sharp contrast to the insig­ nificant coefficient of transformation policy. Put another way, we can say that no statistically significant difference is seen for either the PCC or the t value in esti­ mation results verifying the growth effects of structural change and transformation policy. However, the PCCs and t values for structural change and transformation policy and the three remaining growth determinants exhibit a significant difference, and the values for the latter three factors are much lower than the former two. In fact, if we refer to the means of statistically significant regression coefficients, we find that compared with structural change and transformation policy, the PCCs of the socialist legacy, inflation, and regional conflict are lower, at 0.1234, 0.3504, and 0.2917, respectively, while their t values are also lower, at 1.2901, 4.5346, and 3.2500, respectively. The relationship expressed in the meta-independent variables of growth-determining variable types is highly consistent with the results of metasynthesis reported in Table 3.3. Therefore, we maintain that the discussion in the previous subsection regarding the emergence of a J-curved growth path is a universal policy implication for transition economies beyond the various differences in research conditions. 3.5.3 Meta-analysis of structural change and transformation policy Among the five growth determinants dealt with in this chapter, structural change and transformation policy are of the greatest concern to international organizations and researchers of transition economies. Therefore, we will bring this subsection to a close by performing a meta-analysis focused on these two factors. The literature subjected to our meta-analysis employed five types of indicators for measuring structural changes in a national economy. These are (1) share of private sector in GDP, (2) trade openness, (3) bank credit to private sector, (4) market capitalization, and (5) development of the financial sector. However, the indicators that these previous studies employed for the purpose of examining the relationship between transformation policy and economic growth are more diverse, reflecting the variation in the areas of expertise of and the issues of interest to the researchers. The variable used to express the domains of transformation policy comprises a total of 16 types of indicators. These include indicators relating to economic policy in each area, such as liberalization and price/competition policy, and so on, the reform of institutions/property rights, the reform of government/ politics, democratization, the rule of law and legal reform, and civil rights and society. Furthermore, with respect to the effect of transformation policy on output, against the background of the heated debate on radicalism versus gradualism as alternative transition strategies as discussed in Chapter 2, many researchers have paid a lot of 96

META-ANALYSIS

3.5

attention not only to the degree of success of the transformation policy but also to the speed of policy implementation.9 In fact, of 1702 estimates of transformation policy, 271, or 15.9%, adopted speed of reform (i.e., the interval change rate of a reform indicator) as the benchmark index. For this reason, when we were in the process of coding the estimates of transformation policy variables, we recorded not only the domains targeted for reform but also the categories of reform level and the speed of reform to serve as the benchmark index of the transformation policy variable. Table 3.6 presents the results of meta-synthesis of the estimates of structural change variables and transformation policy variables in accordance with the aforementioned sub-classifications. In addition, regarding the transformation policy variable, it also reports the meta-synthesis results for the classification of estimates according to differences in the benchmark index, namely the reform level and reform speed. As Column a of the table shows, the homogeneity test strongly rejected the null hypothesis; hence, we again employed the coefficient of the random-effects model as a reference value of the synthesized effect size. Table 3.6 indicates that all the variables used in previous research have not effect­ ively or fully captured the growth-enhancing effects of structural change and trans­ formation policy. Actually, in the case of structural change variables, the synthesized effect sizes of bank credit to private sector and market capitalization, while positive, are not statistically significant. Moreover, combined t values weighted by the quality level of the research have not reached a 10% significance level, not only for these two variable types, but also for the development of the financial sector. With respect to transformation policy variables, the synthesized effect sizes for five of the 16 vari­ able types show an insignificant value. Moreover, that for democratization is signifi­ cantly negative. Concerning combined t values that take into account the difference in research quality, 11 of the 16 variable types are insignificant. In addition, the syn­ thesized effect sizes and weighted combined t values of collected estimates, which are used as a measure of the reform level, are significantly positive in both cases; yet, for the reform speed, they are insignificant. The meta-regression results reported in Tables 3.7 and 3.8 provide evidence that supports the findings obtained from the meta-synthesis.10 In other words, in the case of structural change variables, if other research conditions are equal, the three vari­ able types relating to the financial sector are, on average, significantly lower than the share of private sector in GDP in terms of the PCC. Furthermore, the t values of bank credit to the private sector and market capitalization are significantly lower than those for the private-sector GDP share. With respect to transformation policy variables, both the PCCs and t values of the six policy domains, namely comprehen­ sive economic reform, liberalization, financial reform, trade reform, the rule of law/ legal reform, and civil rights/society, are significantly higher than those of compre­ hensive structural reform. In addition, the PCC of political reform/stability and the t values of enterprise reform, government reform, and other transformation policies are significantly higher than the corresponding value of comprehensive structural reform. Moreover, compared with estimates for transformation policy variables that 97

C H A P T E R 3

98

167 453 258 114 48 115 61 66 24 16 30 31 149 93 39 38 1431 271

71 99 40 56 14

Number of estimates (K)

−0.029*** 0.129*** 0.180*** −0.019** 0.030*** 0.025*** 0.131*** 0.200*** 0.111*** 0.017 0.093*** 0.051*** −0.054*** 0.215*** −0.003 0.117*** 0.103*** −0.033***

0.078*** 0.164*** 0.052*** 0.050*** 0.098***

Fixed-effect model a

Notes

a Null hypothesis: The synthesized effect size is zero.

b Null hypothesis: Effect sizes are homogeneous.

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Transformation policy variable Comprehensive structural reform Comprehensive economic reform Liberalization Price and competition reform Enterprise reform Privatization Financial reform Trade reform Institutional quality Property rights reform Government reform Political reform/stability Democratization Rule of law/legal reform Civil rights/society Other transformation policies Reform level Reform speed

Structural change variable Share of private sector in GDP Trade openness Bank credit to private sector Market capitalization Development of financial sector

Subcategory of structural change and transformation policy variable

−0.018 0.141*** 0.207*** −0.006 0.024 −0.003 0.129*** 0.167*** 0.222*** 0.091* 0.084*** 0.130*** −0.061*** 0.243*** 0.089*** 0.043 0.118*** −0.018

0.078*** 0.159*** 0.035 0.028 0.093*** 1353.359*** 5723.992*** 1826.879*** 795.697*** 229.783*** 1278.319*** 370.647*** 608.229*** 146.350*** 53.658*** 49.737*** 127.441*** 550.591*** 406.891*** 226.349*** 362.510*** 13000.000*** 2017.457***

209.465*** 182.413*** 168.848*** 315.364*** 29.343***

Randomeffects Test of homogeneity b model a

−4.226*** 37.321*** 30.379*** −1.363 2.225*** 1.046 11.885*** 16.609*** 6.961*** 2.668*** 4.015*** 5.170*** −7.219*** 16.721*** 3.670*** 5.722*** 47.107*** −5.320***

7.294*** 16.794*** 2.629*** 3.317*** 5.727***

Unweighted combination

−0.914 6.330*** 4.293*** −0.250 0.540 0.196 1.889** 3.030*** 0.830 0.407 0.912 0.832 −0.939 2.808*** 0.597 1.240 7.808*** −0.852

1.363* 3.713*** 0.551 0.700 0.920

Weighted combination

−1.100 2.050 2.110 −0.115 0.290 −1.500 1.376 2.245 1.700 0.875 0.220 1.310 −0.365 1.880 0.462 −0.082 1.325 −0.380

1.000 1.645 0.224 0.796 1.911

Median of t values

(b) Combination of t values

3

(a) Synthesis of PCCs

Table 3.6 Synthesis of estimates by subcategory of structural change and transformation policy variable

935 232712 87733 −36 40 −69 3123 6663 406 26 149 275 2720 9516 155 422 1172038 2564

1325 10220 62 172 156

Failsafe N (fsN)

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

[1]

Clusterrobust OLS

[2]

Cluster-robust WLS [Quality level]

[3]

Clusterrobust WLS [N]

K R2

280 0.210

280 0.273

280 0.222

Structural change variable type (share of private sector in GDP) Trade openness −0.0411 −0.0281 0.0573 Bank credit to pri−0.1444** −0.1366** −0.0523 vate sector −0.0583* −0.0354** Market capitalization −0.0698*** Development of −0.1113** −0.0917** −0.0292 financial sector

Meta-independent variable (default)/ model

Estimator (analytical weight in parentheses)

(a) Dependent variable — PCC

280 0.215

−0.0549*** −0.0626*

−0.0359** −0.0282 280 0.211

0.0166 −0.0889*

[5]

0.0643 −0.0595

[4]

ClusterClusterrobust WLS robust WLS [df] [1/SE]

Table 3.7 Meta-regression analysis using estimates of structural change variable

280 −

−0.0636*** −0.0704*

−0.0342 −0.1028*

[6]

Multilevel mixed-effects RML

280 0.172

−0.0633*** −0.0679*

−0.0335 −0.0994*

[7] a

Cluster-robust random-effects panel GLS

(Continued )

280 0.032

−0.0515** 0.0135

−0.1310*** 0.0355*

[8] b

Cluster-robust fixed-effects panel LSDV

META-ANALYSIS 3.5

99

C H A P T E R 3

100

[9]

[10]

Cluster-robust WLS [Quality level]

[11]

Clusterrobust WLS [N]

280 0.275

280 0.305

280 0.297

280 0.289

−0.5826*** −0.3906

−0.4930** −0.1242 280 0.246

0.5928 −0.7046

[13]

1.1628* −0.6239

[12]

ClusterClusterrobust WLS robust WLS [df] [1/SE]

280 −

−0.6359*** −0.5886*

0.1673 −0.8819*

[14]

Multilevel mixed-effects RML

280 0.265

−0.6331*** −0.5670*

0.1768 −0.8517*

[15] c

Cluster-robust random-effects panel GLS

280 0.060

−0.5095*** 0.2741

−0.9485 0.5601**

[16] d

Cluster-robust fixed-effects panel LSDV

Notes a Breusch–Pagan test: χ2=0.34, p=0.280 b Hausman test: χ2=32.45, p=0.019 c Breusch–Pagan test: χ2=1248.70, p=0.000 d Hausman test: χ2=37.21, p=0.005 Due to space constraints, the estimates of the meta-independent variables of other study conditions and intercept are not presented. Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 3.4 for definition and descriptive statistics of meta-independent variables.

K R2

Structural change variable type (share of private sector in GDP) Trade openness 0.1145 0.0856 1.0139* −1.1265** −0.4260 Bank credit to −1.0577** private sector −0.6196*** −0.4651** Market capitalization −0.6688*** * * Development of −0.7291 −0.6921 −0.1359 financial sector

Meta-independent variable (default)/ model

Clusterrobust OLS

3

Estimator (analytical weight in parentheses)

(b) Dependent variable — t value

Table 3.7 contd.

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

1702 0.226

1702 0.251

1702 0.239

−0.1494***

Benchmark index of transformation policy variable (reform level) −0.1283*** Reform speed −0.1475***

K R2

0.0628** 0.0673 0.1519*** 0.1921*** 0.1019 0.0029 0.1131*** 0.1197** −0.0579* 0.1878*** 0.1042** 0.1351**

0.0659** 0.0654 0.1520*** 0.1971*** 0.1218 0.0072 0.1245*** 0.1408*** −0.0528 0.1916*** 0.1010** 0.1303**

1702 0.247

−0.1535***

0.1425*** 0.0263

[4]

0.1526*** 0.0164

[3]

0.1814***

[2]

1702 0.238

−0.1584***

0.0726** 0.0699 0.1641*** 0.2106*** 0.1591 0.0560 0.1039** 0.1686*** −0.0711 0.2192*** 0.1220* 0.1347**

0.1623*** 0.0411

0.1952***

[5]

1702 −

1702 0.080

−0.0346

−0.0227 −0.0760 0.0524 0.0904 0.1316 0.0691 −0.0899 0.0405 −0.2663** 0.0963 0.0514 −0.0557

−0.0061 −0.0597 0.0693 0.1059 0.1360 0.0855 −0.0707 0.0583 −0.2488** 0.1140 0.0643 −0.0379 −0.0384

0.0328 −0.0808

0.0264

[7] a

0.0519 −0.0641

0.0451

[6]

Cluster-robust ClusterClusterClusterMultilevel Cluster-robust WLS [Quality robust WLS robust WLS robust WLS mixedrandom-effects level] [N] [df] [1/SE] effects RML panel GLS

0.1850***

[1]

Clusterrobust OLS

Transformation policy variable type (comprehensive structural reform) 0.2235*** Comprehensive economic 0.1890*** reform 0.2605*** Liberalization 0.1719*** Price and 0.0329 0.0613 competition reform Enterprise reform 0.0551 0.0437 Privatization 0.0471 0.0252 0.2842*** Financial reform 0.1639*** *** 0.2047*** Trade reform 0.1830 0.3424*** Institutional quality 0.2175** Property rights reform 0.1148 0.1435 Government reform 0.0766 0.0927 0.2655*** Political reform/stability 0.2077*** Democratization −0.0998 0.0032 Rule of law/legal reform 0.2170*** 0.2766*** 0.1892** Civil rights/society 0.1297* Other transformation policies 0.0969 0.2162***

Meta-independent variable (default)/model

Estimator (analytical weight in parentheses)

(a) Dependent variable — PCC

Table 3.8 Meta-regression analysis using estimates of transformation policy variable

(Continued )

1702 0.0003

−0.0303

−0.0802 −0.1321 −0.0043 0.0371 0.1243 0.0053 −0.1591 −0.0242 −0.3423*** 0.0246 −0.0045 −0.1116

−0.0514 −0.1375

−0.0413

[8] b

Cluster-robust fixed-effects panel LSDV

META-ANALYSIS 3.5

101

C H A P T E R 3

102

1702 0.193

1702 0.213

1702 0.238

1702 0.240

−2.4692***

2.6766*** 2.1523*** 0.4861 0.8661* 1.3460 2.2916*** 2.7696*** 0.9061 −0.4576 1.4475*** 1.3662 −1.2164** 2.3678*** 0.9477 2.4121**

[12]

1702 0.221

−2.3879***

2.8576*** 2.1962*** 0.8971 1.2327*** 1.6061* 2.4477*** 3.3316*** 1.6254 0.4688 1.6781*** 1.8255** −0.7249 2.8311*** 1.2645* 2.6002**

[13]

1702 −

−0.5784

0.6284 0.9886 −0.5724 0.0083 −0.3837 1.0170* 1.5133*** 1.6831 0.4175 0.0725 0.6947 −1.5849** 1.1698 0.8237 0.2039

[14]

1702 0.069

−0.5600

0.5691 0.9459 −0.6272 −0.0435 −0.4415 0.9661* 1.4634** 1.7234 0.3727 0.0252 0.6521 −1.6244** 1.1254 0.8010 0.1422

[15] c

1702 0.019

−0.4422

−0.1787 0.2572 −1.3010* −0.6839 −1.1370 0.3514 0.8405 2.3872 −0.2500 −0.6026 0.0713 −2.2600*** 0.4825 0.3600 −0.5453

[16] d

Cluster-robust fixed-effects panel LSDV

Notes a Breusch–Pagan test: χ2=104.94, p=0.000 b Hausman test: χ2=88.44, p=0.000 c Breusch–Pagan test: χ2=132.90, p=0.000 d Hausman test: χ2=51.71, p=0.015 Due to space constraints, the estimates of the meta-independent variables of other study conditions and intercept are not presented. Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 3.4 for definition and descriptive statistics of meta-independent variables.

K R2

−2.4132***

[11]

Benchmark index of transformation policy variable (reform level) −1.8713*** Reform speed −2.0038***

[10]

2.7730*** 2.2832*** 0.5901 1.0416** 1.4127* 2.3973*** 2.9542*** 1.2771 −0.2701 1.8585*** 1.7822** −1.0997** 2.5892*** 0.9165 2.5068**

[9]

Cluster-robust ClusterClusterClusterMultilevel Cluster-robust WLS [Quality robust WLS robust WLS robust WLS mixedrandom-effects level] [N] [df] [1/SE] effects RML panel GLS

Transformation policy variable type (comprehensive structural reform) 2.8988*** Comprehensive economic reform 2.5411*** 2.8215*** Liberalization 2.0723*** Price and competition reform 0.8220 1.3799 Enterprise reform 1.1171** 1.0751* Privatization 1.3168 0.9108 3.0899*** Financial reform 2.1275*** *** 3.1224*** Trade reform 2.8208 Institutional quality 1.9584 2.9520*** Property rights reform 1.1968 1.3636 1.5640*** Government reform 1.4252** *** 2.1430*** Political reform/stability 1.8138 Democratization −0.3760 0.4647 Rule of law/legal reform 2.4150*** 2.8828*** 1.6566** Civil rights/society 1.3050* ** 3.3700** Other transformation policies 1.9264

Meta-independent variable (default)/model

Clusterrobust OLS

3

Estimator (analytical weight in parentheses)

(b) Dependent variable — t value

Table 3.8 contd.

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

ASSESSMENT OF PUBLICATION SELECTION BIAS

3.6

adopt the reform level as the benchmark index, those for the reform speed yielded significantly lower results for both the PCCs and t values.11 As the above has shown, the fact that a fairly large proportion of the variables employed in previous research to empirically examine the growth-enhancing effect of structural change and transformation policy did not produce the expected results is probably related to the results of the meta-analysis based on all of the collected estimates that the effect size of these two factors on growth is small, as pointed out in Subsection 3.5.1. The question of why a significant positive correlation with output growth in the CEE and FSU countries could not be identified from specific fields in structural change, policy scope, and speed of reform is one that warrants further inquiry in the future.12

3.6 ASSESSMENT OF PUBLICATION SELECTION BIAS In this section, we will assess the presence and degree of PSB in the literature of growth determinants in transition economies. Following Babecky and Havranek (2014), Figure 3.5 presents funnel plots for the PCC and degree of freedom by growth-determining variable type. For every variable type, the plot shows a roughly triangular shape and thus does not strongly indicate the presence of type I PSB. We also conducted an additional univariate analysis aimed at determining whether the collected estimates are distributed evenly around the true effect. We actually tested for two cases, the first being where the true effect is assumed to be zero and the second being where the mean of the most precise 10% of estimates is regarded as the approximate value of the true effect.13 The results are shown in Column a of Table 3.9. If the mean of the most precise 10% of estimates is assumed to equal the true effect, the null hypothesis, whereby in the case of structural change variables, the number of PCCs that are lower than the true effect is equal to the number of PCCs that are higher than the true effect, is accepted, whereas in other all cases, the null hypothesis is rejected. Accordingly, there is deemed to be a possibility that type I PSB is present for all variable types. Figure 3.6 shows Galbraith plots for t values and degrees of freedom. In these plots, the two-tail test limits of ±1.96 with a 5% significance level are shown as solid lines. From this figure, we certainly cannot say that, for every variable type, 95% of all the estimates lie between these limits. In other words, if the true effect is assumed to be close to zero, the possibility of type II PSB is regarded to be high for every variable type. The results of a more rigid univariate test are reported in Column b of Table 3.9. As this result indicates, if the true effect is assumed to be zero, a goodness-of-fit test rejects the null hypothesis at a 1% significance level for all vari­ able types. Similarly, if the mean of the most precise 10% of estimates is assumed to be the true effect, the null hypothesis that estimates where the statistic |(k-th estimation result – true effect)=SEk | that does not exceed the threshold of 1.96 accounts for 5% of the total is also strongly rejected in all five cases. These results lead us to infer that,

103

C H A P T E R 3

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

3

Figure 3.5 Funnel plot of estimates by growth-determining variable type Note: Solid line indicates the mean of the top 10% most precise estimates. The values for the structural change variable, transformation policy variable, socialist legacy variable, inflation variable, and regional conflict variable are 0.120, 0.052, -0.078, -0.187, and -0.223, respectively.

104

208 1091

88 77 75

72 611

197 619 241

151 705 167 436 175

−6.4566*** −20.5445*** −9.3382*** 118 260 141

129 997

PCCkx

8.1276*** 11.6349***

Goodnessof-fit test a

Number of estimates

182 766 123 230 103

−2.9025*** −6.6713*** −1.9126*

162 466 213

98 936

|tk|1.96

−1.3148 7.0779***

Goodnessof-fit test b

Number of estimates

40.1567*** 74.9941*** 50.8999***

23.0332*** 94.6350***

Goodnessof-fit test c

Under the assumption that the truth effect size is zero

148 196 158

211 816

a

137 500 158

69 886

|(PCCk |(PCCk −x)/SEk| −x)/SEk| 1.96

Number of estimates

33.3620***

54.8195***

36.7036***

15.0812***

89.0742***

Goodnessof-fit test d

Under the assumption that the truth effect size is the mean value of the top 10 percent most precise estimates (x)

(b) Test of type II publication selection bias

Notes

Null hypothesis: The ratio of the positive versus negative values is 50:50.

b Null hypothesis: The ratio of estimates below x versus those over x is 50:50.

c Null hypothesis: Share of estimates, t value of which is within the range of ±1.96, is 95% in total estimates.

d Null hypothesis: Share of estimates, in which the statistics |(the k-th estimate - the true effect)/SEk | is within the range of ±1.96, is 95% in total estimates.

Figures in parentheses are p values. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

Structural change Transformation policy

Socialist legacy Inflation Regional conflict

Growth-determining variable type PCCk0

Number of estimates

Under the assumption that the truth effect size is zero

Under the assumption that the truth effect size is the mean value of the top 10 percent most precise estimates (x)

(a) Test of type I publication selection bias (funnel asymmetry test)

Table 3.9 Univariate test of publication selection bias by growth-determining variable type

ASSESSMENT OF PUBLICATION SELECTION BIAS 3.6

105

C H A P T E R 3

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

3

Figure 3.6 Galbraith plot of estimates by growth-determining variable type

Note: Solid lines indicate the thresholds of two-sided critical values at the 5% significance level ±1.96.

106

CONCLUSIONS

3.7

irrespective of the difference in growth-determining variable type, the possibility of type II PSB is extremely high in this research field. Table 3.10 reports the results of the meta-regression analysis of publication selec­ tion. If we employ as a judgment criterion the question of whether the null hypoth­ esis is rejected for at least two out of three models for each variable type, as Panel a in the table shows, the FAT results in the rejection of the null hypothesis for the two cases of transformation policy and regional conflict. However, the test for type II PSB shown in Panel b of the same table rejects the null hypothesis for four vari­ able types, excluding inflation. The results of the PET reported in Panel a reject the null hypothesis for the four cases excluding the socialist legacy. This indicates the possibility that the collected estimates contain genuine evidence beyond the publica­ tion bias. Actually, as shown in Panel c of Table 3.10, the PEESE results in a strong rejection of the null hypothesis for four variable types. Furthermore, the coefficient of the inverse standard error (1/SE) implies that the impact on economic growth of structural change and transformation policy is significantly positive, whereas that of inflation and regional conflict is significantly negative. We can also confirm that the mutual relationship between these four factors in terms of effect size is quite consist­ ent with the meta-analysis reported in the previous section. In this sense, the results of the meta-analysis performed in this chapter can be regarded as highly reliable even when the presence of PSB is taken into consideration.

3.7 CONCLUSIONS After experiencing an unprecedented fall in output in the immediate aftermath of the collapse of socialism, the CEE and FSU countries either have recovered or are in the process of recovery. The growth path followed by these nations in the past quarter cen­ tury is quite interesting from a historical perspective and has therefore driven numerous researchers to endeavor to define the determinants of output decline and growth in tran­ sition economies. As argued in the Introduction, we have witnessed the so-called “J-curve phenomenon” not only in the countries having certain progress in the transition to a market economy but also in the non-reforming states. Hence, researchers have per­ ceived that there is a more complex picture beyond economic policies, and, in order to elucidate this multifactorial situation, they conducted extensive empirical analysis. The meta-analysis in this chapter, which employed 3,279 estimates collected from 123 previous studies, made the following findings concerning five factors regarded as being closely connected to the emergence of a J-curved growth path in transition econ­ omies. First, structural changes in a national economy, as well as policies designed to transform the planned system into a market-oriented economy, have only delivered a small growth-enhancing impact, dashing the expectations of policy-makers and researchers. Second, in contrast to these two factors, it is highly likely that the hyperin­ flation and regional conflicts that erupted at the beginning of transition led to a massive reduction in output. Third, the socialist legacy is also thought to have contributed to the

107

C H A P T E R 3

108

−0.3906

0.1265***

280

−0.3906

0.1265***

280

0.0823

Intercept (FAT: H0:

β0=0)

1/SE (PET: H0: β1=0)

K

R2

0.0823

[2]

OLS

Clusterrobust OLS

[1]

Model

Estimator

280

0.0823

0.0020

1702

0.0277*

0.6756***

−0.0274

0.1079*

[4]

OLS

0.0020

1702

0.0277

0.6756*

[5]

Clusterrobust OLS

0.0020

1702

0.2520***

−1.9127**

[6] b

Clusterrobust fixedeffects panel LSDV

Transformation policy

[3] a

Clusterrobust randomeffects panel GLS

Structural change

0.0016

0.0016

285

−0.0280

−0.0280

285

−0.6924

[8]

−0.6924

[7]

OLS

Clusterrobust OLS

0.0016

285

0.0357

−1.3763*

[9] c

Clusterrobust randomeffects panel GLS

Socialist legacy

1.5013

[11]

0.2122

696

0.2122

696

−0.4260*** −0.4260***

1.5013***

[10]

OLS

Clusterrobust OLS

Inflation

0.2122

696

−0.4920**

2.3147

[12] d

Clusterrobust fixedeffects panel LSDV

1.1050*

[14]

0.3041

316

0.3041

316

−0.3605*** −0.3605***

1.1050***

[13]

OLS

Clusterrobust OLS

0.3041

316

−0.4288***

1.7645

[15] e

Cluster­ robust fixedeffects panel LSDV

Regional conflict

3

Estimates to test

(a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

Table 3.10 Meta-regression analysis of publication selection bias by growth-determining variable type

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

280

0.0291

0.0291

280

0.0493**

0.0493***

1/SE

R2

1.2553***

1.2553***

Intercept (H0: β0=0)

K

[17]

OLS

Clusterrobust OLS

1.0773***

−1.3351*

280

0.0291

0.0839

1702

0.1215***

[19]

[18] f

0.2783***

OLS

0.0839

1702

0.1215***

1.0773***

[20]

Clusterrobust OLS

0.0839

1702

0.2330***

−0.2100

[21] g

Clusterrobust fixedeffects panel LSDV

Transformation policy

Clusterrobust fixedeffects panel LSDV

Structural change

[16]

Model

Estimator

Estimates to test

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v)

[23]

0.0062

285

0.0330

0.0062

285

0.0330

2.1427*** 2.1427***

[22]

OLS

Clusterrobust OLS

0.0062

285

0.3830***

−2.0592

[24] h

Clusterrobust fixedeffects panel LSDV

Socialist legacy

0.2048

696

0.3839***

−0.7104**

[25]

OLS

0.2048

696

0.3839***

−0.7104

[26]

Clusterrobust OLS

Inflation

0.2048

696

0.4575**

−1.6168

[27] i

Clusterrobust fixedeffects panel LSDV

0.2317

316

0.2239***

1.0240***

[28]

OLS

316

0.2317

316 0.2317

(Continued )

0.4554***

0.2239***

[29]

−1.2122

[30] j

Clusterrobust OLS

1.0240**

Cluster­ robust fixedeffects panel LSDV

Regional conflict

CONCLUSIONS 3.7

109

C H A P T E R 3

110

280

280

0.2820

K

R2

−

280

0.1046***

0.3044

[33]

Randomeffects panel ML

0.0976

1702

0.0602***

2.6947***

[34]

OLS

0.0976

1702

0.0602**

2.6947**

[35]

Clusterrobust OLS

−

1702

0.1631***

−3.6531***

[36]

Randomeffects panel ML

Transformation policy

0.1124

0.1124

285

−0.0648

285

−2.3802

−2.3802

[38]

−0.0648***

[37]

OLS

Clusterrobust OLS

−

285

−0.0259

−6.0703**

[39]

Randomeffects panel ML

Socialist legacy

5.5720

[41]

0.6132

696

0.6132

696

−0.3487*** −0.3487***

5.5720***

[40]

OLS

Clusterrobust OLS

Inflation

−

696

−0.3683***

7.3092***

[42]

Randomeffects panel ML

4.7254**

[44]

0.5736

316

0.5736

316

−0.3084*** −0.3084***

4.7254***

[43]

OLS

Clusterrobust OLS

−

316

−0.3290***

4.8548**

[45]

Randomeffects panel ML

Regional conflict

Notes a Breusch–Pagan test: χ2=51.89, p=0.000; Hausman test: χ2=0.60, p=0.439 b Breusch–Pagan test: χ2=1621.37, p=0.000; Hausman test: χ2=12.33, p=0.000 c Breusch–Pagan test: χ2=318.12, p=0.000; Hausman test: χ2= 1.50, p=0.220 d Breusch–Pagan test: χ2=2028.26, p=0.000; Hausman test: χ2=3.06, p=0.080 e Breusch–Pagan test: χ2= 170.82, p=0.000; Hausman test: χ2=3.56, p=0.059 f Breusch–Pagan test: χ2=75.76, p=0.000; Hausman test: χ2= 12.69, p=0.000 g Breusch–Pagan test: χ2=2223.53, p=0.000; Hausman test: χ2=8.98, p=0.003 h Breusch–Pagan test: χ2= 130.66, p=0.000; Hausman test: χ2=30.31, p=0.000 i Breusch–Pagan test: χ2=2511.63, p=0.000; Hausman test: χ2= 6.29, p=0.012 j Breusch–Pagan test: χ2=290.13, p=0.000; Hausman test: χ2=4.10, p=0.043 Robust standard errors are used for hypothesis testing except for models [33], [36], [39], [42], and [45]. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

0.2820

−1.2131

0.1049***

−1.2131

0.1049***

[32]

1/SE (H0: β1=0)

[31]

OLS

Clusterrobust OLS

SE

Model

Estimator

Structural change

3

Estimates to test

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Table 3.10 contd.

C H A P T E R

TRANSFORMATIONAL RECESSION AND RECOVERY

CONCLUSIONS

3.7

economic crisis, with an effect size similar to those of structural changes and transform­ ation policy. These results provide a lucid explanation for why the economic recovery that followed the crisis was characterized not by a V-shape but by a slower-paced growth tempo and why marked differences occurred between countries in the rate of output decline during the crisis and the speed of recovery during the rebound. In other words, while interactions among the five factors delivered a J-curved growth path to all of the CEE and FSU countries, the differences among the nations in terms of historical preconditions, political circumstances, reform efforts, and the occurrence of conflicts resulted in major differences in their growth trajectories. In this chapter, we also investigated PSB and the presence of genuine evidence in the existing literature through visual verification using funnel plots and Galbraith plots and estimation of meta-regression models developed specifically for this pur­ pose. The latter results are summarized in Table 3.11. As this table shows, in this research field, the publication frequency of statistically significant empirical findings is unnaturally high. For this reason, it is highly likely that type II PSB is present, though the influence of type I publication bias is not especially serious. Moreover, it is verified that genuine empirical evidence exists in the collected estimates and that the publication-selection-bias-adjusted effect size is significantly different from zero except for the legacy of socialism. In other words, previous research has, on the whole, achieved great success in specifying the true effects of the most important determinants of the growth path in the CEE and FSU countries during the transition period.14 We therefore wish to pay our respects to the generous efforts made by researchers of transition economies from the late 1990s until today.

Table 3.11 Summary of publication selection bias test Test results a

Growth-deter- Number mining vari­ of esti­ able type mates (K)

Funnel asymmetry test for type I PBS (FAT) (H0: β0=0)

Test for type II PBS (H0: β0=0)

Precisioneffect test (PET) (H0: β1=0)

Structural change Transformation policy Socialist legacy Inflation

Not rejected

Rejected

Rejected

Regional conflict

280 1702

Rejected

285 696

Not rejected Not rejected

316

Rejected

Precision-effect esti­ mate with standard error (PEESE) (H0: β1=0) b

Rejected (0.1046/0.1049) Rejected Rejected Rejected (0.0602/0.1631) Rejected Not rejected Not rejected Not rejected Rejected Rejected (−0.3683/−0.3487) Rejected Rejected Rejected (−0.3290/−0.3084)

Notes a The null hypothesis is rejected when more than two of three models show a statistically significant estimate; otherwise, it is not rejected. b Figures in parentheses are PSB-adjusted estimates. If two estimates are reported, the left and right figures denote a minimum and maximum estimate, respectively.

111

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

ACKNOWLEDGMENTS

This chapter is a substantially extended version of Iwasaki and Kumo 2019. We thank Tomáš Havránek, Robert J. Johnston, Martin Paldam, Tom D. Stanley, Manabu Suhara, Paul Wachtel for their helpful comments and suggestions on the earlier version of this paper.

NOTES 1 In fact, according to data published by the European Bank for Reconstruction and Development (http://www.ebrd.com), output in the three CEE countries of Bosnia–Herzegovina, Montenegro, and Serbia and the two FSU countries of Moldova and Ukraine was, in 2013 and 2015, respectively, between 8% and 35% lower than that at the end of socialist period. 2 Babecký and Campos (2011) performed a meta-analysis of 515 estimation results reported in 46 studies, while Babecky and Havranek (2014) employed 537 estimation results from 60 studies. 3 One may find it odd that even among the then-new member states of the Euro­ pean Union, six Central European and Baltic countries with especially strong reputations for promoting reform, as well as three FSU countries, Uzbekistan, Turkmenistan, and Belarus, where the pace of democratization and economic reform has been particularly low, are all included in the same cluster. However, as Iwasaki (2004) pointed out, the macroeconomic performance of these FSU countries, the governments of which dealt with national crises caused by the breakup of the Soviet Union by exercising strong leadership over industry, was not much more unfavorable in comparison with that of the Central European and Baltic countries, especially during the early phase of transition. Putting aside the evaluation of the reform strategy based on a statist, paternal industrial strategy, these facts can be seen as having a big impact on the results of the cluster analysis. 4 The vast majority of studies have employed a dummy variable whereby 1 denotes the year and country in which a conflict occurred. However, for countries that have experienced conflicts, a dummy variable of 1 has been applied to the entire estimation period in some studies. 5 For the details of these 123 selected studies, see Iwasaki and Kumo 2016, tab. 2. 6 Independent variables employed in previous research at frequencies similar to the above five variables are domestic investment and fiscal expenditure, but the number with empirical findings estimated to be statistically significant is much lower than that for the above five factors. The next most frequently used are education level and foreign direct investment (FDI). The empirical results of the growth-promoting effect of education are similar to those of domestic investment and fiscal expenditure, as mentioned in the Introduction. FDI is regarded as a promising factor behind growth in transition economies (see Chapter 9 in this

112

NOTES

7

8

9

10

11

book). Other factors that might significantly affect macroeconomic performance in transition economies include a great shift in the labor force participation rate, large-scale international migration, and dramatic changes in the world oil price (Bah and Brada 2014; Kuboniwa 2014; Bilan and Strielkowski 2016). However, empirical evidence of the growth effects of these factors is extremely limited, and most available estimates do not cover the first decades of the transition period. Therefore, we could not consider these factors, in addition to the factors mentioned above, in the meta-analysis in this chapter. Cohen (1988), who is frequently cited for assessing correlation coefficients, defines a coefficient of 0.3 as the threshold between a “small effect” and a “medium effect” and a coefficient of 0.5 as the threshold between a “medium effect” and a “large effect.” It is argued, however, that Cohen’s guidelines for zero-order correlations are too restrictive when applied to economics. This prompted Doucouliagos (2011) to propose alternative criteria to those of Cohen (1988). According to his new general criteria, the lower thresholds for small, medium, and large effects are set at 0.070, 0.173, and 0.327, respectively. Incidentally, when we performed a meta-synthesis limited to estimates for the 1990s, when almost all of the CEE and FSU countries were either in the midst of crisis or in which output had still failed to recover to the levels at the end of socialism, as the estimation period, the synthesized effect size of structural change using the random-effects model shrank to 0.012, thereby becoming statis­ tically insignificant. However, those of the socialist legacy and inflation both increased dramatically, to −0.206 and −0.413, respectively. Meanwhile, the syn­ thesized effect size of transformation policy and regional conflict changed only slightly, to 0.170 and 0.281, respectively. These results suggest that the timelagged effect of structural change and the time-decay effect of the socialist legacy and inflation have not been adequately captured in the earlier research. Studies that have paid particularly close attention to the relationship between reform speed and economic growth include Heybey and Murrell 1999, Bernardes 2003, Staehr 2005, and Godoy and Stiglitz 2006. Most previous studies have employed temporal differences in the degree of reform as a proxy for reform speed. For instance, see de Macedo and Martins 2008 and Segura-Ubiergo et al. 2010. Due to space constraints, we have left some estimates out, but as was the case with Table 3.5, meta-independent variables that capture various study conditions are simultaneously estimated. In their meta-analysis, Babecky and Havranek focused on the difference in the short-run and long-run growth effect of structural reforms and pointed out that “on average, in the short run reforms lead to significant costs in terms of output growth, while in the long run the effect of reforms on economic performance is positive and substantial” (2014, p. 31). The results of our meta-analysis indicate that the difference between reform level and reform speed may be a more important aspect in empirically examining the effect of transformation policy on output, judging from the findings that the meta-independent variable of reform 113

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY

speed repeatedly shows a significant and negative coefficient, while those of length of estimation period and use of cross-section data are estimated to be insignificant. However, when most meta-independent variables of transformation policy variable type and reform speed are controlled for between-study hetero­ geneity using the multilevel mixed-effects RML or the random/fixed-effects panel estimator, the statistical significance of the regression coefficient drops by a large margin. This makes it likely that some caution should be exercised in the interpretation of estimation results. 12 In this regard, Coricelli and Maurel (2011) gave a key to untangling this puzzle, suggesting that the post-recession performance in transition economies strongly depends on the complementarities among different reform measures, to which most previous studies do not pay sufficient attention. 13 The method for assuming that the mean of the most precise 10% of estimates is the approximate value of the true effect is along the lines of Stanley 2005. 14 We should note, however, that the overwhelming majority of previous studies assume causality between the five factors in question and the macroeconomic performance in CEE and FSU countries as conventional wisdom, and, as a result, they often do not properly deal with the possible endogeneity problem. The distinction between the effects of hyper and moderate inflation on growth is also examined insufficiently. In addition, the impact of resource shift among industrial sectors on productivity is addressed only in a few studies. Addressing these shortcomings remains as a future agenda in this study field.

REFERENCES Akimov, Alexandr, Albert Wijeweera, and Brian Dollery (2009) Financial development and economic growth: Evidence from transition economies. Applied Financial Economics, 19 (10–12), pp. 999–1008. Apolte, Thomas (2011) Democracy and prosperity in two decades of transition. Economics of Transition, 19(4), pp. 693–722. Åslund, Anders, Peter Boone, Simon Johnson, Stanley Fischer, and Barry W. Ickes (1996) How to stabilize: Lessons from post-communist countries. Brookings Papers on Eco­ nomic Activity, 1996(1), pp. 217–313. Babecký, Jan, and Nauro F. Campos (2011) Does reform work? An econometric survey of the reform-growth puzzle. Journal of Comparative Economics, 39(2), pp. 140–158. Babecky, Jan, and Tomas Havranek (2014) Structural reforms and growth in transition: A meta-analysis. Economics of Transition, 22(1), pp. 13–42. Bah, El-hadj, and Josef C. Brada (2014) Labor markets in the transition economies: An overview. European Journal of Comparative Economics, 11(1), pp. 3–53. Beck, Thorsten, and Luc Laeven (2006) Institution building and growth in transition economies. Journal of Economic Growth, 11(2), pp. 157–186. Bennett, John, Saul Estrin, and Giovanni Urga (2007) Methods of privatization and eco­ nomic growth in transition economies. Economics of Transition, 15(4), pp. 661–683.

114

REFERENCES Bennett, John, Saul Estrin, James Maw and Giovanni Urga (2004) Privatization methods and economic growth in transition economies. Discussion Paper No. 4291, Centre for Economic Policy Research: London. Bernardes, Luis G. (2003) Reference-dependent preferences and the speed of economic liberalization. Journal of Socio-Economics, 32(5), pp. 521–548. Bilan, Yuriy, and Wadim Strielkowski (2016) Migration in post-transition economies: Immi­ gration surplus in Visegrad group countries. International Journal of Trade and Global Markets, 9(2), pp. 182–196. Blanchard, Olivier, and Michael Kremer (1997) Disorganization. Quarterly Journal of Eco­ nomics, 112(4), pp. 1091–1126. Brada, C. Josef, and Arthur E. King (1992) Is there a J-curve for the economic transition from socialism to capitalism? Economics of Planning, 25(1), pp. 37–53. Brenton, Paul, Daniel Gros, and Guy Vandille (1997) Output decline and recovery in the transition economies: Causes and social consequences. Economics of Transition, 5(1), pp. 113–130. Campos, Nauro, and Abrizio Coricelli (2002) Growth in transition: What we know, what we don’t, and what we should. Journal of Economic Literature, 40(3), pp. 793–836. Campos, Nauro F., and Yuko Kinoshita (2002) Foreign direct investment as technology transferred: Some panel evidence from the transition economies. Manchester School, 70 (3), pp. 398–419. Capolupo, Rosa, and Giuseppe Celi (2005) Openness and growth in Central-Eastern Euro­ pean countries. Economia Internazionale, 58(2), pp. 141–165. Cernat, Lucian, and Radu Vranceanu (2002) Globalisation and development: New evidence from Central and Eastern Europe. Comparative Economic Studies, 44(4), pp. 119–136. Cerović, Božidar, and Aleksandra Nojković (2009) Transition and growth: What was taught and what happened. Economic Annals, 54(183), pp. 7–31. Cohen, Jacob (1988) Statistical Power Analysis in the Behavioral Sciences. Second edition, Lawrence Erlbaum Associates: Hillsdale, NJ. Cojocaru, Laura, Evangelos M. Falaris, Saul D. Hoffman, and Jeffrey B. Miller (2016) Finan­ cial system development and economic growth in transition economies: New empirical evidence from the CEE and CIS Countries. Emerging Markets Finance and Trade, 52(1), pp. 223–236. Coricelli, Fabrizio, and Mathilde Maurel (2011) Growth and crisis in transition: A comparative perspective. Review of International Economics, 19(1), pp. 49–64. de Macedo, Jorge Braga, and Joaquim Oliveira Martins (2008) Growth, reform indicators and policy complementarities. Economics of Transition, 16(2), pp. 141–164. de Melo, Martha, Cevdet Denizer, and Alan Gelb (1997) From plan to market: Patterns of transition. In Mario Blejer and Marko Škreb (eds.), Macro Economic Stabilization in Tran­ sition Economies. Cambridge University Press: Cambridge and New York, pp. 17–72. de Melo, Martha, Cevdet Denizer, Alan Gelb, and Stoyan Tenev (2001) Circumstance and choice: The role of initial conditions and policies in transition economies. World Bank Economic Review, 15(1), pp. 1–31. de Rocha, Bruno T. (2015) Let the markets begin: The interplay between free prices and privatisation in early transition. Journal of Comparative Economics, 43(2), pp. 350–370. Denizer, Cevdet (1997) Stabilization, adjustment, and growth prospects in transition econ­ omies. Policy Research Working Paper No. 1855, World Bank: Washington DC.

115

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY Djalilov, Khurshid, and Jenifer Piesse (2011) Financial development and growth in transition countries: A study of Central Asia. Emerging Markets Finance and Trade, 47(6), pp. 4–23. Doucouliagos, Hristos (2011) How large is large? Preliminary and relative guidelines for interpreting partial correlations in economics. School Working Paper No. SWP 2011/5, School of Accounting, Economics and Finance, Faculty of Business and Law, Deakin University, Melbourne. Dudian, Monica, and Raluca Andreea Popa (2013) Financial development and eco­ nomic growth in Central and Eastern Europe. Theoretical and Applied Economics, 20(8), pp. 59–68. EBRD (European Bank for Reconstruction and Development) (1999) Transition Report 1999: Ten Years of Transition. EBRD: London. Eicher, Theo S., and Till Schreiber (2010) Structural policies and growth: Time series evidence from a natural experiment. Journal of Development Economics, 91(1), pp. 169–179. Falcetti, Elisabetta, Martin Raiser, and Peter Sanfey (2002) Defying the odds: Initial condi­ tions, reforms, and growth in the first decade of transition. Journal of Comparative Eco­ nomics, 30(2), pp. 229–250. Fidrmuc, Jan (2001) Democracy in transition economies: Grease or sand in the wheels of growth? EIB Papers, 6(2), pp. 25–40. Fidrmuc, Jan (2003) Economic reform, democracy and growth during post-communist transition. European Journal of Political Economy, 19(3), pp. 583–604. Fischer, Stanley, and Ratna Sahay (2001) The transition economies after ten years. In Lucjan T. Orlowski (ed.), Transition and Growth in Post-Communist Countries: The Tenyear Experience. Edward Elger: Cheltenham and Northampton, pp. 3–47. Fischer, Stanley, Ratna Sahay, and Carlos A. Végh (1996a) Economies in transition: The beginnings of growth. American Economic Review, 86(2), pp. 229–233. Fischer, Stanley, Ratna Sahay, and Carlos A. Végh (1996b) Stabilization and growth in transition economies: The early experience. Journal of Economic Perspectives, 10(2), pp. 45–66. Gaffeo, Edoardo, and Petya Garalova (2014) On the finance-growth nexus: Additional evi­ dence from Central and Eastern Europe countries. Economic Change and Restructuring, 47(2), pp. 89–115. Gillman, Max, and Mark N. Harris (2010) The effect of inflation on growth: Evidence from a panel of transition countries. Economics of Transition, 18(4), pp. 697–714. Godoy, Sergio, and Joseph E. Stiglitz (2006) Growth, initial conditions, law and speed of privatization in transition countries: 11 years later. Working Paper No. 11992, National Bureau of Economic Research: Cambridge. Grogan, Louise, and Luc Moers (2001) Growth empirics with institutional measures for tran­ sition countries. Economic Systems, 25(4), pp. 323–344. Halushka, Andrij (1997) Financial system and growth in transition economies. Ukrainian Economic Review, 4/5, pp. 108–122. Havrylyshyn, Oleh (2001), Recovery and growth in transition: A decade of evidence. IMF Staff Papers, 48: Special Issue, pp. 53–87. Havrylyshyn, Oleh, and Thomas Wolf (2001) Growth in transition countries 1990–98: The main lessons. In Oleh Havrylyshyn and Saleh M. Nsouli (eds.), A Decade of Transition: Achievements and Challenges. IMF: Washington DC, pp. 83–128.

116

REFERENCES Heckelman, Jac C. (2010) The connection between democratic freedoms and growth in transition economies. Applied Economics Quarterly, 56(2), pp. 121–146. Heybey, Berta, and Peter Murrell (1999) The relationship between economic growth and the speed of liberalization during transition. Journal of Policy Reform, 3(2), pp. 121–137. Hodgson, Geoffrey M. (2006) Institutions, recessions and recovery in the transitional economies. Journal of Economic Issues, 40(4), pp. 875–894. Iwasaki, Ichiro (2004) Evolution of the government–business relationship and economic per­ formance in the former Soviet states: Order state, rescue state, punish state. Economics of Planning, 36(3), pp. 223–257. Iwasaki, Ichiro, and Kazuhiro Kumo (2016) Decline and growth in transition economies: A meta-analysis. CEI Working Paper No. 2016-9, Center for Economic Institutions, Insti­ tute of Economic Research of Hitotsubashi University: Kunitachi, Tokyo. Iwasaki, Ichiro, and Kazuhiro Kumo (2019) J-curve in transition economies: A large meta-analysis of the determinants of output changes. Comparative Economic Studies, 61(1), pp. 149–191. Iwasaki, Ichiro, and Akira Uegaki (2019) The disinflation effect of central bank independ­ ence: A comparative meta-analysis between transition economies and the rest of the world. In Julien Chevallier, Stéphane Goutte, David Guerreiro, Sophie Saglio, and Bilel Sanhaji (eds.), International Financial Markets. Volume 1, Routledge: Abingdon and New York, pp. 227–287. Josifidis, Kosta, Radmila Dragutinović Mitrović, and Olgica Ivančev (2012) Heterogeneity of growth in the West Balkans and emerging Europe: A dynamic panel data model approach. Panoeconomicus, 59(2), pp. 157–183. Kolodko, Grzegorz (2001) Globalization and catching-up: From recession to growth in tran­ sition economies. Communist and Post-Communist Studies, 34(3), pp. 279–322. Kornai, János (1994) Transformational recession: The main causes. Journal of Comparative Economics, 19(1), pp. 39–63. Kuboniwa, Masaaki (2014) A comparative analysis of the impact of oil prices on oil-rich emerging economies in the Pacific Rim. Journal of Comparative Economics, 42(2), pp. 328–339. Loungani, Prakash, and Nathan Sheets (1997) Central bank independence, inflation, and growth in transition economies. Journal of Money, Credit, and Banking, 29(3), pp. 381–399. Mankiw, Gregory, David Romer, and David Weil (1992) A contribution to the empirics of economic growth. Quarterly Journal of Economics, 107(2), pp. 407–438. Mitrović, Radmila Dragutinović, and Olgica Ivanč ev (2010) Driving forces of economic growth in the second decade of transition. Economic Annals, 55(185), pp. 7–32. Moers, Luc (1999) How important are institutions for growth in transition countries? Discus­ sion Paper No. 99-0004/2, Tinbergen Institute: Amsterdam. Nath, Hiranya K. (2009) Trade, foreign direct investment, and growth: Evidence from transi­ tion economies. Comparative Economic Studies, 51(1), pp. 20–50. Neyapti, Bilin, and Nergiz Dincer (2005) Measuring the quality of bank regulation and supervision with an application to transition economies. Economic Inquiry, 43(1), pp. 79–99. Peev, Evgeni, and Dennis C. Mueller (2012) Democracy, economic freedom and growth in transition economies. Kyklos, 65(3), pp. 371–407. Polanec, Sašo (2004) Convergence at last? Evidence from transition countries. Eastern European Economics, 42(4), pp. 55–80.

117

C H A P T E R 3

C H A P T E R 3

TRANSFORMATIONAL RECESSION AND RECOVERY Popov, Vladimir (2007) Shock therapy versus gradualism reconsidered: Lessons from transi­ tion economies after 15 years of reforms. Comparative Economic Studies, 49(1), pp. 1–31. Próchniak, Mariusz (2011) Determinants of economic growth in Central and Eastern Europe: The global crisis perspective. Post-Communist Economies, 23(4), pp. 449–468. Radulescu, Roxana, and David Barlow (2002) The relationship between policies and growth in transition countries. Economics of Transition, 10(3), pp. 719–745. Redek, Tjaša, and Andrej Sušjan (2005) The impact of institutions on economic growth: The case of transition economies. Journal of Economic Issues, 39(4), pp. 995–1027. Roland, Gérald, and Thierry Verdier (1999) Transition and the output fall. Economic of Tran­ sition, 7(1), pp. 1–28. Rosati, Dariusz (1994) Output decline during transition from plan to market: A reconsideration. Economics of Transition, 2(4), pp. 419–441. Segura-Ubiergo, Alex, Alejandro Simone, Sanjeev Gupta, and Qiang Cui (2010) New evi­ dence on fiscal adjustment and growth in transition economies. Comparative Economic Studies, 52(1), pp. 18–37. Selowsky, Marcelo, and Ricardo Martin (1997) Policy performance and output growth in the transition economies. American Economic Review, 87(2), pp. 349–353. Staehr, Karsten (2005) Reforms and economic growth in transition economies: Complemen­ tarity, sequencing and speed. European Journal of Comparative Economics, 2(2), pp. 177–202. Stanley, T. D. (2005) Beyond publication bias. Journal of Economic Surveys, 19(3), pp. 309–345. Stuart, Robert C., and Christina M. Panayotopoulos (1999) Decline and recovery in transi­ tion economies: The impact of initial conditions. Post-Soviet Geography and Economics, 40(4), pp. 267–280. Sukiassyan, Grigor (2007) Inequality and growth: What does the transition economy data say? Journal of Comparative Economics, 35(1), pp. 35–56. Wolf, Holger (1999) Transition strategies: Choices and outcomes. Princeton Studies in International Finance No. 85, Department of Economics, Princeton University: Princeton, NJ. Wyplosz, Charles (2000) Ten years of transformation: Macroeconomic lessons. Discussion Paper No. 2254, Centre for Economic Policy Research: London.

118

4

Economic transition and poverty Changes in the determinants of poverty Kazuhiro Kumo

4.1 INTRODUCTION The purpose of this chapter is to describe, through previous research, how the factors causing households to fall into poverty in former socialist countries have been explored for more than 25 years since the beginning of the economic transition. It has been widely acknowledged that economic disparities were small and levels of poverty were low in socialist countries (McAuley 1979). Although it was impossible to make detailed studies because hardly any data was made publicly available, it can be said that it was commonly acknowledged that income redistribution, government-set wage rates, and generous social security kept poverty at low levels in the socialist coun­ tries (McAuley 1979). However, it is known that as the economic transition began, this situation changed. The well-known study by Milanovic (1997) employed various types of household survey data to estimate the total number of people with incomes below the poverty line. As shown in Figure 4.1, based on his calculations, in 18 countries located in the former Soviet Union and Southeastern Europe, the number of people in poverty increased by ten times (from 14 million to 147 million people) between 1993 and 1995, the period following the beginning of the economic transition, compared with the period 1987 to 1988, which was before the economic transition. However, this was based on a poverty line of income of US$4 per person per day at 1993 purchasing power parity, so it can be said to be a fairly high estimate. Nevertheless, this does probably not affect the overall trend. In addition, the increase in the number of people in poverty in Russia was striking. Between 1987 and 1988, just 2.2 million (1.5%) of the total population of 146 million people (1987) were in poverty, but after the economic transition began, the number of poor in Russia increased by 30 times to 66 million people, 44% of the total population of 148.5 million people (1993) (Milanovic 1997). Even under socialism, it was not the case that poverty did not exist at all. It needs to be pointed out that it was merely impossible to investigate it due to the inaccess­ ibility of data. At the same time, however, poverty in regions that had been in the

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

4

Figure 4.1 The number of population with income below the poverty line (Million) Source: Author’s illustration based on Milanovic (1997)

socialist bloc increased due to the economic transition, and it can be said that it became more widespread than before. What is interesting here is the impact that “poverty” had as a problem associated with economic transition, and the extent to which the problem is unique to transition economies. Poverty itself is a widely observed phenomenon, so it can be said that the most important task is to determine whether it is actually a problem of “transi­ tion economies.” Therefore this chapter will carefully examine research on poverty in transition economies conducted over the past 25 years or so by exploring trends such as which factors have been studied, how they are different or similar to such factors in other countries, and whether differences are observed among transition economies.

4.2 POVERTY IN TRANSITION ECONOMIES The increase in poverty in transition economies shown by Milanovic (1997), which mentioned in the introduction, has been described as “sudden poverty” in previous research (Ruminska-Zimny 1997). This expression means the rapid increase in pov­ erty in former socialist countries that had established generous systems of social security. Certainly, a big change occurred in the poverty headcount between the socialist era and the after the beginning of the economic transition. Nevertheless, as was mentioned in the introduction, there is hardly any data for the socialist era.

120

POVERTY IN TRANSITION ECONOMIES

4.2

What can be used are various estimated series, such as the one illustrated in Figure 4.2, which shows the poverty headcount (the percentage of the population with incomes below the “cost of maintaining a minimum standard of living”) and the Gini coefficients for per-capita income in Russia from 1980, before the collapse of the Soviet Union, to the 2010s. The poverty headcount, which was 11.4% in 1991, began rising as the economic transition started at the end of 1991, reaching 31.5% in 1993. Similarly, the Gini coefficient, which indicates the level of income disparity, jumped from 0.265 in 1991 to 0.398 in 1993. This can be said to illustrate the occurrence of the “sudden poverty” in the transition economies described by Ruminska-Zimny (1997). However, it is easy to see two contrasting periods, a sudden increase during the 1990s and a decline during the 2000s. It can be pointed out that these trends were closely related to the economic situation. Figure 4.3 shows the poverty headcount in Russia alongside gross national income (GNI) per capita. At the beginning of the 1990s, when the economy shrank in conjunction with the economic transition, the poverty headcount increased sharply. From 1999, however, when the economy began growing on a sustained basis, the poverty headcount trended downwards. The correlation between the poverty headcount and per-capita GNI in Figure 4.3 is −0.82, illustrating that the poverty headcount declines as per-capita GNI increases. Needless to say, this is not something that is limited to Russia. The other socialist countries in Eastern Europe also had almost the same systems, such as social

Figure 4.2 Poverty headcount and Gini coefficients of income in Russia, 1980–2016 Note: Poverty headcount denotes the percentage of the population with incomes below the cost of maintaining a minimum

standard of living.

Source: Author’s illustration based on Braithwaite (1995) and Rosstat, Sotsial’noe polozhenie Iurovenzhisni naseleniya

Rossii (1998, 2001, 2004, 2010, 2011, 2013, 2014, 2015, 2017)

121

C H A P T E R 4

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

4

Figure 4.3 Poverty headcount and GNI per capita in Russia, 1980–2016

Source: Author’s illustration based on Rosstat, Sotsial’noe polozhenie Iurovenzhisni naseleniya Rossii (1998, 2001,2004,

2010, 2011, 2013, 2014, 2015, 2017), World Bank, World Development Indicators (https://data.worldbank.org/indicator/).

security systems providing pensions, healthcare, systems for ensuring employment, and so on (McAuley 1979; Braithwaite et al. 2000). As a result, it can be said that the transition to market economies that took place in these countries exhibited a similar phenomenon in that it made poverty more apparent. However, it must also be pointed out that the situation was not exactly the same in every region. Table 4.1 shows poverty headcounts in the countries that comprised the Soviet Union, the transition economies of Central and Eastern Europe, as well as China and Vietnam. A look at this table enables a number of facts to be confirmed. The top half shows figures for the countries that comprised the Soviet Union, while the bottom half does the same for the transition economies of Central and Eastern Europe and the Asian countries. The figures obtained have been presented, generally and on the whole, poverty headcounts are clearly lower in the bottom half. The aver­ age for the top half is 26.6%, and that for the bottom half is 19.1%, and if China and Vietnam are omitted, the latter is 19.4%. Furthermore, a comparison of the 1990s, 2000s, and 2010s reveals that the poverty headcount trended downwards. The averages for the top half were 46.7% in the 1990s, 31.4% in the 2000s, and 17.2% in the 2010s, while those for the bottom half were 25.2% in the 1990s, 19.9% in the 2000s, and 16.6% in the 2010s. It can be seen that the decline in poverty headcount during the 2000s was most conspicuous in the countries that comprised the Soviet

122

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

Year

(a) FSU

53.5 40.1 30.2 26.4 27.6 34.1 35.8 35 32.4 32 30 29.8 29.4

48.3

55.5

Armenia

15.8 13.2 10.9 9.1 7.6 6

49.6

68.1

Azerbaijan

52.1 54.5 34.3 35.4 36.9 38.8 34.9 34.9 37.3 34.1 30 26.2 23.5 21.6 22

Georgia

46.7 44.5 37.5 33.9 31.6 18.2 12.7 12.1 8.2 6.5 5.5 3.8 2.9 2.8 2.7 2.5

34.6

Kazakhstan

34.3 32 31.3

46.7

53.5

72.4

96

Tajikistan

Table 4.1 Poverty headcount in transition economies, 1993–2016

Turkmenistan

39.9 35 31.7 31.7 33.7 36.8 38 37 30.6 32.1 25.4

Kyrgyz Republic

7.3 6.3 5.5 4.8 5.1 5.7

38.6 32.1 33 46.7 41.9 28.9 30.5 27.1 17.8 12.7 11.1 7.7 6.1 5.4

Belarus

30.2 25.8 26.4 26.3 21.9 17.5 16.6 12.7 11.4 9.6

54.6 40.4 29 26.5

Moldova

24.6 20.3 17.6 17.8 15.2 13.3 13.4 13 12.5 12.7 10.7 10.8 11.2 13.3 13.3

31.4

30.9

Russia

(Continued )

83.3 76.2 65.6 55.3 49.7 12.7 7.1 5.8 8.6 7.8 9 8.3 8.6 6.4 3.8

Ukraine

POVERTY IN TRANSITION ECONOMIES 4.2

123

C H A P T E R 4

124

14.5

1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016

19 19.1 17.3 16.9 17.1 17.6 17.7 17.1 17.3 17 17.6 17.3 15

14.3 14.8 15.6 16.6

14.6

23.8

Poland

11.1

11.2

Croatia

19.4 23.5 21.2 25.9 26.4 20.9 19 19.2 19.4 21.2 22.5

7.5

Latvia

24.6 23.6 22.1 21.6 22.3 22.9 23 25.1 25.4 25.3 23.6

35.9 30.6 28.9 25.1

21.5 25.4

Romania

9 6.6 6.1 6.9 9.2

14.6

14

Serbia

19.1 19.2 18.5 20.4 19

Macedonia

16.9

17.9

18.2

17.7

Bosnia

34.5

37.7 43.7 34.8 45.1

Kosovo

14.3

12.4

18.5

25.4

Albania

Note: In percent.

Source: World Bank, World Development Indicators CD-ROM (2005, 2012), and the World Development Indicators Website (https://data.worldbank.org/indicator/).

13.5 15.9 12.3 12.4 12.4 12.3 14.1 14.3 15 15 14.9

17.3

Hungary

17.2 12.7 10.2 8.5 7.2 5.7 4.5

China

4

Year

(b) CEE and Asia

Table 4.1 contd.

9.8

13.5

17.2

20.7

14.5

16

19.5

28.9

37.4

58.1

Vietnam

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

DETERMINATION OF THE LITERATURE TO BE SURVEYED

4.3

Union. That has actually been pointed out by researchers such as Razumov and Yagodkina (2007) and Bobkov (2007). The collapse of socialism delivered a transitional shock to the regions, and the number of people in poverty increased sharply. The increase was particularly con­ spicuous in the 1990s, and the situation was especially severe in the countries that formerly comprised the Soviet Union. However, the situation changed in the 2000s, and it can be pointed out that the poverty headcount in each country exhibited a clear downward trend. So how was poverty in the transition economies described? During the socialist, or Soviet era, the risk of falling into poverty was regarded as high for households in rural areas and households with children (McAuley 1979; Braithwaite 1995). This view would be in line with the insights provided by general research on poverty. With the appearance of “sudden poverty” (Ruminska-Zimny 1997) during the beginning of the economic transition in 1989 to 1991, poverty also became more widespread in urban areas during the 1990s (Gerry et al. 2008). Later, urban poverty was seen to increase in developing countries worldwide, particularly in Latin Amer­ ica (Ravallion et al. 2007). However, the transition economies in Europe did not exhibit such a trend. On the contrary, the number of people in poverty in urban areas there can actually be said to have declined. Furthermore, in the transition econ­ omies the relative difference in the poverty headcount in urban areas in comparison with that in rural areas can be said to have decreased. Given the above, the 1990s can be perceived as a period in which the poverty headcount increased and stabilized at a high level, while the 2000s can be perceived as a period in which the poverty headcount trended downwards.

4.3 DETERMINATION OF THE LITERATURE TO BE SURVEYED: LITERATURE SEARCH PROCEDURES Before performing the meta-analysis for this chapter and surveying literature to form the basis for that, it was first necessary to identify and list the literatures to be sur­ veyed while avoiding subjective selection biases. This chapter used Econlit, a wellknown electronic database of academic literature, to search for literature published in the 27-year period between January 1989 and December 2015.1 To limit the sub­ jects covered, the author searched for words directly related to the topic, such as “poverty” and “poor.” This chapter also used words that could be related such as “disparity (differential),” and used the “and/or” combination function to extract a wide range of literature. In addition, to search for empirical research on the regions this research should cover, the author used “and/or” to search for keywords such as “transition economies,” “Eastern Europe,” and “Central Europe.” Actually, however, it was impossible to track down a sufficient number of papers. A serious problem was the frequent absence of research on specific countries. Fur­ thermore, although predictable given the size of the country, the usability of data,

125

C H A P T E R 4

C H A P T E R 4

ECONOMIC TRANSITION AND POVERTY

and so on, the search results were incredibly skewed toward Russia. Therefore, in addition to the above, the author performed keyword searches (Econlit Subject searches) using “poverty + [specific country name]”, which produced a total of 1,894 (though some were duplicated) papers and academic writings. This enabled one to gather a reasonably wide range of literature on the transition economies of Central and Eastern Europe (see Figure 4.4). One-page news articles, comments concerning already published papers, corres­ pondence among their writers, reviews, and so on were eliminated from the investi­ gation. The author also decided to exclude papers included in books and discussion papers from international organizations and research organizations such as univer­ sities. This reduced the number of papers surveyed, and there is a risk that important papers have been omitted. However, one also should take account of the fact that many papers contained in books have previously been published in academic jour­ nals, with the books containing revised versions of them, and that while academic journals can be expected to maintain certain standards through processes such as peer review, the same level of quality may not be ensured for papers included in books and discussion papers published by research organizations. Another reason for this decision was that the number of book papers involving quantitative investiga­ tions, at least ones covering the regions this chapter was investigating, is limited.

Figure 4.4 The number of papers by target country, by keyword searches using “poverty” and country names of CEE and FSU Note: 1,894 papers in total, though some were overlapped. 1,714 if without overlapping.

126

DETERMINATION OF THE LITERATURE TO BE SURVEYED

4.3

The author also restricted the investigation to literature written in English, ignor­ ing research conducted in Japanese, Russian, and other languages. In that sense, this chapter follows the conventional approach of systematic review (Borenstein et al. 2009). This decision was also aimed at ensuring a certain level of quality for research results. Over half the literature for some of the countries of the former Soviet Union com­ prised discussion papers from international organizations, particularly the World Bank, papers from books, and so on. Although these could not be included in the author’s investigation, the author collected as many of the 1,025 papers from aca­ demic journals as he could (see Figures 4.5, 4.6).2 The number of studies extracted from the database is shown in Figure 4.5, but of the total of 1,025, the author was only able to obtain 647. However, several hundred of the papers published in the countries of Central and Eastern Europe were written in the local language and could not be included in this chapter’s investigation. Figure 4.5 suggests that there was a steady increase in the amount of poverty research in transition economies after the beginning of the economic transition in 1989. However, the database did not yield even one paper published in an academic journal for 1989 and 1990. This may mean that data that had been kept confidential during the socialist era was increasingly made public and that a certain accumulation of data such as household survey data was needed before research could begin. In fact, the increase in research from 2000 may only have been possible once household

Figure 4.5 The number of research article on poverty, targeting transition economies and published in aca­ demic journals, January 1989–December 2015 Note: No hits in 1989 and 1990

127

C H A P T E R 4

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

4

Figure 4.6 The number of poverty studies in general in academic journals, poverty studies on transition econ­ omies in journals, and the ratio of poverty studies in transition economies to poverty studies in gen­ eral, January 1989 – December 2015

survey data was accumulated. However, it may also be necessary to take into account the increase in the number of journals. A comparison with Figure 4.6, which shows the results of a search performed using “poverty” as the keyword with no other restric­ tions (i.e. no specification of the region, etc.), shows that poverty research as a whole increased sharply from the beginning of the 2000s. It can therefore be said that poverty research on transition economies followed the overall trend for poverty research in general. Nevertheless, it cannot be denied that the accumulation of research progressed steadily. And at the same time, as Figure 4.6 shows, research on poverty in transition economies as a proportion of total, non-region-specific poverty research, increased from less than 2% in the mid-1990s (1996) to over 8% by 2015 (Figure 4.6), which probably indicates that the increase in the number of journals was not the only con­ tributor to the increase in poverty research on transition economies. Looking at the position of poverty research in the field of transition economy research in general (see Figure 4.7), it can be seen that not only has transition economy research itself been increasing in quantitative terms, research dealing with the problem of poverty as a proportion of all transition economy research has increased since the end of the 1990s compared with the beginning of the economic transition. It can therefore be said that “poverty” is gathering interest as a research topic in this field. Next, the author read the titles and abstracts of all the papers he had been able to collect, eliminating those on topics that were obviously different. The author limited the literature to be collected here to that dealing with European transition economies. 128

DETERMINATION OF THE LITERATURE TO BE SURVEYED

4.3

C H A P T E R 4

Figure 4.7 The number of studies on transition economies in general in academic journals, poverty studies on transition economies in journals, and the ratio of poverty studies in transition economies to studies on transition economies in general, January 1989–December 2015

In other words, this research did not include the Asian transition countries, China and Vietnam. There were clear reasons for this. First, China and Vietnam did not experience transitional shock and a subsequent recession, something that all the former socialist countries of Eastern Europe and the Soviet Union were faced with. Table 4.2 shows an index of per-capita GDP in transition economies with 1989 as the base year, and these two countries were the only ones that did not see their per-capita GDP drop after 1989 to below the level they were in that year. It is also difficult to imagine that the factors behind the poverty that occurred in those two countries had the same characteristics as those behind the “sudden poverty” that arose in the transition economies of Europe. Furthermore, a search using the key­ words “China” and “poverty” turned up 3,295 pieces of literature. This figure is far higher than the 1,894 pieces from keyword searches specifying the names of all the European transition economies as regions, which the author mentioned earlier, so there is a lack of balance. In other words, “knowledge from poverty research on China” might be over-representative when investigating “knowledge from poverty research in transition economies as a whole.” For the above reasons, the author deemed it inappropriate to deal simultaneously with research covering China and Vietnam in addition to the European transition economies.

129

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

4

Albania Bulgaria China Estonia Georgia Hungary Kyrgyz Republic Latvia Moldova Mongolia Romania Russian Federation Slovakia Tajikistan Turkmenistan Ukraine Uzbekistan Vietnam

Table 4.2 GDP per capita in transition economies, 1989–2010 1989

1990

1995

2000

2005

2010

100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0 100.0

89.5 92.5 102.3 92.9 85.2 97.5 104.5 92.2 97.2 94.9 94.2 96.6 96.9 96.9 98.1 93.4 99.2 103.1

82.4 84.1 171.8 71.6 24.4 86.9 51.0 56.4 39.2 78.6 86.6 60.1 82.6 33.8 53.5 45.2 72.4 140.3

109.8 87.5 247.9 100.7 34.7 101.6 62.4 77.8 35.2 86.1 82.1 65.9 97.2 31.7 60.8 42.8 80.9 182.0

139.7 120.6 382.4 151.5 50.1 126.2 71.3 118.9 50.1 111.5 112.3 90.7 123.5 49.1 124.6 64.7 99.2 246.3

175.3 141.2 633.5 151.6 63.2 126.0 83.1 118.1 59.2 141.1 131.1 108.5 153.9 63.5 194.1 69.9 138.1 327.7

Note: 1989=100.

Source: World Bank, World Development Indicators CD-ROM (2012)

Of all the 647 academic-journal papers that the author was able to obtain, 15 included results of analysis that could be used to perform a meta-analysis of differ­ ences over time and between regions in the determinants of poverty, and these are listed in Table 4.3. So the author was actually only able to extract results of analysis from fewer than 3% (2.3%) of all the pieces of literature. Not all the 647 academic-journal papers described empirical research. Some of them explained policy trends, and many did not actually constitute poverty research.3 There are reasons why the number of studies from which results of ana­ lysis can be extracted is so small, and they will be discussed these here. No systematic review of all the poverty research conducted in transition econ­ omies exists. However, it is necessary to mention Lokshin 2009 as a previous review of poverty research, albeit one limited to Russia. Lokshin adopted the unusual approach of studying only literature written in the Russian language and investigated the methods used for analyzing poverty in Russia as seen through 250 papers published between 1992 and 2006. He found that whereas 48% of 145 empir­ ical studies published in the top nine American economic journals in 1965 carried out some kind of regression analysis and performed statistical testing by providing standard errors, only 12% of 250 empirical studies in economics published in Rus­ sian journals between 1992 and 2006 carried out a regression analysis, and only 8% of them provided standard errors and performed testing (see Table 4.4). Lokshin’s conclusion was that given external criteria such as whether a paper fea­ tures regression analysis or reports standard errors, it was difficult to say that the

130

Dimova and Wolff

Rhoe, Babu and Reidhead

Gerry, Nivorozhkin and Rigg

Robinson and Guenther

2008

2008

2008

2007

Tajikistan

Russia

Kazakhstan

Bulgaria

Poland

Szulc

2009

2008

Target areas

Brück, Danzer, Ukraine Muravyev and Weisshaar Mills amd Russia Mykerezi

Author

2010

Published year

2003

2004

1996

1995, 97, 2001

2000

1994–98 2000–03

1996

Estimated period

Logit

Logit

Logit

Probit

Probit

Tobit

Tobit

Tobit

Probit

Methods

6

7

2

3

3

6

4

The number of estimation results to be utilized

no − + − − + + − − +

Urban residence Higher education Risk of poverty Number of children Urban residence Higher education Risk of poverty Rural residence Number of children Household size Higher education Risk of poverty Dependency ratio

Significance + + + + − − + − − + − − + − − +

Explaining variables

Risk of poverty Household size Unemployment Urban residence Poverty Ratio Number of children Urban residence Higher education Poverty Ratio Number of children Urban residence Higher education Poverty Ratio Number of children Urban residence Higher education Risk of poverty Number of children Urban residence Higher education Risk of poverty Number of children

Explained variables

Table 4.3 Papers, the results of which would be utilized in meta-analysis: Explained variable—poverty risk/poverty ratio

(Continued )

665–992

53970

1996

2319–2633

35952

2156

2146

1288

22990

Number of samples

DETERMINATION OF THE LITERATURE TO BE SURVEYED 4.3

131

C H A P T E R 4

132

Alexandrova, Hamilton and Kuznetsova

Szulc

Kolev

Bezemer and Lerman Gustafsson and Nivorozhkina

2006

2006

2005

2004

1999

Russia

Armenia

Bulgaria

Poland

Russia

Russia

Bosnia and Herzegovina

Target areas

Commander, Russia Tolstopiatenko and Yemtsov

Bhaumik, Gang and Yun

2006

2004

Kristic and Sanfey

Author

2007

Published year

Table 4.3 contd.

1992–93

1989, 2000

1998

2001

1993, 99

2002

2000

2001–04

Estimated period

Probit

Probit

Ligistic Regression Logit

Probit

Probit

Probit

Probit

Probit

Methods

2

3

1

4

12

1

2

1

The number of estimation results to be utilized

1131 4700

4700

− + + − −

Risk of poverty Higher education Dependency ratio Permanent Dependency ratio poverty Higher education Permanent Dependency ratio non-poverty Higher education

+

1187

2411 1225 1458

32000

3905

416–2101

915

Number of samples

− +

+ − − + + + +

Number of children Risk of poverty Urban residence Higher education Number of children Risk of poverty Existence of children Risk of poverty Existence of children Risk of poverty Household size

4

Risk of poverty Higher education Dependency ratio

− − no − + − −

Urban residence Higher education Risk of poverty Urban residence Higher education Number of children Risk of poverty Urban residence Higher education

Significance +

Explaining variables

Risk of poverty Household size

Explained variables

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

META-ANALYSIS OF POVERTY RESEARCH

4.4

Table 4.4 Reporting style of empirical studies: Journal articles in the US in 1965 vs. Russian journal articles in Russia, 1992–2006 US, 1965

Russia, 1992–2006

100% 53% 48%

75% 8% 12%

C H A P T E R 4

Parameter estimation Report of standard errors Regression analysis Source: Lokshin (2009, Table 3)

poverty studies in Russia met the normal standards for poverty research. Limiting the investigation to literature in the English language means that the final research results tend to also appear as literature in English, which reaches a wider number of readers, so it can be said to be the normal method for meta-analysis (Borenstei et al. 2009). Additionally, Lokshin’s view can be said to support the approach of this chapter, which is to conduct a review focusing on literature in English only. It is also understood that it is quite possible, as was the case with this chapter, that only 2.3% of studies retrieved using the keyword “poverty” include content that can be used for meta-analysis.4

4.4 META-ANALYSIS OF POVERTY RESEARCH IN TRANSITION COUNTRIES The meta-analysis this chapter will perform will be to synthesize partial correlation coefficients (PCCs) and t values. This chapter will synthesize PCCs using the fixedeffect model and random-effects model and determine combined values to be referred by testing for homogeneity (Borenstein et al. 2009). Regarding t values, the author will determine the weights using rankings, impact factors, and so on,5 and present them as combination of t values with or without weighting. Furthermore, by calculating failsafe N (Mullen 1989) at the significant level of 5%, the author will confirm the confidence for the integrated t values calculated here. What needs to be undertaken when performing a meta-analysis is an investigation relating to publication selection (Mullen 1989). In this chapter the author produces a funnel plot to check publication selection bias (PSB). Then the author performs the analysis by estimating a meta-regression model to confirm the existence of true (genuine) effect.6 What one must mention first is the difficulty of grasping poverty dynamics using “transition factors.” In the case of macro-level themes such as the study of economic policy or path dependence, variables such as the degree of progress with privatiza­ tion or the European Bank for Reconstruction and Development’s progress in transi­ tion indicators can also be regarded as explanatory variables. However, to understand the phenomenon of poverty at the individual or household level, such

133

C H A P T E R 4

ECONOMIC TRANSITION AND POVERTY

factors cannot be used in an approach that measures progress in the economic transi­ tion. Having said that, phenomena such as the privatization (shift to private owner­ ship) of housing, at least in the case of Russia, occurred throughout the country at more or less the same time. Basically what happened was that the ownership of the apartments that people lived in at the time was just handed over to their owners almost free of charge. Factors that occur for all agents simultaneously cannot be explanatory variables for phenomena that occur subsequently at the individual level. However, if one traces individual studies, the variables employed in them are the main variables that are widely used in poverty research (including research on devel­ oping countries). In other words, the variables are the education level of wage earn­ ers, the genders of the highest wage earners, whether the household is located in a rural or an urban area, the number of children, the number of pensioners, the industries in which the wage earners work, ethnicity, and so on. Household surveys such as the Russia Longitudinal Monitoring Survey (RLMS) allow ownership (nationalized, privately owned, foreign owned, etc.) of companies at which house­ hold members work to be observed, but no papers employing such attributes as explanatory variables could be found. So here the author will instead investigate how the phenomena changed during the economic transition and whether different phenomena appeared depending on the specific region. This will be based on the examination of the poverty level in transi­ tion economies that this chapter looked at in Section 4.1. This is the recognition that first, the poverty problems in the 1990s and those from 2000 onwards may have been of a different nature (Figure 4.1). Furthermore, the nature of poverty in regions that belonged to the former Soviet Union and the nature of poverty in other regions —that is, Central and Eastern European countries—may have differed (Table 4.1). The above determines a direction for classifying previous research. In addition to synthesizing/combining the results of all the studies, the author will focus on the dif­ ferences of whether the studies concerned former Soviet republics or Central and Eastern European countries and whether they covered the 1990s or the 2000s by synthesizing/combining the data for each separately.7 Furthermore, regarding the explained variable, the author focuses on studies that determine a fixed poverty line and use the qualitative variable of regarding households below that line as having fallen into poverty as the explained variable. The results are shown in Table 4.5. For almost all the analyses, the null hypothesis relating to the assumption of homo­ geneity is rejected, so this chapter will look at the results of the random-effects model. Here the author will discuss Table 4.5, about the synthesis of PCCs first. When all the studies are synthesized, increases in the education level of wage earners reduce the probability of falling into poverty, increases in household size raise the poverty risk, and households located in rural areas are more likely to fall into poverty. These results are fairly typical. The analytical results extracted here are all based on micro data, and simply confirm the understanding obtained not just from studies on transition econ­ omies, but from a wide range of other studies. What the author wants to focus on, however, is the differences when data for the 1990s and the 2000s, and data for the Soviet Union and Central and Eastern Europe 134

Fixed-effect model a

All studies Household 56 0.110*** size Higher 46 −0.050*** education Rural 43 0.044*** residence Soviet Union vs. Central and Eastern Europe. Soviet Union Household 31 0.073*** size Higher 25 −0.063*** education Rural 22 0.063*** residence Central and Eastern Europe Household 25 0.020*** size Higher 21 −0.050*** education Rural 21 0.030*** residence

Number of estimates (K)

Table 4.5 Meta-synthesis of estimates

809.57*** 2152.08*** 1924.09***

131.53*** 183.19*** 790.30***

202.07*** 1898.29*** 909.20***

−0.069*** 0.025***

0.069*** −0.078*** 0.035***

0.030*** −0.059*** 0.015

Test of homogeneity b

0.067**

Randomeffects model a

(a) Synthesis of PCCs

5.94***

−30.57***

15.44**

23.60***

−29.60***

58.18***

28.33***

−42.48***

30.17***

Unweighted combination

0.86

−4.43***

2.36**

3.53***

−4.36***

8.98***

4.17***

−6.21***

4.63***

Weighted combination

1.98

−1.98

1.98

1.75

−4.03

2.85

1.98

−3.37

1.98

Median of t values

(b) Combination of t values

(Continued )

252

7158

2177

4505

8070

9621

17702

30623

19455

Failsafe N (fsN)

META-ANALYSIS OF POVERTY RESEARCH 4.4

135

C H A P T E R 4

136

−0.051***

0.064*** −0.093***

−0.018*** 0.011**

0.060*** −0.088*** 0.070***

26

24

30

20

19

0.043***

0.010

0.036***

0.017***

26

Randomeffects model a

Notes

a Null hypothesis: The synthesized effect size is zero.

b Null hypothesis: Effect sizes are homogeneous.

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

1900s vs. 2000s 1990s Household size Higher education Rural residence 2000s Household size Higher education Rural residence

Fixed-effect model a

1044.69***

934.11***

217.94***

57.93***

297.92***

211.96***

Test of homogeneity b

(a) Synthesis of PCCs

36.10***

−45.10***

59.56***

5.79***

−16.90***

14.69***

Unweighted combination

5.79**

−7.06***

10.16***

0.80

−2.36**

2.05*

Weighted combination

4.28

−5.64

2.85

1.98

−1.98

1.98

Median of t values

(b) Combination of t values

9133

15021

9398

695

2705

2204

Failsafe N (fsN)

4

Number of estimates (K)

Table 4.5 contd.

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

DETECTION OF PUBLICATION SELECTION BIAS

4.5

are synthesized separately. In the 1990s, households in rural areas were no more likely to fall into poverty than those in urban areas. In the 2000s, however, a rural location increased the probability of households falling into poverty. Differences could also be seen when data was synthesized separately for countries that comprised the Soviet Union and countries in Central and Eastern Europe. The above findings applied to Central and Eastern Europe. In other words, in Central and Eastern Europe rural location did not raise the probability of poverty. What needs to be pointed out here is that this result is not due to extreme bias in the sample. When the author checked the effect of the rural domicile variable on poverty probability in the 1990s, the author synthesized the results of 24 analyses, and 10 of these were for countries that comprised the former Soviet Union. Whichever the case, the same can be said concerning the combination of t values. When combining data without weighting them by taking into account third-party evaluations of the academic journals in which papers were published, all variables were significant for all combinations. However, combination of t values that had been weighted were always smaller than those that had been unweighted, and were no longer significant in the above two cases. Failsafe N was fairly large in every case, which can be said to indicate a high level of confidence in the estimated results for the combined t values. The above results indicate that in the 1990s households in urban areas and rural areas had an equal likelihood of falling into poverty, and this situation was due to the transition economies being hit with a recession that occurred in conjunction with the change in the economic system. Compared with that of those in urban areas, the probability of households in rural areas falling into poverty was relatively higher in the countries that comprised the Soviet Union than those in Central and Eastern Europe. However, this situation changed in the 2000s, a phenomenon described by Gerry et al. as a “ruralization of poverty” (2008). It may be said that this, in a sense, describes the process through which the economic turmoil that accompanied transi­ tion came to an end.

4.5 DETECTION OF PUBLICATION SELECTION BIAS AND PRESENCE OR ABSENCE OF GENUINE EFFECT Finally, to check for the existence of PSB, the author will confirm the funnel plots. Additionally, meta-regression analysis will be performed in order to check the presence of genuine effect. Figure 4.8 shows funnel plots of the results of estimat­ ing the impact of each factor on poverty probability. It is difficult to determine whether the plots are horizontally symmetrical or triangular. Therefore, to verify whether PSB exists or not the author will make estimates using a meta-regression model concerning the existence of PSB and the existence of genuine effect. The method follows that explained in Chapter 1.

137

C H A P T E R 4

C H A P T E R

ECONOMIC TRANSITION AND POVERTY

4

For the above estimates, the author will also use the least-squares method, cluster-robust OLS estimation, and unbalanced-panel estimation to confirm the robustness of the results. The results are shown in Tables 4.6, 4.7, and 4.8 Here, the explained variables of poverty probability, the author produced funnel plots concerning three variables (number of family members, education level, and urban domicile), and also made estimates for all of them using a meta-regression model for PSB and genuine effect. According to these results, as is shown in Panels a and b of Tables 4.6, 4.7, and Figure 4.8 Funnel plots for estimation results of various 4.8, except in the case of Table 4.8 (whether rural domicile affects the prob­ factors on the risk of poverty/poverty ratio ability of poverty), the null hypothesis that the intercept β0 is zero is rejected, indicat­ ing that publication selection exists. Regarding the genuine effect, however, in Panel a of Tables 4.6, 4.7, and 4.8 the null hypothesis that the coefficient β1, the reciprocal of the standard error, is zero is rejected, and as shown in Panel c of each table, the coefficient β1, the reciprocal of the standard error, is estimated significantly in at least two of the three models. Therefore, regarding the probability of a household falling into poverty, it can be said that household size and education level have a genuine effect, the former positive and the latter negative. Where poverty “probability” (a two-value variable relating to whether income lies below a fixed poverty line), which attempts to grasp poverty directly, is the explained variable, a genuine effect can be detected with all three models. There is also the problem that PSB has not been eliminated. However, it can probably be said that the results strongly suggest that the factors of household size, education level, and urban

138

CONCLUSIONS

4.6

Table 4.6 Meta-regression analysis on publication biases and the existence of genu­ ine effects of household size on poverty risks: Comparable with Figure 4.8a (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 4

Estimation Model

OLS [1]

Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

2.43** 0.046** 56 0.5

Cluster-robust OLS [2]

Random-effects Panel GLS [3] a

2.43** 0.046** 56 0.5

2.74** 0.043** 56 0.5

Cluster-robust OLS [5]

Random-effects Panel GLS [6] b

2.44** 0.047** 56 0.56

2.69** 0.045** 56 0.56

Cluster-robust OLS [8]

Random-effects Panel ML [9]

0.17** 0.061** 56 0.63

0.103 0.05** 56 −

(b) Test of type II PBS (Equation: |t |=β0+β1(1/SE)+v)

Estimation Model

OLS [4]

Intercept (H0: β0=0) 1/SE K R2

2.44** 0.047** 56 0.56

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Estimation Model

OLS [7]

SE 1/SE (H0: β1=0) K R2

0.17** 0.061** 56 0.63

Notes: a Breusch-Pegan test: χ2=11.13, p=0.00; Hausman test: χ2=2.13, p=0.14 b Breusch-Pegan test: χ2=9.28, p=0.001; Hausman test: χ2=1.20, p=0.27 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coeffi­ cient at the 1%, 5%, and 10% levels, respectively.

domicile, which have been dealt with in this chapter and also investigated in numerous other studies of poverty in transition economies, certainly have an effect on the probabil­ ity of individual households falling into poverty.

4.6 CONCLUSIONS Taking into account the relationship with macro-indicators and research trends in the more than 20 years since transition began, with regard to poverty research in the countries that formerly comprised the Soviet Union and countries in Central and

139

C H A P T E R

ECONOMIC TRANSITION AND POVERTY Table 4.7 Meta-regression analysis on publication biases and the existence of genu­ ine effects of educational attainment on poverty risks: Comparable with Figure 4.8b (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

4 Estimation Model

OLS OLS [1]

Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

−2.62** −0.046** 46 0.37

Cluster-robust Panel GLS [2]

Random-effects

−2.62* −0.046** 46 0.37

−6.84* 0.0057** 46 0.37

Cluster-robust OLS [5]

Random-effects Panel GLS [6] b

[3] a

(b) Test of type II PBS (Equation: |t |=β0+β1(1/SE)+v)

Estimation Model

OLS [4]

Intercept (H0: β0=0) 1/SE K R2

2.62** 0.046** 46 0.37

2.62** 0.046 46 0.37

6.84** −0.0057 46 0.37

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Estimation Model

OLS [7]

SE 1/SE (H0: β1=0) K R2

−4.82** −0.059** 46 0.61

Cluster-robust OLS [8] −4.82** −0.059** 46 0.61

Random-effects Panel ML [9] 1.33 0.019 46 −

Notes: a Breusch-Pegan test: χ2=8.78, p=0.001; Hausman test: χ2=1.20, p=0.27 b Breusch-Pegan test: χ2=8.79, p=0.002; Hausman test: χ2=1.20, p=0.27 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coeffi­ cient at the 1%, 5%, and 10% levels, respectively.

Eastern Europe, this chapter has verified the results of empirical research on the fac­ tors that determine the poverty situation of households by combining them using a basic meta-analytical approach. Research on poverty in this region, which increased as the socialist system col­ lapsed, began shortly after the economic transition began. However, the nature of poverty in the former Soviet Union and Central and Eastern Europe differed, and two phases were observed: a phase of increasing and stabilizing poverty in the 1990s and a phase of declining poverty in the 2000s. Unfortunately, it was impos­ sible to locate any previous research employing transition factors as explanatory variables, so the author attempted a meta-analysis of the impact of household size, 140

CONCLUSIONS

4.6

Table 4.8 Meta-regression analysis on publication biases and the existence of genu­ ine effects of rural residence on poverty risks: Comparable with Figure 4.8c (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 4

Estimation Model

OLS [1]

Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

−0.73 0.093** 43 0.59

Cluster-robust OLS [2]

Random-effects Panel GLS [3] a

−0.73 0.093** 43 0.59

−0.90* 0.094** 43 0.59

Cluster-robust OLS [5]

Random-effects Panel GLS [6] b

0.41 0.087** 43 0.58

0.39 0.087** 43 0.58

Cluster-robust OLS [8]

Random-effects Panel ML [9]

2.35 0.087** 43 0.68

2.00 0.088** 43 −

(b) Test of type II PBS (Equation: |t |=β0+β1(1/SE)+v)

Estimation Model

OLS [4]

Intercept (H0: β0=0) 1/SE K R2

0.41 0.087** 43 0.58

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Estimation Model

OLS [7]

SE 1/SE (H0: β1=0) K R2

2.35 0.087** 43 0.68

Notes: a Breusch-Pegan test: χ2=4.48, p=0.017; Hausman test: χ2=0.50, p=0.48 b Breusch-Pegan test: χ2=3.28, p=0.0035; Hausman test: χ2=0.09, p=0.76 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coeffi­ cient at the 1%, 5%, and 10% levels, respectively.

education level, and urban domicile, which are factors employed in traditional pov­ erty research, taking into account the possibility that their impact may differ depend­ ing on the period or the region. The results generally supported the hypothesis. In the 1990s, there was no difference between urban and rural populations in the probability of falling into poverty. After 2000, however, urban domicile became a significant factor in redu­ cing the probability of falling into poverty. In addition, differences were observed between the former Soviet Union and Central and Eastern Europe in factors affecting the poverty situation. Identification of causes of these differences was beyond the scope of this chapter, but this phenomenon is considered to

141

C H A P T E R 4

ECONOMIC TRANSITION AND POVERTY

indicate one of the directions for research in comparative transition economics in the future. At the same time, however, one must also mention the problem that PSB was detected in all the cases the author put together in Section 4.5 to verify its presence. This may suggest that advances in poverty research in transition economies remains insufficient. However, it was shown that household attribute factors exert a genuine effect on the poverty situation. Although it must be recognized that poverty research in countries that went through the economic transition has not investigated the effects of transition factors directly, the trend with the previous research examined here, which has been to expand the applicability of poverty-level determinants that are employed in stylized household analysis, can probably also be regarded as indi­ cating steady progress in “transition.”

ACKNOWLEDGMENTS This chapter is a substantially revised version of Kumo 2016. In addition to the financial and technical supports acknowledged in the preface, this study is financially supported by the Japan Securities Scholarship Foundation (2018).

NOTES 1 Information from books and journals are not included in the Econlit database as soon as they are published. Taking into account the lag between the publication of information and its inclusion in the database and the reproducibility of the ana­ lysis performed in this chapter, literature published up to approximately one year before this chapter was written is covered. 2 This will be discussed later, but the literature subject to be used for the metaanalysis was not selected arbitrarily. Instead, all the analytical results that could be obtained were collected. 3 The searches were keyword searches, with JEL (Journal of Economic Literature) codes also added. The applicable codes were I300/I320/I390, P360, and P460, which cover subjects such as welfare and consumer economics. The 647 papers retrieved included a lot of papers focused mainly on analysis of education, pen­ sions, and medical care. 4 To give another example, if papers published between January 1989 and Decem­ ber 2015 are searched for using the keywords “poverty” and “Russia,” of these 20% were in the Russian language, 17% were published in the journal Problems of Economic Transition, and 10% were discussion papers. The papers in Russian and the discussion papers were excluded from the study, but 25% of the remain­ ing papers were published in the journal Problems of Economic Transition. This journal is not a typical scientific journal. Instead, its stated role is to describe the

142

REFERENCES

current state of economic research within Russia by carrying English translations of papers published in Russian-language journals. Poverty research in Russia itself is as described by Lokshin (2009), so it is extremely rare for analytical papers to be featured in Problems of Economic Transition. 5 This follows the methods presented in Chapter 1 of the volume for assessing the standard of research. 6 These methods are as defined in Chapter 1. 7 As can be seen from the “estimate period” in Table 4.3, though this was com­ pletely unintentional, the period of analysis of all the studies can be classified as either 1990s or 2000s onwards, as none of them covered both periods.

REFERENCES Alexandrova, Anastassia, Ellen Hamilton, and Polina Kuznetsova (2006) What can be learned from introducing settlement typology into urban poverty analysis: The case of the Tomsk region, Russia. Urban Studies, 43(7), pp. 1177–1189. Bezemer, Dirk, and Zvi Lerman (2004) Rural livelihoods in Armenia. Post-Communist Econ­ omies, 16(3), pp. 333–348. Bisogno, Marcelo, and Alberto Chong (2001) Foreign aid and poverty in Bosnia and Herze­ govina: Targeting simulations and policy implications. European Economic Review, 45 (4–6), pp. 1020–1030. Bobkov, Vyacheslav (2007) (ed.), Quality and Standard of Life of the Population in New Russia. All-Russian Center for Living Standards: Moscow (Russian). Borenstein, Michael, Larry Hedges, Julian Higgins, and Hannah Rothstein (2009) Introduc­ tion to Meta-Analysis. Wiley: Chichester. Braithwaite, Jeanine (1995) The old and new poor in Russia: Trends in poverty. ESP Discus­ sion Paper Series 21227, World Bank: Washington DC. Braithwaite, Jeanine, Christiaan Grootaert, and Branko Milanovic (2000) Poverty and Social Assistance in Transition Countries. Macmillan: London. Brück, Tilman, Alexander Danzer, Alexander Muravyev, and Natalia Weisshaar (2011) Pov­ erty during transition: Household survey evidence from Ukraine. Journal of Comparative Economics, 38(2), pp. 123–145. Commander, Simon, Andrei Tolstopiatenko, and Ruslan Yemtsov (1999) Channels of redis­ tribution: Inequality and poverty in the Russian transition. Economics of Transition, 7(2), pp. 411–447. Dimova, Ralitza, and François-Charles Wolff (2008) Are private transfers poverty and inequality reducing? Household level evidence from Bulgaria. Journal of Comparative Economics, 36(4), pp. 584–598. Gerry, Christopher, Eugene Nivorozhkin, and John Rigg (2008) The great divide: ‘Ruralisation’ of poverty in Russia. Cambridge Journal of Economics, 32(4), pp. 593–607. Gustafsson, Bjorn, and Ludmila Nivorozhkina (2004) Changes in Russian poverty during transition as assessed from microdata from the city of Taganrog. Economics of Transi­ tion, 12(4), pp. 747–776.

143

C H A P T E R 4

C H A P T E R 4

ECONOMIC TRANSITION AND POVERTY Kolev, Alexandre (2005) Unemployment, job quality and poverty: A case study of Bulgaria. International Labour Review, 144(1), pp. 85–114. Kristic, Gorana, and Peter Sanfey (2007) Mobility, poverty and well-being among the infor­ mally employed in Bosnia and Herzegovina. Economic Systems, 31(3), pp. 311–335. Kumo, Kazuhiro (2016) Research on poverty in transition economies: A meta-analysis on changes in the determinants of poverty. Transition Studies Review, 23(1), pp. 37–59. Lokshin, Michael (2009) A survey of poverty research in Russia: Does it follow the scientific method? Economic Systems, 33(3), pp. 191–212. McAuley, Alistair (1979) Economic Welfare in the Soviet Union: Poverty, Living Standards, and Equality. University of Wisconsin Press: Madison. Milanovic, Branco (1997) Income, Inequality, and Poverty during the Transition from Planned to Market Economy. World Bank: Washington DC. Mills, Bradford, and Elton Mykerezi (2009) Chronic and transient poverty in the Russian Federation. Post-Communist Economies, 21(3), pp. 283–306. Mullen, Brian (1989) Advanced Basic Meta-Analysis. Lawrence Erlbaum Associates: Hills­ dale, NJ. Ravallion, Martin, Shaohua Chen, and Prem Sangraula (2007) New evidence on the urban­ ization of global poverty. World Bank Policy Research Working Paper No. 4199, World Bank: Washington DC. Razumov, Aleksandr, and Mariya Yagodkina (2007) Poverty in Modern Russia. Law For­ mula: Moscow (Russian). Rhoe, Valerie, Suresh Babu, and William Reidhead (2008) An analysis of food security and poverty in Central Asia: Case study from Kazakhstan. Journal of International Develop­ ment, 20(4), pp. 452–465. Robinson, Sarah, and Tanya Guenther (2007) Rural livelihoods in three mountainous regions of Tajikistan. Post-Communist Economies, 19(3), pp. 359–378. Ruminska-Zimny, Ewa (1997) Human poverty in transition economies: Regional overview for HDR 1997. Human Development Report Office, United Nations Development Programme: New York. Szulc, Adam (2006) Poverty in Poland during the 1990s: Are the results robust? Review of Income and Wealth, 52(3), pp. 423–448. Szulc, Adam (2008) Checking the consistency of poverty in Poland: 1997–2003 evidence. Post-Communist Economies, 20(1), pp. 33–55. World Bank (2004) From Transition to Development: A Country Economic Memorandum for the Russian Federation. World Bank: Washington DC.

144

5

Social confusion and corruption Investigating the causes and effects of a breakdown of ethics Taku Suzuki and Satoshi Mizobata

5.1 INTRODUCTION In examining the fundamental ethics of human society, Jane Jacobs (1992) looked at the types of moral codes of market ethics and government ethics, pointing out that, in the case of the latter, in particular, transaction avoidance becomes an important factor. In doing so, she both warned against corruption and identified the presence of corruption that is difficult to eradicate amid a mixture of both moral (ethical) codes. It goes without saying that the relationship between these two moral codes brings to mind the contrast between market principles and organization. As a result, while in a democratic society there is a need for a conscious choice between the two moral codes, a socialist economic system under party or state control requires ethics on the part of rulers. Together with stressing the ethical breakdown of communication states, Jacobs asserted: “Former Marxist societies, as they seek to reconstitute themselves, desperately need to clarify right and wrong in business and politics” (1992, p. 446). Truly, in addition to consider­ ing corruption in transition economies to be a phenomenon that was fostered within social systems, we must also look at how values change in preparation for the posttransition economy. Just what is meant by corruption in the first place? Corruption is an ambiguous, broad-ranging concept that encompasses a wide range of matters, as seen in the way it has been described to include the concepts of “fraud, embezzlement, theft, nepotism, cronyism, gifts, tips, donations, clientelism, connections, networks, lobbying, bargain­ ing, mafioso protection rackets, patronage, conflict of interest, kleptocracy” (Offe 2004, p. 77). Accordingly, when seeking a highly versatile conceptual rule of thumb on the subject, Nye’s definition (1967) comes to mind: corruption is “behavior which deviates from the formal duties of a public role … because of private-regarding … wealth or status gains” (Offe 2004, p. 77). This definition includes bribery and nepo­ tism but not acts offensive at a moral level, such as murder of the opposition. At the very least, the process of securing private gains in connection with official duties can only be described as corruption, as in a case of deviating from legal frameworks and systems, including rent-seeking, and “privatizing” state power (ibid., p. 79). Other

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

views of corruption, as extensions of this definition, include a “misuse of public office for private gain” (Treisman 2007, p. 360) or, as advocated by the private think-tank Transparency International,1 “the abuse of entrusted power for private gain.” The widely used international Corruption Perceptions Index (CPI) relies on this last defin­ ition. Based on its monetary amount of losses and the sector of its origins, corruption can be categorized as large-scale corruption, minor corruption, and political corruption. Since its opposite, transparency, concerns rules, plans, processes, and actions, corrup­ tion can be positioned in terms of systematic research aspects. That is, politically speaking, corruption is an impediment to democracy and the rule of law, and, econom­ ically, it decreases national wealth, distorts fair market structures and competition, des­ troys social structures through the loss of people’s trust, and worsens the environment through deficiencies in environmental legal and regulatory systems (Transparency International: www.transparency.org/what-is-corruption#define; Rose-Ackerman and Palifka 2016). Under the assumptions of the specific field of research into transition economies that looks at the transition of systems, corruption comes into view as an extremely complex presence. While corruption did exist as a legacy of a bureaucratic socialist economic system, at the same time, it has taken a new form as the corruption of the market’s moral code. That is, corruption during the transition of systems appears as both a legacy and a collateral development of the transition itself. Furthermore, even if corruption causes both political and economic losses, “under specific conditions”— when, in the process of nation-building (e.g., during the transition of systems), build­ ing market systems on top of the legacy of a socialist economic system that had con­ formed excessively to the bureaucracy—“corruption even improves economic outcomes” (Rose-Ackerman and Palifka 2016, p. 32). This is because corruption makes it possible to reduce transaction costs by avoiding excessive bureaucratic sys­ tems. This is the hypothesis that corruption is a form of greasing the wheels. Campos et al. (2010) systematically reviewed research on corruption (based on quantitative evi­ dence), focusing on the efficacy of the greasing-the-wheels hypothesis. Their metaregression analysis of 460 estimation results extracted from 41 studies showed that 32% did not support the greasing-the-wheels hypothesis, 62% were unrelated to it, and 6% supported it. While existing studies as a whole did not support the greasing-the­ wheels hypothesis, Campos et al. (2010) also suggested that strongly policy-oriented studies and unpublished studies were less likely to support the hypothesis, and when these studies were excluded, the results were ambiguous. From an early stage, the issue of corruption in economies with transitioning systems was proposed as a research subject, and such work has focused mainly on macro- and micro-studies. Macro-studies have included an approach that considers macroeconomic performance during the 1990s to have been underestimated and argues that unofficial economic sectors, including corruption, need to be assessed properly. As a result, according to this approach, actual economic performance during the shock of transition to a market economy was better than it appeared (Lavigne 1995). Micro-research has advanced in a more broad-ranging and complex manner.2 Above all, it was a natural development that such research would be addressed as a focus for researchers on the 146

CORRUPTION LEVELS IN TRANSITION ECONOMIES

5.2

transition of systems, since phenomena equivalent to the selling off of state property appeared in the process of privatizing ownership and management—which should be seen as the star policy of the transition of systems—and as the transition of systems was accompanied by a lack of transparency in reorganizing the bureaucratic structures responsible for approving and authorizing building in the new state. For this reason, the study of corruption also served as proof that research on the transition of systems was developing in an interdisciplinary manner, to encompass not only economics but also other fields, such as political science and sociology. Despite the fact that corruption has been studied considerably in connection with sys­ temic transmission, the findings of such research have not been consistent—in fact, they show a tendency to diverge. Even on the subject of its relationship to growth, the metaanalysis of Campos et al. (2010) does not suffice. There is a need to ascertain whether their conclusions are valid, even in consideration of the special conditions of system tran­ sition. Accordingly, this chapter will attempt a systematic review of corruption in coun­ tries undergoing the transition of systems, using a basic collection of 558 studies3—more than the number used by Campos et al. (2010). In other words, through an unprecedented large-scale systematic review, it will consider the causes and effects of corruption in tran­ sition economies. In addition, the author will verify the correlation between the content of research and the literature attribute of medium of publication to attempt to get a prospect of future debates regarding corruption by considering trends in research on the subject. This chapter will employ the following structure: after first taking an overview of the levels of corruption in transition economies, it will propose theoretical hypotheses for consideration in the systematic review. Then, it will consider the attributes of the basic collection of 558 works, from the literature subject to the systematic review, and then successively test hypotheses concerning causes and effects.

5.2 CORRUPTION LEVELS IN TRANSITION ECONOMIES Although research on corruption under the transition of systems began at the same time as the transitions themselves, only since 2000 has the topic of corruption secured its status within transition research. Initially, the publication by the NGO Transparency International of its Corruption Perceptions Index4 served as the major impetus behind the shift from research inclined towards case studies to empirical research. This is related to the fact that this index began to be used in analysis as an indicator of the degrees of market maturity and transition to a market economy. Dis­ tinguishing features of research on corruption include the facts that research has been led by an international organization rather than a specific individual, that research has advanced since the 2000s, and that the number of quantitative studies increased with the use of the above index in research. Despite the fact that it is difficult to compare Corruption Perceptions Index rankings over the years due to differences in the populations of countries surveyed and the fact that it would be difficult to say that the evaluation criteria used necessarily are objective,

147

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

trends in the index over time do, to some degree, aptly express the properties of coun­ tries undergoing economic transitions. In fact, as Figure 5.1 shows, the countries of Central and Eastern Europe and the Baltics, which successfully joined the EU, have held relatively high positions since the start of the transitions of their systems. In particular, Estonia, which received aid as a model for transition in Europe after quickly stabilizing its currency, is ranked highest among transition economies, at a level that rivals even developed countries. After Estonia comes a series of Central and Eastern European states, among which Poland shows a rising trend while Hungary’s is falling. Next come the EU member states in Southeastern Europe, followed by Southeastern Europe’s non-EU member states. At the lowest level are the former Soviet states, with the low rankings of Russia and the regions of Central Asia and the Caucasus standing out in particular. In general, it must be said that the low levels of corruption in these former Soviet states rival those in Africa. China’s level is in between those of southeast­ ern Europe and the former Soviet Union. It must be noted, however, that the above facts do not mean that the Eastern European states ranked higher in terms of corruption control are ranked among the countries with

Figure 5.1 Change of Corruption Perceptions Index rankings Note: This integrated indicator of corruption, developed in 1995, rated 180 countries in 2018 on a scale of 0 (the highest degree of concern about corruption) to 100 (the lowest). For more details, see n. 3. Source: Transparency International

148

THEORETICAL HYPOTHESES REGARDING CORRUPTION ISSUES

5.3

the lowest levels of corruption globally. According to the Research and Training Insti­ tute of the Ministry of Justice of Japan (2008), the annual rates of victimization by cor­ ruption (i.e., the percentage of survey subjects who reported having encountered corruption by public officials) in 2003/2004 averaged 1.9% in the OECD and was 1.0% or less in the major EU member states, but they were quite high in Poland (4.4%) and Hungary (4.9%). In addition, an awareness survey (on corruption in administrative agen­ cies) by the International Social Survey Programme (ISSP) (2016) showed that, while it must be said that levels were low in Central and Eastern Europe as compared to the original EU member states, corruption levels were relatively high in transition econ­ omies as a whole.5 However, even among these transition economies, where corruption levels are relatively high by global standards, there is a marked range in degrees of corruption. These differences are characterized by lower degrees of corruption in Central and Eastern European states, where the continual period under socialism was shorter, the transition began sooner, and the transplanting of European systems was advanced. However, corruption was high in the former Soviet states, particularly Russia and the Central Asian and Caucasus regions, whose continual periods under socialism were longer, as were their periods of transition, which advanced the formation of their own unique systems. These differences can be restated, largely unchanged, as differences in economic growth and the degree of transition to a market economy.6 Since corruption is a crime, no studies provided ethical support for it, based on the assumption that differences would be apparent in corruption, degree of transition, and economic performance among transition economies. The theoretical hypotheses to be tested in this chapter are described in the following section.

5.3 THEORETICAL HYPOTHESES REGARDING CORRUPTION ISSUES Before proposing the hypotheses, the difficulty of empirically identifying factors that lead to corruption must be pointed out. This is because acts of corruption and other endogeneities of economic and social activities are extremely acute. For example, the relationship between economic growth and corruption could be hypothesized either as one in which economic growth results in lower corruption or one in which a low level of corruption itself leads to economic growth. Put another way: [Corruption’s] many likely determinants interrelate in complicated ways. Some can change quickly and may be caused by corruption as well as the reverse. As with other types of criminal activity, it is hard to observe directly, and so researchers must rely on surveys of corruption’s victims, the accuracy of which is often difficult to assess. … Recent years have seen some major advances. (Treisman 2007, p. 393)

149

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

This has been based on improved data.7 Treisman’s studies (2000, 2007) tested hypotheses concerning the correlation between corruption and economic phenomena in the most systematic way. He identi­ fied and tested 12 hypotheses regarding the correlation between corruption on one hand and legal and political systems and economic growth on the other. The hypoth­ eses are grounded in considerations such as the legacy of colonialism and legal sys­ tems, religious traditions, ethnic categories, resources and rent, economic development, federal structures, democracy, and degree of trade openness. Following the lead of Treisman (2000, 2007), this chapter too will propose theoretical hypotheses, mainly from the aspects of causes, effects, and culture and values, based on the assumptions of research on corruption and rent seeking under the transition of systems—that causes and effects are correlated with each other and that each is characterized by internal connections between cause and effect factors. 5.3.1 Causative factors Corruption arises based on the three factors—specific social and economic systems, economy and welfare, and attributes and environments. A comprehensive economic system must be based on a comprehensive political system, under which power is dis­ tributed broadly across society and its arbitrary exercise is restrained, which serves to prevent the establishment of an exploitative economic system for personal enrichment. In contrast, corruption, a misuse of power, suggests the presence of an exploitative system (Acemoglu and Robinson 2012). Corruption arises in such an environment based on a balance between the expected costs (social, psychological, and monetary) of acts of corruption and their anticipated gains. If the greatest cost of corruption is that of arrest and punishment, then the above balance will depend on factors such as the efficiency of the country’s legal system (Treisman, 2000, 2007). Hypothesis 1.1 Corruption is rarer under an efficient social and economic system. In particular, the nature and consistency of formal and informal systems are decisively important factors. Since corruption is an act for personal gain arising in connection with the exercise of official systems, if such systems are inefficient or are strongly informal in nature, such as one which is customary and depends on the discretion of the parties involved, corruption is more likely to arise. Efficient systems can improve the quality of the market and ensure fairness (Yano 2008).8 Hypothesis 1.2 Corruption is rarer under conditions of democracy and political stability (Iwasaki and Suzuki 2012).9 Democracy is premised on a stable legal system, the free expression of opinion and debate, and the disclosure of information for these purposes. “Democracy to a significant extent reduces corruption” (Kolstad and Wiig 2011, p. 19). In contrast, in countries with frequent electoral irregularities, voters lose trust and governance worsens. In countries where governance has worsened in this way, corruption intended to secure instable government power and resources is relatively more frequent and deeper rooted. For this reason, the former Soviet Union faces greater risks than do Central and 150

THEORETICAL HYPOTHESES REGARDING CORRUPTION ISSUES

5.3

Eastern Europe. The European Bank for Reconstruction and Development (EBRD 2016) argued for a negative correlation between democracy and corruption. However, there is considerable skepticism about the view that holds simply that democracy directly reduces corruption. In fact, in the initial stages of democratization, corruption may increase as a means of securing voters’ support. In such initial stages, democracy may not necessarily be able to perform its role of serving as a check on corruption (Rose-Ackerman 1999). Hypothesis 1.3 Corruption is rarer in developed economies or where wages are high. Put simply, poverty has the effect of encouraging corruption among public officials. It already has been shown that growth reduces corruption, and it is in this sense that cor­ ruption is relatively more common in the former Soviet Union than in other transition economies. Hypothesis 1.4 Corruption is more common in resource-rich nations. This is because gains through corruption are quite large in resource-rich nations due to the large amount of rents associated with resources and the ease of access to resource develop­ ment rights. This is why corruption is so common in the former Soviet states rich in metal and fossil fuels, such as Russia, Kazakhstan, and Turkmenistan. Such states also are characterized by corporate scandals related to resources. Hypothesis 1.5 Privatization of ownership increases the likelihood of corruption. Generally, the privatization of ownership involves opening up public property and public economic activities to private businesspeople. If all other conditions were to remain the same, the risk of corruption would increase, as compared to a case in which no privatiza­ tion of ownership took place. While, individually, transition economies have followed their own methods of ownership privatization—management–employee buyouts, the spontaneous privatization of ownership, or voucher privatization of ownership, which involve low levels of transparency— the likelihood of corruption occurring is greater, “because of their slow pace, high levels of discretion, and lack of transparency” (RoseAckerman and Palifka 2016, p. 160), or more specifically, due to the difficulty of exter­ nally auditing the granting of preferential conditions to related parties through the pricing of properties, selection of methods and recipients of transfer, and internal application. In general, countries that prioritized voucher privatization of ownership, which permitted splurging on state-owned properties (e.g., Russia, the Czech Republic, and Kazakhstan), have higher levels of corruption than those that prioritized the transfer of state-owned properties based on market principles (e.g., Poland, Hungary, and Estonia). Hypothesis 1.6 Liberalization reduces the likelihood of corruption. For example, topdown restrictions, such as protectionist tariffs and trade permits, increase transaction costs as compared to cases in which such restrictions do not apply. This makes corruption pay­ ments to cover such costs more likely. Liberalization reduces corruption by decreasing such transaction costs (Sarwar 2013, p. 185). Of course, this hypothesis is not self-evident, and liberalization could increase corruption due to its relationship with other reforms. Tavares (2007) showed through empirical analysis of experiences with political and 151

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

economic liberalization during the 1980s through the 1990s that, even if democratization reduced corruption, liberalization potentially could increase it. Corruption increased even when liberalization took place five or more years after the experience of democratization. Based on these studies, the effects of liberalization on transition economies may be felt in two ways. 5.3.2 Effect factors Corruption not only affects social and economic systems as well as the economy and welfare but also determines the level of governance as it pertains to traditions or systems exercised by government for public goods, including the pro­ cesses chosen, monitored, or substituted by government (political aspects), govern­ ment’s ability to manage resources effectively and implement sound policies (economic aspects), and respect for the public and national systems (systemic aspects). (Kaufman 2005) Accordingly, when looking at corruption as an effect factor, the following hypoth­ eses can be proposed regarding transition economies. Hypothesis 2.1 Corruption hinders economic growth. It is clear from aspects such as the distortions in systems (monopoly pricing to secure rents) arising as a result of, or in the process of, the misuse of official authority and prioritizing private interests that cor­ ruption has a negative effect on economic growth. However, at the same time, in the absence of trust in public systems within national and social relations, corruption may be chosen as an act intended to support the economy and ensure survival vis-à-vis an untrustworthy state. In such a case, corruption many serve as a factor supporting eco­ nomic growth because failure to take any action would mean that the economic gains lost due to an untrustworthy state would be secured as private gains by a specific stratum of society. That is, it would be worthwhile to study whether the greasing-the-wheels hypothesis applies where an untrustworthy state is present in conditions specific to the transition of systems. Put another way, this theoretical hypothesis concerns whether, amid the confusion of the transition of systems, corruption is a factor that merely hinders growth or contributes to growth as a survival tactic. Hypothesis 2.2 Corruption grows the informal sector. If corruption is an attempt to reduce formal transaction costs, it would be likely to systematize and grow the scale of activities other than formal economic activities (the informal economy). Based on this logic, corruption would correlate positively with the informal sector, so that a decrease in acts of corruption would decrease the size of that sector. Hypothesis 2.3 Corruption increases economic disparity and reduces the level of public welfare. If corruption results in a concentration of gains among an oligarchy or well-connected private capitalists, excluding the general public, then corruption could 152

THEORETICAL HYPOTHESES REGARDING CORRUPTION ISSUES

5.3

promote the concentration of wealth in a specific social group, possibly lowering the level of national welfare as a result. Generally speaking, in the former Soviet Union, the activities of a behind-the-scenes oligarchy increased the maldistribution of wealth, while this phenomenon was rarer in Central and Eastern Europe. The EBRD (2010) pointed out that the value of informal payments—an indicator of the level of corrup­ tion—was highest in the former Soviet Union, followed by Southeastern Europe, and was lowest in Central and Eastern Europe, although it still was higher there than in Western Europe. However, an overall trend toward convergence with the level of Western Europe has been observed (EBRD 2016, p. 28). Hypothesis 2.4 Corruption worsens governance. As the misuse of entrusted power for private gain, corruption is related directly to the worsening of governance because it makes government power and systems less trustworthy. Hypothesis 2.5 Corruption hinders transitional reforms. Since it distorts fair market systems, corruption makes it more difficult to transition to a market economy and implement liberalization policies, resulting in a more difficult transition process. 5.3.3 Culture and values Corruption is connected to culture, customs, and values. Corruption is merely a cultural phenomenon (Barr and Serra 2010), and the efficacy of studying corruption from the cultural perspectives of the value of uncertainty avoidance and the customs of human orientation and group behavior has also been emphasized (Seleim and Bontis 2009). Treisman (2000, 2007) also looked at the historical process by which legal culture, legal systems, and religion are formed, showing that culture and values have major impacts on corruption. From the perspective of transition economies, in particular, the following points are likely to be of importance. Hypothesis 3.1 The degree of permeation of communism is connected to corruption. Corruption has existed at least since the communist era, and the longer that era lasted, the more commonplace the presence of corruption, the more it was accepted as com­ monsense, and the more formal it tended to be. This is why the levels of corruption in the former Soviet Union and Central and Eastern Europe are so high. Hypothesis 3.2 Religion and culture are connected to corruption. Since religious views and culture regulate individual behavior, corruption is regulated strongly by reli­ gion and culture (Seleim and Bontis 2009). In general, Protestants are more tolerant of challenges to authority and individual disagreement, demanding individual responsibil­ ity, while other Christian sects stress human weakness. In particular, Protestants see poverty as being related to idleness, and they stress working hard in life. However, in the Eastern Orthodox Church, the ties between church and state are strong, as are paternalistic values (Treisman 2007). As a result, greater tolerance of corruption was fostered in Russia (Eastern Orthodox) compared to the Baltic states and Central and 153

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

Eastern Europe where, for the most part, the influence of Germany (Protestant) was strong. The same can be observed in Central Asia and elsewhere. In addition, the lessening of legal and political system criteria for EU membership when some Baltic and Central and Eastern European states joined the EU suggests that, where there are differences among the original EU member states regarding public attitudes toward corruption, domestic cultural factors have a strong influence. In fact, many empirical studies deny religion’s influence on corruption (e.g. Shadabi 2013; Ko and Moon 2014). Hypothesis 3.3 Public distrust of society and systems is interrelated with corruption. While the public’s trust in politicians at the highest level, such as presidents, is rela­ tively high in the former Soviet states, trust is low in lower-ranking public officials,10 and even lower in society and systems. This leads to increased corruption. Empirical research by the EBRD (2010) using the Life in Transition Survey (LiTS) showed that trust levels and corruption were inversely related and that corruption in public services negatively affects trust in public officials.

5.4 OVERVIEW OF THE TESTING METHOD, LITERATURE QUERY METHOD, AND BASIC SAMPLING METHOD FOR LITERATURE For the purpose of objectively testing the hypotheses proposed in the preceding sec­ tion, literature was collected through a mechanistic process established in advance. Specifically, literature from 1995 through 2017 was searched using Web of Science, a digital literature database that covers the social sciences as a whole. We conducted this search using combinations of two keywords or terms, one of which was either “corruption” or “rent seeking,” which are core keywords in corruption research. We used another keyword or phrase from among the following: “transition economies,” “Central Europe,” “Eastern Europe,” “former Soviet Union,” or the names of China or any of the countries in Central and Eastern Europe or the former Soviet Union. This resulted in a collection of 676 works. However, as the result of further close review of the literature queried in this mechanistic way, the basic collection was nar­ rowed down to 558 works. A preliminary profile of the literature is described below. As seen in Figure 5.2, while there is some variation in the number of works by year of publication in the basic collection, a trend toward an exponential increase can be observed. This includes similar increasing trends in studies that, instead of simply analyzing the current situation, analyze causes and effects and correlation and in analysis results concerning economic and social systems. With regard to the research content, Figure 5.3 presents an outline of the basic collection by attributes of authorship and publication media. In total, the authors of the 558 works in the basic collection numbered 1,107, of whom 328 were affiliated with research institutions in North America, 136 in the United Kingdom, and 163 in Western Continental Europe, while 154 were affiliated with research institutions in

154

OVERVIEW OF THE TESTING METHOD

5.4

C H A P T E R 5

Figure 5.2 Number of publications in the basic collection by year

Central and Eastern Europe, 58 in the former Soviet Union, and 185 in other transition countries. As such, about two-thirds of researchers were from countries other than former socialist states. Thus, it can be said that researchers in states not directly affected by the implementation of market transition policies were more sensitive to the realities hindering such implementation. At the same time, totaling the number of works in the basic collection in five-year intervals shows that few works were published during the 1990s, but the number has skyrocketed over the years. Although one factor behind this increase might be the availability of the objective index from Transparency International mentioned above as an effective index for research use, it also is affected by factors such as an increase in the number of social surveys and the manifestation of actual large-scale corruption. In addition, a debate on this subject has developed in journals, including those in the field of economics as well as in various other specialized fields, such as sociology, law, political science, and area studies, which clearly shows that the issue of corruption definitely is more than an issue of pure economics. The bulk of the basic collection showed a strong tendency to focus narrowly on certain regions and countries, with about 20% of studies looking at multiple regions, while about 60% focused on specific countries. Even those studies that looked at multiple countries tended strongly to compare countries within the same region. This suggests the high possibility that corruption has become strongly subsumed as a subject of research in certain area studies and is not necessarily being treated as

155

C H A P T E R

SOCIAL CONFUSION AND CORRUPTION

5

(a) Authorship •

Figure 5.3 Breakdown of the basic collection by literature attribute Note: Numbers in parentheses are those of the relevant literature.

156

OVERVIEW OF THE TESTING METHOD

5.4

a subject of comparative research, hence, indicating the importance of the cultural backgrounds of certain countries. While about 60% of all of the literature employed full-fledged quantitative analysis or quantitative backing, they were extremely diverse in terms of their methods of treating indicators in regression analysis, the subjects of testing for a relationship to indicators of corruption, and the directions of cause-and-effect relationships; therefore, it is difficult to identify overall trends in the research. A review of the literature shows that the styles of research in this field can be divided into the two main categories—studies of the present state of corruption and analytical studies intended to verify its causes and effects. Of these, just under 80% of studies focused on causal relationships between corruption and some other factor. These will serve as the basis for the hypothesis testing in this chapter. However, examples of studies of the present state of corruption include Partos 2004, Radin et al. 2011, Votápková and Žák 2013, Yeager 2012, Yessenova 2012, Belas et al. 2015a and 2015b, Jancsics 2015, and Linhartová and Volejniková 2015. While, in light of the point of this chapter, these works will not be described here in detail, each presents highly thought-provoking research results and is likely to serve as a valuable source of information for further research on corruption. The various factors that have been identified as being related to corruption can be divided into the two main categories—decisive factors affecting corruption and factors affected by corruption. As described in the preceding section, decisive factors affecting corruption include traditions and culture as well as trust in society, in addition to social and economic systems, economy and welfare, and attributes and environments. Of these, social and economic systems and their efficiency can be subdivided into a wide range of systemic factors regulating the economy and society, such as foreign exchange, liberalization, privatization of ownership and the methods thereof, size of bureaucratic structures, decentralization of power, system transition reforms, political freedom, property rights, and rule of law. Nearly half of the studies discussing decisive factors affecting corruption focused on these areas. It is clear that, among researchers studying transition economies, systems themselves are considered the most strongly related cause of corruption. Based on this background, a very large number of points are at issue, and four of the hypotheses presented in the preceding section of this chapter—Hypotheses 1.1, 1.2, 1.5, and 1.6—concern this area. However, traditions and customs include a wide range of practices that are not themselves formal systems, such as past practices, historical vestiges, ethics, degree of tolerance for corruption, personal connections, and permeation of Western Euro­ pean culture, as well as culture and social climate. These include numerous items in the category of factors referred to generally as “informal systems,” and they attract the next highest degree of interest as causes of corruption after social and economic systems. Two of the hypotheses presented in the preceding section of this chapter— Hypotheses 3.1 and 3.2—concern this area. The state of the economy and welfare includes the outcomes of economic activities and indicators concerning the level of the public’s standard of living as a result of such activities. Examples include growth expectations, direct investment, technological 157

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

progress, the import ratio, the economic growth rate, returns on investment, and levels of corporate profits, employment, and social security. These economic factors attract less attention than the two factors discussed above and are subjects of about 10% of the studies. The only hypothesis presented in this chapter that concerns this area is Hypothesis 1.3. At the same time, trust in society and systems also is a factor sometimes identified as a cause of corruption in light of this theme (and as a result of corruption, as described below). In addition, when conducting micro analysis at the level of individuals or firms, sometimes the environments of these actors, as well as their own personal attributes— such as reserves of natural resources, corporate size, age of managers, and individuals’ professions and ethnicities—are identified as causes. Of the hypotheses presented in the preceding section of this chapter, Hypotheses 1.4 and 3.3 concern these items. However, factors affected by corruption can be divided into the main categories of social and economic systems and reforms thereof, political governance, trust/mistrust in society and systems, and public welfare/natural environment. Of these, the factor of social and economic systems and reforms thereof partially overlaps with social and economic systems, which are causes of corruption. In addition to the factors of corporate barriers to entry, market reforms, democratization, and state apparatus, non-currency payment systems also belong to this category. This suggests that, as discussed at the start of this chapter from the perspective of endogeneity, while social and economic system factors affect corruption, corruption also affects aspects of social and economic systems. Among factors affected by corruption, transition reforms and the informal economy are attracting attention in particular, and this chapter also pays attention to these tendencies in Hypotheses 2.2 and 2.5. Political governance concerns factors such as social disorder and strife brought about by the state of economic and social systems and reforms therein or the effi­ cient functioning of systems. Specific examples include extreme ethnic disputes, national unity, transparency and fairness, and organized crime and insurrection. Hypothesis 2.4 in this chapter concerns these factors. Economic/corporate performance attracts the most attention among factors affected by corruption. Specific examples include factors related to various economic out­ comes—at both a micro and a macro level—such as corporate earnings, rates of entrepreneurship, foreign direct investment (FDI), economic growth, bank lending, returns on investment, numbers of patents, international trade volume, and income level. The negative effects of corruption on such economic outcomes have long been identified, and this chapter also treats these as one point at issue (Hypothesis 2.1). In addition, as noted above, degrees of trust in and attitudes toward society and systems have also been focused on as outcomes of corruption: in these, a two-way cause-and­ effect relationship with corruption can be discerned. The remaining subjects of national welfare and the natural environment consist mainly of factors that appear to be results of the above-mentioned economic/corporate performance and social and economic systems and reforms therein, such as public welfare, happiness, healthy life expectancy, welfare policies, inequality, healthcare efficiency, and environmental degradation. In light of researchers’ areas of focus regarding these factors, this chapter will summarize conclu­ sions in these areas based on the consideration of Hypotheses 2.3 and 3.3. 158

RESULTS OF TESTING

5.5

This section overviewed the profile of the literature collected systematically in this chapter and confirmed that the factors related to corruption addressed in the basic collec­ tion are consistent with the areas subject to the hypotheses we have proposed. Accord­ ingly, in the next section, we will test the series of hypotheses proposed in Section 5.3 by collecting all results related to the hypotheses proposed in the preceding section.

5.5 RESULTS OF TESTING Part b of Table 5.1 collects the results of previous studies related to the hypotheses proposed above, by cause and effect. This section will introduce the results of hypothesis testing and the main relevant works of the literature in the order in which the hypotheses were proposed in Section 5.3. Hypothesis 1.1: Corruption is rarer under an efficient social and economic system Researchers of transition economies do not dispute the argument that factors such as the quality and consistency of the design of social and economic systems govern levels of corruption. For example, Ahrend (2005) argued that, while it would be dif­ ficult for Russia to break out of its dependency on resources, it would be able to free itself from the spell of resources by preventing corruption through simple, strict, and fair laws. Similarly, Desai and Goldberg (2001) argued that Russia’s lack of property rights is a driving force behind corporate misappropriation, and Gherghina and Chiru (2013) stated that the diversion of national funds, rooted in defects in the law, continues in Romania, in a vicious circle in which the passage of a bill to coun­ ter such diversion merely spurs a search for loopholes to enable further diversion. At the same time, empirical analysis has confirmed that a number of systemic factors are decisive. Results of empirical analysis by Duvanova (2014) of panel data from 26 former communist states for the period from 1999 to 2005, based on a fixed-effects model, lead to the conclusion that forms of ownership, bureaucratic intervention, and a lack of rule of law form a breeding ground for corruption. In addition, Goel et al. (2015) identified factors such as anti-corruption laws, corporate internal ethics rules, and bureaucratic pressure, (and the gender of managers) govern the likelihood of corporate bribes. Of the studies reviews, 48 supported this hypothesis, while none rejected it by arguing against a relationship between systems and corruption. Hypothesis 1.2: Corruption is rarer under conditions of democracy and political stability Only one of the studies reviewed—Sharafutdinova and Steinbuks (2017), which identi­ fied a positive correlation between an administration’s length of time in power and frequency of corruption—clearly rejected this hypothesis. There is strong support for

159

C H A P T E R 5

160

No Yes

0 0 1 0 0

3 4

0 0

1 1

0

0 2

0

0

Other

0

No

(b) Focus on particular regions or countries CEE with EU 8 0 2 0 menbersip (including Baltics, except for Croatia)

(a) Location of affiliated institution CEE (including 5 0 0 Baltics) FSU 3 0 2 Other transi30 0 1 tion countries North America 29 0 6 United 13 0 1 Kingdom Other Western 14 0 2 European countries Other 3 0 1 Author total 97 0 13

Yes

H1.2 Corruption is rarer under conditions of democracy and political stability.

1

5 20

3

7 3

0 2

0

Yes

1

0 8

2

4 0

0 0

2

No

0

0 3

2

0 0

0 1

0

Yes

0

0 0

0

0 0

0 0

0

0

1 6

3

1 1

0 0

0

1

4 12

3

3 0

1 0

1

2

0 3

0

1 2

0 0

0

No Other

H1.5 Privatization of ownership increases the likelihood of corruption.

No Yes

H1.3 Corruption is rarer in H1.4 Corruption developed is more economies or common in where wages resource-rich are high. nations.

0

2 16

1

4 5

0 2

2

Yes

0

0 12

2

4 4

2 0

0

No

1

0 10

4

3 2

1 0

0

Yes

0

0 0

0

0 0

0 0

0

No

H3.1 The degree of the H1.6 Liberaliza- permeation of tion reduces communism the likelihood is connected of corruption. to corruption.

5

3 61

5

18 11

4 13

7

Yes

0

0 0

0

0 0

0 0

0

1

0 2

2

0 0

0 0

0

No Other

H3.2 Religion and culture are connected to corruption.

6

2 30

11

6 2

1 3

5

Yes

0

0 0

0

0 0

0 0

0

28

24 299

54

88 45

14 52

22

No Total

H3.3 Public distrust in society and systems is interrelated with corruption.

5

H1.1 Corruption is rarer under an efficient social and economic system.

(a) Cause of corruption

Table 5.1 Result of vote counting on hypotheses by studies of corruption in transition countries

C H A P T E R

SOCIAL CONFUSION AND CORRUPTION

0

0 0

0

20

0 1

2

0 0 0 0 0

0 0

7 8

1 5 4 7 31

0

2

0 2 1 3 4

0

1 0

1

3 2

1

Total

48

0

10

(d) Intensity of empirical examination Econometric 17 0 0 analysis Using quantita7 0 5 tive data Without quantita24 0 5 tive analysis

(c) Publication year 1995–1999 2000–2004 2005–2009 2010–2014 2015–2017

CEE without EU menbersip (including Croatia) Russia FSU other than Russia and Baltics China and other Asian countries CEE and FSU Multiple areas, including Asia All areas of transition countries

0

0

2

1

0

1

1

0 0 0 0 2

1

0 0

0

0 1

0

1

0 0 0 0 1

0

0 0

0

1 0

0

11

3

1

7

0 1 3 0 7

3

1 0

3

1 1

1

3

1

0

2

0 1 0 1 1

0

0 0

1

1 0

0

3

1

0

2

0 1 1 0 1

1

0 0

1

1 0

0

0

0

0

0

0 0 0 0 0

0

0 0

0

0 0

0

4

2

0

2

2 0 0 1 1

2

0 1

0

1 0

0

7

2

2

3

0 1 4 2 0

0

2 0

1

2 0

1

2

1

0

1

1 1 0 0 0

0

0 0

0

0 0

0

7

3

0

4

1 0 0 2 4

0

1 1

2

0 1

2

9

5

1

3

0 2 1 2 4

0

1 0

4

2 2

0

7

4

1

2

0 2 0 2 3

0

2 0

0

2 2

0

0

0

0

0

0 0 0 0 0

0

0 0

0

0 0

0

32

15

9

8

0 4 7 6 15

3

5 1

9

4 2

3

0

0

0

0

0 0 0 0 0

0

0 0

0

0 0

0

1

0

0

1

0 0 0 0 1

0

0 0

0

0 0

0

0

0 0

0 0 0 0

0 0 0 0 0

0 0 0

0

1

0 2

2 2 0 0

0 0 0 5 8

8 2 3

13

160

69

29

62

5 20 21 31 83

12

15 4

44

25 21

11

RESULTS OF TESTING 5.5

161

C H A P T E R 5

162

0

0 0 0 8

1

2 0

5

7 25

1 0 11 6 0 0

1

5 26

2 136

(b) Focus on particular regions or countries CEE with EU menbersip 11 (including Baltics, except for Croatia) CEE without EU menbersip 2 (including Croatia) Russia 12 FSU other than Russia and 8 Baltics China and other Asian 10 countries

1 2 7 8 2 1

26 6 15 30 24 33

0

0 0

0

2

0 14

1 0 2 4 4 3

0

1 0

0

0

0 2

0 0 0 0 1 1

No

0

1 0

0

0

0 2

0 0 0 0 2 0

Other

Yes

Other

Yes

No

H2.2 Corruption grows the informal sector.

H2.1 Corruption hinders economic growth.

5

2 2

2

9

11 62

5 1 4 19 11 11

Yes

1

0 0

0

0

0 5

0 0 1 0 4 0

No

1

0 0

0

0

0 6

0 0 6 0 0 0

Other

H2.3 Corruption increases economic disparity and reduces the level of public welfare.

9

2 6

3

1

4 39

0 4 12 12 3 4

Yes

0

1 1

0

0

0 2

0 1 0 1 0 0

No

0

2 0

0

0

0 4

0 1 0 3 0 0

Other

H2.4 Corruption worsens governance.

2

2 1

0

0

0 10

1 1 3 5 0 0

Yes

0

0 0

0

0

0 0

0 0 0 0 0 0

No

1

0 0

0

0

0 4

0 0 4 0 0 0

Other

H2.5 Corruption hinders transitional reforms.

14

8 4

1

16

3 87

12 4 20 33 7 8

Yes

0

1 0

0

0

0 1

0 0 0 1 0 0

No

0

0 0

1

1

0 4

0 0 0 1 1 2

Other

H3.3 Public distrust in society and systems is interrelated with corruption.

5

(a) Location of affiliated institution CEE (including Baltics) FSU Other transition countries North America United Kingdom Other Western Europeean countries Other Author total

(b) Effect of corruption

Table 5.1 contd.

56

34 22

10

41

32 429

47 20 85 123 59 63

Total

C H A P T E R

SOCIAL CONFUSION AND CORRUPTION

8 1 0

9

(d) Intensity of empirical examination Econometric analysis 37 Using quantitative data 9 Without quantitative analysis 16

Total

Source: Authors’ calculations

0 0 1 2 6

2 5 6 19 30

62

0 0 0

14 1 4

CEE and FSU Multiple areas, including Asia All areas of transition countries (c) Publication year 1995–1999 2000–2004 2005–2009 2010–2014 2015–2017

10

8 0 2

0 0 1 0 9

1 1 0

5

4 1 0

0 1 0 0 4

2 0 1

2

1 0 1

0 1 1 0 0

0 1 0

1

0 1 0

0 0 0 1 0

0 0 0

30

17 6 7

1 0 2 11 16

5 2 3

1

1 0 0

0 0 0 0 1

0 0 0

2

2 0 0

0 0 0 0 2

0 1 0

21

2 3 16

1 2 2 5 11

0 0 0

2

0 2 0

0 0 1 1 0

0 0 0

2

1 0 1

0 0 0 1 1

0 0 0

6

1 2 3

0 1 1 0 4

0 0 1

0

0 0 0

0 0 0 0 0

0 0 0

1

1 0 0

0 0 0 0 1

0 0 0

48

23 8 17

2 0 5 18 23

4 1 0

1

0 1 0

0 0 0 1 0

0 0 0

26 7 9

6 10 20 61 108

108 34 63 205

0 0 0

0 0 0 2 0

2 0 0 2

RESULTS OF TESTING 5.5

163

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

the hypothesis that, in general, democracy and political stability have restraining effects on corruption. For example, Grzymała-Busse (2003) suggested the efficacy of democracy by pointing out that, in Central and Eastern Europe, cases in which there was a lack of competition among political parties were more likely to experience a diversion of national funds by political parties due to lax legal systems. In addition, Maloney and Kelly (2000), in introducing case studies of efforts in primary and sec­ ondary education to restrain criminal activity in developing countries including Russia, offered the assessment that the development of civil society is a valid countermeasure against corruption, inferring the importance of developing a democratic society. At the same time, since two prior studies identified necessary preconditions for the effi­ cacy of restraints on corruption by political systems, some reservations may be needed when presenting this type of argument. For example, Jetter et al. (2015) argued that, while in countries with per-capita GDP of $2,000 or higher, democratization restrains corruption, in countries with lower levels of GDP, it may actually further corruption, pointing out the pitfalls of haphazard democratization. In contrast, another prior study proposing precondi­ tions for the efficacy of efforts to restrain corruption (Zaloznaya 2015) argued from the case of Belarus that, while under a benevolent dictatorship, corruption may be furthered, under a strict dictatorship, corrupt bureaucrats may be removed from office. This raises the possibility that nondemocratic political systems, which tend to be spoken of most often in critical terms, may, at times, help restrain corruption. However, it may be said that most researchers recognize (although not uncondition­ ally) the presence, overall, of the restraining effects of democracy on corruption. Hypothesis 1.3: Corruption is rarer in developed economies or where wages are high Interpreting the results regarding this hypothesis definitely is not a simple matter. While the hypothesis is supported if wealth is measured simply by the level of national income, it is also possible that the presence of opportunities for rent-seeking in the process of accumulating wealth can encourage corruption. For example, in look­ ing at regional corruption disparities in Russia, Dininio and Orttung (2005) identified a strong negative correlation between the degree of corruption and both the size of the bureaucratic structure and the economic level. In studying decisive factors related to the probability and degree of corporate exploitation and state capture, Iwasaki and Suzuki (2007) pointed out that economic crisis encourages the spread of corruption, as do the degrees of decentralization in relations between government and corporations and of national intervention in corporate management. In contrast to these studies that suggest a negative correlation between stability and wealth in people’s way of life and corruption, some studies have found economic growth to be a hotbed for corruption. A leading example is the study of Safavian et al. (2001), which identified a tendency for small businesses in Russia to be more likely tar­ gets of corruption if their rates of corporate growth were higher. In addition, Wei (2015) warned of the risk that the overheating of rapid urban development in China could encourage the spread of corruption. While, overall, more studies supported than rejected 164

RESULTS OF TESTING

5.5

this hypothesis, some show that, when viewed over the short term, there were different aspects to the relationship between wealth realization and restraint of corruption, and the nonlinear relationship between the two may need to be further considered.

C H A P T E R 5

Hypothesis 1.4: Corruption is more common in resource-rich nations While, as compared to the other hypotheses, relatively few studies argued that the presence of natural resources itself was a decisive factor affecting corruption, no studies rejected this hypothesis. For example, Ahrend (2005) discussed natural resources, together with the sys­ temic factors mentioned above, as causes of corruption in Russia. In addition, Gylfason (2000) argued, based on a correlation coefficient, that natural resource reserves in transition economies spurred corruption, leading both directly and indirectly to low growth. Hypothesis 1.5: Privatization of ownership increases the likelihood of corruption Although more studies rejected than supported this hypothesis, researchers were divided in their views, and some were neutral on the subject. A look at the studies over time shows that, while those published during the 1990s tended to support the hypothesis, over the years there has been increasing advocacy for the effects of privatization of ownership on restraining corruption. Among researchers supporting the hypothesis, Braguinsky (1999) argued that uncertainties inherent to private sector management in Russia induced rentseeking behavior by causing management to adopt extremely short-sighted approaches, suggesting that privatization of ownership in Russia might induce corruption. In addition, Harris and Lockwood (1997) summarized the series of systemic transactions in Russia, China, Vietnam, and Ukraine by arguing that, when the previous system merely collapsed without having built up a functional market economy, and particularly when the privatiza­ tion of ownership was delayed, the result was a nation of rent-seekers. In contrast, among those who rejected this hypothesis, Holmes (2008) identified renationalization as one possible cause of the increased corruption in Russia after a temporary decrease during the Putin era, while Benevolenskaya (2010) analyzed the case of Russia as showing that entrusting the management of state property to the private sector diminished incentives for corruption, mainly through the public disclosure of information. There also is a tendency to argue that the effects of the privatization of ownership will vary depending on the method of privatization and the form of ownership. For example, Bornstein (1999) argued that the method of monetary auctions employed in the Czech Republic, Hungary, and Poland was least likely to induce corruption, while sales through negotiation and management and employee buyouts (MEBOs) were most likely to induce it. Christev and FitzRoy (2002) argued that, while in Poland, outsider firms had the highest increases in productivity and wages, insider firms were more susceptible to rent-seeking behavior through larger wage increases as compared to lower increases in productivity, due to the lower degrees of external pressure on the latter firms. 165

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

In addition, empirical studies, such as those of Iwasaki and Suzuki (2007), mentioned above, and Holtbrügge et al. (2007), who identified a positive correlation between government-related stakeholders and corruption, were more likely to reject the hypothesis (i.e., to support the idea that the privatization of ownership has a restraining effect on corruption). From this, it can be inferred that progress toward private ownership and the spread of corruption are, objectively and over a longer term, more likely to lead to the finding of a negative correlation. There also is a need to pay close attention to conditions and environments. Hypothesis 1.6: Liberalization reduces the likelihood of corruption Overall, there is a struggle between those who support this hypothesis and those who reject it. However, there is a strong tendency for research on the former Soviet Union and China to see liberalization as inducing corruption, and a tendency for analyses of Central and Eastern Europe and multiple regions to conclude that liberalization restrains corrup­ tion. As such, this hypothesis is characterized by a difference in conclusions depending on region. For example, Gokcekus et al. (2015) identified the degree of economic openness in transition economies as a factor governing corruption, while Neshkova and Kostadinova (2012) argued that transition reforms in six Central and Eastern European states were con­ firmed to restrain corruption and induce FDI. However, Popov (2012), citing the Corrup­ tion Perceptions Index, argued that, in 24 transition economies, a transition to smaller government led to systemic breakdown that brought about various negative effects, includ­ ing corruption. Kneen (2000) summarized the experience of Russia as one in which, in the absence of necessary systems and the rule of law, practices of corruption from the Soviet era were spread by a sudden shift to a market economy. These regional disparities also can be interpreted as representing disparities between regions in which liberalization had advanced and those in which it had not, suggesting a nonlinear relationship between the two. Hypothesis 2.1: Corruption hinders economic growth The overwhelming majority of studies supported this hypothesis. There is a kind of consensus, at least among researchers studying the economies of Central and Eastern Europe and the former Soviet Union, that corruption is harmful to the economy. For example, Earle (2000), in recommending countermeasures against corruption in transi­ tion economies, strongly warned of a vicious cycle in which the growth of corruption hinders economic development by driving up the costs of investment. This in turn leads to a loss of support for reforms, and the resulting delay in reforms further worsens the economy. In addition, Ledyaeva et al. (2012, 2013) detected a strong rela­ tionship in which the level of corruption governs FDI. This would be strongly expected to have economic effects resulting from the inflow of funds. Even among the minority of studies that rejected this hypothesis, none argued that, at a macro level, corruption encouraged national economic growth. However, on an excep­ tional basis, some studies did support the greasing-the-wheels hypothesis, arguing, 166

RESULTS OF TESTING

5.5

based on micro-level analysis of the former Soviet Union, that corruption stimulated economic transactions. One such example is the study of Guriev et al. (2010), which confirmed the improved performance of firms in regions adjoining more regions that are captured by multi-regional business groups seeking to remove barriers to distribution transactions. Some analysis results also showed that the impact of corruption on economic growth varied among different social structures. In analyzing the reasons for the dif­ ferences in the effects of corruption in China and Russia, Larsson (2006) mentioned differences in the social structures of the Brezhnev-era Soviet Union and Mao-era China. Larsson argued that, in China, where the economy was centered on low-tech, labor-intensive industries, corruption did not cause trading partners to stay away because most counterparty countries for trade and investment consisted of similar developing countries in which corruption was rampant. In addition, the decentraliza­ tion of power had advanced to the point where bribing certain bureaucrats would not affect national policy, so that corruption did not become a factor hindering economic growth. However, in Russia, where the centralization of power and industrialization already had advanced, corruption severely hindered economic growth. While this hypothesis is supported overall, a very small minority argued that corrup­ tion actually encourages growth. Hypothesis 2.2: Corruption grows the informal sector A small majority supported this hypothesis. While none of the studies that rejected it were of the view that corruption restrains the informal economy, they mainly argued that corruption was not a primary factor or that it had no particular influence. For example, Johnson et al. (2000) detected a strong influence of bureaucratic corruption on informal activities in three eastern European nations, and Williams (2015) argued that the extent of the spread of corruption affected the payment of informal wages in ten Central and Eastern European states. Nesvetailova (2004) argued that the nonmonetary economy in Russia was not caused by previously identified factors, such as corruption, but arose instead as a reaction to deregulated financial markets. While it is not possible to derive a clear conclusion because of the limited number of works on this theme examined in this chapter, since no studies in the literature argued that corruption had positive effects, it is certain at the very least that this is not grounds for justifying corruption. Hypothesis 2.3: Corruption increases economic disparity and reduces the level of public welfare The only study that rejected this hypothesis was that of Hung et al. (2017), which pointed out the possibility that corruption could increase returns in corporate units. The vast majority supported the hypothesis. Based on the results of interviews with public officials, politicians and regulators, and parties related to NGOs, Škrbec and Dobovšek (2013) showed that it has been confirmed through various approaches that 167

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

state capture distorted the rule of law among local governments in Slovenia, resulting in negative effects including negative impacts on trust in government administration and economic outcomes as well as inequality and environmental degradation. Bobak et al. (2007), who studied the causes of worsening health through a logit model, also identified a positive correlation between corruption and poor health. In addition, using a regression model, Minagawa (2013) showed that, among 23 transition economies during the years 2008 and 2009, healthy lifespans were shorter in those with more widespread corruption. While it is clear that this hypothesis is justified since, as seen under Hypothesis 2.1, it already is generally accepted that corruption negatively affects the economy, the fact that it would worsen the public’s standard of living can be considered a natural consequence. Hypothesis 2.4: Corruption worsens governance The overwhelming majority of studies supported this hypothesis. For example, Hagan and Radoeva (1997) concluded that corruption in the upper levels of the social hierarchy in Czechoslovakia prior to transition served as a hotbed for social distrust during the transition and for resulting criminal behavior, and Kolossov and Toal (2007) showed, by surveying people in various positions, that one cause of strife in Russia was the widespread recognition of corruption. However, one study that rejected this hypothesis was that of Darden (2008), who con­ cluded that, in conditions such as those of Ukraine, in which bribery has become a type of informal system, it impedes the development of free politics, but it contributes to stability in tax collection and social order and restrains political opposition. Overall, it is widely recognized that corruption hinders governance, while other results may be demonstrated in extremely specific cases in which corruption deeply permeates society. Hypothesis 2.5: Corruption hinders transitional reforms Perhaps because the relationship described in this hypothesis is considered a natural state of affairs, few of the studies reviewed in this chapter addressed this theme head on. However, no studies rejected the hypothesis. It would appear that this pro­ cess may be viewed in various ways. As one example, Chen (2008), looking at the cases of China and Vietnam, pointed out that transition reforms in which rentseeking is rampant are able to advance no further than the point at which further government reforms would eliminate rents. In addition, as noted previously, Earle (2000) argued that, the vicious cycle between corruption and worsening economic growth apparent under conditions of corruption hinders economic development by restraining investment, leading to a loss of support for reforms. Whatever the case, there is presently no disagreement with the argument that corruption is a serious impediment to the transition of systems.

168

RESULTS OF TESTING

5.5

Hypothesis 3.1: The degree of permeation of communism is connected to corruption No strong opinion was identified in the basic collection opposing the hypothesis that the vestiges of the former era served as a hotbed for corruption. Studies on this theme are very common with regard to the former Soviet Union in particular. For example, AllinaPisano (2010) identified the legacy from the Soviet era as one cause of corruption in an analysis of the capture of political authority by the bureaucratic apparatus in Ukraine, and Obydenkova and Libman (2015) found that corruption was more likely to occur in regions in which rates of membership in the former Communist Party were high. Hypothesis 3.2: Religion and culture are connected to corruption No views could be found that rejected this hypothesis. To the contrary, the vast majority of studies supported it. Regarding the former Soviet Union, Brovkin (2003), in tracing the history of corruption through a review of Russian history since the Soviet era, identified corruption as being rooted in ethical norms and cultural practices. Denisova-Schmidt and Huber (2010), exploring why corruption is more widespread in the eastern part of Ukraine, mentioned commercial practices that have developed in a history free from any battle against corruption. However, many similar points have been made regarding the nations of Central and Eastern Europe as well, with Dimitrova-Grajzl (2007) finding that differences in the severity of corruption in Central and Eastern Europe were strongly influ­ enced by the legacy of political corruption since the later years of the Ottoman Empire. However, it also has been pointed out that this influence of historical vestiges weakens as countries advance further in reform. A typical example of such a study is that of Grosfeld and Zhuravskaya (2015), which argued that, in Poland, factors such as corruption and regional income disparities are becoming more separated from the impacts of historical processes and culture over time. From the above consideration, it can be said that most researchers recognize the reli­ gious and cultural backdrops behind corruption and that the argument for the present may focus on the strength or weakness of their influences. Hypothesis 3.3: Public distrust in society and systems is interrelated with corruption Only one paper, discussed below, expressed a view counter to the arguments that dis­ trust in society breeds corruption and corruption breeds such a sense of distrust. The vast majority identified a relationship between corruption and distrust in society and systems. Among studies included in the literature that identified distrust as a cause of corruption, Ateljevic and Budak (2010) and Giordano (2010) pointed out, through ana­ lyses of Croatian and Serbian societies, respectively, that a lack of mutual social trust was a cause of corruption. In addition, studies that identified trust as a domain of soci­ ety affected by corruption included that of Horne (2012, 2014), which showed that progress on policies to clean up corruption and decreased awareness of corruption in

169

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

Central and Eastern Europe led to a recovery of trust in government and systems. Heusala (2013) pointed out that corruption during the post-perestroika period in Russia was a cause of distrust in public administration. At the same time, Hendley (2010) rejected such an interrelationship. She pointed out that the public viewed the law itself as an impediment to litigation, rather than distrust in the administration of justice as a result of corruption, as the cause of passivity toward liti­ gation on home repair projects in Russia. However, Arnold et al. (2012), in investigating decisive factors affecting the degree of trust in the EU among the 27 EU member states, showed that citizens of countries in which corruption was rampant were more likely to trust the EU, identifying cases in which, ironically, trust in international institutions increased as the mirror image of their distrust of their own countries due to corruption. Thus, these arguments suggest that corruption and the public’s trust in society are two sides of the same coin, showing just how important the social capital of mutual trust among individuals and groups can be.

5.6 CLOSING SUMMARY, IN LIEU OF A CONCLUSION A diverse range of points at issue are involved in corruption in transition economies, reflecting the complex interrelationships between the corruption and delays or distor­ tions in improvements in various social, economic, and cultural aspects in transitioning societies. In light of these circumstances, this chapter posited hypotheses regarding the main points at issue in research on corruption in transition economies, based on a basic collection of 558 works, testing each of these hypotheses by the degree of support for it found in the literature. Table 5.2 briefly summarizes these findings. Although researchers’ views diverged concerning Hypothesis 1.6 (liberalization reduces the likeli­ hood of corruption) and Hypothesis 2.2 (corruption grows the informal sector), in the former case a nonlinear relationship between liberalization and corruption is conceiv­ able. The fact that analyses that included Central and Eastern Europe, where liberaliza­ tion is more advanced, tended to support the hypotheses while those looking at the former Soviet Union and Asia, where the progress of liberalization has been slower, tended to reject it may be described as evidence supporting this concept of a nonlinear relationship. In addition, in the latter case, it is appropriate that researchers’ views should diverge, given the complexity of the decisive factors affecting the informal econ­ omy. One looks forward to seeing the results of further study in the future. However, while in the 1990s most studies supported Hypothesis 1.5 (privatization of ownership increases the likelihood of corruption), since the start of the twenty-first century, there has been a strong tendency to see privatization of ownership as a factor that restrains corruption. This may be symbolic of the fact that transition economies have overcome the disorder arising from reforms and succeeded in restraining corruption. A number of considerations must be noted in interpreting these results since, generally speaking, it is hard to imagine ethical or social support for corruption as a form of misusing a public position for personal gain or to imagine advocating policies that would support corrup­ tion, regardless of the political system, as shown by the systematic review conducted in 170

CLOSING SUMMARY, IN LIEU OF A CONCLUSION

5.6

Table 5.2 Results of examination of hypotheses

Hypotheses H1.1 Corruption is rarer under an efficient social and economic system. H1.2 Corruption is rarer under conditions of democracy and political stability. H1.3 Corruption is rarer in developed economies or where wages are high. H1.4 Corruption is more common in resource-rich nations. H1.5 Privatization of ownership increases the likelihood of corruption. H1.6 Liberalization reduces the likelihood of corruption. H2.1 Corruption hinders economic growth. H2.2 Corruption grows the informal sector. H2.3 Corruption increases economic disparity and reduces the level of public welfare. H2.4 Corruption worsens governance. H2.5 Corruption hinders transitional reforms. H3.1 The degree of the permeation of communism is connected to corruption. H3.2 Religion and culture are connected to corruption. H3.3 Public distrust of society and systems is interrelated with corruption.

Results of examination of hypotheses

○ ○ ○ ○ × ○ ○

△

○ ○ ○ ○ ○ ○

Note: ○ means support for hypothesis, △ means partial support, × means reject.

this chapter. There is unlikely to be an argument against the statement that the dominant view of corruption rejects it, both socially and economically. Certainly, transition econ­ omies have become “normal countries” (Shleifer and Treisman 2005). On further reflec­ tion, however, while the most important central point of this chapter was to verify the degree of support for the greasing-the-wheels hypothesis—assuming conditions in which the markets of transition economies are not functioning fully and democratic pol­ itical systems have not yet taken root, while the psychological legacy of dependency on and fear of the state remains from the previous socialist history, and the level of per­ formance of their duties by the public officials who manage the apparatus of the state is low, then corruption may be tolerated as the second-best solution—for the most part, the basic collection does not support this hypothesis. However, what has been identified is the presence of an interrelationship in which efficient and transparent social, political, and economic systems reduce corruption, and a low level of corruption increases the quality of these systems. Even the testing and review of the theoretical hypothesis that corruption has no benefit whatsoever suggest that the following points should be noted. First, although systems are important, in transitioning to a market economy, the influence of liberal­ ization and privatization of ownership on corruption has a dual nature. That is, while both liberalization and privatization of ownership have the anti-corruption effect of shrinking the domain of government intervention, at the same time, each can increase the corruption that accompanies a market economy by strengthening compe­ tition in the market. The rich will try to control knowledge and information to shape public opinion to their own benefit, making payments to lobbyists and political dona­ tions for this purpose as they attempt to change the system. Truly, “the market 171

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION

economy … has become a corrupter of knowledge” (Crouch 2016, p. 26). If so, then consistency between policies and correlation between political and economic systems are essential subjects in the research of corruption. Second, there are three layers of corruption: pre-transition corruption, corruption in the transition process, and post-transition corruption. Discrepancies in the scale of cor­ ruption among transition economies still remain, depending on their cultures, histories, values, and systems. As a result, even if we reject the greasing-the-wheels hypothesis for transition economies as a whole, studies that support it, even if few in number, can be confirmed in Russia. The continuing differences between the former Soviet Union on one hand and Central and Eastern Europe and the Baltic states on the other are grounded in the scale of differences in the systems they have developed, in addition to the size of the legacies (debts) they have inherited. Third, if the transition is process dependent (Mizobata and Horie 2013), then the strength of the governing factors of culture and values cannot be overlooked. However, at the same time, systems (markets and governments) that are high in quality from the per­ spectives of transparency and fairness reduce corruption, and democratization itself is considered of utmost importance for anti-corruption systems (Roland 2014). The correl­ ation between system reforms that take time and the development of systems that will not take so much time is an important consideration in corruption research (Roland 2012). As symbolized by the Xi Jinping administration’s policies to root out corruption in China in recent years, in nearly all transition economies, corruption is seen as the social phenomenon most symbolic of public dissatisfaction. For this reason, policy­ makers cannot ignore anti-corruption measures. It is impossible to suppress public dis­ satisfaction by forcing acceptance of groundless disparities, due to the recognition that the misuse of public positions for private gain is an existential threat to the legitimacy of the state itself. Even if, at certain specific times and in certain specific regions, corruption may play a role in greasing the wheels, what is necessary in the vast experimental laboratory of the transition to a market economy is not greasing the wheels but the stability and transparency of systems to satisfy both policy-makers and the public, as well as public sympathy for their promotion.

ACKNOWLEDGMENTS This chapter is a substantially extended and updated version of Suzuki and Mizobata 2018. We thank Ichiro Iwasaki and anonymous referees of the Japanese Journal of Comparative Economics for their helpful comments, suggestions, and insights on the earlier version of this paper.

NOTES 1 Headquartered in Berlin and with 100 branch offices worldwide, Transparency International is a large-scale international NGO that aims to solve corruption issues around the world. Its website is https://www.transparency.org. 172

REFERENCES

2 Papers by Partos (2004) and Radin et al. (2011) are leading examples of those sur­ veying the present states of affairs. Another study (Institute of Social and Economic Studies of Population RAS 2003) asked police officers directly about subjects such as their experiences with corruption and the amounts of money involved. 3 The complete list of the basic collection is provided in Suzuki and Mizobata 2019. 4 This integrated indicator of corruption, developed in 1995, rated 180 countries in 2018 on a scale of 0 (the highest degree of concern about corruption) to 100 (the lowest). It was calculated based on 13 reports from 12 international agencies regarding concern about corruption among businesspeople and national experts over the most recent two years. The global average score is 43, with Denmark scoring the highest, at 88, and Somalia the lowest, at 10. 5 ISSP Research Group 2016.

6 See Ledeneva 2018.

7 For example, a measurement of governance by the World Bank uses more than 350

variables from more than 200 countries, analyzing the six aspects of voice and accountability, political stability and absence of violence, government effectiveness, regulatory quality, rule of law, and control of corruption (Kaufman 2005). A succession of other attitude surveys, including the International Social Survey Programme and World Values Survey, which compare opinion surveys internation­ ally, the International Crime Victimization Study, Eurobarometer, and New Russia Barometer, also may be considered opportunities to promote the empirical study of corruption (Richards 2017, p. 6). 8 Systems of ownership and contractual systems are particularly important in regard to the behavior of private economic actors (Frye 2017). Treisman (2000) stressed the strength of common-law ownership protections, while also broadening the per­ spective to include legal culture, colonial (suzerain state) traditions, and religion. 9 Alvarez et al. 1996 is a study on the taxonomy of political systems that categor­ ized democracy and dictatorship based on the selection of top leaders, selection of members of the legislature, and presence of multiple political parties. 10 The frequency of corruption was high among traffic police officers (EBRD, 2011).

REFERENCES Acemoglu, Daron, and James Robinson (2012) Why Nations Fail: The Origins of Power, Prosperity, and Poverty. Profile Books: London. Ahrend, Rudiger (2005) Can Russia break the “resource curse”? Eurasian Geography and Economics, 46(8), pp. 584–609. Allina-Pisano, Jessica (2010) Social contracts and authoritarian projects in post-Soviet space: The use of administrative resource. Communist and Post-Communist Studies, 43(4), pp. 373–382. Alvarez, Mike, José Antonio Cheibub, Fernando Limongi, and Adam Przeworski (1996) Classifying political regimes. Studies in Comparative International Development, 31(2), pp. 3–36.

173

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION Arnold, Christine, Eliyahu V. Sapir, and Galina Zapryanova (2012) Trust in the institutions of the European Union: A cross-country examination. European Integration online Papers: EIoP, 16(Special Mini-Issue 2), Art. 8. Ateljevic, Jovo, and Jelena Budak (2010) Corruption and public procurement: Example from Croatia. Journal of Balkan and near Eastern Studies, 12(4), pp. 375–397. Barr, Abigail, and Danila Serra (2010) Corruption and culture: An experimental analysis. Journal of Public Economics, 94, pp. 862–869. Belas, Jaroslav, Yuriy Bilan, Valer Demjan, and Juraj Sipko (2015a) Entrepreneurship in SME segment: Case study from the Czech Republic and Slovakia. Amfiteatru Economic, 17(38), pp. 308–326. Belas, Jaroslav, Valer Demjan, Jozef Habanik, Maria Hudakova, and Juraj Sipko (2015b) The business environment of small and medium-sized enterprises in selected regions of the Czech Republic and Slovakia. E + M Ekonomie A Management, 18(1), pp. 95–110. Benevolenskaya, Zlata E. (2010) Trust management as a legal form of managing state prop­ erty in Russia. Review of Central and East European Law, 35(1), pp. 59–76. Bobak, Martin, Mike Murphy, Richard Rose, and Michael Marmot (2007) Societal character­ istics and health in the former communist countries of Central and Eastern Europe and the former Soviet Union: A multilevel analysis. Journal of Epidemiology and Community Health, 61(11), pp. 990–996. Bornstein, Morris (1999) Framework issues in the privatisation strategies of the Czech Republic, Hungary and Poland. Post-Communist Economies, 11(1), pp. 47–77. Braguinsky, Serguey (1999) Enforcement of property rights during the Russian transition: Problems and some approaches to a new liberal solution. Journal of Legal Studies, 28(2), pp. 515–544. Brovkin, Vladimir N. (2003) Corruption in the 20th century Russia. Crime, Law and Social Change, 40(2/3), pp. 195–230. Campos, Nauro F., Ralitza Dimova, and Ahmad Saleh (2010) Whither corruption? A quantitative survey of the literature on corruption and growth. Discussion Paper No. 5334, IZA Institute of Labor Economics: Bonn. Chen, Qiangbing (2008) Rent seeking and economic liberalization: Why are China and Vietnam different from Eastern Europe? Journal of the Asia Pacific Economy, 13(2), pp. 168–179. Christev, Atanas, and Felix FitzRoy (2002) Employment and wage adjustment: Insider–out­ sider control in a Polish privatization panel study. Journal of Comparative Economics, 30(2), pp. 251–275. Crouch, Colin (2016) The Knowledge Corrupters: Hidden Consequences of the Financial Takeover of Public Life. Polity Press: Cambridge. Darden, Keith (2008) The integrity of corrupt states: Graft as an informal state institution. Politics & Society, 36(1), pp. 35–59. Denisova-Schmidt, Elena, and Martin Huber (2010) Regional differences in perceived cor­ ruption among Ukrainian firms. Eurasian Geography and Economics, 55(1), pp. 10–36. Desai, Raj M., and Itzhak Goldberg (2001) The politics of Russian enterprise reform: Insiders, local governments, and the obstacles to restructuring. World Bank Research Observer, 16(2), pp. 219–240. Dimitrova-Grajzl, Valentina (2007) The great divide revisited: Ottoman and Habsburg leg­ acies on transition. Kyklos, 60(4), pp. 539–558. Dininio, Phyllis, and Robert Orttung (2005) Explaining patterns of corruption in the Russian regions. World Politics, 57(4), pp. 500–529.

174

REFERENCES Duvanova, Dinissa (2014) Economic regulations, red tape, and bureaucratic corruption in post-communist economies. World Development, 59, pp. 298–312. Earle, Beverley (2000) Bribery and corruption in Eastern Europe, the Baltic States, and the Commonwealth of Independent States: What is to be done? Cornell International Law Journal, 33(3), pp. 483–513. EBRD (European Bank for Reconstruction and Development) (2010) Life in Transition II. EBRD: London. EBRD (2011) Transition Report 2011: Crisis in Transition: The People’s Perspective. EBRD: London. EBRD (2016) Life in Transition III. EBRD: London. Frye, Timothy (2017) Property Rights and Property Wrongs: How Power, Institutions, and Norms Shape Economic Conflict in Russia. Cambridge University Press: Cambridge. Gherghina, Sergiu, and Mihail Chiru (2013) Taking the short route: Political parties, funding regulations, and state resources in Romania. East European Politics and Societies, 27(1), pp. 108–128. Giordano, Christian (2010) Multiple modernities in Bulgaria: Social strategies of capitalist entrepreneurs in the agrarian sector. Eastern European Countryside, 16, pp. 5–24. Goel, Rajeev K., Jelena Budak, and Edo Rajh (2015) Private sector bribery and effectiveness of anti-corruption policies. Applied Economics Letters, 22(10), pp. 759–766. Gokcekus, Omer, Eva Muchova, and Zuzana Brincikova (2015) Level and quality of openness and corruption in the ECA countries. Applied Economics Letters, 22(16), pp. 1340–1344. Grosfeld, Irena, and Ekaterina Zhuravskaya (2015) Cultural vs. economic legacies of empires: Evidence from the partition of Poland. Journal of Comparative Economics, 43(1), pp. 55–75. Grzymała-Busse, Anna (2003) Political competition and the politicization of the state in East Central Europe. Comparative Political Studies, 36(10), pp. 1123–1147. Guriev, Sergei, Evgeny Yakovlev, and Ekaterina Zhuravskaya (2010) Interest group politics in a federation. Journal of Public Economics, 94(9/10), pp. 730–748. Gylfason, Thorvaldur (2000) Resources, agriculture, and economic growth in economies in transition. Kyklos, 53(4), pp. 545–579. Hagan, John, and Detelina Radoeva (1997) Both too much and too little: From elite to street crime in the transformation of the Czech Republic. Crime, Law and Social Change, 28(3/4), pp. 195–211. Harris, Nigel, and David Lockwood (1997) The war-making state and privatisation. Journal of Development Studies, 33(5), pp. 597–634. Hendley, Kathryn (2010) Mobilizing law in contemporary Russia: The evolution of disputes over home repair projects. American Journal of Comparative Law, 58(3), pp. 631–678. Heusala, Anna-Liisa (2013) Changes of administrative accountability in Russian transitions. Review of Central and East European Law, 38(3/4), pp. 267–293. Holmes, Leslie (2008) Corruption and organised crime in Putin’s Russia. Europe-Asia Stud­ ies, 60(6), pp. 1011–1031. Horne, Cynthia M. (2012) Assessing the impact of lustration on trust in public institutions and national government in Central and Eastern Europe. Comparative Political Studies, 45(4), pp. 412–446. Horne, Cynthia M. (2014) Lustration, transitional justice, and social trust in post-communist countries. Repairing or wresting the ties that bind? Europe-Asia Studies, 66(2), pp. 225–254.

175

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION Hung, Chi-Hsiou D., Yuxiang Jiang, Frank Hong Liu, Hong Tu, and Senyu Wang (2017) Bank political connections and performance in China. Journal of Financial Stability, 32, pp.57–69. Holtbrügge, Dirk, Nicola Berg, and Jonas F. Puck (2007) To bribe or to convince? Political stakeholders and political activities in German multinational corporations. International Business Review, 16(1), pp. 47–67. Institute of Social and Economic Studies of Population RAS (2003) Economic Activity of Police Officers. Volumes 1, 2, Higher School of Economics: Moscow (Russian). ISSP Research Group (2016) International social survey programme: Citizenship II: ISSP 2014 (available at https://dbk.gesis.org/dbksearch/sdesc2.asp?ll=10¬abs=&af=&nf= &search=&search2=&db=e&no=6670). Iwasaki, Ichiro, and Taku Suzuki (2007) Transition strategy, corporate exploitation, and state capture: An empirical analysis of the former Soviet states. Communist and PostCommunist Studies, 40(4), pp. 393–422. Iwasaki, Ichiro, and Taku Suzuki (2012) The Determinants of Corruption in Transition Economies. Economics Letters,114(1), pp. 54–60. Jacobs, Jane (1992) Systems of Survival: A Dialogue of the Moral Foundations of Com­ merce and Politics. Random House: New York. Jancsics, David (2015) “A friend gave me a phone number”: Brokerage in low-level corrup­ tion. International Journal of Law Crime and Justice, 43(1), pp. 68–87. Jetter, Michael, Alejandra Montoya Agudelo, and Andres Ramirez Hassan (2015) The effect of democracy on corruption: Income is key. World Development, 74, pp. 286–304. Johnson, Simon, Daniel Kaufmann, John McMillan, and Christopher Woodruff (2000) Why do firms hide? Bribes and unofficial activity after communism. Journal of Public Econom­ ics, 76(3), pp. 495–520. Kaufman, Daniel (2005) Back to basics:10 myths about governance and corruption. Finance and Development, 42(3) (available at http://www.imf.org/external/pubs/ft/fandd/2005/09/ basics.htm). Kneen, Peter (2000) Political corruption in Russia and the Soviet legacy. Crime, Law and Social Change, 34(4), pp. 349–367. Ko, Kilkon, and Seong-Gin Moon (2014) The relationship between religion and corruption: Are the proposed causal links empirically valid? International Review of Public Adminis­ tration, 19(1), pp. 44–62. Kolossov, Vladimir, and Gerard Toal (2007) An empire’s fraying edge? The North Caucasus instability in contemporary Russian geopolitical culture. Eurasian Geography and Eco­ nomics, 48(2), pp. 202–225. Kolstad, Ivar, and Arne Wiig (2011) Does democracy reduce corruption? Working Paper No. WP2011-4, Chr. Michelsen Institute: Bergen. Larsson, Tomas (2006) Reform, corruption, and growth: Why corruption is more devastating in Russia than in China. Communist and Post-Communist Studies, 39(2), pp. 265–281. Lavigne, Marie (1995) The Economics of Transition: From Socialist Economy to Market Economy. Macmillan Press: London. Ledeneva, Alena (2018) (ed.), The Global Encyclopaedia of Informality. Volume 1: Towards Understanding of Social and Cultural Complexity. UCL Press: London. Ledyaeva, Svetlana, Päivi Karhunen, and Riitta Kosonen (2012) The global economic crisis and foreign investment in Russia from the EU: Empirical evidence from firm-level data. Eurasian Geography and Economics, 53(6), pp. 772–789.

176

REFERENCES Ledyaeva, Svetlana, Päivi Karhunen, and Riitta Kosonen (2013) Birds of a feather: Evidence on commonality of corruption and democracy in the origin and location of foreign invest­ ment in Russian regions. European Journal of Political Economy, 32, pp. 1–25. Linhartová, Veronika, and Jolana Volejniková (2015) Quantifying corruption at a subnational level. E + M Ekonomie A Management, 18(2), pp. 25–39. Maloney, Wendy Hall, and Robert J. Kelly. (2000) Notes on school-based crime fighting: International lessons in moral education. Journal of Social Distress and the Homeless, 9(2), pp. 71–90. Minagawa, Yuka (2013) Inequalities in healthy life expectancy in Eastern Europe. Population and Development Review, 39(4), pp. 649–671. Mizobata, Satoshi, and Norio Horie (2013) The Transition to a Market Economy and Process Dependency. Economic Review, 64(4), pp. 338–352. Neshkova, Milena I., and Tatiana Kostadinova (2012) The Effectiveness of administrative reform in new democracies. Public Administration Review, 72(3), pp. 324–333. Nesvetailova, Anastasia (2004) Coping in the global financial system: The political economy of nonpayment in Russia. Review of International Political Economy, 11(5), pp. 995–1021. Nye, Joseph S. (1967) Corruption and political development: A cost-benefit analysis. Ameri­ can Political Science Review, 61(2), pp. 417–427. Obydenkova, Anastassia, and Alexander Libman (2015) Understanding the survival of post-communist corruption in contemporary Russia: The influence of historical legacies. Post-Soviet Affairs, 31(4), pp. 304–338. Offe, Claus (2004) Political corruption: Conceptual and practical issues. In Janos Kornai and Susan Rose-Ackerman (eds.), Building a Trustworthy State in Post-Socialist Transi­ tion. Palgrave Macmillan: New York, pp. 77–99. Partos, Gabriel (2004) Serbia: Apathy rules. World Today, 60(1), pp. 18–19. Popov, Vladimir (2012) Russia: Austerity and deficit reduction in historical and comparative perspective. Cambridge Journal of Economics, 36(1), pp. 313–334. Radin, Dagmar, Aleksandar Džakula, and Vanesa Benković (2011) Health care issues in Cro­ atian elections 2005–2009: Series of public opinion surveys. Croatian Medical Journal, 52(5), pp. 585–592. Research and Training Institute of the Ministry of Justice of Japan (2008) An Overview of the International Crime Victims Survey (2004/2005). Research Department Report 39: International Comparison of the Second Crime Victims Survey (Report No. 2) (Centered on Developed Countries), pp. 7–62 (available at http://www.moj.go.jp/content/ 000049668.pdf). Richards, Lindsay (2017) Using survey methods to research corruption, in Schwicker­ ath, Anna. K., Aiysha Varraich and Laura-Lee Smith (eds.), How to Research Corrup­ tion? Conference Proceedings, Interdisciplinary Corruption Research Network, pp. 4–12 (available at https://www.icrnetwork.org/what-we-do/icr-forums/icr-forum­ amsterdam-2016). Roland Gérard (2012) The long-run weight of communism or the weight of long-run history? In Gérard Roland (ed.), Economies in Transition: The Long-Run View. Palgrave Macmil­ lan: London, pp. 153–171. Roland Gérard (2014) Transition in historical perspective. In Anders Åslund and Simeon Djankov (eds.), The Great Rebirth: Lessons from The Victory of Capitalism over Communism. Peterson Institute for International Economics: Washington DC, pp. 251–271. Rose-Ackerman, Susan (1999) Political corruption and democracy. Connecticut Journal of International Law, 14(2), pp. 363–378.

177

C H A P T E R 5

C H A P T E R 5

SOCIAL CONFUSION AND CORRUPTION Rose-Ackerman, Susan, and Bonnie J. Palifka (2016) Corruption and Government: Causes, Consequences, and Reform. Second edition, Cambridge University Press: Cambridge. Safavian, Mehnaz S., Douglas H. Graham, and Claudio Gonzalez-Vega (2001) Corruption and microenterprises in Russia. World Development, 29(7), pp. 1215–1224. Sarwar, Saima (2013) An empirical investigation between trade liberalization and corruption: A panel data approach. Journal of Economics and Sustainable Development, 4(3), pp. 179–189. Seleim, Ahmed, and Nick Bontis (2009) The relationship between culture and corruption: A cross-national study. Journal of Intellectual Capital, 10(1), pp. 165–184. Shadabi, Leila (2013) The impact of religion on corruption. Journal of Business Inquiry, 12, pp. 102–117. Sharafutdinova, Gulnaz, and Jevgenijs Steinbuks (2017) Governors matter: A comparative study of state-business relations in Russia’s regions. Economics of Transition, 25(3), pp. 471–493. Shleifer, Andrei, and Daniel Treisman (2005) A normal country: Russia after communism. Journal of Economic Perspective, 19(1), pp. 151–174. Škrbec, Jure, and Bojan Dobovšek (2013) Corruption capture of local self-governments in Slovenia. Lex Localis-Journal of Local Self-Government, 11(3), pp. 615–630. Suzuki, Taku, and Satoshi Mizobata (2018) Lapse of morality and corruption in transition economies: A systematic review. Japanese Journal of Comparative Economics, 55(1), pp. 23–43 (Japanese). Suzuki, Taku, and Satoshi Mizobata (2019) Social Confusion and corruption: Investigating the causes and effects of a breakdown of ethics. IER Discussion Paper Series No. A690, Institute of Economic Research, Hitotsubashi University: Tokyo. Tavares, Samia Costa (2007) Do rapid political and trade liberalizations increase corruption? European Journal of Political Economy, 23(4), pp. 1053–1076. Treisman, Daniel (2000) The causes of corruption: A cross-national study. Journal of Public Economics, 76, pp. 399–457. Treisman, Daniel (2007) The causes of corruption: A cross-national study. In Erik Berglof and Gérard Roland (eds.), The Economics of Transition: The Fifth Nobel Symposium in Economics. Palgrave Macmillan: New York, pp. 251–271. Votápková, Jana, and Milan Žák (2013) Institutional efficiency of selected EU & OECD coun­ tries using DEA-like approach. Prague Economic Papers, 22(2), pp. 206–223. Wei, Yehua Dennis (2015) Zone fever, project fever: Development policy, economic transition, and urban expansion in China. Geographical Review, 105(2), pp. 156–177. Williams, Colin C. (2015) Evaluating cross-national variations in envelope wage payments in East-Central Europe, Economic and Industrial Democracy, 36(2), pp. 283–303. Yano, Makoto (2008) Competitive fairness and the concept of a fair price under Delaware law on M&A. International Journal of Economic Theory, 4(2), pp. 175–190. Yeager, Matthew G. (2012) The CIA made me do it: Understanding the political economy of corruption in Kazakhstan. Crime, Law and Social Change, 57(4), pp. 441–457. Yessenova, Saulesh (2012) The Tengiz oil enclave: Labor, business, and the state. PoLAR: Political and Legal Anthropology Review, 35(1), pp. 94–114. Zaloznaya, Marina (2015) Does authoritarianism breed corruption? Reconsidering the rela­ tionship between authoritarian governance and corrupt exchanges in bureaucracies. Law and Social Inquiry, 40(2), pp. 345–376.

178

6

Privatization, corporate ownership, and enterprise restructuring Ichiro Iwasaki and Satoshi Mizobata

6.1 INTRODUCTION “Privatization is transition” (Brada 1996). While this phrase is obviously an exagger­ ation, in the sense that without privatization, the former socialist transition economies in Central and Eastern Europe (CEE) and the former Soviet Union (FSU) could not have changed to a market-oriented system, no one particularly objects to accepting it as part of the truth. This is because state ownership and state planning were positioned at the heart of the socialist economic system. Systemic transformation is the process of replacing these elements with those of the capitalist economic system, namely private ownership and market principles. If the comprehensive nationalization of the means of production was the point of transformation to a socialist economy, their re-privatization can be regarded as a giant leap toward a market economy. However, the “leaps” observed in CEE and FSU countries exhibited their own diversity, being strongly affected by historical preconditions in each country, international circumstances, and the motives of foreign governments and multinational firms. It is probably no exaggeration to say that the question of whether a privatization policy would improve the performance of former socialist firms has been the issue of most interest and a source of debate in the field of transition economics since 1989. Theoretically speaking, if the dysfunction of state-owned firms could be regarded as the fundamental cause of the stagnation and demise of the socialist planned system, their privatization should have improved firm performances. In fact, as Figure 6.1 shows, according to assessments of the transition progress by the European Bank for Reconstruction and Development (EBRD), at the level of national economies, a close, positive correlation can be observed between the degree of success of the privatization policy and the degree of progress with enterprise restructuring. In other words, accord­ ing to the line of best fit shown in the figure, a marginal increase of 1 in the mean of

C H A P T E R

PRIVATIZATION

6

Figure 6.1 Relationship between privatization policy and enterprise reform in transition economies Notes: a

Country abbreviations: AL — Albania; AM — Armenia; AZ — Azerbaijan; BA — Bosnia and Herzegovina; BG — Bulgaria; BY — Belarus; CZ — Czech Republic; EE — Estonia; GE — Georgia; HR — Croatia; HU — Hungary; KG — Kyrgyz Republic; KZ — Kazakhstan; LT — Lithuania; LV — Latvia; MD — Moldova; ME — Montenegro; MK — FYR Macedonia; PL — Poland; RO — Romania; RS — Serbia; RU — Russia; SI — Slovenia; SK — Slovakia; TJ — Tajikistan; TM — Turk­ menistan; UA — Ukraine; UZ — Uzbekistan; YK — Kosovo. ●, ■, and ▲ represent CEE EU member countries, CEE nonEU member countries, and FSU states, respectively.

b

The index takes the range between 1.00 (representing little or no change from a rigid, centrally planned economy) and 4.33 (representing the standards of an industrialized market economy). The figure for the Czech Republic is from 2007. The figure for other countries is from 2013.

c

Figures in parentheses beneath the regression coefficients of the approximate straight line are standard errors. *** and ** denote statistical significance at the 1% and 5% levels, respectively.

Source: Authors’ illustration based on data obtained from the EBRD website (http://www.ebrd.com/pages/homepage. shtml)

180

INTRODUCTION

6.1

the small-scale and large-scale privatization indicator leads to a 1.108-point increase in the enterprise restructuring indicator, which surpasses the statistical significance level of 1%. Nevertheless, microeconomic empirical results produced by a series of earlier studies do not seem to support that theory. For instance, a systematic review of the Russian lit­ erature on enterprise restructuring and corporate governance by Iwasaki (2007) revealed that, while numerous studies have found that private and former state-owned privatized firms are relatively superior to state enterprises in terms of productivity and financial performance (Kapelyushnikov 2000; Linz 2002), other studies have not confirmed a statistically significant correlation between post-privatization ownership and firm per­ formance (Jones et al. 1998; Judge et al. 2003). Furthermore, an analysis of many stud­ ies produced the surprising finding that companies still under state control are actually performing better than privatized firms (Bevan et al. 2001; Brown and Earle 2004). Faced with the reality that solid empirical evidence of the privatization effect on firm restructuring cannot be obtained from a simple comparison of state firms and private firms, researchers have turned their attention to the diversity of the “leap” process. Although numerous factors led to this diversity, the one they focused on first was the diversity of new owners following enterprise privatization. As we will discuss later, the possibility that differences between insider and outside investors, differences between insider managers and employees, and differences in the types and nationalities of outside investors could affect firm performance is a serious issue in the field of corporate finance theory, where it has led to vigorous debate. In the field of transition economics, researchers had strongly recognized the importance of this viewpoint by the mid-1990s, when the initial phase of the privatization of state companies in almost all CEE and FSU countries was drawing to a close. As a result, empirical studies have challenged the comparative analysis of a variety of company owners from around this time. Empirical evidence has grad­ ually accumulated in this way, which has greatly promoted the development of knowledge about which types of owner most improve firm performance. However, it is also true that the greater the empirical evidence, the more ambiguous the big pic­ ture becomes. Djankov and Murrell 2002 and Estrin et al. 2009 are systematic reviews that attempt to overcome the limitations of piecemeal empirical studies. In this chapter, we will use the largest database of literature, which includes numerous previous studies not covered by these two articles, and a more methodologically thorough and refined meta-analysis to shed light on the overall conclusions that can be reached by studies of the interrelationship between post-privatization ownership and firm performance in transition economies over the past quarter century. This is the primary objective of this chapter. Numerous researchers have also studied differences between countries as seen in the nature of their privatization policies. CEE and FSU countries privatized firms using a combination of four main methods: (1) the voucher system, (2) management and employee buyouts (MEBOs), (3) direct sales to strategic investors, and (4) the auction system. As shown in Table 6.1, the priority given to each privatization method and the ways in which the methods were combined differed greatly from 181

C H A P T E R 6

C H A P T E R

PRIVATIZATION Table 6.1 Privatization method and private sector size in transition economies Privatization method (I: primary; II: secondary)

6 Country name

Vouchers

MEBOs

Direct sales

Albania Armenia Azerbaijan Belarus Bosnia and Herzegovina Bulgaria Croatia Czech Republic Estonia Macedonia Georgia Hungary Kazakhstan Kyrgyz Republic Latvia Lithuania Moldova Montenegro Poland Romania Russia Serbia Slovakia Slovenia Tajikistan Turkmenistan Ukraine Uzbekistan

II

I II

I

II II I II II I II

I I II I I

I I II II I II I I I

I

I

II I II II I I

II

II I

II II

Auctions

I II I II I

I II II II I II II II I I II II

I

Private sector share in GDP in 2010 (%) 75 75 75 30 60 75 70 80 80 70 75 80 65 75 70 75 65 65 75 70 65 60 80 70 55 25 60 45

Source: EBRD (2004) and EBRD website (http://www.ebrd.com)

country to country. Furthermore, as the figures for the private sector’s share of gross domestic product (GDP) in 2010 illustrate, striking differences also emerged among countries in their speed of implementing the privatization policy. To address these points, comparisons of a wide range of transition countries are indispensable. It is, therefore, obvious that empirical studies targeting only specific countries or regions cannot deliver firm conclusions. To overcome this difficulty, Djankov and Murrell (2002) and Estrin et al. (2009) attempted systematic reviews of the earlier literature focusing on differences between the CEE and FSU regions. In this chapter, we will also try to investigate the possibility that differences in the pri­ vatization policy methods and the speed of implementation affected the empirical findings of the previous research. This is the secondary objective of this chapter.1 To achieve the above two objectives, we will conduct a meta-analysis using 2,894 estimates drawn from 121 previous studies. As a result, we found that the baseline

182

POST-PRIVATIZATION OWNERSHIP

6.2

estimation of a meta-regression model indicated the superior impact of foreign own­ ership on firm performance as compared to state and domestic private entities. How­ ever, it did not go so far as to comprehensively verify the series of hypotheses concerning the interrelationship between different ownership types. The estimation of an extended meta-regression model that explicitly controls for the idiosyncrasies of transition economies and privatization policies suggested that differences between countries in location, privatization method, and policy implementation speed strongly influences the link between post-privatization ownership structure and firm performance. We also found that these factors not only cause the remarkable gap between countries in terms of ex post improvement in firm performance but also significantly affect the inter­ relationship between foreign investors, domestic outsider owners and firm managers, and the relative superiority of various domestic outsiders. Definitive evidence of the harm caused to ex post firm performance by voucher privatization is one of the most noteworthy empirical findings obtained from the meta-analysis in this work. The remainder of this chapter is organized as follows: in the next section, we pre­ sent testable hypotheses to verify by meta-analysis. In Section 6.3, we describe how we retrieved and selected literature for the meta-analysis, provide an overview of the collected estimates, and describe the methodology of meta-analysis. In Section 6.4, we attempt a meta-synthesis of the collected estimates, while in Section 6.5, we esti­ mate a meta-regression model to examine the possible heterogeneity of the extant literature. In Section 6.6, an extended meta-regression model that takes into account the idiosyncrasies of transition countries and privatization policies is estimated. In Section 6.7, we assess the presence and degree of publication selection bias (PSB) in this research field and, finally, in Section 6.8, we summarize major findings obtained from the meta-analysis and conclude.

6.2 POST-PRIVATIZATION OWNERSHIP AND FIRM PERFORMANCE: LITERATURE REVIEW AND TESTABLE HYPOTHESES In this section, through a comprehensive literature review, we present the hypotheses regarding the relationship between post-privatization ownership and firm performance to be verified in this chapter. To tackle the two research objectives stated in the Introduction, in Subsection 6.2.1, we focus on the general debate concerning the relative superiority/ inferiority of different owner types, while in Subsection 6.2.2, we explore various factors that are peculiar to the privatization of enterprises carried out in transition economies. 6.2.1 Ownership and firm performance To complete a systemic transformation from a planned system to a market economy, the large-scale transfer of ownership from the state to the private sector is unavoidable. Therefore, the positioning of the interrelationship between ownership and firm per­ formance as one of the focal points of transition economics was a natural development

183

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

(IMF 2014). At the root of this so-called “privatization debate” is a belief that has also been supported by the outcome of research in comparative economic systems and corporate finance. In other words, there is a strong belief that the management and production activities of privately owned firms are far more efficient than those of state-owned enterprises (Roland 2008). Therefore, researchers of transition economies initially focused their efforts on showing that the performance of private firms in CEE and FSU countries is vastly superior to that of firms remaining under state ownership. This is because, based on the premise of the collapse of socialism and the transition to capitalism in this region, the proposition that “an economy in which all major deci­ sions on investment, employment, and production are left to private firms will outper­ form a mixed economy where governments play a significant role in such decisions” (Quiggin 2010, p. 189) was self-evident to them. Actually, the notion that state enter­ prises cannot achieve high efficiency has been advocated repeatedly in the fields of public choice theory and financial economics. However, this proposition has not necessarily been completely proven in the field of economics, even in research on advanced economies. In fact, Bös argued: Many empirical studies comparing private and public firms confirm that private enterprises are more efficient than public enterprises producing the same goods or very close substitutes, given the same or very similar technology, regulatory con­ straints, and financial capabilities. As could be expected, however, some counterevidence also exists which shows exactly the opposite. Moreover, some further empirical studies report ambiguous results: the public firm is more efficient according to one indicator, whereas the private firm is more efficient according to another indicator. (Bös 1991, p. 7) Furthermore, regardless of the distinction between public and private ownership, a “separation of ownership and control,” in the sense that business execution is per­ formed by professional managers, is observed in both cases. In addition, the nature of the organizational structure is also vital. Because of the above-mentioned facts, the proposition that private ownership is superior to state ownership did not neces­ sarily match the expectations for transition countries (Stiglitz 1994). Actually, as stated in the Introduction to this chapter, while numerous studies have been pub­ lished that identify relatively good firm performance by private companies as com­ pared with state-owned enterprises, there have also been studies that have not found a significant difference between the two and studies that have demonstrated the rela­ tive superiority of state enterprises. Even though such findings have not threatened the dominant view that private owners are superior to the state, by the mid-1990s, it had already become fairly clear that approaches based on a “state versus private” dichotomy have limits. Therefore, since the late 1990s, most researchers have begun to focus on the diversity of firm owners in the post-privatization period. Four categories of firm ownership can be applied to CEE and FSU economies: (1) the state, (2) insiders: that is, firm managers and employees (workers’ associations), (3) domestic private-sector investors, and (4) foreign investors. Of these types, 184

POST-PRIVATIZATION OWNERSHIP

6.2

insiders garnered the most attention. One main reason for this was that privatizations favoring MEBOs or insiders took place in many transition countries. This type of firm ownership by insiders is a full-fledged form of private owner­ ship. There is no doubt that by making their property rights clear, insiders are strongly motivated to make profits. However, the impact on firm performance is not entirely positive. For example, insiders tend to feel negatively toward restructurings involving the mass dismissal of workers. There is also a risk that managers will go along with the employees’ wishes and adopt a short-term, opportunistic outlook, choosing to raise wages rather than invest for the future. However, there is also counterevidence to the above “insider inefficiency hypoth­ esis.” In fact, the histories of employee stock ownership plans (ESOPs) in the USA, a common decision-making system in Germany, and the in-house labor market in Japan strongly suggest that employee ownership and management participation by insiders do not directly worsen performance (Frydman and Rapaczynski 1994). In other words, in these advanced economies, an “incentive compatibility,” established due to the connection between the promotion of insiders and improved profits and firm performance, clearly enhances management. Therefore, if the positive impact of the incentive compatibility exceeds the negative impact of possible inappropriate management decisions, insiders may make better owners than the state. When investigating the impact of insider ownership on firm performance, a strict distinction must be made between the two, except in cases where man­ agers and rank-and-file employees (workers’ associations) collude with each other. Managers from the socialist era (so-called “red executives”) may not be adequately suited to the new management environment in the transition period; however, at the very least, they are clearly more skilled than ordinary workers. Furthermore, it is easy to imagine that the financial and material benefits and the social reputation received by managers when they improve performance will be more than it would be for rank-and-file employees. In other words, all else being equal, the motivation for managers to restructure their firms is much higher than it is for employees. For this reason, the hypothesis that insider managers are superior to insider employees as owners of privatized firms is widely accepted (Earle and Estrin 1996). However, the motivation with regard to management by insider employees will differ depending on the scale of their ownership. Gener­ ally, controlling owners with a majority stake are more interested in management than are minority owners. In contrast to this ambiguous effect of insider ownership on firm performance, the effect of ownership by outsiders has a clearer and more active significance. Frydman et al. concluded that their research results were consistent with the hypothesis that the superior results of product restructuring by firms privat­ ized to outside owners are a function of their greater willingness to accept risks and their freedom to make decisions without having to justify them to employee owners or a hierarchy of state officials. (Frydman 2006, p. 218)

185

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

As can be seen from this remark, economists widely believe that outside investors, who are free of the internal self-protective interests that restrict insiders, inevitably try harder than the state or insiders to improve the management of the companies they invest in. Nevertheless, because outside investors are extremely economic entities, the degree of impact on firm performance will vary greatly depending on their identity (Frydman et al. 2007). Particularly in the context of transition economies, a great deal of attention has been paid to two matters: (1) differences between individual investors and institutional investors and (2) differences based on nationality. It has been argued that while individual investors are only minority shareholders, institutional investors have a strong tendency to be major shareholders; arguably, institutional investors are also more motivated by profit than are individual investors and, therefore, apply more pressure on firm managers to improve performance. As a result, the prediction holds that institutional investors will behave more proactively and effectively as corporate restructurers than individual investors (Vittas and Michelitsch 1996; Stark and Bruszt 1998). However, quite a number of researchers hold reserved views concerning financial institutions. In CEE and FSU countries, financial institutions, mainly commercial banks, by becoming shareholders in or creditors of privatized companies, have been expected to impose hard budget constraints on the companies and, therefore, strongly stimulate firm restructuring. However, it was extremely difficult in these countries to have well-performing commercial banks under the two-layered banking system (Iwa­ saki and Uegaki 2019), while direct and indirect government protection meant that a paternalistic relationship between state-owned banks and large formerly stateowned firms often remained. Due to the above process, financial institutions in tran­ sition economies did not succeed in obtaining the skills and incentives they needed to perform financial intermediary functions, monitoring functions, and asset manage­ ment. As a consequence, it has been argued that, far from becoming capable players of restructuring privatized firms, they ended up forming financial groups that were dependent on collusion with firms (Frydman and Rapaczynski 1994; Dittus and Prowse 1996). Nevertheless, because a series of empirical studies of financial and industrial groups in Russia have given high marks to commercial banks as restructur­ ing promoters (Brown et al. 1999; Perotti and Gelfer 2001; Dolgopyatova et al. 2009), it is highly likely that the validity of the above problems with financial insti­ tutions will differ considerably by country and era. Therefore, the foundations of the generally accepted theory concerning the relative superiority of institutional investors over individual investors as corporate restructuring promoters in the post-privatization period are not regarded to have been overturned. A certain level of consensus has been established among researchers of transition economies concerning differences between domestic and foreign investors. In fact, from the initial phase of the transition, it was strongly believed that foreign investors could have a greater impact on firm restructuring than domestic investors. This is because by bringing in not only vast sums of capital but also advanced production technology and management know-how, as well as other forms of tacit knowledge, 186

POST-PRIVATIZATION OWNERSHIP

6.2

foreign investors have a lot of potential for dramatically improving the productivity and efficiency of the firms they own (see Chapter 9 in this book). Furthermore, in the process of being incorporated into the international division-of-labor structure of multinational firms through direct investment, domestic firms have the potential to secure a robust operational foundation that cannot be compared with what they had during the socialist era (Dunning 1986; Blomstrom and Wolff 1994; Kogut 1996). These clearly could not have been achieved by domestic investors in those countries under socialism, so many researchers predicted that the superiority of foreign invest­ ors over domestic investors would be universally observed in every transition economy. This debate can be summarized as follows: researchers agree on three hypotheses concerning the impact of post-privatization ownership structure on firm performance, namely (1) that private ownership is superior to state ownership, (2) that outside investors outperform insiders, and (3) that foreign investors work better than domes­ tic investors. Furthermore, two hypotheses have slightly less support than the former three, namely (4) that managers reform their owned companies more intensively than employees would and (5) that domestic institutional investors surpass domestic individual investors in improving the performance of companies in which they have invested. Therefore, the main purpose of the meta-analysis in this chapter is to verify whether these five predictions have been empirically proven in the previous research on CEE and FSU countries. 6.2.2 Specific factors in transition economies Privatization in CEE and FSU countries was extremely different from that in advanced economies in terms of its breadth and depth. In other words, the privatiza­ tion policies in these countries were more than just the transfer of ownership from the state to the private sector. It represented a process through which a system of private ownership was reintroduced to the society and, at the corporate level, encom­ passed the elimination of the command-economy system and the infiltration of the principle of decision making based on economic rationality and the profit motive (Frydman and Rapaczynski 1994; Shleifer and Vishny 1994). It also involved a process whereby systems and structures—including legal systems, rules, and cus­ toms—were reconstructed in a broad-based fashion (Dewatripont and Roland 1996). In other words, in transition countries, privatization was an extremely complex social process that would fundamentally transform the economic system. The objectives of the privatization policy also became ambiguous. In addition to the original policy goals of establishing a class of private owners and developing firm managers adapted to a market economy, other goals, such as securing tax rev­ enue to fund structural reform and achieving macroeconomic stability, were incorp­ orated into the implementation objectives. Furthermore, the privatization policy was heavily used as a means for reformers to obtain political support from citizens and, conversely, for anti-reformers, such as former communist party officials, to reclaim power (Åslund 2013). As a result of these ambiguous political intentions, the 187

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

privatization methods adopted by the governments of CEE and FSU countries exhib­ ited a great deal of diversity: significant differences between these countries in the speed of policy implementation also emerged. Moreover, the foundations for imple­ menting the privatization policies, namely preconditions such as proximity to the EU and maturity as a civic society, were decidedly different among the transition coun­ tries. These factors are highly likely to have had some influence on the incentive structure and effort level of company-owning entities. Therefore, in the context of transition economies, it could be extremely important to consider these points when examining the relationship between ownership structure and firm performance. In this subsection, therefore, we will begin by discussing the nature of region-specific factors for transition countries, after which we will explore differences in privatiza­ tion methods and the speed of policy implementation. Finally, we will present an additional testable hypothesis for the meta-analysis of this chapter. With regard to region-specific factors for transition economies, numerous researchers have focused their attention on differences between CEE and FSU regions. This is because there would be major differences in terms of the imple­ mentation process and results of the privatization policy between CEE and Baltic countries, where the nature of transformation was heavily influenced by the eastern EU enlargement process, and FSU countries, which were not a part of this and fol­ lowed their own paths to becoming market economies. The results of EBRD assessment, which are presented in Figure 6.1 and Table 6.1, clearly illustrate this. In fact, the CEE/Baltic states took advantage of their favorable geopolitical situ­ ation, namely their proximity to Western Europe. By establishing legal and other systems that met EU standards, they created a stable foundation for policies, including privatization, aimed at establishing themselves as market economies. They also paved the way for drawing foreign investors, most notably Western multinational companies, into the privatization process. A dramatic improvement in fairness in system design and transparency in the policy decision-making pro­ cess was also effective in reducing information asymmetry between foreign invest­ ors and insiders. Furthermore, in these countries, where civic society had reached a certain level of maturity, the management abilities of company owners were well respected irrespective of differences in their nationalities and other background details. In this regard, Djankov insisted that “foreign investors and workers became better owners in Eastern Europe than in the former Soviet Union” (2014, p. 191). This may be a result of synergies among these factors. In contrast, the political environment in the non-Baltic FSU countries was fragile. Unlike CEE countries, they were under no outside pressure to meet EU membership criteria, and, as a result, practices such as property rights were not properly estab­ lished, and rule changes were very frequent. In that sense, the situation in the FSU countries was very unstable. Furthermore, the system design and implementation process under privatization policies were extremely opaque, and opportunistic, rentseeking behavior by politicians and bureaucrats, as well as state capture by managers and entrepreneurs, had a huge impact. Because of this, the selection of acquirers for state-owned assets was unlikely to be fair and fully achievable (Frye 2002; Iwasaki 188

POST-PRIVATIZATION OWNERSHIP

6.2

and Suzuki 2007, 2012). These factors combined with other problems, such as a lack of strategic investors, including foreign investors, and chronic corruption throughout society, probably served to significantly reduce the impact of privatization in many privatized firms in Russia and other FSU countries (Johnson et al. 2000; Radygin 2014). From this point of view, it is fully understandable that Djankov and Murrell (2002) and Estrin et al. (2009) conducted their systematic reviews of the transition literature with a prime focus on differences between the two regions. Regarding privatization methods, two points must be given particular attention from the standpoint of the selection of players engaging in corporate restructuring. The first is whether state-owned assets were transferred free of charge or sold; the second is to what extent financial and managerial capabilities were considered during the process of select­ ing acquirers. Turning once again to Table 6.1, we see that the most favored privatiza­ tion methods adopted by the majority of CEE and FSU countries were (1) vouchers, (2) MEBOs, and (3) direct sales to strategic investors. As Table 6.2 illustrates, these three methods are, from the perspective of the two points above, highly contrasting policy techniques: during the privatization period, these differences probably decisively impacted the ownership structure as well as the incentive structure and effort of the new owning entities. The evidence for this is as follows. The voucher system was the most favored privatization method in 9 of the 28 CEE and FSU countries. Particularly, in two countries—the Czech Republic and Russia—the adoption of the voucher system was so heavily promoted because they needed a policy response to the absolute shortage of domestic capital, on one hand, and, on the other, reformers made populist political decisions. However, the coun­ tries differed in terms of the number of vouchers issued and the way they were dis­ tributed and used in the voucher systems (Miller 2013). In the Czech Republic, vouchers (privatization coupons) were mainly held by investment privatization funds. Because they were owned and operated by banks under the direct influence of the government, the state remained the ultimate owner. As a result, public ownership was, in effect, revived (Stark and Bruszt 1998). In Russia, by contrast, vouchers (pri­ vatization checks) with a face value of 10,000 rubles were distributed “equally” to all citizens, and the investment funds hardly fulfilled the role of producing outsider investors at all. Instead, the majority of state-owned firms were basically transferred free of charge to insiders (Boycko et al. 1995; Mizobata 2005, 2008).

Table 6.2 Characteristics of privatization methods in terms of mode of distribution of state properties and selection of their acquisitors

Distribution of state property Selection of state-property acquisitors by availability of funds and management capability

Vouchers

MEBOs

Direct sales

Acquisition for free No

Acquisition for counter value No

Acquisition for counter value Yes

189

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

While the implementation methods differed between countries, the policy outcome for all countries that adopted the voucher system was that state-owned firms were transferred free of charge or extremely cheaply to the public or specific groups. This approach meant that hardly any privatized firms secured capable and adequately motivated owners and managers, and the government raised no revenue from the pro­ cess. Although the voucher system confers political advantages, in that it is easy to obtain the support of the public because the vouchers are provided free of charge, it dilutes the interest in and sense of responsibility for firms among their new owners. As a result, it fails to adequately encourage improvements in firm performance, a drawback that became apparent soon after firms were privatized. “A policy of people’s capitalism could easily fail because shares are sold to lower-income-earners who are not prone to buy assets such as shares” (Bös 1991, p. 25). Furthermore, and this was especially true in Russia, the transfer of state-owned assets was carried out for reasons other than financial or managerial competence (rather, the main objectives were political or personal gain), so the exclusion of economic entities that were most desirable as corporate reformers from the ownership and management of privatized firms—a kind of “adverse selection effect”—had a wide-ranging and noticeable impact on countries that utilized the voucher system. MEBOs were the most favored method for privatizing firms in eight transition countries, because these countries were in line with the principle of self-management by workers that had existed in the socialist era and the generally accepted notion that workers should be involved in managing their firms (Thompson and Valsan 1999). The fact that this system was particularly preferred in the former Yugoslavia is symbolic from this viewpoint (Mencinger 1996). There is no doubt that MEBOs have a primary focus on transferring state-owned properties to insiders. In that sense, the method neglected the selection of owning entities based on their level of financial and managerial competence. Furthermore, due to the reasons given in Sub­ section 6.2.1, the majority of privatized firms failed to avert the adverse effects of insider ownership. However, because MEBOs normally involve the sale of assets, the negative effect on firm restructuring activity stemming from free-of-charge trans­ fers in relation to the voucher system might be avoided. Furthermore, if, due to reasons such as the immaturity of capital markets or inadequate government regula­ tions concerning corporate information disclosure, there is serious information asym­ metry between outsider investors and firm managers, insider control can result in a relatively more effective ownership structure. Because of this, it is possible that the adverse impact of using MEBOs on the performance of privatized firms could have been restricted over the short term (Wright et al. 1989). Direct sales to strategic investors were the most favored method in nine countries. Among them, in three countries—Estonia, Hungary, and Poland—sale by auction was preferred, and the transfer of assets to foreign investors was actively encour­ aged. Furthermore, in the case of Hungary, even the largest manufacturing firms and commercial banks were generously sold off to strategic investors, particularly to Western firms (Iwasaki et al. 2012). In the case of direct sales, regardless of who the buyer is, the owners are compelled to restructure the firm they have invested in so 190

POST-PRIVATIZATION OWNERSHIP

6.2

as to ensure that the total value of the assets acquired and the cash flow generated from the firm’s operations exceeds the purchase price. Direct sales lead to the emer­ gence of owners and managers whose top priority is to recover the money they have invested and earn additional profits: in macroeconomic terms, they greatly contribute to the creation of a competitive market environment. Furthermore, the positive effects of limiting acquirers of state-owned assets to strategic investors who look for ways to run their companies successfully over the long term are worth emphasizing as an advantage of this method. Differences in privatization methods aside, there were also big differences in the speed of policy implementation among CEE and FSU countries. In Table 6.1, we see that the average private sector share of GDP in 2010 for the 28 transition countries was 66.6% (median: 70%). While some countries far exceeded this figure, the percentage was much lower in several countries. In some FSU countries, in particular, government leaders were extremely cautious about instituting large-scale structural reforms; even now, little progress with privatization has been made in these countries. The view that, if other conditions are held constant, the speed of privatization policy implementation and the effect of firm restructuring should be negatively cor­ related, has been put forward by Radygin (2014). A privatization policy that priori­ tizes implementation speed not only leads to an excessive dispersion of ownership but also delays the formation of capital markets and actually hinders the establish­ ment of a market system. This is because it greatly harms the stability of ownership rights and confidence in markets, which will probably also negatively impact the operating activities of privatized firms. Taking a similar viewpoint, Roland (2000) argued that excessively fast privatization leads to massive asset stripping, which may result in a weak effect of post-privatization ownership on firm performance. These arguments mesh with the views of Arrow (2000), who expressed serious concerns about the side effects of radical transformation. Nevertheless, the correlation between the speed of policy implementation and the firm restructuring effect could be positive. Because expanding the private sector in conjunction with implementing privatization policy leads to the creation of a competitive market environment, it is possible that owners and managers exposed to severe selection pressure from the market, regardless of the nature of the owner­ ship structure that emerges after privatization, could be stimulated to restructure their firms (Åslund 2013). Therefore, the faster privatization occurs in a country, the greater the effect on firm performance. Furthermore, hard competition among firms may reduce gaps in the effort level that stem from differences in the attributes of owners more efficiently in countries with high-speed privatization than in countries where privatization policy has stagnated, leading to the preservation of firms that have weak management foundations. The theoretical arguments in this subsection lead to three hypotheses regarding the impacts of specific factors relating to transition economies on firm performance in the post-privatization period: (1) CEE countries do better than FSU countries in enterprise restructuring, (2) the voucher system is the worst privatization method, and (3) direct sales are superior to MEBOs. However, although the impact of privatization speed is 191

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

difficult to predict theoretically, a fourth hypothesis could be put forward, namely that (4) the speed of policy implementation is related to the progress in firm restructuring, which is in accordance with the fact that the speed of implementation of other market­ ization policies is seen to positively correlate with the degree of economic restructuring. In the following sections, therefore, we will perform a meta-analysis of the previous lit­ erature to empirically verify our testable hypotheses presented in this section.

6.3 PROCEDURE OF LITERATURE SELECTION AND OVERVIEW OF SELECTED STUDIES FOR META-ANALYSIS In this section, we first describe the procedure for literature selection. Next, we over­ view the estimates drawn from the selected studies, and then we explain the method­ ology of the meta-analysis performed in this chapter. As a first step toward identifying literature that has empirically examined the impacts of post-privatization ownership structure on firm performance in CEE and FSU countries, we used the EconLit and Web of Science databases of academic lit­ erature to search for studies published during the 26-year period between 1989 and 2015. When using these electronic databases, we employed as search terms combin­ ations of one of “privatization,” “ownership,” “restructuring,” or “firm performance” and one of “transition economies,” “Central Europe,” “Eastern Europe,” “former Soviet Union,” or the actual name of a CEE or FSU country. This generated around 800 hits. We used the bibliographies in these mechanically searched articles to obtain as many similar research works as possible that were published during the same period. As a result, we obtained more than 1,000 publications, which contained a large number of unempirical papers. As a next step, we closely examined the contents of these research works and limited our literature list to those containing estimates that could be subjected to meta-analysis for this chapter. In the end, we selected a total of 121 studies.2 From these selected works, we extracted a total of 2,894 estimates (mean: 23.9 per study; median: 13). These estimates came from studies covering 29 countries, meaning that almost the entirety of the CEE and FSU regions were included. However, there are large differences between countries in how frequently they were subjected to empir­ ical analysis. Actually, 36 studies dealt with the Czech Republic, and 31 dealt with Russia. These were followed by studies of Hungary (23), Poland (22), Romania (21), Estonia (20), Ukraine (18), Slovenia (17), Bulgaria (15), and Slovakia (11), with 10 or fewer studies addressing each of the remaining 19 countries. Regarding the industries studied, the selected works can be roughly divided into two categories, with 65 studies covering the mining and manufacturing industry and 57 covering a broad range of industries. Just six studies focused on the service sector. If the 121 studies are taken as a whole, the estimation period covers a period of 27 years from 1985 to 2011, with a mean estimation period for the collected estimates of 4.16 years (median: 4 years).

192

PROCEDURE OF LITERATURE SELECTION

6.3

The variables of firm performance (i.e., dependent variables) used in the selected studies can be classified into five types: (1) sales/output indicators; (2) efficiency indicators, such as ROA; (3) productivity indicators, including labor productivity or total factor productivity; (4) firm value indicators represented by stock price and Tobin’s Q; and (5) other firm performance indicators. Each type as a percentage of the total collected estimates is 26.6% (771 estimates), 30.8% (890), 24.3% (703), 12.8% (369), and 5.6% (161), respectively.3 There are 15 ownership variables (i.e., independent variables), ranging from a variable for unspecified government to that for employees. Hereinafter, we refer to these 15 variable types collectively as the “basic category of ownership variable.” Figure 6.2 gives a breakdown of the collected estimates in accordance with this

Figure 6.2 Breakdown of collected estimates by basic category of ownership variable Note: Total number of collected estimates is 2894.

193

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

category. Furthermore, corresponding to the discussion in the previous section, we IV. All insiders, also employ an aggregate I. State, 597, 21% 477, 16% category: namely, we condensed the three variable types from the unspecified government ownership variable to the regional/local Total estimates: 2894 government ownership variIII. Foreign able, the eight types from investors, the unspecified domestic 874, 30% II. All domestic outsider investor ownership outsider investors, variable to the other domes946, 33% tic non-financial company ownership variable, and the three types from the unspecified insider ownership variable to the employee ownership variable into Figure 6.3 broader variable types called Breakdown of collected estimates by aggregated category of the “state ownership variownership variable able,” the “all domestic outNote: Values following category name denote number of collected estimates and share in total estimates, respectively. sider investor ownership variable,” and the “all insider ownership variable,” respectively. We collectively call these three variable types plus the foreign investor variable the “aggregated category of ownership variable” in the remainder of this chapter. Figure 6.3 puts the collected estimates into these four aggregated variable types. In this chapter, we will mainly rely on the aggregated category to conduct a comparative meta-analysis of the effect size and statistical significance of owning entities with different attributes and to assess the presence and degree of PSB in the extant literature. However, we will also utilize the basic categories, depending on the need for hypothesis verification.

6.4 META-SYNTHESIS In this section, we will perform the meta-synthesis employing the aforementioned 2,894 collected estimates according to the methodology described in Chapter 1. Table 6.3 presents the results from the meta-synthesis of the collected estimates. These results are based not only on the aggregated category of ownership variable but also on the basic category mentioned in the previous section. As reported in

194

0.036*** −0.022**

0.026*** 0.043*** −0.021*** 0.011*** 0.014*** 0.042*** 0.027*** 0.024*** 0.035*** 0.037*** 0.046*** 0.007*

168 98 123 95 144 77 132 874 477 163 187 127

0.050*** 0.055*** 0.009

0.043***

0.047***

0.030***

0.037***

0.011* 0.015***

0.027***

0.041***

0.032***

109

0.004 0.005* 0.003 −0.006 0.021***

734.615*** 537.153*** 300.608***

1651.600***

33000.000***

282.481***

183.738***

247.594*** 357.946***

696.014***

635.112***

789.334***

213.126***

7115.438*** 4731.900*** 508.632*** 327.686*** 3799.439***

Random-effects Test of model a homogeneity b

−0.018*** −0.003*** −0.038*** −0.067*** 0.026***

Fixed-effect model a

597 493 60 44 946

Number of estimates (K)

(a) Synthesis of PCCs

24.197*** 13.714*** 1.888**

23.706***

90.120***

9.883***

8.818***

3.047*** 7.444***

−5.793***

17.152***

13.859***

9.948***

−3.280*** 0.794 −4.617*** −9.347*** 22.726***

Unweighted combination

a

4.734*** 3.168*** 0.517

5.323***

14.478***

2.096**

1.762**

0.628 2.153**

−0.955

3.452***

2.573***

2.413***

−0.687 0.152 −4.617*** −9.347*** 4.660***

Weighted combination

1403 231 2805 2136 4632 2622279

−0.260 −0.030 0.453 0.910 0.710 1.558

98580

10556

1.087

35105 12810 40

11757

0.635

1.700 0.999 0.174

3877

0.858

C H A P T E R

0.993

1777 −378 413 1377 179610

Failsafe N (fsN)

0.003 0.067 −0.043 −0.141 0.562

Median of t values

(b) Combination of t values

Notes

Ownership variable types with Arabic numerals belong to the basic category, while those with Roman numerals belong to the aggregated category.

b Null hypothesis: The synthesized effect size is zero.

c Null hypothesis: Effect sizes are homogeneous.

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

IV. All insiders 13. Unspecified insiders 14. Managers 15. Employees

III (12). Foreign investors

II. All domestic outsider investors 4. Unspecified domestic outsider investors 5. Domestic outsider individual investors 6. Unspecified domestic outsider institutional investors 7. Unspecified domestic financial institutions 8. Domestic banks 9. Domestic non-bank financial institutions 10. Domestic company groups and holdings 11. Other domestic nonfinancial companies

I. State 1. Unspecified government 2. Central government 3. Regional/local government

Ownership variable type a

Table 6.3 Synthesis of estimates

META-SYNTHESIS

6

6.4

195

C H A P T E R 6

PRIVATIZATION

Column a of this table, for both category types, the homogeneity test rejects the null hypothesis in every case. Hence, we adopt the estimate from the random-effects model as a reference value. Meanwhile, Column b of the same table demonstrates the com­ bination of t values. The result shows that, in 16 of the 18 cases, the combined t value weighted for the quality level of the study is much lower than the unconditionally combined t value. In other words, the statistical significance of the empirical findings depends greatly on the research quality and/or its background study conditions. The major findings from Table 6.3 can be summarized in four points: first, as compared with the state, the private sector has, on the whole, made a greater contri­ bution to improved firm performance in the post-privatization period. However, with respect to the effect size, the collected estimates that use the unspecified domestic financial institution variable, and with respect to statistical significance, the collected estimates that use the unspecified domestic financial institution ownership variable, the domestic bank ownership variable, and the employee ownership variable exhibit results that buck this overall trend. Second, the empirical assessment of domestic outsider investors reveals that, contrary to our predictions, they are generally inferior to insiders. Third, while the effect of foreign ownership on firm performance is far more statistically significant than that of the state or other private entity ownership, there is hardly any difference as compared to insiders with respect to effect size. Fourth, in terms of both effect size and statistical significance, company managers are seen to be clearly superior to employees, which supports our hypothesis. Figure 6.4 plots the collected estimates in chronological order. As shown in this figure, the PCC of the state ownership variable, the foreign investor ownership variable, and the all insider ownership variable exhibit downward trends along the time axis. In fact, according to the approximate straight lines drawn in the figure, with each one-year increase in the average estimation period, the PCC drops by 0.0045 for the state owner­ ship variable, 0.0048 for the foreign investor ownership variable, and 0.0069 for the all insider ownership variable, with statistical significance at the 1% level. In contrast, the all domestic outsider investor ownership variable exhibits an upward trend as the time period increases. Actually, with each one-year increment in the average estimation period, the PCC climbs by 0.0028. We found that the t values show the same time trend. These results indicate that the relative superiority/inferiority of different owners as cor­ porate restructurers can change depending on the period of time covered by a study. As discussed above, while the results of the meta-synthesis reported in this section provide supporting evidence for the hypotheses of the superiority of the private sector over the state as a firm-owning entity and that company managers outperform employees, they do not back up our theoretical predictions concerning the relative superiority of certain private company owners over others. As indicated by the chronological order of the collected estimates displayed in Figure 6.4 and the remarkable difference between the weighted and unconditionally combined t values reported in Column b of Table 6.3, it is highly likely that the empirical results of the previous literature were heavily affected by the research conditions and the quality level of the studies. Accordingly, in the next section, we will perform a metaregression analysis to test our hypotheses in a more rigorous manner. 196

META-REGRESSION ANALYSIS: BASELINE ESTIMATION

6.5

C H A P T E R 6

Figure 6.4 Chronological order of partial correlation coefficients by aggregated category of ownership variable Note: Figures in parentheses beneath the regression coefficients of the approximate straight line are standard errors. *** and ** denote statistical significance at the 1% and 5% levels, respectively.

6.5 META-REGRESSION ANALYSIS: BASELINE ESTIMATION In this section, we will perform a meta-regression analysis (MRA) to examine whether the results of the meta-synthesis reported in the previous section can be reproduced even when other research conditions are simultaneously controlled for. To this end, we introduce the PCC or the t value into the left-hand side of the regression equation, while on its righthand side, we adopt a series of meta-independent variables designed to capture differences in the ownership variable type,4 target countries and industries, estimation periods, and benchmark indexes of firm performance variables that we mentioned in Section 6.3. It also captures differences in the type and source of data used, the estimator, equation type, pres­ ence of treatment for selection bias of privatized firms, presence of various control vari­ ables that would significantly affect estimation results, and degrees of freedom and quality of the study. The names, definitions, and descriptive statistics of these meta-independent variables are shown in Table 6.4. 197

C H A P T E R

PRIVATIZATION Table 6.4 Name, definition, and descriptive statistics of meta-independent variables Descriptive statistics

6 Variable name

Definition

All domestic outsider investors

1 = if ownership variable used for estimation belongs to the aggregated category of all domestic outsider investors, 0 = otherwise 1 = if ownership variable used for estimation belongs to the category of foreign investors, 0 = otherwise 1 = if ownership variable used for estimation belongs to the aggregated category of all insiders, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of central government, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of regional/local government, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of unspecified domestic outsider investors, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of domestic outsider individual investors, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of unspecified domestic outsider institutional investors, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of unspecified domestic financial institutions, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of domestic banks, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of domestic non-bank financial institutions, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of domestic company groups and holdings, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of other domestic non-financial companies, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of unspecified insiders, 0 = otherwise 1 = if ownership variable used for estimation belongs to the basic category of managers, 0 = otherwise

Foreign investors

All insiders

Central government

Regional/local government Unspecified domestic outsider investors Domestic outsider individual investors Unspecified domestic outsider institutional investors Unspecified domestic financial institutions Domestic banks

Domestic non-bank financial institutions Domestic company groups and holdings Other domestic non-financial companies Unspecified insiders

Managers

Mean

Median

S.D.

0.327

0

0.469

0.302

0

0.459

0.165

0

0.371

0.021

0

0.143

0.015

0

0.122

0.038

0

0.190

0.058

0

0.234

0.034

0

0.181

0.043

0

0.202

0.033

0

0.178

0.050

0

0.217

0.027

0

0.161

0.046

0

0.209

0.056

0

0.231

0.065

0

0.246

(Continued )

198

META-REGRESSION ANALYSIS: BASELINE ESTIMATION

6.5

Table 6.4 contd. Descriptive statistics Variable name Employees

Definition

Mean

Median

1 = if ownership variable used for estimation 0.044 0 belongs to the basic category of employees, 0 = otherwise Dummy variable 1 = if ownership variable is a dummy variable, 0.541 1 0 = otherwise Lagged variable 1 = if a lagged ownership variable is used for 0.101 0 estimation, 0 = otherwise With an interaction 1 = if estimation is carried out with an interaction 0.085 0 term(s) term(s) of the ownership variable, 0 = otherwise Efficiency 1 = if efficiency is adopted as the benchmark 0.308 0 index of the firm performance variable, 0 = otherwise Productivity 1 = if productivity is adopted as the benchmark 0.243 0 index of the firm performance variable, 0 = otherwise Firm value 1 = if firm value is adopted as the benchmark 0.128 0 index of the firm performance variable, 0 = otherwise Other firm performance 1 = if a performance measure other than sale/ 0.056 0 output and the above indices is adopted as the benchmark index of the firm performance variable, 0 = otherwise Mining and manufactur- 1 = if target industry is the mining and manufactur0.388 0 ing industries ing industries, 0 = otherwise Service industry 1 = if target industry is the service industry, 0.021 0 0 = otherwise First year of estimation First year of estimation period 1995.892 1995 Length of estimation Years of estimation period 4.166 4 Cross-section data 1 = if cross-section data is employed for empirical 0.452 0 analysis, 0 = otherwise Commercial database 1 = if data employed for empirical analysis is 0.357 0 based on a commercial database, 0 = otherwise Original enterprise 1 = if data employed for empirical analysis is 0.276 0 survey based on an original enterprise survey, 0 = otherwise FE 1 = if fixed-effects panel estimator is used for esti0.180 0 mation, 0 = otherwise RE 1 = if random-effects panel estimator is used for 0.072 0 estimation, 0 = otherwise Robust 1 = if robust estimator is used for estimation, 0.058 0 0 = otherwise GMM 1 = if GMM estimator is used for estimation, 0.023 0 0 = otherwise Other estimators 1 = if an estimator other than OLS and the above 0.050 0 estimators is used for estimation, 0 = otherwise IV/2SLS/3SLS 1 = if instrumental variable method or 2SLS or 0.127 0 3SLS is used for estimation, 0 = otherwise

S.D. 0.205

0.498 0.301 0.278 0.462

0.429

0.334

0.229

0.487 0.144 3.897 2.932 0.498 0.479 0.447

0.385 0.258 0.235 0.150 0.219 0.333

(Continued )

199

C H A P T E R 6

C H A P T E R

PRIVATIZATION

6

Variable name

Definition

Difference model

1 = if difference model is used for estimation, 0 = otherwise 1 = if translog model is used for estimation, 0 = otherwise 1 = if estimation treats for the selection bias of privatized companies, 0 = otherwise 1 = if estimation simultaneously controls for the degree of market competition, 0 = otherwise 1 = if estimation simultaneously controls for location fixed effects, 0 = otherwise 1 = if estimation simultaneously controls for industry fixed effects, 0 = otherwise 1 = if estimation simultaneously controls for time fixed effects, 0 = otherwise Proportion of Russian firm samples in observations used for estimation Proportion of Polish firm samples in observations used for estimation Proportion of Hungarian firm samples in observations used for estimation Proportion of Ukrainian firm samples in observations used for estimation Proportion of firm samples other than Czech Republic and the above countries in observations used for estimation Proportion of CEE firm samples in observations used for estimation a Proportion of firm samples in CEE EU member states in observations used for estimation Proportion of firm samples in countries with voucher privatization as the primary method in observations used for estimation a Proportion of firm samples in countries with MEBO privatization as the primary method in observations used for estimation a Proportion of firm samples in countries with direct-sale privatization as the primary method in observations used for estimation a Proportion of firm samples in countries where the private sector share in GDP is less than 70% in 2010 in observations used for estimationa Root of degree of freedom of the estimated model Ten-point scale of the quality level of the study b

Table 6.4 contd. Descriptive statistics

Translog model Treatment for selection bias Market competition Location fixed effects Industry fixed effects Time fixed effects Russia Poland Hungary Ukraine Other CEE and FSU countries CEE countries EU member states Voucher privatization countries MEBO privatization countries Direct-sale privatization countries Slow-speed privatization countries √Degree of freedom Quality level

Notes

a Countries in this category correspond with those in Table 6.1.

b See Chapter 1 for more details.

200

Mean

Median

S.D.

0.155

0

0.362

0.168

0

0.374

0.071

0

0.257

0.119

0

0.324

0.295

0

0.456

0.627

1

0.484

0.481

0

0.500

0.208

0

0.398

0.069

0

0.234

0.067

0

0.230

0.068

0

0.247

0.220

0

0.402

0.705

1.000

0.450

0.698

1.000

0.454

0.662

1.000

0.455

0.119

0.000

0.308

0.218

0.000

0.388

0.300

0.000

0.452

50.226 4.272

26.842 63.555 4 3.034

META-REGRESSION ANALYSIS: BASELINE ESTIMATION

6.5

To begin with, we conducted estimations using the aggregated category of owner­ ship variables. Table 6.5 shows the results. As this table illustrates, the estimates are sensitive to the choice of estimator. Therefore, hereinafter, we will interpret the regression results under the assumption that the meta-independent variables that are statistically significant and have the same sign in at least four of seven models con­ stitute statistically robust estimation results. According to the estimates shown in Panel a of Table 6.5, for which the depend­ ent variable is the PCC, collected estimates reporting a statistically significantly larger positive effect size than the state ownership variable are limited to the foreign investor ownership variable. In fact, a meta-independent variable that captures esti­ mates of the foreign investor ownership variable by a value of 1 is positive at the 1% significance level for all seven models, indicating that it is a highly robust esti­ mate. Put another way, if other research conditions are held constant, the PCC of the foreign investor ownership variable is, with a range of 0.0406 to 0.0730, higher than that of the state ownership variable. In contrast, the meta-independent variable, which takes a value of 1 for estimates of the all domestic outsider investor owner­ ship variable and the all insider ownership variable, shows a positive sign for almost all of the models, but the vast majority are insignificant. According to Panel b of the same table, which shows the estimation results with the t value on the left-hand side in the regression equation, once again, the metaindependent variable of the foreign investor ownership variable is positive and sig­ nificant at the 1% level in all seven models. In other words, the statistical certainty of the effect of foreign ownership on firm performance is higher than that of state ownership, with a range of 1.3910 to 7.8449. In contrast, while the coefficient of the all domestic outsider investor ownership variable and the all insider ownership vari­ able is positive for all seven models, only two models exhibit statistically significant estimates. Therefore, it is difficult to assert that there is a remarkable difference in statistical significance between the ownership effect of the state and domestic private owners. Keeping the above findings in mind, our next estimation was performed using the basic category of ownership variable. The results are reported in Table 6.6. Due to space limitations, we have omitted the estimates for other research condi­ tions and the intercept, but otherwise the table is configured in exactly the same way as Table 6.5. According to Panel a of Table 6.6, in the case of the MRA with the PCC as the dependent variable, the meta-independent variable, which assigns a value of 1 to estimates of the unspecified domestic outsider investor variable and the other domes­ tic non-financial company variable, shows a significant and positive sign in five or more models in addition to the foreign investor ownership variable. Moreover, the estimation results in Panel b of the same table, which take the t value as the depend­ ent variable, exhibit significant and positive coefficients in five or more models for the other non-financial company ownership variable and the unspecified insider own­ ership variable as well as the foreign investor ownership variable.

201

C H A P T E R 6

202

Data type (Panel data) Cross-section data

Estimation period First year of estimation Length of estimation

0.0285

−0.0039*** −0.0042*

Target industry (Various industries) Mining and manufacturing 0.0136 industries −0.0072 Service industry

0.0287

0.0042

−0.0029** −0.0028**

0.0084

0.0097 −0.0030** −0.0041**

−0.0001

0.0257*

0.0615*

0.0109

−0.0016 −0.0010

0.0040

−0.0059 −0.0031 0.0020

−0.0059

−0.0092 −0.0204 −0.0224 −0.0040

−0.0066 −0.0152

−0.0030

0.0459*** 0.0073

0.0100

[5]

Multilevel mixedeffects RML

−0.0158

0.0343* 0.0352* 0.1555*** 0.0877**

−0.0025 −0.0076 0.0623*** 0.0302

Firm performance variable type (Sales/output) Efficiency 0.0053 Productivity −0.0114 Firm value 0.0382** Other firm performance 0.0234 −0.0058 −0.0297*** 0.0141 −0.0024

0.0079 −0.0074

−0.0064 −0.0195**

0.0269 −0.0065

0.0730*** 0.0141 −0.0180

0.0406*** 0.0242**

−0.0056

[4]

−0.0160***

0.0648*** 0.0141

0.0596*** 0.0204

0.0174

[3]

−0.0226**

0.0161

[2]

0.0187*

[1]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

Other characteristics of ownership variable Dummy variable (Owner−0.0130 ship share) Lagged variable 0.0260 With an interaction term(s) −0.0023

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Cluster-robust WLS [Quality level]

0.0104

−0.0017 −0.0011

0.0040

−0.0057

−0.0091 −0.0203 −0.0214 −0.0030

−0.0061 −0.0151

−0.0033

0.0461*** 0.0074

0.0101

[6] a

Cluster-robust random-effects panel GLS

0.0285***

0.0005 0.0032

−0.0059

−0.0191

−0.0106 −0.0211 −0.0336 −0.0206

−0.0131*** −0.0144

0.0023

0.0427*** 0.0057

0.0088

[7] b

Cluster-robust fixed-effects panel LSDV

6

Estimator (Analytical weight Clusterin parentheses) robust OLS

(a) Dependent variable — PCC

Table 6.5 Meta-regression analysis using the aggregated category of ownership variable: Base-line estimation

C H A P T E R

PRIVATIZATION

0.0229** 0.0224* −0.0127 −0.0224 0.0256 −0.0120

0.0230 0.0005

−0.0007 0.0083 0.0181 0.0001 0.0087 0.0099 0.0269* −0.0033

−0.0065

−0.0064 −0.0002

0.0336*** 0.0315*** −0.0267* −0.0196 0.0152 −0.0072

0.0130 −0.0123

K R2

2894 0.154

Degree of freedom and research quality √Degree of freedom −0.0001 Quality level −0.0013 Intercept 7.7500*** 2894 0.229

−0.0001** − 6.0188**

Proportion of sample firms in observations (Czech Republic) Russia 0.0075 0.0085 Poland 0.0056 0.0017 Hungary 0.0167 −0.0028 Ukraine 0.0274 0.0287 Other CEE and FSU 0.0151 0.0112 countries

Control variable Market competition Location fixed effects Industry fixed effects Time fixed effects

Treatment for selection bias of privatized firms Treatment for selection −0.0008 bias

Equation type (Models other than listed below) Difference model −0.0064 Translog model −0.0026

Estimator (OLS) FE RE Robust GMM Other estimators IV/2SLS/3SLS

Data source (Official statistics) Commercial database Original enterprise survey

2894 0.364

−0.0001 −0.0012 5.9049***

0.0108 0.0307* −0.0014 −0.0084 0.0029

−0.0008 0.0151 0.0234*** −0.0023

0.0096

0.0070 −0.0035

−0.0002 0.0090 −0.0216 −0.0183 −0.0150* 0.0091

−0.0123 −0.0289

2894 0.529

−0.0002 −0.0049** 6.1382

−0.0378 0.0346 0.0109 0.0447* −0.0048

−0.0124 −0.0227 0.0257 0.0170

−0.0461

0.0346 0.0409*

0.0215 0.0112 −0.0844*** −0.0643** 0.1236*** −0.0110

−0.0246 0.0277

2894 −

−0.0002*** 0.0017 3.1985

−0.0046 0.0241 0.0216 0.0197 0.0125

0.0056 0.0001 0.0178 −0.0160

0.0128

−0.0553** 0.0006

−0.0019 0.0150* 0.0071** −0.0113 0.0069 −0.0219**

0.0091 −0.0017

2894 0.049

−0.0002*** 0.0017 3.3363

−0.0044 0.0231 0.0219 0.0206 0.0125

0.0060 0.0004 0.0175 −0.0159

0.0126

−0.0540** 0.0007

−0.0018 0.0149* 0.0070** −0.0118 0.0070 −0.0219**

0.0098 −0.0015

(Continued )

2894 0.012

−0.0002** dropped −0.9048

−0.0023 0.0439 0.0081 −0.0063 0.0125

−0.0079 −0.0049 0.0236 −0.0236**

0.0095

−0.0865** −0.0065

−0.0009 0.0168* 0.0079 −0.0040 0.0069 −0.0225**

−0.0358*** dropped

META-REGRESSION ANALYSIS: BASELINE ESTIMATION 6.5

203

C H A P T E R 6

204

Data source (Official statistics) Commercial database Original enterprise survey

Data type (Panel data) Cross-section data

−0.4609 −0.5335

0.5236 −0.6958 −1.8566*

0.5979

−0.1929*** −0.2300*

Estimation period First year of estimation Length of estimation

−0.1813** −0.1106

−1.2178

1.1417*

−0.2873 −0.4674 1.1774 0.1551

Target industry (Various industries) Mining and manufacturing 0.0190 industries −2.8602 Service industry

Firm performance variable type (Sales/output) Efficiency −0.3797 Productivity −0.5390 Firm value 0.9711 Other firm performance −0.2276

−8.9647*** −6.3031**

−6.1465*

−0.2204 −0.4318**

−2.3693

−5.5161***

−2.3972** −6.6233*** −5.2279*** −1.2889

−3.6659** 0.6950

0.6833

−0.4037*** −0.0283

−3.5724*

−1.9021*

0.1356 −0.1231 4.2780*** 2.5699

−0.8980 0.0260

−0.3927 −6.3883***

0.6813 0.1695

4.1089*** 0.6887 −0.9045

7.8449*** 3.6765**

0.6633

[11]

−4.3007***

2.2623*** 0.4122

2.7076*** 0.6971

3.3549*

[10]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−0.2935

0.5100

[9]

0.7370*

[8]

Other characteristics of ownership variable Dummy variable (Owner−0.7162 ship share) Lagged variable 0.6150 With an interaction term(s) 0.1743

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Cluster-robust WLS [Quality level]

−0.8434 −1.0309

0.6044

−0.1964 −0.1277

−1.1892

−0.5920

−0.3781 −0.4277 −0.8732* −0.3101

−0.3448** −1.7996*

−0.5982

1.5008*** 0.6344*

0.4266

[12]

Multilevel mixedeffects RML

−0.8639 −1.0346

0.5797

−0.1964* −0.1278

−1.1147

−0.5432

−0.3838 −0.4328 −0.8578* −0.2722

−0.3243** −1.7513*

−0.6156

1.5294*** 0.6368

0.4335

[13] c

Cluster-robust random-effects panel GLS

−0.7836 dropped

0.7544

−0.1905 −0.0628

−2.2170**

−1.4062

−0.3537 −0.4101 −0.9237 −0.4151

−0.4071*** −2.0074*

−0.5134

1.3910*** 0.6275

0.3964

[14] d

Cluster-robust fixed-effects panel LSDV

6

Estimator (Analytical weight Clusterin parentheses) robust OLS

(b) Dependent variable — t value

Table 6.5 contd.

C H A P T E R

PRIVATIZATION

0.8154 1.5846** −0.1547 −2.0299** 0.3274 −0.0463

2894 0.216

2894 0.291

0.0041 − 384.3074*** 2894 0.590

−0.0113 −0.0049 451.7466

2894 0.393

−0.0051 −0.2780* 806.4356***

−0.7303 3.5384 3.0367 −0.3649 0.7880

−1.9755 −0.4430 2.5885** 0.6769

−3.9125**

0.9313 2.2487

2.5346** 2.4264 −3.0400** −3.9123*** 3.4005* 0.6552

2894 −

0.0064 0.0738 392.8722

0.6829 2.9005** 2.6147* 0.1825 2.3616

−1.1516* 0.0078 0.6050 0.0260

0.3218

−1.9892 0.0498

−0.9682 1.3254* 0.1697* −1.7765 −1.3309 −0.2160

2894 0.095

0.0059 0.0755 392.6808*

0.7602 2.7353** 2.5345* 0.1944 2.2209

−1.1191 −0.0485 0.6810 0.0904

0.3256

−1.9283 0.1421

−0.9341 1.3266* 0.1689* −1.7576 −1.3569 −0.2123

2894 0.040

0.0098 dropped 381.4948

−0.1395 4.2817** 3.5396* 0.5003 3.6953

−1.2321* 0.3891 0.1376 −0.5722

0.2369

−2.3950 −0.4770

−1.0773 1.3177* 0.1775* −1.8722 −1.1938 −0.2273

a

Notes Breusch–Pagan test: χ2=1120.40, p=0.000 b Hausman test: χ2=68.48, p=0.000 c Breusch–Pagan test: χ2=1277.86, p=0.000 d Hausman test: χ2=155.95, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

Degree of freedom and research quality √Degree of freedom 0.0084 Quality level −0.0005 Intercept 360.4910**

6.2227** 6.2682*** −0.2073 −1.4166 1.6270

Proportion of sample firms in observations (Czech Republic) Russia 0.9441 1.7882 Poland 1.3689 0.9148 Hungary 1.8178 0.9609*** Ukraine 0.4974 0.0838 Other CEE and FSU 1.3532 0.8091 countries

2.5709

0.9768 0.9594

−3.2907 0.6319 −3.2753 −5.0272* −7.8662*** 3.8150***

−2.3155 −0.1574 4.8539*** −1.2020

−0.2186 0.2210 1.4270** 0.2344

−0.6751

−0.7736 0.2681

0.9854* 2.5419** 0.3998 −1.5349 0.8027 0.3626

0.2052 −0.0459 1.4888** 0.2417

Control variable Market competition Location fixed effects Industry fixed effects Time fixed effects

Treatment for selection bias of privatized firms Treatment for selection −0.7540 bias

Equation type (Models other than listed below) Difference model 0.0512 Translog model 0.4030

Estimator (OLS) FE RE Robust GMM Other estimators IV/2SLS/3SLS

META-REGRESSION ANALYSIS: BASELINE ESTIMATION 6.5

205

C H A P T E R 6

206

[1]

K R2

2894 0.177

−0.0022

0.0166 0.0144 0.0346* 0.0223* 0.0448* −0.0152

0.0514* 0.0323* 0.0659* 0.0211 0.0262 −0.0134

2894 0.379

0.0044

0.0371* −0.0022

−0.0076 0.0009

0.0299 0.0145

2894 0.253

0.0022 0.0078

−0.0910*

−0.0237

−0.0339

2894 0.545

0.0726* 0.0043 0.0290* −0.0185

0.0231*

2894 −

0.0477* 0.0168 0.0339 −0.0286

0.0326*

−0.0077

0.0065

−0.0012

0.0268*

0.0120

−0.0067

0.0165

0.0317* 0.0054 0.0413*

0.0029 0.0282 −0.0091

0.0183*

[5]

0.0216

[4]

Multilevel mixedeffects RML

−0.0264 −0.0472* −0.0040

[3]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

0.0151 0.0118 0.0557*

[2]

Cluster-robust WLS [Quality level]

Ownership variable type (Unspecified government) Central government 0.0084 Regional/local government 0.0010 Unspecified domestic outsider 0.0390* investors Domestic outsider individual 0.0177 investors Unspecified domestic outsider 0.0211 institutional investors Unspecified domestic financial −0.0208 institutions Domestic banks 0.0154 Domestic non-bank financial 0.0201 institutions Domestic company groups 0.0409 and holdings Other domestic non-financial 0.0378* companies Foreign investors 0.0605* Unspecified insiders 0.0282* Managers 0.0427 Employees −0.0188

Meta-independent variable (Default)/Model

Clusterrobust OLS

2894 0.060

0.0479* 0.0169 0.0340 −0.0286

0.0326*

0.0048

0.0023 0.0079

−0.0078

0.0065

0.0120

0.0317* 0.0055 0.0414*

[6] a

Cluster-robust random-effects panel GLS

2894 0.019

0.0441* 0.0150 0.0330 −0.0297

0.0320*

−0.0014

0.0004 0.0057

−0.0075

0.0052

0.0114

0.0318* 0.0029 0.0388*

[7] b

Cluster-robust fixed-effects panel LSDV

6

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

Table 6.6 Meta-regression analysis using the basic category of ownership variable: Base-line estimation

C H A P T E R

PRIVATIZATION

[8]

Clusterrobust OLS

2894 0.224

1.7363*

−2.8489 −1.6690 5.1475* 2.2903 2.6387 −1.2742

1.2466 0.6415 2.1458* 1.0049* 0.3475 −1.1535 2894 0.633

−1.5055

−2.9715 0.0180

0.0954 −0.2552

2894 0.300

1.4162* −0.6149

1.9979

−0.1534

2894 0.414

3.6867* −0.2122 1.0620 −2.0662*

0.6347

2894 −

1.5764* 0.9635* 1.0769* 0.0609

1.0796*

0.3650

0.0141 0.2292

−0.0291

0.8009

0.6885

1.5350* −0.0706 1.0507*

[12]

Multilevel mixedeffects RML

2894 0.098

1.6086* 0.9649* 1.0603* 0.0528

1.0724*

0.3727

0.0088 0.2264

−0.0252

0.8021

0.6867

1.4635* −0.1255 1.0431*

[13] b c

Cluster-robust random-effects panel GLS

2894 0.041

1.4906* 0.9561* 1.1289* 0.0810

1.0903*

0.3347

0.0193 0.2276

−0.0439

0.7943

0.6917

1.7026* 0.0608 1.0646*

[14] d

Cluster-robust fixed-effects panel LSDV

Notes a Breusch–Pagan test: χ2=1127.38, p=0.000 b Hausman test: χ2=77.97, p=0.000 c Breusch–Pagan test: χ2=1231.14, p=0.000 d Hausman test: χ2=178.46, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

−1.1509

3.9893

0.9711

0.5010

2.2489

0.6514

−4.6959 −10.7130 −2.9302

[11]

−12.0835* −22.7647* −5.2403*

[10]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−1.6244 −1.7309 0.1368

[9]

Cluster-robust WLS [Quality level]

Ownership variable type (Unspecified government) Central government −1.0083 Regional/local government −1.4657 Unspecified domestic outsider 0.1977 investors Domestic outsider individual 0.8849* investors Unspecified domestic outsider 0.8698 institutional investors Unspecified domestic financial 0.3081 institutions Domestic banks −0.1210 Domestic non-bank financial −0.0552 institutions Domestic company groups 0.8224 and holdings Other domestic non-financial 0.9476* companies Foreign investors 2.5057* Unspecified insiders 0.9464* Managers 0.6779 Employees −0.4188

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

META-REGRESSION ANALYSIS: BASELINE ESTIMATION 6.5

207

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

The above results suggest that, with regard to the estimation results based on the aggregated category of ownership variable shown in Table 6.5, in which the metaregression coefficient of the all domestic outsider investor variable is almost insig­ nificant, the previous studies that empirically examined the impact on firm perform­ ance of six owner types from domestic outsider individual investors to domestic company groups and holdings largely failed to detect an economically meaningful and statistically significant ownership effect. Meanwhile, the insignificant estimates of the all insider ownership variable in Table 6.5 are mainly due to the ownership effect of the sample of employee owners being extremely small. To sum up, the results of the baseline estimation reported in Tables 6.5 and 6.6 prove only the superiority of foreign investors as compared with state or domestic private owners. Consequently, they do not provide comprehensive support for the entire series of hypotheses concerning differences between ownership types pre­ sented in Section 6.2, as in the case of the meta-synthesis conducted in the previous section. One of the major reasons for these disappointing results is that ownership variables designed to verify the idiosyncrasies of specific countries/regions and pri­ vatization policies, the domestic outsider ownership variables in particular, did not deliver expected results in many extant works. Accordingly, in the next section, we will attempt to estimate an extended model that takes the idiosyncrasies of transition countries and privatization policies into account to examine whether certain order lies within the opaqueness seen in the empirical results of the previous literature.

6.6 META-REGRESSION ANALYSIS CONCERNING THE IDIOSYNCRASIES OF TRANSITION ECONOMIES In this section, with reference to the discussion in Subsection 6.2.2 and Table 6.1, we will perform an MRA focusing on the specific aspects of transition countries and privat­ ization policies, namely (1) the idiosyncrasies of CEE countries as compared with FSU countries, (2) the idiosyncrasies of countries that favored privatization using the voucher system, (3) the idiosyncrasies of countries that favored MEBOs, and (4) the idiosyncra­ sies of countries that favored direct sales to strategic investors, as well as (5) differences in the speed of implementing privatization policy. More concretely, we will classify the countries studied based on the above five aspects and estimate an interaction term between an ownership variable type and the proportion of the transition country group at issue in the total collected estimates. The discussion that follows will focus mainly on estimation results of the extended model using the aggregated category of ownership variable, but we also refer from time to time to estimates based on the basic category. 6.6.1 Idiosyncrasies of CEE countries Table 6.7 presents the estimation results of the extended model, which introduced the proportion of CEE country observations and its interaction term with the meta­

208

K R2

Interaction term All domestic outsider investors × CEE countries Foreign investors × CEE countries All insiders × CEE countries CEE countries 0.0082

−0.0416

2894 0.163

2894 0.248

0.0281**

−0.0886**

−0.0737**

0.0340**

−0.0455**

0.1367** 0.0127

0.1143** 0.0479

−0.0454**

0.0487**

[2]

0.0480**

[1]

Cluster-robust ClusterWLS [Quality robust OLS level]

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent vari­ able (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

2894 0.373

0.0444**

−0.0015

−0.0603**

−0.0253

0.0898** 0.0227

0.0337

[3]

Clusterrobust WLS [N]

2894 0.519

0.0309

−0.0259

−0.0565**

−0.0158

0.1209** 0.0300

0.0071

[4]

2894 −

0.0399**

−0.0268

−0.0544**

−0.0113

0.0865** 0.0231

0.0182

[5]

Multilevel Cluster-robust mixed-effects WLS [1/SE] RML

2894 0.048

0.0391**

−0.0269

−0.0546**

−0.0116

0.0868** 0.0233

0.0185

[6] a

Cluster-robust random-effects panel GLS

(Continued )

2894 0.008

0.0662

−0.0259

−0.0565

−0.0084

0.0851** 0.0208

0.0152

[7] b

Cluster-robust fixed­ effects panel LSDV

Table 6.7 Meta-regression analysis of the idiosyncrasy of CEE countries: Estimation using the aggregated category of ownership variable

META-REGRESSION ANALYSIS CONCERNING 6.6

209

C H A P T E R 6

210

−0.3403 1.0500

−1.0029

1.4257 2894 0.302

−4.2192**

−3.7021**

2894 0.221

−1.3355**

5.6708** 0.9194**

5.6981** 1.5219**

−1.4603**

1.4006**

[9]

1.5947**

[8]

2894 0.601

9.6088**

−3.2844

−16.5619**

−4.9372

21.2447** 5.6822

5.9944

[10]

Clusterrobust WLS [N]

2894 0.387

3.1031

−1.3911

−3.2993

−1.3505

6.8343** 1.5556

1.5255

[11]

2894 −

1.3799

−0.2741

−0.8187

−0.1590

2.1282** 0.8122**

0.5396**

[12]

Multilevel Cluster-robust mixed-effects WLS [1/SE] RML

2894 0.078

1.2871

−0.3003

−0.8744

−0.1894

2.2072** 0.8350**

0.5688**

[13] c

Cluster-robust random-effects panel GLS

2894 0.028

2.0263

−0.1893

−0.7204

−0.0749

1.9391** 0.7406

0.4531

[14] d

Cluster-robust fixed­ effects panel LSDV

b

a

Notes Breusch–Pagan test: χ2=896.85, p=0.000 Hausman test: χ2=73.65, p=0.000 c Breusch–Pagan test: χ2=1243.91, p=0.000 d Hausman test: χ2=149.01, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

Interaction term All domestic outsider investors × CEE countries Foreign investors × CEE countries All insiders × CEE countries CEE countries

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent vari­ able (Default)/Model

Cluster-robust ClusterWLS [Quality robust OLS level]

6

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

Table 6.7 contd.

C H A P T E R

PRIVATIZATION

META-REGRESSION ANALYSIS CONCERNING

6.6

independent variable for each ownership variable type into the right-hand side of the regression equation. Due to space constraints, we have left some estimates out, but as was the case with Table 6.6, meta-independent variables that capture various research conditions are simultaneously estimated.5 As can be seen from Panel a of Table 6.7, while the variable of CEE countries is estimated to be significant and positive in five of the seven models, the interaction term with the foreign investor ownership variable is significant and negative in six models. This result hints at the possibility that, while empirical studies on CEE countries have generally reported a greater effect size than have those on FSU countries, the effect size of foreign ownership is significantly smaller than it is for FSU countries. The relative superior­ ity of foreign investors over the state and domestic private owners might not be as conspicuous in CEE countries as it is in FSU countries.6 According to Panel b of Table 6.7, if the idiosyncrasies of CEE countries are controlled for, the ownership variable types are estimated to be relatively robust and positive. In other words, studies of CEE countries might not produce support­ ing evidence for theoretical predictions regarding the effect on firm performance of ownership by domestic private owners and foreign investors as opposed to state ownership, as studies of FSU countries did, from the standpoint of statistical significance. According to the estimation results of the extended model using the basic category of ownership variable (not reported),7 marked differences between CEE countries and FSU countries in terms of both effect size and statistical significance are seen, particularly in the case of estimates for domestic non-bank financial institutions and firm managers. The results also confirm that studies of FSU countries deliver a more positive empirical assessment of the impact of these two ownership types on firm performance. Moreover, from the point of view of statistical significance, a similar tendency is seen for the ownership effect of domestic company groups and holdings. 6.6.2 Idiosyncrasies of countries that favored privatization using the voucher system Table 6.8 shows meta-regression models designed to identify the idiosyncrasies of countries that favored voucher privatization. These results are particularly worthy of attention. They show that, if the distinctive effects on the empirical results for vou­ cher privatization countries are dissociated by the interaction term, the metaindependent variables of ownership variable types—regardless of the difference in dependent variables—are given a significant and positive coefficient in five or more of the seven models. In addition, as compared with the baseline estimation in Table 6.5, the coefficient of the all domestic outsider investor ownership variable is esti­ mated to be much higher; for all the models with which significant estimates are obtained, it surpasses the all insider ownership variable. Moreover, in six models, it also exceeds even the coefficient of the foreign investor ownership variable. Mean­ while, the interaction term between the variable of voucher privatization countries

211

C H A P T E R 6

6

212

K R2

Interaction term All domestic outsider investors × Voucher privatization countries Foreign investors × Voucher privatization countries All insiders × Voucher privatization countries Voucher privatization countries

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

−0.0510*** 0.0157

0.0056

−0.0008 2894 0.235

0.0048

0.0090

2894 0.159

−0.0456***

0.0511*** 0.0660*** 0.0439***

[2]

Cluster-robust WLS [Quality level]

−0.0440***

0.0529*** 0.0564*** 0.0194***

[1]

Clusterrobust OLS

2894 0.377

−0.0329***

0.0013

0.0431***

−0.0128

0.0228*** 0.0261*** 0.0212***

[3]

Clusterrobust WLS [N]

2894 0.517

0.0031

2894 −

−0.0263

−0.0126

0.0350

−0.0072 0.0120

−0.0521***

0.0527*** 0.0283*** 0.0173***

[5]

Multilevel mixed-effects RML

−0.0308

0.0205 0.0844*** 0.0078***

[4]

Clusterrobust WLS [1/SE]

2894 0.047

−0.0244***

−0.0127

0.0338***

−0.0522***

0.0528*** 0.0294*** 0.0175***

[6] a

Cluster-robust random-effects panel GLS

2894 0.011

−0.0576***

−0.0110

0.0482***

−0.0507***

0.0509*** 0.0172*** 0.0150

[7] b

Cluster-robust fixed-effects panel LSDV

Table 6.8 Meta-regression analysis of the idiosyncrasy of voucher privatization countries: Estimation using the aggregated category of ownership variable

C H A P T E R

PRIVATIZATION

0.0677

−0.6493 2894 0.286

−1.4793***

−0.5166

2894 0.219

0.1505

−1.8192***

1.8615*** 2.2100*** 1.3551***

[9]

Cluster-robust WLS [Quality level]

−0.0096

−1.9097***

2.1523*** 2.8218*** 1.0585***

[8]

Clusterrobust OLS

2894 0.594

−8.0449***

1.0759

10.6914***

−0.5576

3.4696*** 3.8810*** 2.5830

[10]

Clusterrobust WLS [N]

2894 0.390

−0.0667

−0.8256

−2.5233

−2.3252

2.4447 6.1283*** 1.3920

[11]

Clusterrobust WLS [1/SE]

2894 −

−1.5370

−1.2179

−0.0219

−2.0283***

2.0389*** 1.6552*** 1.5045***

[12]

Multilevel mixed-effects RML

2894 0.098

−1.3988

−1.2074

−0.0403

−2.0275***

2.0446*** 1.6990*** 1.4987***

[13] c

Cluster-robust random-effects panel GLS

2894 0.039

−2.2777***

−1.2308***

0.1078

−2.0128***

2.0043*** 1.4715*** 1.5069***

[14] d

Cluster-robust fixed-effects panel LSDV

a

Notes Breusch–Pagan test: χ2=1098.35, p=0.000 b Hausman test: χ2=91.10, p=0.000 c Breusch–Pagan test: χ2=1324.70, p=0.000 d Hausman test: χ2=47.72, p=0.159 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

Interaction term All domestic outsider investors × Voucher privatization countries Foreign investors × Voucher privatization countries All insiders × Voucher privatization countries Voucher privatization countries

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

META-REGRESSION ANALYSIS CONCERNING 6.6

213

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

and the all domestic outsider investor variable is significantly negative in five models in Panels a and b of Table 6.8. These results strongly suggest that, compared with studies of CEE and FSU coun­ tries that did not favor voucher privatization, studies covering other transition econ­ omies that adopted a voucher system as the primary method of enterprise privatization include a much larger number of empirical results that do not support theoretical predictions concerning the interrelationship of different owning entities. One of the reasons these two study types generated such highly asymmetrical empir­ ical findings is that the performance of firms owned by domestic outsider investors fell far short of expectations in voucher privatization countries. In that sense, we conjecture that the indiscriminate transfer of state assets free of charge did not adequately inspire these owners to make an effort to restructure the privatized firms. In addition, the estimation using the basic category of ownership variable (not reported) makes it clear that estimates for the unspecified domestic institutional investor ownership variable and the unspecified domestic financial institution owner­ ship variable in studies of voucher privatization countries are far inferior to those in studies of other countries in terms of both effect size and statistical significance. These results are also noteworthy for elucidating why domestic private company owners might not have remarkably improved firm performance. 6.6.3 Idiosyncrasies of countries that favored MEBOs Estimations that treat the idiosyncrasies of transition countries that favored MEBOs are represented in Table 6.9. In this table, the interaction terms show no robust coef­ ficients, regardless of differences in the dependent variable. This result implies that the policy of encouraging managers or rank-and-file employees to buy out their own companies as the most favored privatization method did not lead to marked differ­ ences in the empirical results as compared with those of transition countries that emphasized other privatization methods. However, in Panel b of Table 6.9, the vari­ able of a MEBO privatization country itself is estimated to be significant and posi­ tive in four models, suggesting that the statistical significance of estimates reported in studies on countries that made MEBOs a priority is higher on average than those in studies of other transition countries. According to estimation results that employed the basic category of ownership variable (not reported), the interaction term with the unspecified domestic financial institution ownership variable and the domestic bank ownership variable shows a robust and positive coefficient for both effect size and statistical significance, while the interaction term with the domestic institutional investor variable does so for statistical significance. This result hints that domestic institutional investors in MEBO-favoring countries, which were mainly financial institutions, more favorably affected the restructuring of privatized firms they owned than did those in other tran­ sition countries. In contrast, the interaction term with the employee ownership vari­ able is given a negative coefficient in six and five models for effect size and statistical significance, respectively, which clearly illustrates that employee insider 214

K R2

0.0120 0.0220

−0.0382

0.0313 2894 0.231

−0.0397

−0.0462

2894 0.159

0.0248

0.0133 0.0694* 0.0090

[2]

Cluster-robust WLS [Quality level]

0.0242

0.0156 0.0655* 0.0264

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Interaction term All domestic outsider investors × MEBO privatization countries Foreign investors × MEBO privatization countries All insiders × MEBO privatization countries MEBO privatization countries

[1]

Clusterrobust OLS

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

2894 0.363

0.0041

0.0020

−0.0166

0.0200

0.0061 0.0430* 0.0210

[3]

2894 0.515

0.0022

−0.0167

−0.0032

0.0152

−0.0061 0.0804* 0.0189

[4]

ClusterClusterrobust WLS robust WLS [N] [1/SE]

2894 −

0.0864*

−0.0295

−0.0835*

0.0367

0.0080 0.0578* 0.0114

[5]

Multilevel mixed-effects RML

2894 0.044

0.1347*

−0.0310

−0.1070*

0.0361

0.0073 0.0577* 0.0104

[6] a

(Continued )

2894 0.007

0.0833*

−0.0293

−0.0819*

0.0367

0.0081 0.0579* 0.0115

[7] b

Cluster-robust random-effects panel Cluster-robust fixed­ GLS effects panel LSDV

Table 6.9 Meta-regression analysis of the idiosyncrasy of MEBO privatization countries: Estimation using the aggregated category of ownership variable

META-REGRESSION ANALYSIS CONCERNING 6.6

215

C H A P T E R 6

216

1.2591

2.1539* 2894 0.289

0.7337

−0.3624

2894 0.223

−1.6721

−2.1237

2894 0.578

0.5309

2.9605

−4.3168

3.8744

0.8801 8.0491* 1.3559

[10]

2894 0.381

0.6406

0.6802

−0.4557

0.8927

0.6199 4.7872* 0.6518

[11]

ClusterClusterrobust WLS robust WLS [N] [1/SE]

2894 −

4.1625*

0.1769

−1.8739*

1.3825

0.3273 1.7929* 0.5883

[12]

Multilevel mixed-effects RML

2894 0.096

3.9137*

0.1331

−1.8051

1.3706

0.3367 1.8296* 0.5970

[13] c

2894 0.022

5.0217*

0.2999

−2.1461*

1.4114

0.3008 1.7132* 0.5611

[14] d

Cluster-robust random-effects panel Cluster-robust fixed­ GLS effects panel LSDV

a

Notes Breusch–Pagan test: χ2=1103.65, p=0.000 b Hausman test: χ2=94.95, p=0.000 c Breusch–Pagan test: χ2=1348.50, p=0.000 d Hausman test: χ2=439.83, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

1.8745

0.2616 2.4301* 0.2414

[9]

Cluster-robust WLS [Quality level]

1.5012

0.4687 3.1397* 0.6463

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Interaction term All domestic outsider investors × MEBO privatization countries Foreign investors × MEBO privatization countries All insiders × MEBO privatization countries MEBO privatization countries

[8]

Meta-independent variable (Default)/Model

Clusterrobust OLS

6

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

Table 6.9 contd.

C H A P T E R

PRIVATIZATION

META-REGRESSION ANALYSIS CONCERNING

6.6

ownership is harmful. This is an interesting finding in terms of understanding the idiosyncrasies of the MEBO method. 6.6.4 Idiosyncrasies of countries that favored direct sales

C H A P T E R 6

Table 6.10 strongly suggests that direct sales to strategic investors was an extremely effective privatization method, particularly for actualizing the effect of domestic out­ sider ownership on firm performance. We see this in Panels a and b of the table, where the interaction term between the all domestic outsider investor variable and the variable of direct-sale privatization countries is estimated with a significant and positive sign in five models. However, the interaction term with the foreign investor ownership variable is insignificant for all models except one. These results imply that, in countries that executed direct sales as the primary method of privatization, the differences between foreign investors and domestic outsider investors in terms of the ownership effect on firm performance were much smaller than in other transition countries. The strict screening process of selecting acquirers of state assets in coun­ tries such as Hungary and Poland may have greatly contributed to the discovery of domestic investors whose competence is on a par with that of foreign investors. With respect to the estimation that uses the basic category of ownership variable (not reported), in five or more models that take the PCC as the dependent variable, the inter­ action term of the variable of direct-sale privatization countries with the domestic out­ sider individual investor ownership variable, the unspecified domestic institutional investor ownership variable, and the domestic non-bank financial institution variable are given significant and positive coefficients. Meanwhile, in the case of the estimation with the t value on the left side, the unspecified domestic financial institution ownership vari­ able, the domestic non-bank financial institution ownership variable, the unspecified insider ownership variable, and the employee ownership variable show significant and positive estimates. These results indicate that domestic institutional investors and insiders are more active restructurers in direct sales-favoring countries than in other tran­ sition countries, which is consistent with the above discussion. 6.6.5 Differences in privatization policy implementation speed To investigate the impact of the speed of implementing privatization policy on the empirical results of previous studies, we sorted the transition countries into higher and lower groups on the basis of the medium value of 70% in terms of the private sector share of GDP in 2010 as reported in Table 6.1. We then estimated the variable of slow-speed privatization and its interaction terms with the ownership variable types. Table 6.11 shows the results. In Panel a of this table, the variable of slow-speed privatization is estimated with a significant and negative coefficient in six of the seven models. However, the interaction term with the foreign investor ownership variable is given a significant and positive estimate in six models. In other words, the effect size reported in studies of countries in which progress with privatization tended to be slow is, with a range of 0.0258 to 0.0705, lower than that reported in 217

6

218

K R2

0.0568** −0.0381**

0.0278

−0.0261 2894 0.231

0.0231

0.0246

2894 0.154

0.0526**

0.0072 0.0616** 0.0021

[2]

0.0521**

0.0111 0.0553** 0.0175

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Interaction term All domestic outsider investors × Direct-sale privatization countries Foreign investors × Direct-sale privatization countries All insiders × Direct-sale privatization countries Direct-sale privatization countries

[1]

Cluster-robust ClusterWLS [Quality robust OLS level]

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

−0.0084

−0.0009

2894 0.363

2894 0.516

−0.0092

0.0136

−0.0263

0.0243

0.0482

−0.0073 0.0769** 0.0145

[4]

0.0008

0.0224** 0.0533** 0.0296**

[3]

ClusterClusterrobust WLS robust WLS [N] [1/SE]

2894 −

−0.0193

0.0476

0.0103

0.0581**

0.0027 0.0467** −0.0004

[5]

Multilevel mixed-effects RML

2894 0.045

−0.0198

0.0473

0.0111

0.0579**

0.0028 0.0468** −0.0002

[6] a

Cluster-robust random-effects panel GLS

2894 0.009

−0.0150

0.0493

0.0021

0.0585**

0.0018 0.0455** −0.0023

[7] b

Cluster-robust fixed-effects panel LSDV

Table 6.10 Meta-regression analysis of the idiosyncrasy of direct-sale privatization countries: Estimation using the aggregated category of ownership variable

C H A P T E R

PRIVATIZATION

−1.1446

−0.9745 2894 0.280

1.2875

0.8524

2894 0.209

0.9601

1.4413 −0.9933

−5.1380**

2894 0.577

2894 0.391

−1.1174

3.6916

−6.4624**

6.4581**

3.5369

0.3448 3.5933** 0.8981

[11]

−2.1961

4.9810** 10.5036** 6.1062**

[10]

ClusterClusterrobust WLS robust WLS [N] [1/SE]

2894 −

−1.7406

1.9606**

1.0623

2.3581**

0.1175 1.3678** 0.3306

[12]

Multilevel mixed-effects RML

2894 0.071

−1.6865

1.9499**

1.0606

2.3502**

0.1242 1.3998** 0.3360

[13] c

Cluster-robust random-effects panel GLS

2894 0.011

−2.0098

1.9813**

1.0597

2.3778**

0.0897 1.2377** 0.3115

[14] d

Cluster-robust fixed-effects panel LSDV

a

Notes Breusch–Pagan test: χ2=1090.06, p=0.000 b Hausman test: χ2=76.09, p=0.000 c Breusch–Pagan test: χ2=1373.35, p=0.000 d Hausman test: χ2=47.34, p=0.051 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

1.5999**

0.1795 2.0010** 0.3138

[9]

1.8244**

0.4198 2.4751** 0.7271

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Interaction term All domestic outsider investors × Direct-sale privatization countries Foreign investors × Direct-sale privatization countries All insiders × Direct-sale privatization countries Direct-sale privatization countries

[8]

Cluster-robust ClusterWLS [Quality robust OLS level]

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

META-REGRESSION ANALYSIS CONCERNING 6.6

219

C H A P T E R 6

6

220

K R2

Interaction term All domestic outsider investors × Slow-speed privatization countries Foreign investors × Slow-speed privatization countries All insiders × Slow-speed privatization countries Slow-speed privatization countries

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

2894 0.163

−0.0308*** 2894 0.249

−0.0258***

−0.0104

0.0876***

0.0727*** 0.0405

0.0494***

0.0017 0.0482*** 0.0215

[2]

Cluster-robust WLS [Quality level]

0.0490***

0.0014 0.0410*** 0.0071

[1]

Clusterrobust OLS

2894 0.372

−0.0433***

0.0024

0.0598***

0.0281

0.0083 0.0297*** 0.0210***

[3]

Clusterrobust WLS [N]

2894 0.519

−0.0310

0.0234

0.0568***

0.0159

−0.0087 0.0644*** 0.0047

[4]

Clusterrobust WLS [1/SE]

2894 −

−0.0414***

0.0307

0.0578***

0.0180

0.0049 0.0311*** −0.0047

[5]

Multilevel mixed-effects RML

2894 0.049

−0.0403***

0.0307

0.0579***

0.0183

0.0049 0.0313*** −0.0045

[6] a

Cluster-robust random-effects panel GLS

2894 0.009

−0.0705***

0.0306

0.0608

0.0157

0.0048 0.0275*** −0.0063

[7] b

Cluster-robust fixed-effects panel LSDV

Table 6.11 Meta-regression analysis of the idiosyncrasy of slow-speed privatization countries: Estimation using the aggregated cat­ egory of ownership

C H A P T E R

PRIVATIZATION

0.1307 −0.7717

0.9118

−1.2734 2894 0.302

4.0240***

3.6388***

2894 0.221

1.3111***

0.0687 1.5061*** 0.6494

[9]

Cluster-robust WLS [Quality level]

1.4589***

0.1343 2.0256*** 0.5564

[8]

Clusterrobust OLS

2894 0.602

−9.5625***

3.3611

16.6145***

5.0914

1.0421 4.6994*** 2.3723

[10]

Clusterrobust WLS [N]

2894 0.387

−3.0870

1.3154

3.3199

1.3621

0.1819 3.5412*** 0.1885

[11]

Clusterrobust WLS [1/SE]

2894 −

−1.3523

0.3227

0.8445

0.2288

0.3593 1.3013*** 0.5232

[12]

Multilevel mixed-effects RML

2894 0.077

−1.2504

0.3465

0.8981

0.2575

0.3582 1.3256*** 0.5203

[13] c

Cluster-robust random-effects panel GLS

2894 0.028

−2.0658

0.2475

0.7610

0.1517

0.3567 1.2077*** 0.5350

[14] d

Cluster-robust fixed-effects panel LSDV

a

Notes Breusch–Pagan test: χ2=883.91, p=0.000 b Hausman test: χ2=79.94, p=0.000 c Breusch–Pagan test: χ2=1247.71, p=0.000 d Hausman test: χ2=153.35, p=0.000 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 6.4 for definition and descriptive statistics of meta-independent variables.

K R2

Interaction term All domestic outsider investors × Slow-speed privatization countries Foreign investors × Slow-speed privatization countries All insiders × Slow-speed privatization countries Slow-speed privatization countries

Ownership variable type (State) All domestic outsider investors Foreign investors All insiders

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

META-REGRESSION ANALYSIS CONCERNING 6.6

221

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

studies of rapidly privatizing countries. At the same time, it is also confirmed that, in slow-speed privatization countries, the gap in the ownership effect between foreign investors and other owning entities is much greater. As discussed in Subsection 2.2, in countries where enterprise privatization progressed rapidly, factors such as more intense interfirm competition and the crowding out of poorly performing domestic firms from the market may have led to a decline in the relative superiority of foreign ownership.8 According to the estimation results using the basic category of ownership variable (not reported), in the meta-regression models that use the PCC as the dependent vari­ able, the interaction term of the variable of slow-speed privatization countries with the unspecified domestic financial institution ownership variable, the domestic nonbank financial institution ownership variable, and the managerial ownership variable are estimated to be significant and positive in four or more models. Meanwhile, in models that take the t value on the left-hand side, the same is true for the domestic non-bank financial institution ownership variable, the domestic company groups and holdings ownership variable, and the managerial ownership variable. These results imply the relative superiority of domestic institutional investors and firm managers as corporate owners in slowly privatizing countries.9 In sum, the estimation results of the extended meta-regression model reveal that the differences among CEE and FSU countries in geographical conditions, privatization methods, and policy implementation speed strongly influence the link between postprivatization ownership structure and firm performance. At the same time, however, we notice that the real picture is more complex than we predicted in Subsection 2.2. In fact, these specific factors in transition economies not only cause the remarkable gap between countries in terms of ex-post improvement in firm performance but also significantly affect the interrelationship between foreign investors, domestic outsider owners. and firm managers, and the relative superiority among different domestic outsiders. From this viewpoint, the empirical findings reported in this section may provide a new interpretation of the historic privatization policy experiments conducted in the CEE and FSU region.

6.7 ASSESSMENT OF PUBLICATION SELECTION BIAS As the final step of our meta-analysis, in this section we will test the PSB and the presence of genuine empirical evidence in this research field. Figure 6.5 shows funnel plots for the four ownership variable types of the aggre­ gated category. The plots employ the PCCs and inverse standard errors of the col­ lected estimates. According to statistical theory, if PSB is absent, the effect sizes reported by independent studies should be distributed randomly and symmetrically around the true effect. Furthermore, the dispersion of the effect size is predicted to be negatively correlated with the precision of the estimate. Therefore, the shape of this scatter plot should look like an inverted funnel. With that in mind, an examination of the funnel plots in Figure 6.5 reveals that even if the true effect is assumed to be zero, and even if the mean of the most precise 10% of

222

ASSESSMENT OF PUBLICATION SELECTION BIAS

6.7

C H A P T E R 6

Figure 6.5 Funnel plot of estimates by aggregated category of ownership variable Note: Solid line indicates the mean of the top 10% most-precise estimates. The values for state, all domestic outsider investors, foreign investors, and all insiders are -0.012, 0.027, 0.017, and 0.019, respectively.

estimates—as denoted by the solid line in the figure—is regarded as the approximation value of the true effect, it is difficult to assert that the data match the prediction from statistical theory, namely, that for all ownership variable types, the collected estimates are not distributed with a bilateral symmetry and in a triangular shape.10 The asymmetry is particularly marked in the case of the foreign investor ownership variable. Let us assume that the true effect is zero. The ratio of positive to negative PCCs is 306:291 for the state ownership variable, 627:319 for the all domestic outsider investor ownership variable, 641:233 for the foreign investor ownership variable, and 341:136 for the all insider ownership variable. Therefore, the null hypothesis that the ratio of positive

223

C H A P T E R 6

PRIVATIZATION

and negative values is the same is rejected at the 1% level for the three variable types other than the state ownership variable. Furthermore, when the true effect is assumed to be close to the mean of the most precise 10% of estimates and the collected estimates are divided by two, with this value being the threshold, the ratio for each ownership variable type becomes 244:353, 519:427, 402:472, and 188:289, respectively. Hence the null hypothesis that the ratio of the above-mean values and the below-mean values is equal is rejected for all four variable types. These results therefore suggest that regardless of differ­ ences in the ownership variable types, it is highly probable that type I PSB is present. Figure 6.6 displays Galbraith plots using the t values and the inverse of standard errors of the collected estimates. The figure strongly suggests that type II PSB is

Figure 6.6 Galbraith plot of estimates by aggregated category of ownership variable

Note: Solid lines indicate the thresholds of two-sided critical values at the 5% significance level ±1.96.

224

CONCLUSIONS

6.8

present for all of the ownership variable types. In fact, the percentage of collected estimates for which the t value is within the range of ±1.96 or the two-sided critical values of the 5% significance level is 75.7% for the state ownership variable, 73.4% for the all domestic outsider investor ownership variable, 48.0% for the foreign investor ownership variable, and 66.0% for the all insider ownership variable. Accord­ ingly, the null hypothesis that the ratio is 95% is strongly rejected for all variable types. Even if we assume that the mean of the most precise 10% of estimates is the true effect, the percentage of estimates where the statistic |(k-th estimation result – true effect)=SEk | does not exceed the threshold of 1.96 accounts for 70.5%, 75.2%, 48.9%, and 73.%, respectively, and thus the null hypothesis is rejected once again for all of the variable types. These results indicate that, irrespective of the differences in variable types, the likelihood of type II PSB is considerably high in this study area. Table 6.12 reports the estimation results of meta-regression models, which are designed to test for two types of PSB and the presence of genuine empirical evidence. If we employ as a judgment criterion the question of whether the null hypothesis is rejected for at least two out of three models for each variable type, then Panel a of this table shows that the FAT strongly rejects the null hypothesis for the foreign investor ownership variable, the funnel plot for which exhibits marked asymmetrical distribution. Hence, type I PSB is strongly suspected. In the case of the remaining three types of ownership variables, the null hypothesis is accepted, suggesting that the effect of type I PSB is slight. However, the results of the type II PSB test shown in Panel b of the table strongly reject the null hypothesis for all of the ownership variable types, which backs up the impression obtained from the Galbraith plots. Further, according to the results of the PET reported in Panel a of Table 6.12, we find that the null hypothesis is rejected except for the state ownership variable. It is therefore highly likely that in the case of three private ownership variables, the collected estimates contain genuine evidence beyond any publication bias. In fact, Panel c of the same table shows that the PEESE approach resulted in a strong rejection of the null hypothesis for these three variable types, and judging from the coefficient of β1, we can ascertain that the true effect of all of the private ownership variables is significantly positive. Table 6.13, in addition to a summary of the above test results based on the aggre­ gated category of ownership variable, also presents a summary of results based on the basic category. As this table shows, the presence of type I PSB is confirmed for 5 of the 18 cases, while the type II PSB is detected in 15 of the 18 cases. At the same time, according to the PET and PEESE results, a publication selection bias-adjusted effect size is obtained in 10 of the 18 cases. These outcomes prove a reasonable suc­ cess in identifying the real impacts of post-privatization ownership structure on the performance of privatized enterprises in the formerly socialist transition economies.

6.8 CONCLUSIONS The privatization of state-owned enterprises in CEE and FSU countries constituted a social experiment on a scale never seen before in the economic history of the 225

C H A P T E R 6

226

K R2

Intercept (FAT: H0:

β0=0)

1/SE (PET: H0: β1=0)

Model

Estimator

597 0.1615

597 0.1615

−0.0289

−0.0289**

597 0.1615

0.0025 946 0.1572

0.0298**

−0.2177

−0.2404

1.0997

[4]

[3] a

OLS

[2]

1.0997**

[1]

OLS

Clusterrobust OLS

946 0.1572

0.0298**

−0.2177

[5]

Clusterrobust OLS

946 0.1572

−0.0076

0.9822

[6] b

Clusterrobust fixedeffects panel LSDV

II. All domestic outsider investors

Clusterrobust fixedeffects panel LSDV

I. State

874 0.0425

0.0142**

1.7987**

[7]

OLS

874 0.0425

0.0142**

1.7987**

[8]

Clusterrobust OLS

874 0.0425

0.0085

2.2801**

[9] c

Clusterrobust randomeffects panel GLS

III. Foreign investors

477 0.1827

0.0296**

0.2914**

[10]

OLS

477 0.1827

0.0296**

0.2914

[11]

Clusterrobust OLS

477 0.1827

0.0269**

0.3130

[12] d

Cluster­ robust random­ effects panel GLS

IV. All insiders

6

Estimates to test

(a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

Table 6.12 Meta-regression analysis of publication selection bias by aggregated category of ownership variable

C H A P T E R

PRIVATIZATION

597 0.1911

597 0.1911

0.0266

0.0266**

K R2

0.8153

0.8153**

Intercept (H0: β0=0)

1/SE

[14]

[13]

OLS

Clusterrobust OLS

I. State

Model

Estimator

Estimates to test

597 0.1911

0.0133**

1.3811**

[15] e

Clusterrobust fixedeffects panel LSDV

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v)

946 0.1815

0.0249**

0.7554**

[16]

OLS

946 0.1815

0.0249**

0.7554**

[17]

Clusterrobust OLS

946 0.1815

0.0227**

0.9493**

[18] f

Clusterrobust randomeffects panel GLS

II. All domestic outsider investors

874 0.0820

0.0176**

2.5518**

[19]

OLS

874 0.0820

0.0176**

2.5518**

[20]

Clusterrobust OLS

874 0.0820

0.0129**

2.4931**

[21] g

Clusterrobust randomeffects panel GLS

III. Foreign investors

477 0.2058

0.0236**

1.0974**

[22]

OLS

477 0.2058

477 0.2058

0.0242**

1.0854**

[24] h

(Continued )

0.0236**

1.0974**

[23]

Clusterrobust OLS

Cluster­ robust random­ effects panel GLS

IV. All insiders

CONCLUSIONS 6.8

227

C H A P T E R 6

228

597 0.1262

10.2009** −0.0205**

[25]

OLS

597 0.1262

597 −

946 0.2402

−0.1554 0.0261**

−7.9188 −0.0269**

10.2009 −0.0205

[28]

[27]

[26]

OLS

946 0.2402

−0.1554 0.0261**

[29]

Clusterrobust OLS

946 −

−0.2518 0.0224**

[30]

Randomeffects panel ML

a

Notes Breusch–Pagan test: χ2=402.59, p=0.000; Hausman test: χ2=17.09, p=0.000 b Breusch–Pagan test: χ2=285.06, p=0.000; Hausman test: χ2=5.89, p=0.015 c Breusch–Pagan test: χ2=4754.29, p=0.000; Hausman test: χ2=0.75, p=0.387 d Breusch–Pagan test: χ2=835.36, p=0.000; Hausman test: χ2=0.39, p=0.535 e Breusch–Pagan test: χ2=464.07, p=0.000; Hausman test: χ2=13.04, p=0.001 f Breusch–Pagan test: χ2=542.97, p=0.000; Hausman test: χ2=0.27, p=0.606 g Breusch–Pagan test: χ2=1609.92, p=0.000; Hausman test: χ2=0.02, p=0.889 h Breusch–Pagan test: χ2=424.53, p=0.000; Hausman test: χ2=0.21, p=0.645 Robust standard errors are used for hypothesis testing except for models [27], [30], [33], and [36]. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

K R2

SE 1/SE (H0: β1=0)

Model

Estimator

Randomeffects panel ML

II. All domestic outsider investors

Clusterrobust OLS

I. State

874 0.2071

20.8845** 0.0228**

[31]

OLS

874 0.2071

20.8845** 0.0228**

[32]

Clusterrobust OLS

874 −

10.2603 0.0124**

[33]

Randomeffects panel ML

III. Foreign investors

477 0.3584

2.4375** 0.0330**

[34]

OLS

477 0.3580

2.4375 0.0330**

[35]

Clusterrobust OLS

477

−

0.5854 0.0302**

[36]

Random­ effects panel ML

IV. All insiders

6

Estimates to test

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Table 6.12 contd.

C H A P T E R

PRIVATIZATION

CONCLUSIONS

6.8

C H A P T E R 6

Table 6.13 Summary of publication selection bias test Test results b

Ownership variable type a I. State 1. Unspecified government 2. Central government 3. Regional/local government II. All domestic outsider investors 4. Unspecified domes­ tic outsider investors 5. Domestic outsider individual investors 6. Unspecified domes­ tic outsider institu­ tional investors 7. Unspecified domes­ tic financial institutions 8. Domestic banks 9. Domestic non-bank financial institutions 10. Domestic company groups and holdings 11. Other domestic non­ financial companies III (12). Foreign investors

Funnel asymmetry test for type I PBS (FAT) (H0: β0=0)

Test for type II PBS (H0: β0=0)

Precisioneffect test (PET) (H0: β1=0)

Precision-effect estimate with standard error (PEESE) (H0: β1=0) c

597

Not rejected

Rejected

493

Not rejected

Rejected

Rejected (−0.0267/−0.0205) Not rejected

60

Rejected

44

Rejected

Not rejected Rejected

Not rejected Not rejected Rejected Rejected

946

Not rejected

Rejected

Rejected

109

Rejected

Rejected

168

Not rejected

Rejected

Not rejected Rejected

98

Not rejected

Not rejected

Rejected

123

Not rejected

Rejected

Not rejected

Not rejected

95

Not rejected

Rejected

Not rejected

144

Not rejected

Rejected

Not rejected Rejected

77

Not rejected

Number of esti­ mates (K)

132

Rejected

Not rejected Rejected

874

Rejected

Rejected

Not rejected Rejected

477

Not rejected

Rejected

Rejected

13. Unspecified insiders

163

Not rejected

Rejected

Rejected

14. Managers

187

Not rejected

Rejected

15. Employees

127

Not rejected

Rejected

Not rejected Not rejected

IV. All insiders

Rejected

Rejected (−0.0459/−0.0384) Rejected (−0.0748/−0.0743) Rejected (0.0224/0.0261) Rejected (0.0137/0.0193) Rejected (0.0251/0.0265) Rejected (0.0375/0.0426)

Rejected (0.0112) Rejected (0.0537/0.0689) Rejected (0.0222) Rejected (0.0124/0.0228) Rejected (0.0302/0.0330) Rejected (0.0340/0.0363) Rejected (0.0284/0.0261) Not rejected

Notes a Ownership variable types with Arabic numerals belong to the basic category, while those with Roman numerals belong to the aggregated category. b The null hypothesis is rejected when more than two of three models show a statistically significant estimate. Otherwise not rejected. c Figures in parentheses are PSB-adjusted estimates. If two estimates are reported, the left and right figures denote the minimum and maximum estimate, respectively.

229

C H A P T E R 6

PRIVATIZATION

world. Studying the design of the privatization methods, their implementation pro­ cess, and their outcomes has been a key task for those who study transition econ­ omies. Due to their great efforts over the past quarter century, the study of privatization has expanded to such an extent that it probably boasts more studies than any other area in the field of transition economics. These studies provide valu­ able and plentiful insights not only for understanding the formerly socialist transition economies but also from the standpoint of corporate finance and organizational economics. This trend has also produced numerous studies that empirically examined the rela­ tionship between post-privatization ownership and firm performance. Reflecting pro­ gress in the implementation of privatization policies in the CEE and FSU regions, the volume of these works peaked in the first half of the 2000s, although such stud­ ies have steadily continued to be published until the present day. The accumulation of empirical evidence has gradually come to satisfy the thirst of researchers for an answer to the question of what sort of owners are most desirable for the restructur­ ing of firms formerly owned by the state. Nevertheless, it remains extremely difficult to gauge the big picture of empirical findings revealed in the existing literature. This is because the number of studies is so large and their empirical results are too mixed to determine whether the experiences of transition economies support the standard theory regarding the interrelationship of different types of corporate ownership. To tackle this problem, in this chapter, we employed a total of 2,894 estimates drawn from 121 relevant studies published from 1996 to 2015 to perform a metaanalysis of the impact of post-privatization ownership on firm performance. The col­ lected estimates encompass almost all CEE and FSU countries, and their estimation period covers 27 years from 1985 to 2011. Therefore, they are ideal for conducting a comprehensive assessment of the study of CEE and FSU economies in the transi­ tion period. Findings from the meta-synthesis of the collected estimates performed in Section 6.4 provided support for the theoretical predictions concerning the superiority of the private sector over the state and the inefficiency of employees as compared with firm managers as owners of their firms. However, it did not offer full support for predictions concerning interactions between private entities, including foreign invest­ ors. Meanwhile, an MRA that accounted for the heterogeneity of literature in Section 6.5 proved that, as compared with other corporate owners, the size of the effect and the statistical significance are much higher for foreign investors. This corresponds with our argument. Nevertheless, as was the case with the synthesis results in Sec­ tion 6.4, the MRA results did not provide comprehensive proof of the series of hypotheses proposed in Section 6.2. All of these results may reflect the high degree of complexity of privatization policies in CEE and FSU countries. In Section 6.6, with the aim of identifying factors that have caused disorderliness in this research field, we attempted to estimate an extended model that explicitly controlled for the idiosyncrasies of transition economies. The first noteworthy finding from this analysis was that the effect size and statistical significance of the foreign investors as compared with those of state and domestic private owners were much 230

CONCLUSIONS

6.8

larger in studies of FSU countries than in those of CEE countries. This suggests that foreign investors operating in FSU countries behaved, compared to their domestic coun­ terparts, as superior company owners to a greater extent than they did in CEE countries. Second, because meta-independent variables that exclude the influence of the idio­ syncrasies of voucher privatization countries clearly support the theoretical hypoth­ eses presented in Section 6.2, we infer that there is a strong possibility that countries that dished out state assets indiscriminately and free of charge through a voucher system failed to motivate the citizens who benefitted to make particularly striking efforts to restructure their firms. In other words, this result backs up the perceptive notion by Megginson and Netter that “most countries’ actual experience with vou­ chers has been poor” (2001, p. 345). Third, in countries that used direct sales to strategic investors as their core approach to privatization, the impact on firm performance of domestic outsider ownership was clearly higher than that in other countries in terms of both effect size and statistical sig­ nificance. Therefore, it is evident that the combination of restricted screening of acquir­ ers of state assets and their transfer for counter value was a highly effective means of discovering domestic company owners comparable to foreign investors. Fourth, in countries where enterprise privatization progressed swiftly, the differen­ tial between domestic owners and foreign investors as seen in their effect on firm performance was, in terms of effect size, much narrower than it was in nations that moved more slowly toward privatization. This finding indicates that rapid progress toward privatization, by intensifying interfirm competition and the exit of underper­ forming firms from the market, might have served to eliminate gaps among different company owners. In the above sense, the estimation results of the extended meta-regression model reported in Section 6.6 inform us that a comprehensive comparative analysis of dif­ ferences between nations as manifested in locations, privatization methods, and policy implementation speed is an effective way to shake off the opaqueness of empirical findings in the extant literature and to derive clear and important theoret­ ical implications concerning the impact of ownership structure on firm performance in the post-privatization period.11 Furthermore, according to the assessment of PSB in Section 6.7, while there is a high probability that type II PSB exists in this field of research as a whole, the risk of type I bias is relatively low. As a result, we revealed that for 10 of the 18 ownership variable types, the estimates collected from the previous literature are highly likely to contain genuine empirical evidence. With regard to the remaining eight ownership variable types, more empirical studies are needed to understand their actual effect on firm performance. In light of the meta-analysis results summarized above, we would like to empha­ size the following two important lessons from the privatization study of CEE and FSU economies. First, in the former socialist transition economies, the private sector is more desir­ able than the state as a firm-owning entity. Because of this, privatization policy was a vital element for the restructuring of domestic firms in every country. Indeed, 231

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

“privatization is transition” (Brada 1996). However, the selection of owners is more important than privatization itself. In fact, results of the meta-analysis in this chap­ ter provide strong support for Kornai’s assertion that “state property must be squandered by distributing it to one and all merely out of kin. … The point now is not to hand out the property, but rather to place it into the hands of a really better owner” (1990, pp. 81–82). It also supports Stiglitz’s reminder that “property rights are more important, [and] how property rights are assigned may be more import­ ant” (1994, p. 176). We found that the restructuring effect of enterprise privatization is heavily influ­ enced by the policy method and the speed of implementation, as well as countryspecific factors. From this point of view, mass privatization via vouchers was extremely problematic. This is due to the fact that, in voucher-privatization coun­ tries, there was a high risk that post-privatization owners would dramatically reduce their efforts to restructure their firms, irrespective of the differences in their attri­ butes. This was an obvious side effect of mass privatization through the use of a voucher system being carried out with the primary political aims of obtaining the support of citizens and adhering to the Washington Consensus, with the economic goal of restructuring privatized firms being secondary. In contrast to the bitter experiences of the voucher-privatization countries, it is almost certain that direct sales to strategic investors was quite an effective method from the viewpoint of improved post-privatization firm performance. In this case, the profit-seeking motivation seems to have served as a highly effective tool for inspiring new owners to restructure. As shown in Table 6.1, direct sales were car­ ried out in 21 of the 28 CEE and FSU countries. It is extremely interesting that this method also came to be emphasized in countries that had initially conducted voucher privatization. The second biggest point at issue in the debate on enterprise privatization in CEE and FSU countries has been whether insiders and domestic outsider investors are superior. According to the results of our meta-analysis, the series of empirical studies over the past quarter century have not necessarily arrived at a single conclusion with regard to this point. In most transition countries, privatization policies were designed and implemented to benefit insiders and were, in a sense, natural political choices, given that it was necessary to find people to take over more than 150,000 large and medium-sized state-owned firms and hundreds of thousands of small state-owned firms (Åslund 2013). In the case of insider ownership, there was a strong tendency for employees to remain in their posts and for the payment of wages to take priority over investment. This tendency posed a risk of diminishing the effect of restructuring. The large-scale implementation of privatization policies that favored insiders may, therefore, have had a large negative impact on the entire national economy. However, with regard to the question of whether the ownership of firms by outsider investors was definitely effective, Frydman et al. (2007) made the following point: if managers, who are agents, endeavor to satisfy the speculative motives of outsider investors, who are principals, by doing everything they can to maximize short-term profits, company 232

NOTES

management with a long-time horizon may be neglected. As a consequence, the ini­ tially expected restructuring effect will not be adequately realized. Furthermore, in cases where there is extremely serious information asymmetry between outsider investors and firm managers, it is impossible to reject the possibility that managerial ownership will have a more favorable impact on restructuring than ownership by outsider investors by solving the problem caused by the separation of ownership and control. From this point of view, the results of our meta-analysis, which imply com­ petitive impacts between insiders and domestic outsider investors on firm perform­ ance in the post-privatization period, are noteworthy. Nevertheless, it is difficult to make a rigorous distinction between insiders and outsider investors, either theoretically or in practice. For example, the bank or hold­ ing company at the center of a business group, while formally an outsider investor from the point of view of its subsidiaries, actually behaves a lot more like an insider (Frydman and Rapaczynski 1994; Aoki et al. 2007). Thus, a future task is to further refine approaches for the comparative analysis of insider ownership and outsider investor ownership.

ACKNOWLEDGMENTS This chapter is an extended version of Iwasaki and Mizobata 2018. We thank Masato Hiwatari, Shuichi Ikemoto, Robert J. Johnston, Ryo Kambayashi, Evžen Kočenda, Marco A. Marini, and Tom D. Stanley for their helpful comments and sug­ gestions on the earlier version of this paper.

NOTES 1 To accomplish this objective effectively, the meta-analysis in this chapter limits its focus to CEE and FSU countries, not dealing with China and other transition economies. 2 For details of these 121 studies, see Table 3 and the reference list in Iwasaki and Mizobata 2017. 3 In this chapter, estimates of variables related to restructuring activity such as reorganization and capital investment are not used at all in the meta-analysis; rather, we focus on firm performance in the narrow sense, i.e., the efficiency and profitability of management and production activities. 4 Interaction terms with an ownership variable are not included in the collected estimates because they do not indicate any pure effect of the ownership structure itself. However, in the course of MRA, we will examine how the simultaneous estimation of an interaction term(s) affects estimates of the ownership variable. 5 Needless to say, the proportion of sample firms in observations for each country is excluded from the estimation.

233

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION

6 This result corresponds well with the empirical findings in Sabirianova et al. 2012, which demonstrate a larger efficiency gap between foreign and domestic firms in Russia as compared to that in the Czech Republic. They argue that the different findings for these two countries are caused by differences in the polit­ ical and institutional environments rather than the level of economic develop­ ment or the geographic proximity to Western business culture. 7 The estimation results using the basic category of ownership variable are avail­ able in the supplemental material in Iwasaki and Mizobata 2017. 8 As described above, we classified some transition economies as slow-speed pri­ vatization countries in terms of ex post policy implementation speed, referring to the private sector share of GDP in 2010. However, even if CEE and FSU countries are divided into two country groups in terms of initial implementation speed, referring to the private sector share of GDP in 1995, or in terms of the privatization year (see Bennett et al. 2007, tab. 1, p. 670), the results of metaregression analysis are not greatly different from those in Table 6.11. This is mainly because the composition of the slow-speed privatization countries is almost unchanged. 9 To deal with estimates derived from studies of multiple countries, we employed a proportion of the sub-sample group. However, the majority of the literature subject to meta-analysis is made up of single-country studies; hence, in most cases, this variable takes a value of 1. Even if multiple-country studies are com­ pletely excluded and a binary dummy variable for the countries concerned is used in place of the proportion of the subsample group, the conclusions drawn are not all that different from the meta-regression results in this section. Further­ more, the simultaneous estimation of all of the intercepted variables in question shows similar results to those reported in Tables 6.7 to 6.11. 10 The method for assuming that the mean of the most precise 10% of estimates is the approximate value of the true effect is along the lines of Stanley 2005. 11 Another factor that led to the unclear results of the meta-analysis performed in Sections 6.4 and 6.5 is the fact that study-specific research conditions, such as definitions of ownership variables and data-processing methods, underlying the literature covered by the meta-analysis cannot be excluded. However, the results presented in Section 6.6 indicate that even if they did have an effect, it was probably a minor one.

REFERENCES Aoki, Masahiko, Gregory Jackson, and Hideaki Miyajima (2007) Corporate Governance in Japan: Institutional Change and Organizational Diversity. Oxford University Press: Oxford. Arrow, Kenneth J. (2000) Economic transition: Speed and scope. Journal of Institutional and Theoretical Economics, 156(1), pp. 9–18.

234

REFERENCES Åslund, Anders (2013) How Capitalism was Built: The Transformation of Central and Eastern Europe, Russia, and Central Asia. Second edition, Cambridge University Press: New York. Bennett, John, Saul Estrin, and Giovanni Urga (2007) Methods of privatization and eco­ nomic growth in transition economies. Economics of Transition, 15(4), pp. 661–683. Bevan, Alan A., Saul Estrin, Boris Kuznetsov, Mark E. Schaffer, Manuela Angelucci, Julian Fennema, and Giovanni Mangiarotti (2001) The determinants of privatised enter­ prise performance in Russia. Working Paper No. 452, William Davidson Institute, Univer­ sity of Michigan: Ann Arbor. Blomstrom, Magnus, and Edward Wolff (1994) Multinational corporations and productivity convergence in Mexico. In William Baumol, Richard Nelson, and Edward Wolff (eds.), Convergence of Productivity: Cross-National Studies and Historical Evidence. Oxford University Press: New York, pp. 243–259. Bös, Dieter (1991) Privatization: A Theoretical Treatment. Clarendon Press: Oxford. Boycko, Maxim, Andrei Shleifer, and Robert Vishny (1995) Privatizing Russia. MIT Press: Cambridge, Mass. Brada, Josef C. (1996) Privatization is transition – or is it? Journal of Economic Perspec­ tives, 10(2), pp. 67–86. Brown, David J., and John S. Earle (2004) Does privatization raise productivity? Evi­ dence from comprehensive panel data on manufacturing firms in Hungary, Romania, Russia and Ukraine. Discussion Paper No. 4791, Center for Economic Policy Research (CEPR): London. Brown, David, Sergei Guriev, and Natalya Volchkova (1999) Financial-industrial groups in Russia: Virtue or vice. Russian Economic Trends, 8(3), pp. 6–12. Dewatripont, Mathias, and Gérard Roland (1996) Transition as a process of large-scale insti­ tutional change. Economics of Transition, 4(1), pp. 1–30. Dittus, Peter, and Stephen Prowse (1996) Corporate control in Central Europe and Russia: Should banks own shares? In Roman Frydman, Cheryl Gray, and Andrzej Rapaczynski (eds.), Corporate Governance in Central Europe and Russia. Volume 1, CEU Press: Budapest and London, pp. 20–67. Djankov, Simeon (2014) The microeconomics of post-communist transformation. In Anders Åslund and Simeon Djankov (eds.), The Great Rebirth: Lessons from the Victory of Capitalism over Communism. Peterson Institute for International Economics: Wash­ ington DC, pp. 187–204. Djankov, Simeon, and Peter Murrell (2002) Enterprise restructuring in transition: A quantitative survey. Journal of Economic Literature, 40(3), pp. 739–792. Dolgopyatova, Tatiana, Ichiro Iwasaki, and Andrei A. Yakovlev (2009) (eds.), Organization and Development of Russian Business: A Firm-level Analysis. Palgrave Macmillan: Basingstoke. Dunning, John (1986) Japanese Participation in British Industry. Croom Helm: London and Dover, NH. Earle, John S., and Saul Estrin (1996) Employment ownership in transition. In Roman Frydman, Cheryl Gray, and Andrzej Rapaczynski (eds.), Corporate Govern­ ance in Central Europe and Russia. Volume 2, CEU Press: Budapest and London, pp. 1–61. EBRD (European Bank for Reconstruction and Development) (2004) Transition Report 2004: Infrastructure. EBRD: London.

235

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION Estrin, Saul, Jan Hanousek, Evžen Koč enda, and Jan Svejnar (2009) The effects of privatiza­ tion and ownership in transition economies. Journal of Economic Literature, 47(3), pp. 699–728. Frydman, Roman, and Andrzej Rapaczynski (1994) Privatization in Eastern Europe: Is the State Withering Away? CEU Press: London. Frydman Roman, Marek Hessel, and Andrzej Rapaczynski (2006) Why ownership mat­ ters: Entrepreneurship and the restructuring of enterprises in Central Europe. In Merritt B. Fox, and Michael A. Heller (eds.), Corporate Governance Lessons from Transition Economy Reforms. Princeton University Press: Princeton, NJ, and Oxford, pp. 194–227. Frydman Roman, Cheryl Gray, Marek Hessel, and Andrzej Rapaczynski (2007) When does privatization work? The impact of private ownership on corporate performance in the transition economies. In Erik Berglöf and Gérard Roland (eds.), The Economics of Transi­ tion: The Fifth Nobel Symposium in Economics. Palgrave Macmillan: Basingstoke and New York, pp. 37–69. Frye, Timothy (2002) Capture or exchange? Business lobbying in Russia. Europe-Asia Stud­ ies, 54(7), pp. 1017–1036. IMF (International Monetary Fund) (2014) 25 Years of Transition: Post-communist Europe and the IMF. Regional Economic Issues, special report, IMF: Washington DC. Iwasaki, Ichiro (2007) Enterprise reform and corporate governance in Russia: A quantitative survey. Journal of Economic Surveys, 21(5), pp. 849–902. Iwasaki, Ichiro, and Satoshi Mizobata (2017) Post-privatization ownership and firm perform­ ance: A large meta-analysis of the transition literature. Working Paper No. 2016-13, Center for Economic Institutions, Institute of Economic Research, Hitotsubashi Univer­ sity: Tokyo. Iwasaki, Ichiro, and Satoshi Mizobata (2018) Post-privatization ownership and firm perform­ ance: A large meta-analysis of the transition literature. Annals of Public and Cooperative Economics, 89(2), pp. 263–322. Iwasaki, Ichiro, and Taku Suzuki (2007) Transition strategy, corporate exploitation, and state capture: An empirical analysis of the Former Soviet States. Communist and PostCommunist Studies, 40(4), pp. 393–422. Iwasaki, Ichiro, and Taku Suzuki (2012) The determinants of corruption in transition economies. Economics Letters, 114(1), pp. 54–60. Iwasaki, Ichiro, and Akira Uegaki (2019) The disinflation effect of central bank independ­ ence: A comparative meta-analysis between transition economies and the rest of the world. In Julien Chevallier, Stéphane Goutte, David Guerreiro, Sophie Saglio, and Bilel Sanhaji (eds.), International Financial Markets. Volume 1, Routledge: Abingdon, pp. 227–287. Iwasaki, Ichiro, Csaba Makó, Miklos Szanyi, Péter Csizmadia, and Miklos Illéssy (2012) Eco­ nomic Transformation and Industrial Restructuring: The Hungarian Experience. Maruzen Publishing: Tokyo. Johnson, Simon, Daniel Kaufmann, John McMillan, and Christopher Woodruff (2000) Why do firms hide? Bribes and unofficial activity after communism. Journal of Public Econom­ ics, 76(3), pp. 495–520. Jones, Derek, Mark Klinedinst, and Charles Rock (1998) Productive efficiency during transi­ tion: Evidence from Bulgarian panel data. Journal of Comparative Economics, 26(3), pp. 446–464.

236

REFERENCES Judge, William, Irina Naoumova, and Nadejda Koutzevol (2003) Corporate governance and firm performance in Russia: An empirical study. Journal of World Business, 38(4), pp. 385–396. Kapelyushnikov, R. (2000) The largest and dominant shareholders in the Russian industry: Evidence of the Russian economic barometer monitoring. Russian Economic Barometer, 9(1), pp. 9–46. Kogut, Bruce (1996) Direct investment, experimentation, and corporate governance in tran­ sition economies. In Roman Frydman, Cheryl Gray, and Andrzej Rapaczynski (eds.), Cor­ porate Governance in Central Europe and Russia. Volume 1, CEU Press: Budapest and London, pp. 293–332. Kornai, János (1990) Road to a Free Economy: Shifting from a Socialist System: The Example of Hungary. Norton: New York. Linz, Susan J. (2002) Ownership and employment in Russian industry: 1992–1995. Inter­ national Journal of Manpower, 23(1), pp. 32–62. Megginson, William L., and Jeffry M. Netter (2001) From state to market: A survey of empir­ ical studies on privatization. Journal of Economic Literature, 39(2), pp. 321–389. Mencinger, Joze (1996) Privatization experiences in Slovenia. Annals of Public and Cooperative Economics, 67(3), pp. 415–428. Miller, James (2013) Privatization. In Paul Hare and Gerard Turley (eds.), Handbook of the Economics and Political Economy of Transition. Routledge: London, pp. 131–137. Mizobata, Satoshi (2005) Evolution of Russian corporate governance. Journal of Compara­ tive Economic Studies, 1, pp. 25–58. Mizobata, Satoshi (2008) Diverging and harmonizing corporate governance in Russia. In John Pickles (ed.), State and Society in Post-Socialist Economies. Palgrave Macmillan: Basingstoke and New York, pp. 111–139. Perotti, Enrico C., and Stanislav Gelfer (2001) Red barons or robber barons? Governance and investment in Russian financial-industrial groups. European Economic Review, 45 (9), pp. 1601–1617. Quiggin, John (2010) Zombie Economics: How Dead Ideas Still Walk Among Us. Princeton University Press: Princeton, NJ. Radygin, A. (2014) (ed.) Privatization in the Contemporary World. Volume 2, Delo: Moscow (Russian). Roland, Gérard (2000) Transition and Economics: Politics, Markets, and Firms. MIT Press: Cambridge, Mass. Roland, Gérard (2008) (ed.) Privatization: Successes and Failures. Columbia University Press: New York. Sabirianova, Klara Peter, Jan Svejnar, and Katherine Terrell (2012) Foreign investment, cor­ porate ownership, and development: Are firms in emerging markets catching up to the world standard? Review of Economics and Statistics, 94(4), pp. 981–999. Shleifer, Andrei, and Robert Vishny (1994) Politicians and firms. Quarterly Journal of Eco­ nomics, 109(4), pp. 995–1025. Stanley, T. D. (2005) Beyond publication bias. Journal of Economic Surveys, 19(3), pp. 309–345. Stark, David, and László Bruszt (1998) Postsocialist Pathways: Transforming Politics and Property in East Central Europe. Cambridge University Press: Cambridge and New York. Stiglitz, Joseph E. (1994) Whither Socialism? MIT Press: Cambridge, Mass.

237

C H A P T E R 6

C H A P T E R 6

PRIVATIZATION Thompson, G. Rodney, and Calin Valsan (1999) Early privatization in Romania: The period of management and employee buyouts, 1991 to 1995. Eastern European Economics, 37 (6), pp. 35–53. Vittas, Dimitri, and Roland Michelitsch (1996) The potential role of pension funds: Lessons from OECD and developing countries. In Roman Frydman, Cheryl Gray, and Andrzej Rapaczynski (eds.), Corporate Governance in Central Europe and Russia. Volume 1, CEU Press: Budapest and London, pp. 242–292. Wright, M., R. S. Thompson, and K. Robbie (1989) Privatisation via management and employee buyouts: Analysis and U.K. experience. Annals of Public and Cooperative Economy, 60(4), pp. 399–429.

238

7

Human resource

management in

transition

Norio Horie and Kazuhiro Kumo

7.1 HUMAN RESOURCE MANAGEMENT IN EUROPEAN TRANSITION ECONOMIES In human resource management (HRM) studies, the Central and Eastern European countries, which experienced the socialist system, are the new frontier for developing new research. With the emphasis on the context of the transition process from a socialist economy to a capitalist economy, Zupan and Kase (2005*)1 named the Central and Eastern European countries the European Transition Economies (ETEs). This chapter attempts to investigate how socialist HRM as institutional and cultural legacies have been intentionally assessed in the HRM studies on ETEs. Enterprises in the former socialist countries shared a centralized corporate structure and socialistic corporate culture under strong state control. These socialist enterprises had to take survival measures to transform their corporate structure and culture into new ones adapted to the market economy. The transplanting of modern HRM practices into their management is also a very important precondition for their survival strategy. Their lack of modern HRM encouraged them to learn Western HRM practices in the process of transition to the market economy. However, the transition to the market economy did not derive from an “institutional vacuum” but depended on “a dense and complex institutional legacy such that the (often invisible) remnants of previous eco­ nomic and political orders still shape expectations and patterns of conduct” (Nielsen et al. 1995, p. 4). The legacies of the socialist personnel management functions that socialist enterprises used to have conventionally should be focused on in order to understand how they adapt their management to Western-style HRM functions. Thirty years have passed since the fall of the Berlin Wall. Do the institutional legacies of socialist personnel management in ETEs still attract researchers? Is it still a problem for us? Does it continue to be an important subject of consideration as a distinctive feature that differentiates the HRM studies in ETEs from those in other regions? The main concerns of this chapter derive from these questions. The special issue of the Baltic Journal of Management in 2010 focused on HRM in Central and Eastern Europe. It addressed Central and Eastern Europe as a region “characterized by significant structural/institutional and configurational differences,

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

along with significant practice differences in HRM, compared with other regions and territories” (Brewster et al. 2010, p. 146). The special issue of the Human Resource Management Journal in 2011 (Cooke et al. 2011) also focused on HRM in the region, and Horwitz (2011*), who contributed to this issue, stressed the lingering effects of the previous institutional environment both in external and internal con­ texts of the post-socialist countries. Garavan et al. (1998*) observed Polish firms’ struggle to adapt the strategic HRM used in the United States and the United King­ dom, with limited success due to the communist contextual legacies inherited from the Soviet model of management. The communist legacies have been in focus in HRM studies in this region. However, the importance of institutional and cultural legacies in HRM in this region has not been sufficiently reviewed.2 Brewster et al. (2010) argue that it is not clear what the transition countries are transforming to, although it is clear what they are transforming from. However, the starting point of the transition, which they consider as a matter of course, is the undefined framework of the old regime called a Soviet model of management. It is not always clear in the discussions in HRM research specifically what they are attempting to abandon and consider as strategic resources, and what form of HRM they aim to pursue. It is, therefore, necessary to conduct analytical survey research on HRM that places the transition experienced by ETEs contextually. There has been almost no analytical survey in which the trends in HRM studies in ETEs are analyzed with a special focus on the institutional and cultural legacies of the Soviet model of management. This chapter attempts to systematically analyze how the tran­ sition as a contextual perspective has been included within the framework of HRM studies in ETEs and determine the importance of studying institutional legacies of socialist personnel management.

7.2 LITERATURE REVIEW 7.2.1 Methodology of the literature survey In order to identify studies related to HRM in ETEs, we first examined the Web of Science and EconLit databases for published literature registered between 1989 and 2015 that contained a combination of two terms: one from “human resource manage­ ment” or “personnel management” and another from “transition,” “transition econ­ omies,” “post-socialist,” “post-communist,” or any of the names of the 21 Central and Eastern European countries (Albania, Bosnia–Herzegovina, Croatia, the Czech Republic, Hungary, Poland, Romania, Serbia, Slovakia, Slovenia, Bulgaria, Kosovo, Macedonia, Montenegro, Moldova, Estonia, Latvia, Lithuania, Russia, Ukraine, and Belarus). We excluded the non-English articles, those not related to HRM, and those in which less than one-third of the countries of the comparison subject were Central or Eastern European countries. In addition, articles that presented only a review and the research concepts but did not conduct any analysis of HRM in each country, either descriptively or empirically, were also excluded. The analysis subjects are 240

LITERATURE REVIEW

7.2

English articles published in academic journals but do not include those published in books. As a result, the number of listed references is 309.3 Among the listed articles, we identified those that specifically focused on the context of transition from socialist economy or management to capitalist economy or manage­ ment. We call the articles extracted in this way “the extracted basic references” in this chapter. The extraction was carried out as described below. Dividing the respective art­ icles into three structural parts: an introductory part, an analytical part, and a concluding part, the reference related to the transition in each part was taken up, and the articles that mention factors related to the transition in even one of the three structures were selected. The transition-related factors specifically mean the references related to culture, values, systems, and management practices that were inherited from the period of social­ ism; those that are related to culture, values, systems, and management practices that had been lacking in the period of socialism; and those that refer to major topics related to the transition to the market economy (privatization of state-owned enterprises, socio­ economic-environmental changes in the transition period, and other aspects). As a result of this extraction, 97 articles were specified as extracted basic references. 7.2.2 Research attributes of the extracted basic references This chapter does not require a narrow definition of HRM.4 For ETEs, HRM, which was introduced from the Western world during the transition to a market economy, was completely different from the management practices implemented under the socialist system (Pieper 1992). The personnel department in socialist enterprises carried out a limited range of personnel management practices, most of which were performed at the state level in accordance with a centralized planned economy. The personnel depart­ ment mainly conducted administrative tasks (not management), such as employee data collection and recording (Kazlauskaitė and Buciuniene 2010*), and roles in the rigid work organization were characterized by narrowly defined state job classification (Gurkov and Settles 2013*).5 Within the scope of our literature review, an article by Pieper (1992) was the first to discuss the term “HRM” in ETEs. He discussed socialist HRM, which is completely different in concept from Western HRM. Bangert and Poor (1993) dealt with the effects on HRM of the expansion of multinational enterprises into Hungary. Luthans et al. (1993*) argued that Russian factory managers emphasized traditional management prac­ tices. 1993 is assumed to be the first year in which articles on HRM in the context of transition in ETEs were published. As presented in Figure 7.1, the extracted basic references have been discussed constantly since the beginning of the transition. However, in consideration of the substantial number of extracted basic references after 2000 in comparison with the 1990s, it is clear that articles related to HRM based on the context of the transition have been published steadily and more constantly in the 2000s. With respect to the location of the institutional affiliations of the authors of the extracted basic articles, including co-authors (total number), researchers from institutions in the Central and Eastern European transition countries account for 36.5% of the total, 241

C H A P T E R 7

C H A P T E R

HUMAN RESOURCE MANAGEMENT

7

Figure 7.1 Number of extracted basic references by publication year

and those based in Western countries account for approximately 60%. In the former Soviet Union countries except for Russia, no authors were involved in the extracted basic references. Among the new EU member countries, the researchers affiliated with universities in Slovenia (11 persons), researchers in Poland (9 persons), and those in Lithuania (8 persons) are prominent. The researchers in non-EU-member countries were only from Serbia. With the focus only on the first authors (Table 7.1), researchers from affiliated institutions based in the United States account for the majority. Adding those from affiliated institutions based in the United Kingdom to these based in the United States, they account for approximately half of the total number. Approximately 30% are from affiliated institutions in ETEs. Among the ETEs, Poland, the Czech Republic, Slovakia, Hungary, the Baltic states, and Slovenia joined the EU in 2004. After 2007, Bulgaria, Romania, and Croatia among the Southeast European countries also joined the EU. The rest of the Southeast European countries—Albania, Bosnia–Herzegovina, Serbia, Kosovo, Macedonia, and Montenegro—are non-EU­ member countries. The difTable 7.1 Location of the institutional affiliations of the ferences in their period of first authors EU participation and those Region U.S. U.K. ETEs Others between EU members and non-members may greatly 22 20 37 18 Number of references affect the process of dealing Percentage 22.68% 20.61% 38.14% 18.55% with the problem of the

242

LITERATURE REVIEW

transition in HRM research. As presented in Table 7.2, however, a high proportion of the extracted basic refer­ ences target the Central European countries among the EU-member countries in their research, and there are 12 articles targeting Slovenia, which is included in Southeast Europe in this chapter.6 There are fewer studies targeting the late comers to the EUmembership and the non­ member countries. It can be easily imagined that socialist legacies have been most firmly embedded in the industries that were dominant in the period of the former socialist regime. Under socialist planned economies, in consideration of the fact that basic industries such as machinery production, electronic engineering, defense, and communication equipment were directly supervised by ministries of the central government, the manufac­ turing industry7 and the mining industry8 can be considered as traditional industries of the period under the socialist regime. In addition, the service industry can be also divided into the traditional service industry and the modern one. In the socialist period, there were also retail, wholesale, transportation,

7.2

Table 7.2 Research target regions Subtotal of the number of Number of references by references target region

Region

Country

Former Soviet Union

Russia Belarus Ukraine Moldova

31 2 4 1

38

30.2

Central Europe

Poland Hungary Czech Republic East Germany Slovakia

18 5 13

44

34.9

Rumania Bulgaria Slovenia Serbia BosniaHerzegovina Former Yugoslavia

8 5 12 3 1

31

24.6

5 6 1

12

9.5

1

1

0.1

Southeast Europe

Baltic States

%

1 7

2

Estonia Lithuania Latvia

All the European transition economies Total

126

100

Table 7.3 Research target industries Industrial classification Mining industry Manufacturing industry Traditional tertiary industry Modern tertiary industry Others (Industry: not specified)

Total number of articles

Percentage of the total number of articles

3 34

2.91% 33.01%

21

20.39%

24

23.30%

21

20.39%

243

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

accommodation, and food service industries, which come under traditional tertiary indus­ tries, including public services and education. However, the telecommunications indus­ try, financial and insurance businesses, real estate, and other enterprises formed the service industry that developed only after the beginning of the transition to a market economy. In this chapter, they are considered to form the modern tertiary industry that also includes professional, scientific, and technical services, business management and support services, and health and social work. As articles targeting the manufacturing industry and the traditional tertiary industry account for more than half of the total number of articles (see Table 7.3), it can be assumed that there is a tendency to cover the dominant industries during the socialist period as research subjects in articles featuring the transition economies, particularly their socialist legacies.

7.3 SOCIALIST LEGACIES IN HRM RESEARCH IN ETES 7.3.1 Focus on the issues of the transition Our interest is whether the socialist legacies have thus far been an important factor in distinguishing the transition economies in HRM research on ETEs. Therefore the transition matters. It is reasonable that those who are interested in the transition explore the HRM practices that had been functioning in the period of socialism and the culture that embedded the behavior of managers and employees. Therefore those who focus on the transition tend to explore how they have been changed to form more appropriate HRM in the market economy and how they were replaced by European and American HRM as well as their appropriate values and behaviors adaptable to the market economy. From these perspectives, the extracted basic references were divided into three categories. Articles in the first category address a group of problems widely and directly related to the transition of HRM policies and practices such as industrial relations, employee participation, and education and training of employees and managers. This category is named “issues on institutional legacies” in this chapter. Mainly, the institutional legacies of socialist personnel management are the research subject here. The second category is called “issues on cultural legacies,” which includes research targeting national and corporate culture stemming from socialism, and the values and attitudes of employees and managers based on a culture originating from their socialist experiences. The third category compiled those references that cannot be assigned to the above two categories and is named “other issues.” As a whole, the literature is proportionally divided into the three major categories (Table 7.4). Interestingly, the number of articles has increased equally in all the categor­ ies since 2004, when the major Central European countries joined the EU. This implies that path-dependent features of their HRM still remain even in the EU countries. For instance, Skuza et al. (2013*) discussed Poland and the former socialist cultural legacies, which still continue to dominate in Polish-owned companies, where the HR 244

SOCIALIST LEGACIES IN HRM RESEARCH IN ETES

7.3

Table 7.4 Distribution of the determinants of the transition Group of problems on Group of Group of human resource problems on other management policies cultural legacies problems

Transition mode Major European transition countries before participating in the EU (before 2003)

No. of items of literature

10

11

10

Percentage

29.40%

40.70%

45.50%

No. of items Major European transition countries after participating in of literature the EU (after 2004) Percentage

33

21

12

76.74%

65.63%

54.50%

Total number of references

43

32

22

function is particularly underdeveloped compared with other organizational functions. Zientara and Kuczynski (2009*) also examined the Polish local public administration and described the aspects in which HRM practices in the period of socialism have remained just as they were. They concluded that modern HRM practices have remained underdeveloped, and socialist bureaucracy is still evident in the modern Polish administration. 7.3.2 Institutional legacies of HRM The issues of institutional legacies are divided into those regarding the transition of HRM practices, those considering the changes in and strengthening of employment rela­ tions and employee involvement during the transition, and those considering institutional capacity and the learning of various practices as well as adaptation to the new manage­ ment. In the processes of management modernization, the introduction of Western HRM practices, which did not exist in socialist personnel management, and the tran­ sition from socialist personnel management to modern HRM are the common central targets of this issue. The survey shows the arguments can be divided into two types. The first type argues for the persistence of traditional HRM practices of the socialist period, and the second type examines the extent of the introduction of Western HRM practices and their development on the assumption that HRM practices and HRM professional managers were lacking or absent in ETEs. The former emphasizes the path-dependent feature of the current HRM in ETEs. The later emphasizes the absence of modern HRM practices and skills. The socialist legacies represented by issues of institutional legacies are the main topic we aim to focus on in this chapter. There were 23 articles that discussed the transition of HRM policies, among which eight examined Russia and two examined the other former Soviet Union countries. There were seven articles examining Central European countries such as 245

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

Hungary, Poland, and the Czech Republic, and six examined the Southeast European countries. Among the six articles examining the Southeast European countries, three listed Zupan as one of the authors, examining Slovenia. These articles often indicate that the personnel management ETEs implemented during the period of socialism is still firmly entrenched. Furthermore, the characteristics of socialist personnel management are common to almost all ETEs. Lupina-Wegener (2013*) and Kazlauskaitė and Buciuniene (2010*) examined Poland and Lithuania, and they described the characteristics of personnel management of both countries in the socialist period based on the common understanding of socialist HRM characterized by Pieper (1992), whose understanding is based on the experience in East Germany. Lucas et al. (2004*), Kazlauskaitė and Buciuniene (2010*), and Gurkov and Settles (2013*) also described the general characteristics of socialist personnel management. They described socialist personnel management as not in fact “management” but “administra­ tion,” and personnel management as primarily derived from the centralized planned economy. Individual socialist enterprises had no initiative to manage personnel as a resource. The personnel planning, job design, and remuneration schemes were imple­ mented by central governments, and there was very little room to manage at the corpor­ ate level. Central government set the number of employees and the number of school and university graduates to be absorbed. Therefore, neither recruitment nor selection of personnel arose as a problem to be tackled at the corporate level. The department in charge of socialist personnel management at the corporate level, called the “cadre department,” was an instrument of supervision by the Communist Party. It dealt with such practices as recording hiring and dismissal, employee data collection and reporting to the government, and allocating employee training. There were no free labor markets, and, under the principle of full employment, the evaluation and motivation of employees were not an important factors for corporate personnel management. The remnants of the socialist personnel management in transition economies are major interest of the articles identified in this category. Among the extracted basic references, Milikic et al. (2008*) also point out that there is a tendency to implement only the personnel management practices of the socialist period in the HRM of modern Serbian enterprises and suggest that the HRM characteristics remain unchanged from the old regime. Cyr and Schneider (1996*), Gurkov and Settle (2013*), Fey et al. (1999*), Kazlauskaitė and Buciuniene (2010*), Lucas et al. (2004*), Zupan and Kase (2005*), and Zupan and Ograjensek (2004*) also discussed the remaining unchanged socialist-typed HRM in ETEs. Clarke (2004*) examined unchanged Soviet-type person­ nel management, shop floor management, payment systems, and discipline, that have been inherited and maintained in modern Russian enterprises, particularly focusing on shop-floor management in the manufacturing industry.9 However, research with a focus on the lack of Western HRM practices, rather than the remaining socialist personnel management in the transition economies, is also an important area in HRM research in transition economies. For example, Bjorkman et al. (2007*) took up the historical absence of capitalist-style business as an issue, instead of focusing on the remaining legacies, and built a hypothesis that the absence of HRM practices such as training, performance appraisal systems, 246

SOCIALIST LEGACIES IN HRM RESEARCH IN ETES

7.3

performance-based compensation systems, and performance-based promotion sys­ tems require foreign-capital enterprises in Russia to emphasize these practices. With respect to the articles that examine the extent to which Western HRM practices were introduced, functioned, and developed, Fey and Bjorkman have actively contributed. They examined the effects of Western HRM practices on corporate performance in Russia with the use of several factors, such as internal communication, knowledge transfer, employee motivation, training, job security, and so on, and have analyzed the adaptability of Western HRM (Fey and Denison 2003*; Fey et al. 2000, 2009; Bjorkman and Ehrnrooth 2000; Bjorkman et al. 2007*). These studies also started their analyses on the basis of the historical absence of HRM policies in the former socialist countries. Those who focus on institutional legacies are assumed to employ a contextual approach to HRM. The contextual approach aims to understand what is contextually unique and to explain why (Brewster 2006, 2007). The institutional legacies are context­ ually unique in ETEs so that those who pick up socialist institutional legacies in ETEs are assumed to have high concerns on the contextual approach Brewster advocates. His articles are actively cited in the studies that focus on institutional legacies. Articles that include Brewster in the references of all the basic extracted references account for 23.7% (23 articles) of the extracted basic references. In this sense, we expected researchers who focus on institutional legacies in the universalist paradigm to be few, because their study is intended to be “about generating understanding in order to improve the way that human resources are managed within the organization, with the ultimate aim of improving organizational performance” (Brewster 2007, p. 241). Contrary to the contextual approach, those who focus on institutional legacies are not assumed to employ an American-style HRM approach: that is, a universalist approach. As we expected, there are very few extracted basic references on analyses related to the relationship between HRM practices and corporate performance. Among the extracted basic references, Judge et al. (2009*), Zupan and Kase (2005*), and Zupan and Ograjen­ sek (2004*) have conducted analyses related to the relationship between HRM practices and corporate performance. Besides the extracted basic references, Fey et al. (2000, 2009), Fey and Bjorkman (2001), Kazlauskaite et al. (2012), and Buciuniene and Kazlauskaitė (2012) focus on the relationship between HRM practices and corporate performance. However, they do not mention the transition factors and thus were excluded from the extracted basic references. 7.3.3 Cultural legacies in HRM The issues of cultural legacies take up cultural aspects such as the national, corpor­ ate, and management cultures as well as the attitudes and values of the managers and employees. There are 11 articles that focus on the attitudes and values of employees and man­ agers, among which six articles target the attitudes and values of employees and five target those of managers. With respect to the attitudes and values of employees as leg­ acies of the socialist system, the characteristics that differentiate the behavior of 247

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

employees in Western countries become the starting point of the transition. Alas and Rees (2006*) conducted a comparative analysis of work-related attitudes and values between capitalist and former socialist countrties. Zientara and Kuczynski (2009*) focused on public administration employees in Poland, which is assumed to be unchanged from communist times. Kazlauskaitė et al. (2009*) tackled the impact of organizational empowerment on work-related attitudes using a comparative analysis between Denmark as an old EU member and Lithuania as a transition country. They showed that no significant differences between the Danish and Lithuanian hotels were found in the levels of organizational empowerment and job satisfaction. In contrast, research on the relationships between the personality traits of employees or psycho­ logical contracts and employee performance like that by Linz and Semykina (2009*) and Kase and Zupan (2007*) also show the lack of attitudes and values of the employ­ ees, which are prerequisites in the West. These studies focus on unique attitudes among employees in the former socialist countries that inhibit performance improvement. While there are studies that discuss new attitudes toward the new market environment and policies, such as the market orientation of the managers and the attitude of managers toward the introduction of Western HRM, we also find studies that analyze the old remaining values of collectivism of the socialist period and the convergence of the trad­ itional values of collectivism with the values of European and American managers in terms of the attitudes of the managers. The former analyze how different the attitude of managers is toward the introduction of Western-style HRM in the period of the transi­ tion. For example, Constantin et al. (2006*) found that modern HRM practices are not a priority for managers in Romanian firms and that they have not changed their trad­ itional leadership behavior or their attitudes toward HRM practices. However, the latter studies focus on the traditional values of collectivism as a starting point for transition and demonstrate how they converge on individualism (Stan and Evans 1999*), beliefs of the new entrepreneurial managers (Puffer et al. 1997*), and American-style manager­ ial values (Alexashin and Blenkinsopp 2005*). Stan and Evans (1999*) argued that collectivist values and individualist values are complementary and discussed trad­ itional values as positive ones that should not be abandoned.10 National and corporate culture are also often discussed in the basic extracted refer­ ences. The cultural aspects of HRM embedded in national and organizational culture under the socialist system have wide variety: for example, a collectivist value system with strong socialist characteristics (Fey and Denison 2003*; Giacobbe-Miller et al. 2003*; Puffer et al. 1997*; Stan and Evans 1999*); the extreme clientelism of the communist system (Letiche 1996*); the old authoritarianism, hierarchical and conser­ vative mentalities (Dalton and Druker 2012*); a belief in a hierarchical, inequitable power system supporting inherited autocratic superiors (Kustin 2006*); communist egalitarianism (Dalton and Druker 2012*; Crow 1998*; Giacobbe-Miller et al. 2003*; Woldu and Budhwar 2011*); a unique time perspective and a unique set of subcultures embedded in socialism (Fey and Denison 2003*); the Soviet mentality and personality to which the term “Homo Sovieticus” can be applied (Szalkowski 1996*; Szalkowski and Jankowicz 1999*); and low reliability derived from the transition period (Pucetaite et al. 2010*). 248

SOCIALIST LEGACIES IN HRM RESEARCH IN ETES

7.3

Perceptions of business culture in different countries are important factors when multinational companies apply their principles of international HRM in their host countries (Brewster and Bennett 2010*). Hofstede (1993) identified four dimensions as elements of a common structure in cultural systems of the surveyed countries, such as power distance, collectivism versus individualism, femininity versus masculinity, and uncertainty avoidance. Hofstede’s surveys are actively cited in the studies of those countries after the beginning of the transition to a market economy (e.g. Alexashin and Blenkinsopp 2005*; Kustin 2006*). Articles that include Hofstede in the refer­ ences account for 21.6% (21 articles) of the extracted basic references. The studies linking national and organizational culture to the socialist experience are based on the perspective that the Central and Eastern European countries are still in tran­ sition. Moreover, they tend to stress how the socialist legacies in the culture have not been swept away in transition. The socialist cultural factors embedded in the Central and Eastern European countries still remain an important point of argument regarding the transition: see, for example, the article by Dalton and Druker (2012*) on Romania and the one by Hirt and Ortlieb (2012*) on Bosnia–Herzegovina. 7.3.4 Other issues The group “other issues” covers the issues that are not included in the previous two types of categories. External environment and institutional changes in economy, man­ agement style, and management turnover are discussed in the articles in this group. There are eight articles related to changes in external environment and institutional changes outside corporate organizations, such as ownership structures and labor markets. The articles by Jarvalt and Randma-Liiv (2010*),11 Karhunen (2008*), and Weinstein and Obloj (2002*) can be identified as articles that feature the rapid changes in the macro-level institutional, social, and economic environments or contexts due to transition. Although privatization is expected to be frequently taken up in the context of transition economies, there were surprisingly very few articles that focus on the differ­ ence in ownership structure as a determinant of the transition. Ivanova (2007*) dis­ cussed the effects on motivations and decision-making of middle managers resulting from the difference in ownership. Russell (2002*) also focused on privatization and ownership, and Wright et al. (2002*) examined how the privatization led by manage­ ment buy-out and employee buy-out encourages restructuring and is likely to be associ­ ated with a greater degree of employee-oriented HRM strategies. Jones et al. (1995*) discussed the underdevelopment of the managerial labor market. Eriksson (2005*), Mur­ avyev (2001*, 2003*), and Ryan (2006*) argued the underdevelopment of the manager­ ial labor market in transition is strongly related to the lack of managerial incentive pay schemes or executive compensation and the lack of professional HR managers. Management style and leadership theory are also discussed in this group of issues. The managers in the period of socialism (red executives) symbolized the management style of the socialist period, and their behavior patterns are intrinsically different from the behavior of managers in capitalist countries. Some attempted to clarify what kind of leadership the managers from the days of the socialist period exerted (Linz 1996*; 249

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

Fidrmuc and Fidrmuc 2006*) and how they exhibited leadership appropriate to the market economy (Barton and Barton 2011*), and some attempted to clarify how the socialist behavior patterns impede such leadership (Solomakhin and Ekaterinoslavsky 1994*). There were seven articles that discussed the management turnover in this group of issues. Three articles were authored by Muravyev, who mainly focused on Russia and Ukraine. With respect to Russia, based on the recognition that ownership struc­ tures greatly affect the frequency of executive turnover, he clarified the trend that both state and private insider ownership of firms restrain CEO turnover and raised the fact that performance plays an important role in the turnover process in Russia (Muravyev 2003*), and found managers of Russian firms faced the threat of dismis­ sal if the firms performed inefficiently (Muravyev 2001*). Regarding Ukraine, Mur­ avyev et al. (2010*) confirmed a strong relationship between executive turnover and performance. This is based on the recognition that the executives of the socialist period lacked the ability and skills required in relation to the market economy and that management turnover is indispensable in making corporate governance efficient. Other than the former Soviet Union, the relationship between management turnover and firm performance is discussed with regard to the Czech Republic, Slovakia, and Slovenia (Claessens and Djankov 1999*; Eriksson 2005*; Knezevic and Pahor 2004*). Claessens and Djankov (1999*) analyzed the Czech Republic in the early stage of transition. Their finding is that profitability and labor productivity are both positively related to appointments of new managers, and this shows that enterprise restructuring requires new human capital, which occurs through managerial turnover in transition economies. Eriksson (2005*) also focused on the underdeveloped man­ agerial labor market and found that changes in performance do not give rise to changes in managerial pay in the Czech Republic. With a focus on the strong insider ownership of Slovenian enterprises, Knezevic and Pahor (2004*) also argued that ownership changes have the greatest impact on management turnover, and the changes in management are strongly associated with a consequence of the ongoing transition process. These studies share common criteria that market transition should be associated with firm restructuring in terms of managerial turnover with a strong relationship to the firm performance. 7.3.5 Hypotheses Our aim in this chapter is to identify how the attributes of the extracted basic refer­ ences relate to each group of issues. The first attribute is the difference between the basis of American-style HRM research and that of European comparative HRM research. In contrast with the universalist paradigm that considers American-style HRM practices to be the best, Brewster (2006, 2007) placed the European compara­ tive HRM research as the contextual paradigm. Socialist legacies in HRM are very contextual, and those who explore socialist legacies as very important determinants to understanding HRM in ETEs are assumed to be very conscious of the contextual paradigm and are expected to cite Brewster’s articles. In terms of cultural legacies, 250

SOCIALIST LEGACIES IN HRM RESEARCH IN ETES

7.3

Hofstede’s cultural dimension theory is popular and his articles can be raised as the most influencial ones, when those who explore socialist legacies are very important determinants to understanding HRM in ETEs. Therefore the first attribute we employ is if the extracted basic references cite Brewster’s articles and Hofstede’s articles. The second attribute is the location of the affiliated institutions of authors. The universalist approach dominates research in the United States (Brewster 2007). We employed the rough assumption that when the locations of the affiliated institu­ tions of researchers are the US and the UK, their research can be identified as based on American-style HRM research. We would like to examine whether the extent of the relative focus on the diversity of HRM is strong in the research on institutional legacies and that on cultural legacies. Therefore, we examined the effect of the dummy variables such as the citations of Brewster or Hofstede and the locations of the institutional affiliations of the first authors. Thirdly we examined the effect of research target regions. The prototype of socialist personnel management practices is Soviet-type personnel management. We can expect the Soviet tradition to have remained more firmly in the former Soviet states. Contrary to the institutional legacies, cultural legacies are more deeply connected with the socialist ideology, which prevailed in the former socialist countries, than with the Soviet tradition. If it is assumed to be a phenomenon found commonly in the countries that have experienced socialism, it can be supposed that the legacies can be observed widely in the European Transition Economy countries regardless of whether they are the research target regions. The perspective of cultural legacies, therefore, is an issue involving all the European Transition Economy countries, not only those limited to the former Soviet states, and hypotheses can be devised such that the extent of the focus on them should not depend on the research target industries. Fourthly, it can be assumed that socialist HRM should become apparent in the traditional industries inherited from the period of socialism. Conversely, enterprises in industries that have emerged after the transition to a market economy do not have the traditional socialist personnel management practices to be observed, and it can be assumed that the effects of socialist personnel management should be limited.12 This can be verified by exam­ ining not only the relationships between the research target regions and the groups of issues but also those between the research target industries and the groups of issues in the extracted basic references. Fifthly we employed EU membership as an attribute to examine. We expected that the focus on socialist legacies in research should decline due to the convergence of management toward the EU. This can be verified by whether the research subjects are EU members at the time of article publication and the relationship with the groups of issues the articles target. The hypotheses are summarized as the following: Hypothesis 1 The research on socialist legacies intends to explain the socialist person­ nel management in transition economies by further focusing on the relative diversity of capitalism and that of personnel and HRM.

251

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

Hypothesis 2 In the HRM research on ETEs, the research on the former Soviet states focuses more strongly on institutional legacies than the other research does. Hypothesis 3 In the HRM research on ETEs, the research on traditional industries inherited from the period of socialism focuses more strongly on institutional legacies. Hypothesis 4 In the HRM research on ETEs, the interest in cultural legacies are shared by all the ETEs, which are not limited to the former Soviet states, and the cul­ tural legacies are focused on without depending on the research target industries. Hypothesis 5 Participation in the EU by the research subject countries diminishes the interest in socialist legacies. We examine these hypotheses. The examination is based on the above-mentioned attributes of the extracted basic references. Table 7.5 shows a summary of attributes of the extracted basic references.

Table 7.5 Descriptive statistics on the variables introduced into the analyses Mean Min. Max. Type of research Institutional legacies Cultural legacies Others Locations of the first authors (Default category: Europe other than the U.K. and the U.S.) The UK or the US Transition countries Target regions (Default category: Baltic States) Former Soviet Union Central Europe Southeast Europe Target industrial fields (Default category: Non-targeted industries [normative, etc.]) Mining, electricity, gas, heat supply, and water supply Manufacturing and construction industries Traditional tertiary industry (wholesale, retail, accommodation and food services, and public services) Modern tertiary industry (IT, finance, real estate, science and technology, education, and insurance and hygiene) Cited references (Default category: not-cited) Brewster Hofstede Non-EU member countries at the time of publication are included in the target.

0.44 0.33 0.23

0 0 0

1 1 1

0.43 0.38

0 0

1 1

0.48 0.31 0.25

0 0 0

1 1 1

0.031 0.31

0 0

1 1

0.11

0

1

0.23

0

1

0.24 0.22 0.64

0 0 0

1 1 1

Note: Descriptive statistics of the variables that were introduced into the analyses. Number of observations = 97

252

HYPOTHESES VERIFICATION

7.4

7.4 HYPOTHESES VERIFICATION The extracted basic references were classified into three thematic categories: institutional legacies, cultural legacies, and other issues in this chapter. Regarding examination of the relationships between each category and the article attributes, these are identified as Model 1 for institutional legacies, Model 2 for cultural legacies, and Model 3 for other issues. Table 7.6 shows the results of examining the relationship between each category and the article attributes. In order to analyze them, logit analysis was adopted, as the explained variables are binary (Panel a). However, in consideration of the fact that all the explanatory variables are dummy variables, OLS estimation was also conducted (Panel b). The variables that obtained significant coefficients as a result of the logit analysis became significant in the OLS estimation as well, and the results came out qualitatively in the same way. It can therefore be considered that the robustness of the results could be generally confirmed. According to the results of the logit analysis, the following interpretation is made. With respect to the group of “other issues” among the three groups of issues, it should be noted in advance that it was not significant for the entire estimated specifi­ cation (Prob > Chi2 = 0.20), although we found the negative consequences that manu­ facturing industry cannot be a explanatory factor for the group of “other issues” and that it is also negative to explain this group with the researches targeting Central Euro­ pean countries. Among the three groups of issues that we focus on, the interesting results were obtained in relation to the groups of issues on institutional and cultural legacies. While the articles citing Brewster have a relatively high frequency in relation to their focus on the legacies of socialist personnel management, for those citing Hof­ stede, the frequency is low. In addition, the articles citing Hofstede to a considerable extent focus on the group of issues on cultural legacies, although the frequency of the focus on the legacies of socialist personnel management was low compared to those not citing Hofstede. What is important here is that the articles citing Brewster to a significantly greater extent focus on socialist personnel management, while no correlations with cultural legacies were found. We had in advance the assumption that the articles whose first authors work in the United States or the United Kingdom pay more attention to the convergence to the global standard (the American values), and the articles whose first authors work in other countries pay more attention to the institutional and cultural divergence of HRM. However, the results show that there is no clear correlation between the authors’ affiliation and the issues discussed. This suggest that it is not so important for researchers in all countries to confirm whether they should address convergence to the American style of HRM or divergence from the American style of HRM. As a result, Hypothesis 1 should be partly rejected. It is obvious that wherever the first author lives, the authors who pay more attention to institutional legacies of socialism tend to focus on the discourse led by Brewster, and those who pay more attention to cultural legacies of socialism tend to focus on the discourse led by Hofsted, whereas those who focus on socialist institutional and cultural legacies tend to pay much

253

C H A P T E R 7

C H A P T E R

HUMAN RESOURCE MANAGEMENT Table 7.6 Estimation results (a) Analysis 1: Logit analysis

7

Articles that explain with institutional legacies Explanatory variables (article

attributes)/Model

[1]

Articles that explain Articles that with cultural explain with other legacies factors

[2]

Locations of the first authors (Default category: Europe other than the UK and the US) The UK or the US −1.11 1.23 Transition countries −0.99 1.72+ Target regions (Default category: Baltic States) Former Soviet Union −0.74 1.04 Central Europe −0.76 2.34* Southeast Europe −0.49 1.32 Target industrial fields (Default category: Non-targeted industries) Mining, electricity, gas, heat −2.65+ 3.19* supply, and water supply Manufacturing and construction 1.9** −0.98 industries Traditional tertiary industry −0.86 1.23 Modern tertiary industry −0.43 0.75 Cited references (Default category: not-cited) Brewster 2.1** −1.27 Hofstede −2.09* 2.79** Non-EU member countries at the −0.257 0.78 time of publication are included in the target. Constant term 1.13 −4.62 Log likelihood No. of observations Prob > Chi2 Pseudo R2

−51.39 97 0.002 0.23

−42.67 97 0.002 0.31

[3]

0.0064 −0.39 −0.63 −1.76+ −1.11 0 −1.34+ −0.43 −0.34 −1.46+ −0.51 −0.39

0.97 −43.79 94 0.02 0.14

(Continued )

attention to the variety or divergence of capitalism or HRM practices. But each researcher’s stance against the global standard is not a “landmark” to divide the research interests as time goes by. Hypothesis 2, which assumes that HRM in the former Soviet states is explained by the institutional legacies, was rejected. Contrary to the result that Hypothesis 2 was not supported, it can be also understood that the legacies of socialist personnel management could serve as a research subject in any country of the ETEs. We specifically verified that the group of issues of institutional legacies, in compari­ son with other groups, is very frequently explained in the manufacturing industry, which is classified as a traditional industry from the period of socialism. However, the

254

HYPOTHESES VERIFICATION

7.4

C H A P T E R

Articles that explain Articles that with cultural explain with other legacies factors

7

Table 7.6 contd. (b) Analysis 2: OLS Articles that explain with institutional legacies Explanatory variables (article attributes)/Model

[1]

[2]

Locations of the first authors (Default category: Europe other than the UK and the US) The UK or the US −0.18 0.19 Transition countries −0.17 0.26** Target regions (Default category: Baltic States) Former Soviet Union −0.094 0.14 Central Europe −0.12 0.34** Southeast Europe −0.073 0.19 Target industrial fields (Default category: Not-targeting industries) Mining, electricity, gas, heat −0.5+ 0.59* supply, and water supply Manufacturing and construction 0.34** −0.16 industries Traditional tertiary industry −0.16 0.2 Modern tertiary industry −0.1 0.13 Cited references (Default category: not-cited) Brewster 0.36** −0.19+ Hofstede −0.44** 0.49** Non-EU member countries at the −0.38 0.1 time of publication are included in the target. Constant term 0.67 −0.22 Log likelihood 0.15 0.26 No. of observations 97 97 0.0001 0.009 Prob > Chi2 2.44 3.81 Pseudo R2

[3]

0.0089 −0.091 −0.045 −0.21+ −0.12 −0.085 −0.19+ −0.04 −0.029 −0.17 −0.059 −0.065 −0.55 0.013 97 0.36 1.11

group of issues on institutional legacies could not be explained in the other traditional non-manufacturing industries. Therefore Hypothesis 3, which assumes that the institu­ tions in traditional industries are frequently explained by institutional legacies, was supported only in the manufacturing industry. Our examination shows that cultural legacies attract attention in research on Central European countries, although they are not an explanatory factor in research on Southeast Europe, where some countries lately joined the EU and some have not yet joined the EU. Cultural legacies of socialism are still widely discussed and are still maintained as one of the key issues for their research. And we could also confirm that the issues on cultural legacies tended to be discussed widely across the industries. Therefore Hypoth­ esis 4 is supported, as it assumed that cultural legacies would be an issue not only for the former Soviet states but also for all the other ETEs and draw attention without depending on the research subject industries.

255

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

When we focus only on cultural legacies, Hypothesis 5, which assumes that the research subject countries’ EU participation diminishes the research on socialist legacies, is supported. When we focus on institutional legacies, EU participation of targeting countries was not significant. Therefore we can suggest that institu­ tional legacies are an important research subject as common characteristics of the HRM of the ETEs, which make the ETEs diverge from the European and Ameri­ can HRM.

7.5 CONCLUSIONS Socialist institutional and cultural legacies are still important factors in the study of HRM in ETEs. The manufacturing industry especially was an important industrial sector where socialist institutional legacies could be found in HRM. However, socialist cultural legacies are widely researched in ETEs. Therefore our examination suggests that socialist legacies are still important not only for the former Soviet Union but also for all the other ETEs. However, in the modern industry that has risen after the beginning of the transi­ tion to a market economy, the interest in socialist legacies is low, and the evaluation of the effects of socialist legacies varies between institutional legacies and cultural ones, even when the same countries are targeted. There is a possibility that discus­ sions on HRM divergence in ETEs based on socialist cultural legacies may disappear in accordance with their deepening integration with the EU economy. Even 25 years after the transition of ETEs, and even after many countries have joined the EU, socialist legacies are still discussed actively and continuously in ETEs. The socialist institutional legacies of HRM in ETEs are especially important factors in understanding the uniqueness of HRM in this region. It is certain that the focus on socialist institutional legacies of HRM in ETEs contributes to understand­ ing the diversity of European HRM.

NOTES 1 In terms of the extracted basic references as mentioned below, this chapter indi­ cates them with * after the publication years. All the extracted basic references are listed in Horie and Kumo 2019. 2 Horwitz (2011*) conducted a survey with a focus on the HRM context of multi­ national enterprises in Central and Eastern European countries. Napier and Vu (1998) reviewed the international HRM of developing countries and transition coun­ tries, and Puffer and McCarthy (2011) and McCarthy and Puffer (2013) presented a broad viewpoint in terms of business and management research in Russia without limiting it to HRM. Furthermore, Michailova et al. (2009) presented a survey related to the context of the transition. 256

NOTES

3 This chapter only targets ETEs and does not include socialist countries such as China and Vietnam. There are reasons they should not be included. First, this chapter is based on the context of European and American HRM research exam­ ining the transplantation of Western HRM in ETEs or the transition of Socialist HRM to Western-style HRM. In discussing HRM in China and Vietnam, not only European- and American-style HRM but also the Japanese one tends to be raised as a topic. There are few studies that focused on the transplantation of the HRM practices of Japanese multinational corporations into ETEs. We avoided the conceptual gap between Western and Eastern HRM. Second, when we tested “China” as a survey term, a significant larger number of articles were addition­ ally extracted, compared with the number of articles targeting only ETEs. We avoided the imbalance. 4 In this chapter, personnel and human resource management means the broad system that includes both traditional personnel management and modern HRM. In particular, when personnel management before the rise of HRM is indicated, it is called traditional personnel and human resource management, and the personnel management in the period of socialism is named socialist personnel management. 5 Other than these articles, the characteristics of the personnel and human resource management of socialist enterprises are discussed distinctively in Pieper 1992, Gurkov 2013, Gurkov and Settles 2013*, Lucas et al. 2004*, and Weinstein and Obloj 2002*. 6 According to the definition by the European Bank for Reconstruction and Devel­ opment (EBRD), Slovenia is classified as a Central European country and is not included in Southeastern Europe. In this chapter, however, Slovenia is included in Southeastern Europe in order to place the former Yugoslavian countries, which had a tradition of self-management in the former socialist period, with the Southeastern European countries. 7 In this chapter, the construction industry is included in the manufacturing for convenience. 8 Electricity, gas, water, and heat supply are included in this category for convenience. 9 It is widely observed in Russia. Trappman also refers to the fact that HRM in a traditional Russian steel company is “all as it was” (2007, p. 138). 10 Among the extracted basic references, there are articles that include the conver­ gence and/or divergence debate (e.g. Alexashin and Blenkinsopp 2005*; Horwitz 2011*; Poor and Milovecz 2011; Svetlik et al. 2007*). However, the arguments are diverse as to what they converge and by what they are diverged. 11 Jarvalt and Randma-Liiv (2010*) analyzed decentralization and the absence of a sector-wide strategic HRM in the public sector in Estonia, but this absence is not explicitly explained in terms of the socialist experience. Therefore, we put this article into the category of other issues. 12 It can be assumed that the fact that research subject enterprises are foreign­ capital-including multinational enterprises may also have some effect, but no sig­ nificant results were obtained. 257

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT

REFERENCES

Alas, Ruth, and Christopher Rees (2006) Work-related attitudes, values and radical change in Post-Socialist contexts: A comparative study. Journal of Business Ethics, 68(2), pp. 181–189. Alexashin, Yuri, and John Blenkinsopp (2005) Changes in Russian managerial values: A test of the convergence hypothesis? International Journal of Human Resource Management, 16(3), pp. 427–444. Bangert, David, and Jozsef Poor (1993) Foreign involvement in the Hungarian economy: Its impact on human resource management. International Journal of Human Resource Management, 4(4), pp. 817–840. Barton, Harry, and Lisa Barton (2011) Trust and psychological empowerment in the Russian work context. Human Resource Management Review, 21(3), pp. 201–208. Bjorkman, Ingmar, and Mats Ehrnrooth (2000) HRM in western subsidiaries in Russia and Poland. Journal of East-West Business, 5(3), pp. 63–79. Bjorkman, Ingmar, Carl Fey, and Hyeon Jeong Park (2007) Institutional theory and MNC subsidiary HRM practices: Evidence from a three-country study. Journal of International Business Studies, 38(3), pp. 430–446. Brewster, Chris (2006) Comparing HRM policies and practices across geographical bor­ ders. In G. Stahl and I. Bjorkman (eds.), Handbook of Research in International Human Resource Management. Edward Elgar Publishing: Cheltenham, pp. 68–90. Brewster, Chris (2007) A European perspective on HRM. European Journal of International Management, 1(3), pp. 239–259. Brewster, Chris, and CliveViegas Bennett (2010) Perceptions of business cultures in eastern Europe and their implications for international HRM. International Journal of Human Resource Management, 21(14), pp. 2568–2588. Brewster, Chris, Michael Morley, and Ilona Buciuniene (2010) The reality of human resource management in Central and Eastern Europe: A special issue to mark the 20th Anniver­ sary of Cranet (the Cranfield network on comparative human resource management). Baltic Journal of Management, 5(2), pp. 145–155. Buciuniene, Ilona, and Rūta Kazlauskaitė (2012) The linkage between HRM, CSR and per­ formance outcomes. Baltic Journal of Management, 7(1), pp. 5–24. Claessens, Stijn, and Simeon Djankov (1999) Enterprise performance and management turnover in the Czech Republic. European Economic Review, 43(4-6), pp. 1115–1124. Clarke, Simon (2004) A very Soviet form of capitalism? The management of holding com­ panies in Russia. Post-Communist Economies, 16(4), pp. 405–422. Constantin, Ticu, Daniela Pop, and Ana Stoica-Constantin (2006) Romanian managers and human resource management. Journal of Organizational Change Management, 19(6), pp. 760–765. Cooke, Fang, Geoffrey Wood, Alexandros Psychogios, and Leslie Szamosi (2011) HRM in emergent market economies: Evidence and implications from Europe. Human Resource Management Journal, 21(4), pp. 368–378. Crow, Mary (1998) Personnel in transition: The case of Polish women personnel managers. Personnel Review, 27(3), pp. 243–261. Cyr, Dianne, and Susan Schneider (1996) Implications for learning: Human resource man­ agement in east–west joint ventures. Organization Studies, 17(2), pp. 207–226.

258

REFERENCES Dalton, Kevin and Janet Druker (2012) Transferring HR concepts and practices within multi­ national corporations in Romania: The management experience. European Management Journal, 30(6), pp. 588–602. Eriksson, Tor (2005) Managerial pay and executive turnover in the Czech and Slovak Republics. Economics of Transition, 13(4), pp. 659–677. Fey, Carl, and Ingmar Bjorkman (2001) The effect of human resource management practices on MNC subsidiary performance in Russia. Journal of International Business Studies, 32(1), pp. 59–76. Fey, Carl, and Daniel Denison (2003) Organizational culture and effectiveness: Can Ameri­ can theory be applied in Russia? Organization Science, 14(6), pp. 686–706. Fey, Carl, Ingmar Bjorkman, and Antonina Pavlovskata (2000) The effect of human resource management practices on firm performance in Russia. International Journal of Human Resource Management, 11(1),pp.1–18. Fey, Carl, Pontus Engstrom, and Ingmar Bjorkman (1999) Doing business in Russia: Effect­ ive human resource management practices for foreign firms in Russia. Organizational Dynamics, 28(2), pp. 69–80. Fey, Carl, Sergey Morgulis-Yakushev, Hyeon Joeng Park, and Ingmar Bjorkman (2009) Opening the black box of the relationship between HRM practices and firm performance: A comparison of MNE subsidiaries in the USA, Finland, and Russia. Journal of International Business Studies, 40(4), pp. 690–712. Fidrmuc, Jana, and Jan Fidrmuc (2006) Can you teach old dogs new tricks? On comple­ mentarity of human capital and incentives. Journal of International Money and Finance, 25(3), pp. 445–458. Garavan, Thomas, Michael Morley, Noreen Heraty, Jacinta Lucewicz, and Adam Suchodolski (1998) Managing human resources in a post-command economy: Personnel administra­ tion or strategic HRM. Personnel Review, 27(3), pp. 200–212. Giacobbe-Miller, Jane, Douglas Miller, Wei-Bin Zhang, and Vladimir Victorov (2003) Country and organizational-level adaptation to foreign workplace ideologies: A comparative study of distributive justice values in China, Russia and the United States. Journal of International Business Studies, 34(4), pp. 389–406. Gurkov, Igor (2013) HRM in Russian industrial companies: An overview of past studies and new observations. Journal of Comparative Economic Studies, 8, pp. 21–33. Gurkov, Igor, and Alexander Settles (2013) A dominant archetype of the HRM system in Russian industrial companies in post-recession times. International Journal of Human Resource Management, 24(19), pp. 3630–3643. Hirt, Christian, and Renate Ortlieb (2012) Cultural standards of Bosnia and Herzegovina: Empirical findings and implications for strategic human resource management. Journal for East European Management Studies, 17(2), pp. 205–225. Hofstede, Geert (1993) Cultural constraints in management theories. Academy of Manage­ ment Exective, 7(1), pp. 81–94. Horie, Norio, and Kazuhiro Kumo (2019) Socialist legacies and human resource management in European transition economies: An analytical survey. CEI Working Paper No. 2019-7, Center for Economic Institutions, Institute of Economic Research of Hitotsubashi Univer­ sity: Kunitachi, Tokyo. Horwitz, Frank (2011) Future HRM challenges for multinational firms in Eastern and Central Europe. Human Resource Management Journal, 21(4), pp. 432–443.

259

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT Ivanova, Yuliya (2007) The motivation of managerial choices in the state-controlled econ­ omy of Belarus. Journal of East-West Business, 13(4), pp. 5–36. Jarvalt, Jane, and Tiina Randma-Liiv (2010) Public sector HRM: The case of no central human resource strategy. Baltic Journal of Management, 5(2), pp. 242–256. Jones, Derek, Takao Kato, and Svetlana Avramov (1995) Managerial labour markets in tran­ sitional economies: Evidence from Bulgaria. International Journal of Manpower, 16(10), pp. 14–24. Judge, William, Irina Naoumova, and Thomas Douglas (2009) Organizational capacity for change and firm performance in a transition economy. International Journal of Human Resource Management, 20(8), pp. 1737–1752. Karhunen, Päivi (2008) Managing international business operations in a changing institu­ tional context: The case of the St. Petersburg hotel industry. Journal of International Management, 14(1), pp. 28–45. Kase, Robert, and Nada Zupan (2007) Psychological contracts and employee outcomes in transition to market economy: A comparison of two Slovenian companies. Problems and Perspectives in Management, 5(4), pp. 16–27. Kazlauskaite, Ruta, Ilona Buciuniene, and Linas Turauskas (2012) Organisational and psy­ chological empowerment in the HRM–performance linkage. Employee Relations, 34(2), pp. 138–158. Kazlauskaitė, Rūta, and Ilona Buciuniene (2010) HR function developments in Lithuania. Baltic Journal of Management, 5(2), pp. 218–241. Kazlauskaitė, Rūta, Ilona Buciuniene, Linas Turauskas, and Laura Salciuviene (2009) A comparative study of employee empowerment in Danish and Lithuanian hotels. Trans­ formations in Business and Economics, 8(17), pp. 66–85. Knezevic, Ljubica, and Marko Pahor (2004) The influence of management turnover on enter­ prise performance and corporate governance: The case of Slovenia. Economic and Busi­ ness Review for Central and South-Eastern Europe, 6(4), pp. 297–324. Kustin, Richard (2006) A study of Hofstede’s culture value survey in a post-Soviet country: Perspectives from Belarus. Journal of Transnational Management, 11(4), pp. 3–17. Letiche, Hugo (1998) Transition and human resources in Slovakia. Personnel Review, 27(3), pp. 213–226. Linz, Susan (1996) Red executives in Russia’s transition economy. Post-Soviet Geography and Economics, 37(10), pp. 633–651. Linz, Susan, and Anastasia Semykina (2009) Personality traits as performance enhancers? A comparative analysis of workers in Russia, Armenia and Kazakhstan. Journal of Eco­ nomic Psychology, 30(1), pp. 71–91. Lucas, Rosemary, Milena Marinova, Jana Kucerova, and Milota Vetrokova (2004) HRM practice in emerging economies: A long way to go in the Slovak hotel industry? Inter­ national Journal of Human Resource Management, 15(7), pp. 1262–1279. Lupina-Wegener, Anna (2013) Human resource integration in subsidiary mergers and acquisitions Evidence from Poland. Journal of Organizational Change Management, 26(2), pp. 286–304. Luthans, Fred, Dianne Welsh, and Stuart Rosenkrantz (1993) What do Russian managers really do? An observational study with comparisons to U.S. managers. Journal of inter­ national Business Studies, 24(4), pp. 741–761.

260

REFERENCES McCarthy, Daniel, and Sheila Puffer (2013) Business and management in Russia: A review of the post-soviet literature and future research directions. European Journal of Inter­ national Management, 7(1), pp. 74–111. Michailova, Snejina, Noreen Heraty, and Michael Morley (2009) Studying human resource management in the international context: The case of Central and Eastern Europe. In Michel Morley, Noreen Heraty, and Snejina Michailova (eds.), Managing Human Resource Management in Central and Eastern Europe. Routledge: London, pp. 1–24. Milikic, Biljana, Nebojša Janicijevic, and Mirjana Petkovic (2008) HRM in transition econ­ omies: The case of Serbia. South East European Journal of Economics and Business, 3(2), pp. 75–88. Muravyev, Alexander (2001) Turnover of top executives in Russian companies. Russian Economic Trends, 10(1), pp. 20–24. Muravyev, Alexander (2003) Turnover of senior managers in Russian privatised firms. Com­ parative Economic Studies, 45(2), pp. 148–172. Muravyev, Alexander, Oleksandr Talavera, Olga Bilyk, and Bogdana Grechaniuk (2010) Is corporate governance effective in Ukraine? A crude test using chief executive officer turnover data. Eastern European Economics, 48(2), pp. 5–24. Napier, Nancy, and Van Tuan Vu (2001) International human resource management in devel­ oping and transitional economy countries: A breed apart? Human Resource Manage­ ment Review, 8(1), pp. 39–77. Nielsen, Klaus, Bob Jessop, and Jerzy Hausner (1995) Institutional Change in Post-Socialism. In Jerzy Hausner, Bob Jessop, and Klaus Nielsen (eds.), Strategic Choice and PathDependency in Post-Socialism: Institutional Dynamics in the Transformation Process. Edward Elgar: Aldershot, pp. 3–44. Pieper, Rüdiger (1992) Socialist HRM: An analysis of HRM theory and practice in the former socialist countries in Eastern Europe. International Executive, 36(6), pp. 499–516. Poor, Jozsef, and Agnes Milovecz (2011) Management consulting in human resource man­ agement: Central and Eastern European perspectives in light of empirical experiences. Journal of Service Science and Management, 4(3), pp. 300–314. Pucetaite, Raminta, Anna-Maija Lamsa, and Aurelija Novelskaite (2010) Organizations which have the strongest potential for high-level oganizational trust in a low-trust societal context. Transformations in Business and Economics, 9(2B), pp. 318–334. Puffer, Sheila, and Daniel McCarthy (2011) Two decades of Russian business and manage­ ment research: An institutional theory perspective. Academy of Management Perspec­ tives, 25(2), pp. 21–36. Puffer, Sheila, Daniel McCarthy, and Alexander Naumov (1997) Russian managers’ beliefs about work: Beyond the stereotypes. Journal of World Business, 32(3), pp. 258–276. Russell, Raymond (2002) The influence of ownership and organizational conditions on employee participation in Russian enterprises. Economic and Industrial Democracy, 23 (4), pp. 555–584. Ryan, Leo (2006) Current ethical issues in Polish HRM. Journal of Business Ethics, 66(2–3), pp. 273–290. Skuza, Agnieszka, Hugh Scullion, and Anthony McDonnell (2013) An analysis of the talent management challenges in a post-communist country: The case of Poland. International Journal of Human Resource Management, 24(3), pp. 453–470.

261

C H A P T E R 7

C H A P T E R 7

HUMAN RESOURCE MANAGEMENT Solomakhin, Dmitry, and Yury Ekaterinoslavsky (1994) Managers for the Russian market-economy: Their selection, training and employment from the perspective of busi­ ness culture. International Journal of Technology Management, 9(8), pp. 851–855. Stan, Simona, and Kenneth Evans (1999) Small business retailing in privatizing economies: The influence of managers’ individualistic and collectivist values. Journal of East-West Business, 5(1–2), pp. 123–143. Svetlik, Ivan, Tonu Kaarelson, Ruth Alas, and Andrej Kohont (2007) The development of the personnel function in transition countries: Slovenian and Estonian experience. TramesJournal of the Humanities and Social Sciences, 11(1), pp. 35–53. Szalkowski, Adam (1996) Human resource management in the process of transformation of the Polish economy, International Journal of Social Economics, 23(9), pp. 51–60. Szalkowski, Adam, and Devi Jankowicz (1999) The ethical problems of personnel man­ agement in a transition economy. International Journal of Social Economics, 26(12), pp. 1418–1427. Trappman, Vera (2007) Human resource management at a steel giant in Russia. In Michel Domsch, and Tatjana Lidokhover (eds.), Human Resource Management in Russia. Ashgate: Aldershot, pp. 133–149. Weinstein, Marc, and Krzysztof Obloj (2002) Strategic and environmental determinants of HRM innovations in post-socialist Poland. International Journal of Human Resource Management, 13(4), pp. 642–659. Woldu, Habte, and Pawan Budhwar (2011) Cultural value orientations of the former com­ munist countries: A gender-based analysis. International Journal of Human Resource Management, 22(7), pp. 1365–1386. Wright, Mike, Trevor Buck, and Igor Filatotchev (2002) Post-privatization effects of man­ agement and employee buy-outs. Annals of Public and Cooperative Economics, 73(3), pp. 303–352. Zientara, Piotr, and Grzegorz Kuczynski (2009) Human resources practices and work-related attitudes in Polish public administration. Eastern European Economics, 47 (5), pp. 42–60. Zupan, Nada, and Robert Kase (2005) Strategic human resource management in European transition economies: Building a conceptual model on the case of Slovenia. International Journal of Human Resource Management, 16(6), pp. 882–906. Zupan, Nada, and Irena Ograjensek (2004) The Link between human resource management and company performance. Journal of East-West Business, 10(1), pp. 105–119.

262

8

The collapse of

the COMECON

system and trade in transition countries Akira Uegaki and Kazuhiro Kumo

8.1 INTRODUCTION During the socialist era, international trade involving the Soviet Union, some Central and Eastern European countries, as well as Mongolia, Cuba, and Vietnam, was under­ pinned by two systems: state monopolization of trade and the Council for Mutual Eco­ nomic Assistance (COMECON).1 Under the former, decisions on the goods to be traded, the quantities to be traded, and countries to trade with were made by state bodies. Actual trade procedures were also under the strict control of state bodies. Under the latter, goods, quantities, and prices for trade among member countries were prescribed in trade agreements determined in advance by COMECON coun­ tries. Over the medium to long term, in particular, discussions also covered a conscious international division of labor among the countries. Furthermore, the international settlement of payments under the former initially involved bilateral settlement, but this was later replaced by transfers between accounts held with the International Bank for Economic Cooperation (COMECON Bank)2 (Uegaki 2011). In this chapter, we will refer to this system of international trade among socialist countries as the “COMECON system,” and employ the techniques of meta-analysis to explore the changes that the collapse of this system brought to the trade of former Soviet, and Central and Eastern European countries.3 As mentioned above, the COMECON system embodied the intent to confine trade within the framework of economic planning. It also constituted an attempt to tran­ scendentally determine a basic structure for socialist countries and their international organizations that was based on their values. As a result, its collapse freed the coun­ tries’ economies from what had been an extremely rigid system. This would allow the countries to demonstrate their “genuine economic potential” and was expected to lead to significant changes in the volume and structure of trade. Furthermore, under the COMECON system, domestic economies and the international economy were, in regulatory terms, cut off from each other due to the state monopolization of trade

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

and setting of exchange rates for each product, in particular. This meant that the col­ lapse of the system led to the domestic economies of member countries becoming directly affected by the global economy. In response to this development, researchers in each country began to focus on empirically proving a relationship between trade in post-systemic-transformation former socialist states and structural changes in domestic economies. In this chapter, we will employ the techniques of meta-analysis to statistically synthesize/combine the results of analysis performed in the numerous such studies (books, papers, etc.) that have attempted to shed light, using economet­ ric methods, on the relationship between trade and national economic phases in the former Soviet, and Central and Eastern European countries. The remainder is organized as follows: in Section 8.2 we will provide an overview of the collapse of the COMECON system and the transformation of the trade struc­ ture. In Section 8.3 we will state the method we employed to select the studies sub­ ject to our survey and also classify the studies. In Section 8.4, we will describe the specific procedures we used when putting the studies on the chopping board for the meta-analysis. In Section 8.5 we will describe the claims made in the studies that used the “gravity model” that we actually performed the meta-analysis on in this chapter. In Section 8.6, we will present the results of the meta-analysis. Finally, in Section 8.7, we will state the conclusions of this chapter.

8.2 THE COLLAPSE OF THE COMECON SYSTEM AND THE TRANSFORMATION OF THE TRADE STRUCTURE The collapse of the COMECON structure actually brought significant changes to the trade patterns of former Soviet (FSU), and Central and Eastern European (CEE) countries. As Table 8.1 shows, CEE exports and imports, which had slumped in the 1980s, increased rapidly following the systemic transformation.4 In the former Soviet Union, however, while exports have climbed since the systemic transformation,5 imports have slumped. This reflects the situation in Russia, the largest of the former Soviet states, where the economic shock caused by systemic transformation and the breakup of the union was severe, resulting in a slump in demand for the entire state economy, yet where exportTable 8.1 Increase in the trade volume (Billion US$) competitive products (oil 1980 1985 1990 1995 Region and natural gas) were abundant. CEE The composition of coun­ Export 56.367 55.020 61.733 79.893 Import 65.443 53.875 63.408 100.187 tries traded with also changed FSU dramatically. As Table 8.2 Export 57.942 57.317 59.056 83.274 shows, in CEE, the weight of Import 52.218 54.763 64.963 49.603 trade with other CEE countries Source: UNECE, Economic Survey of Europe in 1995–1996 (1996, pp. and with the Commonwealth 188–189)

264

THE COLLAPSE OF THE COMECON SYSTEM

8.2

Table 8.2 Regional structure of trade partners of CEE/FSU, changes by year, in per cent 1980

1990

1995

2000

2001

2002

C H A P T E R 8

CEE Export

Import

FSU/Russia Export

Import

with CEE, CIS with Developed Economies with Less-Developed Economies with CEE, CIS with Developed Economies with Less-Developed Economies

48.5 35.7 15.8 42.0 38.7 19.3

38.1 49.5 12.4 26.6 53.3 20.1

28.6 62.6 8.8 25.8 65.6 8.5

20.7 73.0 6.3 23.6 66.6 9.8

21.0 72.9 6.1 23.0 66.1 10.9

20.8 72.6 6.6 24.7 66.2 9.1

with CEE, CIS with Developed Economies with Less-Developed Economies with CEE, CIS with Developed Economies with Less-Developed Economies

34.5 42.2 23.3 31.5 46.4 22.1

21.8 49.5 28.7 24.7 52.9 22.4

16.8 60.6 22.6 15.5 69.5 15.0

20.0 55.6 24.4 10.9 69.3 19.8

19.4 55.2 25.4 10.1 67.6 22.3

17.2 55.6 27.2 10.1 65.4 24.5

Source: UNECE, Economic Survey of Europe (2003, No.1, p. 236)

of Independent States (CIS), as well as trade with less-developed countries, has declined quickly, while the weight of trade with advanced countries has increased rapidly. In the former Soviet Union and Russia, as has been the case in Central and Eastern Europe, the weight of trade with CEE countries and with other CIS countries has declined while that with advanced countries has increased. Unlike in CEE, how­ ever, the weight of trade with less-developed countries has not fallen. Furthermore, the increase in the weight of trade with advanced countries has not been rapid, either in terms of exports or imports. The structure of goods traded in the former Soviet Union and Eastern Europe also changed with the collapse of the COMECON system. Although we will not present detailed data here, Russia, which exported a considerable amount of machinery during the Soviet era, became a mono-cultural state in the 2000s as oil and natural gas came to account for more than two-thirds of exported goods. Eastern European countries, meanwhile, as they went through the process of joining the EU, saw each country acquire a role in the intra-EU division of labor, so the changes in the goods traded are important (Uegaki 2011). However, when examined closely, the nature of the aforementioned changes in trade structure (volume, trading partners, and goods) exhibits various patterns, and it would be dangerous to try to identify a general trend from the above information alone. In fact, most of the studies performed until now have focused on trying to identify the causes of this diversity in the nature of the changes in trade structure, and have also explored the impact of this diversity on other economic indicators (e.g. GDP) in the countries. In this chapter, we will also pay attention to this diver­ sity as we carefully examine studies that have been published up to the end of 2014.

265

C H A P T E R

THE COLLAPSE OF THE COMECON SYSTEM

8

We began by using the keywords presented in Table 8.3 to search EconLit for Eng­ lish-language studies of the relationship between trade in transition countries and the economies of the countries, with the goal of obtaining as wide a range of stud­ ies as possible.6 This machine search produced a total of 579 studies as hits. The two authors examined the titles of the studies, and excluded those that were clearly at odds with our objectives (ones that did not contain economics-based analysis, ones that did not involve statistical analysis, ones that did not refer to trade, ones that did not study the economies in transition, etc.). This left 120 studies to be used as the subject of investigation. We read through all 120 of these studies, and selected studies that matched the purpose of our inquiry, namely ones that explored structural changes that had occurred in the relationship between trade in Russia and Eastern Europe and their domestic economies, and also employed quantitative analysis. This left 77 studies.7

8.3 OVERVIEW OF THE STUDIES SURVEYED

8.3.1 Selection of studies

Table 8.3 Steps of literature selection Survey Round Period

Keywords

First survey

“COMECON” January 1989– “COMECON” and “transition economies” December 2010 “collapse of COMECON” “trade” and “transition economies” “international economic relationship” and “tra­ sition economies” “economic integration” and “transition economies” “CMEA” and “transition economies” “division of labor” and “transition economies” “transferable Ruble” “transferable Ruble” and “transition economies” “COMECON bank” “COMECON bank” and “transition economies” “socialist integration” and “transition economies”

August 2011

Second survey November 2015 The same as the first survey

Third survey

266

June 2016

Target Period

January 1989– December 2014

“transition economies” and “GDP” and “trade” January 1989– “transition economies” and “growth” and December 2014 “trade” “socialist economies” and “GDP” and “trade” “socialist economies” and “growth” and “trade” “East(ern) europe” and “GDP” and “trade” “East(ern) europe” and “growth” and “trade”

OVERVIEW OF THE STUDIES SURVEYED

8.3

8.3.2 Trends in the studies subject to analysis The studies ultimately selected are presented in Table 8.4. We classified the studies selected into two main types. These were analysis that included “trade-related indicators,”8 such as the value of exports in the independent variables (Group A) and analysis where the dependent variable was a “trade-related indicator” (Group B). Because the sample size is small, it may be not easy to identify a clear trend from it, but in the 1990s there were few studies in Group A. Almost all the papers were in Group B. As the years passed, however, the number of studies in the former group increased, and from 2006 such studies actually seem to have become predominant. Nevertheless, it is not the case that analysis in the vein of Group B has disappeared. It is difficult to predict future research trends, so here simply confirming that the 77 papers presented in Table 8.4 can be classified into two main groups should suffice. Group A includes a wide variety of types of analysis. From our point of view, those studies where the dependent variable is GDP, economic growth, or productivity can be said to be interesting. This is because if the COMECON system intentionally cut off the relationship between domestic economies and the international economy, the collapse of that system meant the creation of a new relationship between the two. For example, the potential for exports to drive economic growth, as they had in Hong Kong, Singapore, South Korea, Taiwan, and so on, may also now be open to Russia and Eastern Europe. As can be seen from Table 8.4, 20 studies, which com­ prise over half of Group A, constitute analysis that uses GDP (total or per capita), Table 8.4 Classification of literatures using econometric analysis (A) Trade-related indicators in independent variables a

(B) Trade-related indicators in dependent variables

Among them GDP, Growth rates or productivity is a dependent variable: 1992–2000 (average by year) 2001–2005 (average by year) 2006–2010 (average by year) 2011–2014 (average by year) 1992–2014 b (average by year)

2 0.22 5 0.56 20 2.22 11 1.22 38 1.65

1 0.11 2 0.22 10 1.11 7 0.78 20 0.87

Among them those utilized gravity models: 10 1.11 8 1.60 14 2.80 10 2.50 42 1.83

6 0.67 5 1.00 5 1.00 3 0.75 19 0.83

Notes a Trade-related indicators are: export volume, import volume, total of import and export volume, the total of current account balance, per-capita total of these values, or ratio of them against GDP. b The total number of literatures surveyed was 77. Three papers conducted analysls of both (A) and (B) types (38+42−3).

267

C H A P T E R 8

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

economic growth rate, productivity, and so on as the dependent variable. Having said that, these papers exhibited variation in their treatment of “trade-related indicators” as independent variables. In many cases, they used them to focus on the relationship between other independent variables and economic growth, or as fairly insignificant control variables. As a result, we found that not many of the 20 papers would relate directly to the issue we were interested in. All the papers in Group B, however, addressed which factors determined the nature of post-systemic-transformation trade in Russia and Eastern European states, so they can all be said to be deeply connected with the issue we were interested in. This is because as we stated at the beginning of this chapter, if the COMECON system is seen as a system that artificially determined the volume of trade, the goods traded, and the countries traded with, its collapse constituted a transformation to a situation in which trade emerged as a result of natural economic forces in each country. Obviously, the interest of researchers is probably in shedding light on what these “natural economic forces” are. Furthermore, a noteworthy aspect of these papers is that almost half of them (19 papers) employed the gravity model, which will be discussed later.

8.4 POLICY FOR META-ANALYSIS For this chapter, which of the two groups of papers, each has one of the two direc­ tions described above, should be made the subject of our analysis? Group A or Group B? Of course, both research directions are significant in the field of transition economics, and interesting conclusions can probably be drawn from both. In this chapter, however, we made papers in Group B the object of our meta-analysis for the reasons given below. One of the reasons for choosing Group B is technical. As mentioned earlier, the dependent variables in Group A include various types of factor.9 In such a situation, it is likely that we would be unable to obtain a sufficient number of observations for each category from the empirical findings for the meta-analysis to be attempted in this chapter, which involves the combination of t values, meta-regression analysis, and an investigation of publication selection bias (PSB), if the classification is per­ formed in accordance with differences in the analysis. Furthermore, while the papers in Group A include “trade-related indicators” in their independent variables, it must be pointed out that in most cases those “trade-related indicators” are either at odds with our purpose or the authors had little interest in trade and merely used “traderelated indicators” as control variables for making other assertions. Another, and more important, reason is that the papers in Group B, as will be explained below, clearly matched the issue we were interested in, namely what sort of changes the collapse of the COMECON system brought to the trade structure of each transition country. Table 8.5 shows changes in the per-capital value of goods exported (in other words, service exports were not included) from four Eastern

268

POLICY FOR META-ANALYSIS

8.4

Table 8.5 Export volume of goods per capita (Million US$)

Bulgaria Czech Republic Hungary Poland

1995

2000

2005

2010

2015

639.4 1529.6 1093.9 648.9

603.1 1914.7 2322.9 933.3

1535.3 6308.4 5806.5 2281.2

3323.3 10948.9 8757.8 4058.6

3871.8 12433.5 9071.3 4939.8

Note: The figure for Bulgaria in 2015 is substituted by that in 2014. Source: IMF, International Financial Statistics, various issues

European countries. An examination of the data reveals that the role played by trade in the national economy varied greatly, even in countries that had already joined the EU in 2004 and 2007. Of course, there was already variation at the starting point (1995 in this case), but the growth that followed also differed. For example, in the Czech Republic and Hungary, the per-capita value of goods exported jumped approxi­ mately ninefold in 20 years, while in Bulgaria it increased sixfold. Gaps between states in terms of trade performance actually widened, and the impact of the collapse of the so-called COMECON system has varied from country to country. To find out why this has occurred, it is necessary to examine the papers in Group B, which may have investigated the factors that determine the scale of trade. However, a more detailed examination reveals that the papers in Group B employ a variety of different methodologies, and it is not the case that every paper in Group B can be synthesized/combined in a meta-analysis. For this chapter we therefore decided to use as the objects of our analysis only papers that employed the gravity model and papers that did not employ the gravity model per se but employed a similar method. Here, a “method similar to the gravity model” refers to a method that does not employ the “distance” between trading partners as an independent vari­ able, but that includes in the independent variables other quantitative variables that are used in typical gravity models, such as GDP and population.10 By adopting this definition, we can perform an analysis that is methodologically consistent from start to finish and that matches the purpose of this chapter. The gravity model is based on the ideas of Newtonian mechanics and is premised on the assumption that the value of trade between two countries is determined by the economic strength of the coun­ tries (e.g. the product of the countries’ GDPs) and the distance (square thereof) between them. It also takes other factors into account, and attempts to determine the factors and the weight thereof that determine the volume of trade in all fields between the two countries. This matches our purpose, as it asks to what extent a situation arose in which trade was determined by “natural economic forces” that emerged in conjunction with the collapse of the COMECON system. There were also papers that we did not use in the meta-analysis for this chapter even though they employed the gravity model. These were papers where the bulk of the data used was on non-transition countries, which was at odds with the analytical objectives of this chapter. There were also papers that performed analysis, such as 269

C H A P T E R 8

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

the Granger causality test and the unit root test, that could not be incorporated into the meta-analysis conducted here, which involves the combination of t values and synthe­ sized partial correlation coefficients (PCCs). These papers were also excluded.11 So we ultimately selected 21 papers.12 The estimate results reported in these 21 studies will be the object of this chapter’s meta-analysis. Before we move on to the analysis, however, let us first provide some excerpts from them and explain their content to illustrate what specific claims they make.

8.5 CLAIMS MADE BY THE PAPERS SUBJECT TO THE SURVEY Let us begin by introducing some papers that employed the gravity model. The paper by Gros and Gonciarz (1996) is an example of an early application of the gravity model, which appeared in the 1990s.13 The authors used the gravity model to investigate the ways in which trade in former socialist countries in Central and Eastern Europe changed as a result of the liberalization of external trade. They con­ cluded that the widely circulated view that trade in Central and Eastern Europe would expand was being overemphasized. They contended that the expansion and change in direction, which had been bound to occur, were already underway, and the importance of trade among the peripheries was being overemphasized. Egger et al.’s paper (2007) is another example of research employing the gravity model. The authors classified trade into three types: west–west trade, west–east trade, and east–east trade. They estimated a gravity model for each of them, and concluded that west–east trade and east–east trade exhibit different characteristics. More specific­ ally, they asserted that the degree of reaction to trade friction was faster in the case of west–east trade (due to presence of large numbers of multinational firms), that in the case of east–east trade the main economic determinant was differences in factor endowment, and that west–east trade and east–east trade were more sensitive over the medium term to changes in the economic cycle and market size. They also concluded that west–east trade and east–east trade, the latter in particular, were more sensitive to distance and trading costs than west–west trade. Ghatak et al. (2009) used an augmented gravity model to examine the impact of migration (in terms of both flow and stock) from Central and Eastern Europe to the UK on the value of trade between the countries concerned and the UK. They made estimates after controlling for “distance, “size of economy,” “common language,” “common borders,” “historical combination,” and “free trade agreement (FTA) par­ ticipation.” The authors discovered a surprising fact, namely that with the augmented gravity model, the sign for “migration” was negative, but that even when the fixedeffects panel structure model is employed, the sign for “migration” was only positive for imports. The authors then added UK per-capita GDP to the UK import model and Central and Eastern European country per-capita GDP to the UK export model, and concluded that migration from a country to the UK more significantly affects

270

CLAIMS MADE BY THE PAPERS SUBJECT

8.5

exports from the country concerned (a country that is the source of migrants) to the UK than imports by the country concerned. Marti et al. (2014) employed the gravity model to analyze the impact of logistics and transportation on trade in emerging markets. They used the World Bank’s Logis­ tics Performance Index (2007, 2010, and 2012, covering 150 countries), and showed that the LPI obtained a positive and significant coefficient for both exporters and importers. In transition countries, logistics and transportation systems remain inad­ equate, and, given that many hold the view that this inadequacy could be hampering economic development, this can be said to be a significant study. Next we will introduce some papers that cannot be said to have employed the gravity model in a strict sense because they did not include the distance between trading partners in their independent variables. In spite of this, the issues these papers explored matched the purpose of this chapter, and the estimates they reported can be synthesized/combined with those from papers that did employ the gravity model. Toole and Lutz (2005) performed their analysis for the purpose of exploring a dilemma faced by transition countries, namely the degree to which they should open up their countries to foreign trade. They studied former centrally planned econ­ omies and 22 Western countries, and using the per-capita value of imports, which was used to denote the “degree of trade openness,” as the dependent variable and population and wealth (per-capita GNP) as independent variables, they performed a multiple regression. They also considered geographical proximity14 to Western Europe and the degree of political openness separately. They asserted that the degree of trade openness was a product of the degree of openness of the political system (which was affected by geographical proximity). Aristovnik (2007) applied analytical models employing the following independent variables: this year’s current-account balance, the previous year’s current-account balance, the GDP growth rate, per-capita income as a proportion of that of the EU15, general government fiscal balance, the degree of trade openness, external debt, and the real economic growth rate of the EU15. Aristovnik stated that the results were more or less as expected and the same as those presented in previous research. He then compared the current-account balance that could be computed from the model (i.e. for which a simulation could be performed) with the actual cur­ rent-account balances of Central and Eastern European countries (average for 2000 to 2003 in both cases) and found that Albania, Bulgaria, Latvia, and Tajikistan were running excessive current-account deficits.15 In other words, these countries had trade deficits that were larger than estimated from the model, which suggests that in these countries there were unique factors generating the deficit that could not be grasped using the model. Kancs (2007) discussed trade liberalization policies in Southeastern European countries. He claimed that in Southeastern Europe it was not increases in the average price of exports, but rather increases in the number of exporting companies and increases in the types of imported goods that served to increase bilateral trade. This led him to the following policy conclusion: if the policy goal is to increase the total 271

C H A P T E R 8

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

value of exports, the most effective policy is to improve road and rail infrastruc­ ture in order to reduce transportation costs per unit (the author referred to this as “reducing variable trading costs”). If, however, the primary policy goal is to maintain or expand export markets for the small number of companies that are internationally competitive, the most effective trade policy will be to reduce fixed trading costs: that is non-tariff barriers and bureaucratic costs associated with crossing borders. According to the author, the maximum benefit will be obtained if trading costs in Bulgaria, Croatia, and Romania are reduced and a free trade area is established. Stare and Andreja (2008) examined the effects of regulatory reform on and pat­ terns concerning service trade exports in Central and Eastern Europe between 1993 and 2004. They claimed that although the impact of regulatory reform on service exports was sufficiently large, there was still room for development. According to the authors, external demand is a secondary determinant of exports. This is a unique paper in that its focus is on the service trade in Central and Eastern Europe.

8.6 META-ANALYSIS 8.6.1 Meta-synthesis and combination of selected estimate results Table 8.6 shows the synthesized PCCs computed by the authors based on 216 esti­ mation results selected from the 21 papers that we picked out for analysis, which include the ones described above. It also presents combination of t values. Here, in addition to using “GDP,” “population,” and “distance,” which are typically employed in the gravity model, as independent variables, we have also introduced a number of transition-factor variables. We classified these transition-factor variables into three types: “structural change variables” (variables for share of value added by the ser­ vice sector, private-sector share of GDP, FDI inflow, depth of finance, and degree of trade openness), “structural reform variables” (variables for transition index, EBRD liberalization index, political risk index, economic risk index, and capital liberalization index), and “EU-factor variables” (variables for whether signatory to EU treaties, whether candidate for EU membership, and value of trade with the EU). For each of these, we performed a meta-synthesis of PCCs and combination of t values.16 We syn­ thesized PCCs using a fixed-effect model and a random-effects model, and determined the synthesized/combined values to reference using a heterogeneity test (Borenstein et al. 2009). Regarding t values, we determined weights based on academic journal rankings, impact factors, and so on and presented these weighted combination of t values alongside unweighted combined t values. We also confirmed the reliability of the combination of t values calculated here by computing a failsafe number (Mullen 1989) with a 5% confidence level. The meaning of the results of the meta-synthesis/combination presented in Table 8.6 is as follows: first, for all the analyses except that for the “financial depth” variable, we will look at the random-effects model, as the null hypothesis concerning the 272

127 102 124 115 43 10 12 72 19 39 86

−0.002*** 0.188*** 0.014*** 0.001* 0.087*** 0.178*** 0.010 0.001 0.010*** −0.009*** 0.228***

Fixed-effect model a 0.187*** 0.248*** −0.183*** 0.023*** 0.223*** 0.178*** 0.186** 0.017*** 0.018*** 0.010** 0.206***

7945.45*** 1281.17*** 7452.78*** 3040.02*** 321.30*** 5.02 79.53*** 2643.97*** 417.20*** 1304.36*** 337.49***

RandomTest of homoeffects model a geneity b

(a) Synthesis of PCCs

45.32*** 43.52*** −40.01*** 18.24*** 12.03*** 7.56*** 3.74*** 13.75*** 25.36*** −5.33*** 34.16***

Unweighted combination

b

a

Notes Null hypothesis: The synthesized effect size is zero. Null hypothesis: Effect sizes are homogeneous. c Structural Change: Share of Service Value-Added, Private Sector Share in GDP, FDI Inflows, Financial Deepening, Trade Openness d Structural Reform: Transition Index, EBRD Liberalization Index, Political Risk Index, Economic Risk Index, Capital Liberalization e EU factors: EU Agreement, EU candidate, Trade with EU ***, **, and * denote statistical significance at the 1%, 5%, 10% level, respectively.

1. GDP 2. Population 3. Distance 4. Transition factors 4.1 Structural change c 4.1.1 Financial deepening 4.1.2 Trade openness 4.2 Structural reform d 4.2.1 Economic risk 4.2.2 Political risk 5. EU factors e

Number of estimates utilized in meta-analysis (K)

Table 8.6 Meta-synthesis of partial correlation coefficient and t values

8.12*** 6.47*** −6.72*** 3.44*** 4.55*** 7.56*** 3.74*** 2.15** 3.62*** −0.76 5.36***

Weighted combination

Failsafe N (fsN) 96283 71280 73197 14016 2256 201 50 4958 4495 371 36988

Median of t values 2.7 4.28 −3.66 2.35 2.601 2.61 2.69 1.96 6.7 1.62 3.05

(b) Combination of t values

META-ANALYSIS 8.6

273

C H A P T E R 8

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

assumption of heterogeneity is rejected. According to the results of the synthesized PCCs presented in Column (a) of the table, positive and significant effect size was obtained for GDP and population. Furthermore, a negative and significant value was obtained for the distance between trading partners. This means that the general rule of the gravity model, which has been repeatedly proven in past research on advanced countries and developing countries and states that “the larger the GDPs (either the sum or product of GDP) and populations (sum or product) of countries that trade with each other is, the greater the trade between the two countries will be, also holds for transition countries, and that the statistical grounds for asserting this are adequate. This chapter’s area of attention is transition economies, and its focus of interest is “transition-factor variables, namely “structural change variables,” “structural reform variables,” and “EU-factor variables.” Looking at these, the results of our meta­ synthesis/combination clearly show that the effect size of these transition factors on trade activity is extremely large. In other words, they show that as transition pro­ gresses, and, for example, the greater the degree of “financial depth” becomes and the more the “economic risk index” declines, the greater the trade between transition countries will become. The same can be said for the combined t values. If they are combined unconditionally —that is, without taking into account peer reviews of the academic journals in which they are published—all the variables are significant for all the combinations, but weighted t values are uniformly smaller than unweighted combination of t values and become insignificant in the case of the “political risk index.” The failsafe number is fairly large in every case, which can be said to attest to the reliability of the estimation results for the combination of t values. However, a look at the failsafe numbers for the “degree of trade openness” and the “political risk index” reveals that they are smaller than for the other variables. This may mean that not enough research to obtain stable results has been accumulated yet. Let us summarize the above results once again. The results clearly showed that the basic variables in the gravity model, namely “GDP,” “population,” and “dis­ tance,” were all consistent with theoretical predictions. This in itself is not a new discovery. Rather, our focus in this chapter is on “financial depth, “trade depth,” and indicators that capture “structural change” as a whole by combining the former two, and our results show that these indicators had a clear and significantly positive impact on the expansion in the volume of trade in transition economies. It was also shown that “economic risk,” indicators of “structural reform” that include that, and “EU factors, contribute the expansion of trade with an effect size that is neither superior or inferior to the aforementioned basic consituent variables of the gravity model. It is safe to assume that with regard to trade in the former socialist countries,

274

META-ANALYSIS

8.6

which had been confined to the economic planning framework under the COM­ ECON system, the predicted result, namely that the transition process, which involved reform of and changes in the economic structure, as well as the process of integration with the European economy, would directly lead to an expansion in the scale of trade among transition economies, did indeed occur. 8.6.2 Detection of publication selection bias and presence of genuine effects As a means of further scrutinizing the reliability of the meta-analytical results reported in Subsection 8.6.1, in this subsection we will investigate PSB (Mullen 1989). We will then perform an analysis using estimates from a meta-regression model used to confirm the genuine effect. With regard to publication selection bias that can occur as a result of assuming a specific sign relationship (positive or nega­ tive) (type I PSB), we will perform a FAT test and a precision-effect test (PET). We will also test publication selection that can occur due to the fact that papers that produce significant results are published more frequently (type II PSB), and by employing a precision-effect estimate with standard error (PEESE), we will confirm the presence of PSB and verify the presence of a genuine effect. As in the other chapters, the methodology employed will be based on that presented in Chap­ ter 1 of this volume. When estimating the meta-regression model for the purposes of investigating PSB, we will employ the least-squares method and make Cluster-robust OLS estimates and unbal­ anced-panel estimates in order to confirm the statistical robustness of the model. Here, we performed funnel plots of “structural change variables,” “structural reform variables,” “GDP,” “distance,” “EU factors,” and for all of these we estimated a meta-regression model concerning PSB and genuine effects. The results are reported in Tables 8.7 to 8.11 and Figure 8.1. As shown in Panels (a) and (b) of each table except Table 8.11, the null hypothesis, which contends that the intercept of specification (1) and specification (2) is zero, is rejected, indicating the presence of PSB in all cases except “EU factors.” A look at the genuine effect, how­ ever, reveals a slight difference. In the case of “structural change” as well as “GDP” and “distance,” the null hypothesis that the coefficient of the recipricol of the standard error β1 is rejected in all cases in Panel (a) of Tables 8.7, 8.9, and 8.10. This can be said to show that “structural change” as well as “GDP” and “distance” do indeed have a positive genuine effect on the scale of trade. With regard to “EU factors,” however, while the PEESE results in Table 8.11 suggest that there may be an genuine effect, with regard to the coefficient of the reciprocal of the standard error for Panel (a) of Table 8.11, the null hypothesis is only rejected for the random-effects GLS. For the other analysis results in Panel (a), the null hypothesis cannot be rejected.

275

C H A P T E R 8

C H A P T E R

THE COLLAPSE OF THE COMECON SYSTEM Table 8.7 Meta-regression analysis of publication selection bias and genuine effects by structural change variables (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

8 Estimation

OLS

Cluster-robust OLS

Random-effects Panel GLS

[1]

[2]

[3] a

Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0)

2.81*** −0.053***

2.81*** −0.053***

2.81*** −0.053***

K R2

43 0.29

43 0.29

43 0.29

Model

(b) Test of type II PSB (Equation: |t |=β0+β1(1/SE)+v) Estimation Model Intercept (H0: β0=0) 1/SE K R2

OLS

Cluster-robust OLS

Random-effects Panel GLS

[4]

[5]

[6] b

2.58*** 0.0067

2.58*** 0.0067

2.29*** 0.0076

43 0.014

43 0.014

43 0.014

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v) Estimation

OLS

Cluster-robust OLS

Random-effects Panel ML

[7]

[8]

[9]

SE 1/SE (H0: β1=0)

2.23*** −0.00058

2.26** −0.00058

1.39 −0.029

K R2

43 0.15

43 0.15

43 −

Model

Notes: a Breusch-Pegan test: χ2=8.35, p=0.00; Hausman test: χ2=0.08, p=0.78 b Breusch-Pegan test: χ2=8.46, p=0.00; Hausman test: χ2=0.19, p=0.67 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

276

META-ANALYSIS

8.6

Table 8.8 Meta-regression analysis of publication selection bias and genuine effects by structural reform variables (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 8

Estimation Model Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

OLS

Cluster-robust OLS

Random-effects Panel GLS

[1]

[2]

[3] a

1.55* 0.01

1.55** 0.01

1.55*** 0.01

72 0.033

72 0.033

72 0.033

OLS

Cluster-robust OLS

Random-effects Panel GLS

[4]

[5]

[6] b

4.06*** 0.0033

4.06** 0.0033

4.06*** 0.0033

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v) Estimation Model Intercept (H0: β0=0) 1/SE K R2

72 0.019

72 0.019

72 0.019

OLS

Cluster-robust OLS

Random-effects Panel ML

[7]

[8]

[9]

1.36* 0.015

1.36 0.015

1.36 0.015

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v) Estimation Model SE 1/SE (H0: β1=0) K R2

72 0.09

72 0.09

72 −

Notes: a Breusch-Pegan test: χ2=3.23, p=0.07; Hausman test: χ2=3.08, p=0.08 b Breusch-Pegan test: χ2=1.88, p=0.17; Hausman test: χ2=4.47, p=0.03 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

277

C H A P T E R

THE COLLAPSE OF THE COMECON SYSTEM Table 8.9 Meta-regression analysis of publication selection bias and genuine effects by GDP variables (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

8 Estimation Model Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

OLS

Cluster-robust OLS

Random-effects Panel GLS

[1]

[2]

[3] a

4.02*** −0.01*

4.02*** −0.01*

4.13*** −0.002

127 0.07

127 0.07

127 0.0003

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v) Estimation

OLS

Cluster-robust OLS

Random-effects Panel GLS

Model

[4]

[5]

[6] b

Intercept (H0: β0=0) 1/SE

4.98*** 0.01***

4.98*** 0.01*

5.54*** 0.02***

K R2

127 0.09

127 0.09

127 0.23

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v) Estimation Model SE 1/SE (H0: β1=0) K R2

OLS

Cluster-robust OLS

Random-effects Panel ML

[7]

[8]

[9]

0.23 −0.002

0.23 −0.002

−0.06 −0.0006

127 0.01

127 0.01

127 −

Notes: a Breusch-Pegan test: χ2=28.96, p=0.00; Hausman test: χ2=0.99, p=0.32 b Breusch-Pegan test: χ2=3.59, p=0.03; Hausman test: χ2=1.26, p=0.26 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

278

META-ANALYSIS

8.6

Table 8.10 Meta-regression analysis of publication selection bias and genuine effects by distance variables (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 8

Estimation Model Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

OLS

Cluster-robust OLS

Random-effects Panel GLS

[1]

[2]

[3] a

−5.95*** 0.09***

−5.96*** 0.09***

−6.73*** 0.10***

124 0.54

124 0.54

124 0.50

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v) Estimation

OLS

Cluster-robust OLS

Fixed-effects Panel LSDV

Model

[4]

[5]

[6] b

Intercept (H0: β0=0) 1/SE

4.79*** 0.06***

4.79*** 0.06***

3.58*** 0.10***

K R2

124 0.36

124 0.36

124 0.61

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v) Estimation Model SE 1/SE (H0: β1=0) K R2

OLS

Cluster-robust OLS

Random-effects Panel ML

[7]

[8]

[9]

−6.09*** 0.06***

−6.09*** 0.06***

124 0.33

124 0.33

2.21 0.10*** 124 −

Notes: a Breusch-Pegan test: χ2=16.37, p=0.00; Hausman test: χ2=0.64, p=0.42 b Breusch-Pegan test: χ2=31.26, p=0.00; Hausman test: χ2=14.21, p=0.0002 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

279

C H A P T E R

THE COLLAPSE OF THE COMECON SYSTEM Table 8.11 Meta-regression analysis of publication selection bias and genuine effects by EU factors variables (a) FAT (type I PBS)-PET test (Equation: t=β0+β1(1/SE)+v)

8 Estimation Model

OLS

Cluster-robust OLS

Random-effects Panel GLS

[1]

[2]

[3] a

Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0)

4.23** −0.13

4.23 −0.13

K R2

86 0.0075

86 0.0075

86 0.075

OLS

Cluster-robust OLS

Random-effects Panel GLS

[4]

[5]

[6] b

0.93 0.602**

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v) Estimation Model Intercept (H0: β0=0) 1/SE

4.44** −0.17

4.44 −0.17

1.52 0.53**

K R2

86 0.013

86 0.013

86 0.013

OLS

Cluster-robust OLS

Random-effects Panel ML

[7]

[8]

[9]

8.3** 0.32**

8.3 0.32

0.84 0.67**

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v) Estimation Model SE 1/SE (H0: β1=0) K R2

86 0.64

86 0.64

86 −

Notes: a Breusch-Pegan test: χ2=67.61, p=0.00; Hausman test: χ2=0.83, p=0.36 b Breusch-Pegan test: χ2=76.90, p=0.00; Hausman test: χ2=0.97, p=0.32 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively.

280

META-ANALYSIS

8.6

C H A P T E R 8

Figure 8.1 Funnel plots of estimates by aggregated category of factors affecting trade volumes

281

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

One unexpected finding was that with regard to “structural reform” (Table 8.8), the results of the meta-regression show that a genuine effect has not been identi­ fied for the determination of the scale of trade. This result has to be stated. It is a fact that the number of studies from which “structural reform” variables can be gathered is limited, and the same goes for “structural change.” One possible interpretation for this is that with regard to the impact of “structural reform,” insufficient research has been accumulated to obtain the genuine effect. Couple this with the fact that with all variables except “EU factors” the presence of PSB has been detected, this may suggest that transition economists actually tend to publish papers that emphasize the importance of “structural reform” in causing an expansion in the scale of trade. Nevertheless, while there exists the problem that PSB cannot be avoided, it can be said that our results strongly suggest that “structural reform” factors, which have been the focus of previous research, and which this chapter has focused on in the context of transition economics., have a positive effect on trade expansion. However, the results for “EU factors” were inadequate, so we hope that more research on them will be conducted in the future.

8.7 CONCLUSIONS For this chapter, we selected 216 estimation results from 21 papers that had quantitatively analyzed “trade-related factors” as dependent variables as a means of investigating changes in the structure of trade in transition countries following the collapse of the COMECON system. We then performed a meta-synthesis/ combination and investigated PSB. The conclusions drawn from this are as fol­ lows: it was made clear that the factors that are assumed under the gravity model to be determinants of trade volume, namely GDP and distance, also had a significant effect in transition countries. Transition factors defined as “struc­ tural change variables” serve to expand the volume of trade. In other words, it was shown that as “transition” progresses, the scale of trade increases. With respect to “structural reform variables” and “EU-factor variables,” it was not possible to identify a genuine effect. One possible interpretation of this is that not enough research has been conducted to allow a genuine effect to be obtained. It may also suggest a tendency for most researchers to publish papers that stress the import­ ance of “structural reform” in expanding the scale of trade. That being said, the fact that strong results were obtained for “structural change variables” probably owes much to the important contributions made by the pion­ eers in this field. The fact that PSB is strongly suggested except in the case of “EU-factor variables” could indicate that an adequate volume of research has still not been accumulated, so it is hoped that further progress is made with such research.

282

NOTES

NOTES

1 With the official English acronym “CMEA,” COMECON was an international economic organization covering most socialist countries. 2 The decision to establish it was made in 1963, and it began activities in 1964. Its headquarters were in Moscow. 3 China (a non-COMECON country) and Vietnam (a COMECON member coun­ try) are excluded from this inquiry. 4 The political transformation in Eastern Europe occurred at the end of 1989. 5 The political transformation in the former Soviet Union occurred at the end of 1991. 6 We combined the “title” and “subject” search fields in various ways as appropriate. 7 See Table 8.4, n. b. 8 For more details, see Table 8.4, n. a. 9 The list of dependent variables used includes productivity, GDP level, GDP growth rate (economic growth rate), environment-related variables such as energy consump­ tion and CO2 emissions, structure of demand shocks and supply shocks in each country, FDI inflow, wage premium of skilled workers over unskilled workers, degree of COMECON convergence, external assets/liabilities, regional disparities, immigration, changes in employment structure, narrowness of interest-rate margins, sales by domestic companies, convergence among countries, conditions for negotiat­ ing EU membership, and share of employment for each sector. 10 There were six such papers: Toole and Lutz 2005, Aristovnik 2007, Kancs 2007, Mandel and Tomšík 2008, Stare and Jaklič 2008, and Beckmann and Fidrmuc 2012. 11 Awokuse 2007, which was one of the papers excluded, performs analysis in both the A and B directions. The paper explains that there are three hypotheses: the ELG (export-led growth) hypothesis, which holds that export expansion and openness to foreign markets are key determinants of economic growth, the GLE (growth-led export) hypothesis, which emphasizes the causal relationship of eco­ nomic growth on exports, and the ILG (import-led growth) hypothesis, which holds that economic growth is initially fueled by imports. It then investigates which hypothesis applies to each Eastern European country. It is an interesting paper that also relates to the purpose of this chapter. That being said, it performs analysis, such as the Granger causality test and the unit root test, that cannot be employed in the meta-analysis conducted for this chapter. 12 We used 15 of the 19 studies listed in the “utilized gravity models” section in Group B of Table 8.4 and 6 of the 23 studies that were included in Group B, but not in the “utilized gravity models” section. 13 Wang and Winters (1992) (and Hamilton and Winters 1992) is an analysis of pre-systemic-transformation data, so it is not included in the studies subject to this chapter’s meta-analysis. Nevertheless, it is an important paper as it was the first to suggest that the gravity model could be useful for research on the structure of trade in Central and Eastern Europe. It can be said that they have laid the foundation for the trend seen in subsequent research.

283

C H A P T E R 8

C H A P T E R 8

THE COLLAPSE OF THE COMECON SYSTEM

14 Because the paper does not employ “geographical proximity” as an independent variable in the model, it is not gravity-model paper. 15 They also estimated other models after changing the independent variables. 16 As Table 8.7 shows, we separately calculated synthesized/combined values for “finan­ cial depth” and “degree of trade openness,” which are “structural change variables,” and “economic risk index” and “political risk index,” which are “structural reform variables.”

REFERENCES Aristovnik, Aleksander (2007) Are current account deficits in Eastern Europe and former Soviet Union too high? Transformations in Business and Economics, 6(1), pp. 32–52. Awokuse, Titus (2007) Causality between exports, imports, and economic growth: Evidence from transition economies. Economics Letters, 94, pp. 389–95. Beckmann, Elisabeth, and Jarko Fidrmuc (2012) Oil price shock and structural changes in CMEA Trade: Pouring oil on troubled waters? European Journal of Comparative Eco­ nomics, 9(2), pp. 31–49. Borenstein, Michael, Larry Hedges, Julian Higgins, and Hannah Rothstein (2009) Introduc­ tion to Meta-Analysis. Wiley: Chichester. Egger, Peter, Michael Pfaffermayr, and Roland Schmidt (2007) Trade in Western and East­ ern Europe in the aftermath of COMECON: An assessment of behavioral change. Oxford Economic Papers, 59(1), pp. 102–126. Ghatak, Subrata, Monica Ioana Pop Silaghi, and Vince Daly (2009) Trade and migration flows between some CEE countries and the UK. Journal of International Trade and Eco­ nomic Development, 18(1), pp. 61–78. Gros, Daniel, and Andrzej Gonciarz (1996) A note on the trade potential of Central and East­ ern Europe. European Journal of Political Economy, 12(4), pp. 709–721. Hamilton, Carl G., and Alan Winters (1992) Opening up international trade with Eastern Europe. Economic Policy, 7(14), pp. 77–116. Kancs, d’Artis (2007) Trade growth in a heterogeneous firm model: Evidence from South Eastern Europe. World Economy, 30(7), pp. 1139–1169. Mandel, Martin, and Vladimir Tomšík (2008) External balance in a transition economy. Eastern European Economics, 46(4), pp. 5–26. Marti, Luisa, Rosa Puertas, and Leandro Garcia (2014) The importance of the logistics per­ formance index in international trade. Applied Economics, 46(22–24), pp. 2982–2992. Mullen, Brian (1989) Advanced Basic Meta-Analysis. Lawrence Erlbaum Associates: Hillsdale. Stare, Metka, and Andreja Jaklič (2008) Transition, regulation and trade in services. Service Industries Journal, 28(3), pp. 277–290. Toole, James, and James Lutz (2005) Trade policies of the former centrally planned economies. Global Economy Journal, 5(3), pp. 1–22. Uegaki, Akira (2011) International economic relationships. In Masahiko Yoshii and Satoshi Mizobata (eds.), Modern Russian Economics. Minerva: Kyoto, pp. 193–213 (Japanese). Wang, Z. K., and Alan Winters (1992) The trading potential of Eastern Europe. Journal of Economic Integration, 7(2), pp. 113–36.

284

9

Foreign direct

investment in

transition

economies

Its determinants and macroeconomic impacts Ichiro Iwasaki and Masahiro Tokunaga

9.1 INTRODUCTION When the transition to a market economy began in Central and Eastern Europe (CEE) and the former Soviet Union (FSU), policy-makers and academic researchers widely expected that foreign direct investment (FDI) could play a significant role in the eco­ nomic recovery of this region (Bangert and Poór 1993; Carlin and Landesmann 1997; Jensen 2006). Nevertheless, as Sinn and Weichenrieder pointed out, “the low level of FDI has been a big disappointment” (1997, p. 180) except in a few reforming countries. In fact, according to Figure 9.1, the stock value of FDI in the region during the 1990s reached only US$141 billion, and just three countries—the Czech Republic, Hungary, and Poland—represented 54%, or US$76 billion, of the total investment. This gloomy situation changed dramatically in the 2000s. With a background of remarkable progress in systemic transformation to a market economy and high economic growth in the region, the investment of foreign capital and advancement of multinational enterprises from the old EU member countries and other advanced economies were greatly activated. It is worth stressing in this regard that, as shown in Figure 9.1, FSU countries exceeded CEE countries in the total amount of FDI inflow in the period from 2008 to 2017, and the gap between these two country groups is now remarkable. As a matter of fact, the CEE countries, including the Baltic states, received a total of US$35.3 billion from abroad as direct investment in 2017, while the gross inflow of FDI into 12 FSU countries reached US$412.3 billion in the same year. Consequently, as illustrated in Panel a of Figure 9.2, Russia became the host country of the largest FDI by the end of 2017 among 28 CEE and FSU countries, and Kazakhstan and Ukraine were also ranked in the top ten recipients, together with six new EU member countries from Poland to Bulgaria.

C H A P T E R

FOREIGN DIRECT INVESTMENT

9

Figure 9.1 The dynamics of FDI into CEE and FSU countries in 1990–2017 (Billion US$) Note: The line graph (right axis) and the bar graph (left axis) illustrate the annual inflow and stock value of foreign direct invest­ ment, respectively. CEE 10 EU countries denote the Central and Eastern European countries that joined the European Union

either in 2004 or 2007. FSU excludes the Baltic countries.

Source: UNCTAD website (http://unctadstat.unctad.org/)

Nevertheless, experts of transition economies pay attention to the trend of FDI inflow, taking into account the size of each country. In other words, when the scale of FDI is discounted by the total population, we are faced with a completely differ­ ent picture from this viewpoint. In fact, as Panel b of Figure 9.2 shows, in terms of FDI stock per capita in 2017, EU countries from Estonia to Latvia are all top five recipients of the 28 countries. In contrast, many FSU countries, including Russia and Ukraine, hold a subordinate position. The notable differences in country rank between these two panels give researchers a great hint for considering the determin­ ants and economic impacts of FDI in CEE and FSU countries. As pointed out above, due to unsatisfactory trends in foreign capital inflow in the 1990s combined with various technical constraints, including limited data availability and accessibility, empirical studies of FDI were far from adequate in terms of both quality and quantity throughout the first decade of transition. However, this shortage of studies was greatly ameliorated thanks to active research conducted in the 2000s and onward. Now we have a bulk of studies on this topic, and thus we may be able to draw a general picture regarding the determinants and impacts of FDI in transition economies.

286

INTRODUCTION

9.1

C H A P T E R 9

Figure 9.2 FDI stock and per capita value in CEE and FSU countries in 2017 Note: ■, ■, and ■ represent CEE 10 EU member countries that joined the European Union either in 2004 or 2007, other

CEE countries, and FSU countries, respectively.

Source: UNCTAD website (http://unctadstat.unctad.org/)

287

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

With regard to the determinants of FDI, most of the researchers focus on the effect of economic reforms, among many factors to be considered. As Panel a of Figure 9.3 suggests, the economic transition from a socialist economy to a market economy realizes more FDI in countries with a lack of savings and driving forces for the restructuring of extremely inefficient Soviet-type command economies, and many researchers expect a positive correlation between FDI performance and market economy reforms related to the processes of economic transition. The assumption that progress in economic transition matters for FDI supports the view, in an implicit way, that a series of market economy reforms opens opportunities for profitable investment and motivates foreign investors to take advantage of these new business chances: that is to say, FDI inflows go beyond macroeconomic stability without arbi­ trary bureaucracy and require unremittingly any recipient country to create effective and sound business institutions compatible with the market economy (EBRD 1998, pp. 81–82). In most cases, FDI studies use transition indicators of the European Bank for Reconstruction and Development (EBRD) and/or their sub-indicators by area as proxies for the extent of the economic transformation; thus, the classifica­ tion reflects in principle how the EBRD categorizes the transition process into these indicators. Some scholars, however, have been critical and skeptical of an econometric approach to measuring the FDI-inducing effect of transition from the early stage of market economy reforms: according to Myant and Drahokoupil (2012), a high score in quantified transition indicators does not necessarily imply that an efficient modern economy has been established, as the indicators are based on a narrow concept of private ownership rather than on a broader perspective of economic development that is truly indispensable for transition countries. As was acknowledged both by the EBRD, which formulated transition indicators, and Nicolas Stern, who served as the chief economist in the 1990s, the simple approach to transition indicators leaves out what seems to be important to the functioning of the market economy; although state authorities must be sufficiently strong and well organized to secure wellregulated and efficiently operational market mechanisms, these overarching and basic considerations are reflected only in a limited way in quantifying the economic transformation process in CEE and FSU countries (Stern 1997). Therefore, transition indicators show how far an economy has moved from a planned or command regime to a market economy; however, they do not fully indicate how and to what extent a country has worked to carry forward its market reforms. At the same time, Djankov and Murrell (2002) warned that the empirical research on transition economies paid little attention to how to make sense of transition in the wider context of economic development. Another issue of great interest to experts is whether FDI produced a sufficient effect to encourage economic growth in the former socialist states. The economic theory, however, does not support the positive effect of FDI in this respect. In fact, according to the neoclassical growth theory, where FDI is deemed to be a pure factor input, its effect on economic growth in the long term is neutral, although it does affect the national income level. This is because the growth rate will converge 288

a

a

Notes: Country abbreviations: AL — Albania; AM — Armenia; AZ — Azerbaijan; BA — Bosnia and Herzegovina; BG — Bulgaria; BY — Belarus; CZ — Czech Republic; EE — Estonia; GE — Georgia; HR — Croatia; HU — Hungary; KG — Kyrgyzstan; KZ — Kazakhstan; LT — Lithuania; LV — Latvia; MD — Moldova; MK — FYR Macedonia; MO - Montenegro; PL — Poland; RO — Romania; RU — Russia; SB — Serbia; SI — Slovenia; SK — Slovakia; TJ — Tajikistan; TM — Turkmenistan; UA — Ukraine; UZ — Uzbekistan. ●, ■, and ▲ represent CEE 10 EU member countries that joined the European Union either in 2004 or 2007, other CEE countries, and FSU countries, respectively.b The average of EBRD transition scores in 2014 on large scale privatizaation, small scale privatization, governance and enterprise restructuring, price liberalization, tra de and forex sys tem, and competition policy. The figure for the Czech republic is in 2007. The index takes the range between 1.00 (representing little or no change from a rigid centrally planned economy) and 4.33 (representing the standards of an industrialized marke t economy).c Compu ted using data in US 2010 constant price Source: EBRD (http://www.ebrd.com/pages/homepage.shtml) and UNCTAD (http://unctadstat.unctad.org/) websites

Relationship between FDI inflow, transition reform, and economic growth in the CEE and FSU countries

Figure 9.3

INTRODUCTION 9.1

289

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

in the long run as the marginal product of capital diminishes its returns over time, even if the exogenous increase in capital realized in the form of capital inflow from foreign countries may temporarily expand production (Solow 1956). In contrast, according to the endogenous growth theory, where attention with regard to FDI is focused on its function as a delivery vehicle for transferring the excellent technology, knowledge, and know-how accumulated in developed econ­ omies, FDI has a positive effect on long-term economic growth. This is true as long as it brings improvements in technology systems and/or human capital to the recipi­ ent countries through the contributions of foreign participation in management, the establishment of local subsidiaries by multinational enterprises, the outsourcing of contracts between local and foreign firms and so on (Grossman and Helpman 1991; Aghion and Howitt 1997). As Borensztein et al. (1998) and Durham (2004) argued, the growth-enhancing effect of FDI largely depends on the absorption capacity of local entities (i.e. domestic firms and workers). Nevertheless, based on the assump­ tion of high levels of education and sufficient penetration of modern rationalism in the former socialist bloc, many researchers anticipated that the possibility of such an effect would never be small in transition economies (UNECE 2001). However, FDI could rather negatively affect economic growth in the recipient countries if it hampers domestic investment. Indeed, Mišun and Tomšík (2002) reported that FDI crowded out domestic investment in Poland during the period of 1990 to 2000. Moreover, Kosová (2010) also found that, in the Czech Republic, the new entry of foreign-affiliated firms significantly pushed up the ex post exit rate of domestic firms from 1994 to 2001. Taking into account the weak management base and backward production technology of former socialist enterprises compared with multinational corporations based in developed economies, it is highly likely that such negative external effects occurred in many transition economies. Moreover, as pointed out by Easterly (1993), exemptions from corporate income tax and other FDI-friendly policies to attract foreign firms might negatively affect economic growth if these measures heavily distort incentives for domestic entities. It is a well-known fact that CEE countries launched extremely preferential policies to induce FDI in a competitive manner (Cass 2007). Hence, we cannot rule out the possibility that what Easterly (1993) has called the “adverse incentive effect” might actually have had a negative impact on domestic firms in these states. As mentioned above, FDI has the potential to bring about both positive and nega­ tive macroeconomic effects for the recipient countries; however, it is extremely diffi­ cult to theoretically predict the respective degree of these countervailing effects. Furthermore, as indicated in Panel b of Figure 9.3, the scale of FDI inflow and the economic growth rate during the transition period do not portray a definite positive correlation in a simple scatter plot as FDI and transition reforms do in Panel a. Thus, economists examined this issue by performing a multivariate econometric analysis that considers various determinants of economic growth simultaneously. Nevertheless, as we report later, the empirical results in the extant literature regard­ ing the causality between FDI and macroeconomic growth in CEE and the FSU are too mixed to draw a conclusion simply by looking at them. 290

METHODOLOGY OF LITERATURE SELECTION

9.2

To overcome the above research issues, in this chapter, we conduct a metaanalysis of the literature that empirically examines the determinants and macroeco­ nomic impacts of FDI in transition economies. More specifically, we asked the fol­ lowing questions: what do existing studies tell us about the determinants and macroeconomic impacts of FDI as a whole? What determines the differences in the empirical evidence reported in these studies? Is there any artificial bias in their pub­ lication, and, if there is, are the relevant studies sufficient for identifying the true effect beyond such a bias? From our meta-analysis of relevant studies on transition-specific determinants of FDI in transition economies, we found that the composition of target countries in terms of both FDI donors and recipients, the data type, the estimator, and the degree of freedom bring out the heterogeneity of the empirical evidence of the original papers. Furthermore, the results of our meta-regression analysis (MRA) reveal that the pertinent literature has provided limited empirical evidence to prove a nonzero FDI-inducing effect of economic transition, or a tiny true effect if it exists at all, partially because of the existence of publication selection bias (PSB). With respect to the macroeconomic impacts of FDI, we confirmed that existing studies indicate a growth-enhancing effect of FDI in the region as a whole. The results of our meta-regression analysis suggest that the effect size of the reported estimates depends on study conditions. In particular, the composition of target coun­ tries and the type of FDI variable are important factors that explain the heterogeneity of the empirical results. The degree of freedom also greatly affects the magnitude of the FDI variable. We also found that the relevant studies present genuine evidence of a nonzero FDI effect on macroeconomic growth, and its true effect size is esti­ mated to be positive but small. The remainder of this chapter is organized as follows: the next section describes the methodology of literature selection. Section 9.3 conducts a meta-analysis of the determinants of FDI, focusing on the effect of systemic transformation, while Sec­ tion 9.4 examines the macroeconomic impacts of FDI. Based on the results obtained from the meta-analysis, Section 9.5 summarizes the major findings and draws conclusions.

9.2 METHODOLOGY OF LITERATURE SELECTION In this section, we describe our methods for selecting and coding relevant studies and for meta-analysis based on the empirical evidence collected. In order to identify studies related to FDI in CEE and FSU countries as a base collection, we first searched the EconLit and Web of Science databases for research works that had been registered in the 30 years from 1989 to 2018 that contained a combination of two terms including one from “foreign direct investment,” “FDI,” or “multinational enterprise” and another from “transition economies,” “Central Europe,” “Eastern Europe,” “the former Soviet Union,” or the respective names of

291

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

each CEE and FSU country. From approximately 600 studies that we found at this stage, we actually obtained nearly 400 studies, or about 67%, of the total. We also searched the references in these 400 studies and obtained about 80 additional papers. As a result, we collected nearly 480 studies.1 These approximately 480 studies include various papers other than empirical studies on the determinants and macroeconomic impacts of FDI. Hence, as the next step, we closely examined the contents of these works and narrowed the literature list to those containing estimates that could be subjected to meta-analysis in this chapter. In the next sections, we report the results of our literature selection in detail. During this process, we decided to exclude all unpublished research works. According to Doucouliagos et al. (2012), unpublished working papers might present estimates that are not final; moreover, these manuscripts are more likely to be insufficient since, at that time, they had not yet gone through the peer review process. In our judgment, the same concerns also apply to unpublished works we obtained for this study. Another reason to exclude unpublished works is that we use the quality level of each paper that we evaluate based on external indicators as a weight for a combination of statistical significance levels and as an ana­ lytical weight or a meta-independent variable for the MRA. In addition, the following facts also motivate us to take this measure: first, the number of working papers is not very large in our case; second, these unpublished works are not heavily concentrated in recent years. The latter fact led us to decide that there is no particular concern in over­ looking the latest research results due to their exclusion. For the study in this chapter, we adopt an eclectic coding rule to simultaneously mitigate the following two selection problems: the arbitrary-selection problem caused by data collection in which the meta-analyst selects only one estimate per study and the overrepresentation caused by data collection in which all estimates are taken from every study without any conditions. More specifically, we do not necessarily limit the selection to one estimate per study, but multiple estimates are collected if, and only if, we can recognize notable differences from the viewpoint of empirical methodology in at least one item of the target regions/countries, data type, regression equation, estima­ tion period, or estimator. Hereinafter, K denotes the total number of collected estimates (k = 1, 2, …, K). To analyze the collected estimates from selected studies using metaanalytic techniques, we follow the methodology described in Chapter 1.

9.3 DETERMINANTS OF FDI IN TRANSITION ECONOMIES In this section, we attempt to see the relationship between economic transition and FDI performance in the CEE and FSU regions over the past quarter century. First, Subsection 9.3.1 gives an overview of the studies selected for meta-analysis. Next, Subsection 9.3.2 demonstrates our synthesis of the collected estimates, and Subsec­ tion 9.3.3 performs meta-regression analysis to explore the heterogeneity observed between studies. Finally, Subsection 9.3.4 assesses the publication selection bias in the relevant literature.

292

DETERMINANTS OF FDI

9.3

9.3.1 Overview of selected studies for meta-analysis We will begin to give a brief review of the selected studies for a meta-analysis of the determinants of FDI in the CEE and FSU countries during the transition period. Among various key FDI-enhancing factors being discussed so far, a central preoccupa­ tion of scholars and policy makers in the region is the extent to which FDI inflow has been influenced by market economy reforms such as liberalization, enterprise restruc­ turing, competition policy, and privatization. Some empirical works were in place by the mid-1990s, and all of these studies found a positive correlation between FDI per­ formance and market economy reforms related to the processes of economic transition that were represented by EBRD transition indicators, among other things (Lankes and Venables 1996, Lansbury et al. 1996, Selowsky and Martin 1997, EBRD 1998, ch. 4). Then, a rapidly increasing FDI inflow in the ensuing years and the growing availabil­ ity of statistical data for econometric analysis enabled researchers to accelerate their study of FDI determinants in the transition economies, a large part of which drew the conclusion that more progress in the economic transition led to greater FDI received. In accordance with the method of literature selection described in the previous section, we selected a total of 44 studies that contain estimates suitable for our meta-analysis.2 Note that we removed those studies that (1) do not provide empirical results in quantitative way, such as descriptive studies specifically; (2) involve only one explanatory variable in simple regression models; (3) adopt binary dependent variables with probit and/or logit estimators, of which the explanatory variables’ effect sizes are not comparable to those of linear regression models;3 and (4) focus on spatially limited areas or specific industrial subsectors in a host country, of which the research design seems to be fundamentally different from those of country-level studies. Finally, we trimmed our samples by eliminating some outliers that bring out quite a large inverse of standard errors of the partial correlation coefficients (PCCs). In the 44 selected studies, non-EU CEE countries and FSU countries, excluding the Baltics, with less opportunity to participate in the process of EU accession despite high FDI performance or potential, are found less among the quantitative studies. An exception is Croatia, which joined the EU in 2013. Derado 2013 is a good example of works driven by the perspective of a country’s EU accession process (see also Deich­ mann 2013). Also, recent studies try to fill the knowledge gap, focusing on the deter­ minants of FDI location in Southeastern Europe and the Balkans (Hengel 2011; Estrin and Uvalic 2014; Dauti 2015a, 2015b; Lee 2015; Shukurov 2016). On the whole, except for Döhrn (2000) and Jensen (2002), who do not report the composition of FDI recipients, the total number of host country observations is 541, of which 59.7% (323 observations) deal with CEE EU countries. Meanwhile, the share of non-EU CEE countries and FSU countries, excluding the three Baltic states, account for only 13.9% (75 observations) and 19.2% (104 observations), respectively. Empirical analysis in the selected studies above covers the 23 years from 1989 to 2011 as a whole. The average estimation period of collected estimates is 10.7 years (median: 10; standard deviation: 3.9). Twenty-two studies employ the total FDI model with all FDI received from the world as a dependent variable, while 20 studies rely on

293

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

the bilateral FDI model that uses an amount of FDI from a specific home country as a dependent variable. The remaining two estimate both models. We see a recent upward trend in the number of studies adopting the bilateral model, which reflects the intention of those who have been analyzing FDI determinants in general to attach more weight to the gravity model as a basic research design. Reflecting the reality that a large portion of inward FDI to CEE and FSU countries comes from advanced coun­ tries within the EU, the bilateral FDI model makes Western Europe a main target for analysis. A few advanced non-EU countries and leading emerging market economies, including those in the former socialist bloc, are also added to the list of investors in Bandelj (2002, 2008) and Estrin and Uvalic (2014). As for data type, studies using panel data make up over 80% of the total; other­ wise they employ cross-sectional data or, in only one case, rely on time series data. Some researchers were conducting empirical analyses with cross-sectional data until the mid-2000s. This is probably due to the limited availability of longitudinal data as well as the volatility of FDI inflow to the region during the first decade of transi­ tion. Next, the FDI indicators to be introduced as dependent variables on the lefthand side of regression equations can be subdivided into seven groups. The annual net FDI inflow is the most commonly used indicator: 17 of the 44 studies count upon this type of variable. Cumulative gross FDI value or FDI stock comes next: nine studies use these. The FDI variable chosen seems to depend both on purely technical considerations and a priori selection of the specific variables, given the research interest of each study. In the case of the first issue, when one applies pub­ lished and widely used FDI datasets that are often extracted from the UNCTADStat, OECD StatExtracts, the World Economic Outlook database of the IMF, and the World Development Indicators provided by the World Bank Group, a negative value would be found. This is because these datasets express the annual net value of FDI flow or a difference between inbound FDI and outbound FDI based on the balance of payment statistics of each country. That poses a serious obstacle to performing log-transformed linear regression. In fact, we have seen a negative bilateral invest­ ment flow in CEE and FSU countries explicitly during the two financial crises of the mid-1990s and 2008 to 2009. In Russia, among others, “capital flight” continues to be a macroeconomic problem even now, despite its largest FDI volume received in absolute terms. Besides that, the unevenness of FDI inflow has the potential to make for noisier relationships with other flows, such as GDP, to which they are often scaled (Claessens et al. 2000). To avoid this problem, for example, Garibaldi et al. (2001) used the gross value of FDI inflow without any deduction for outflow, and Botrić and Škuflić (2006) cited the FDI stock from a direct investment position data­ base. As for a priori selection of FDI indicators, although not often expressly stated in the papers, it is highly predictable that the authors prefer a specific FDI variable for their research design and tasks. To give an example, Overesch and Wamser (2010) argued for the conceptual advantages of the number of investments (count variable) as a result of location choice by multinational enterprises because a usual form of binary choice model (to go or not to go) is incapable of taking into account that multinational enterprises often have multiple affiliates in a host country. 294

DETERMINANTS OF FDI

9.3

Meanwhile, transition-specific explanatory variables that are incorporated into the right-hand side of regression equations can be classified according to their contents with six indicators. In most cases, the selected studies use EBRD transition indicators and/or their sub-indicators by area as proxies for the extent of the economic transformation; thus, the classification reflects in principle how the EBRD categorizes the transition process into these indicators. However, the privatization indicators stipulated herein include the large- and small-scale privatization indexes provided by the EBRD as well as other privat­ ization-related variables, such as private sector share and privatization revenues in each country. We found that studies using these privatization indicators as transition-specific explanatory variables are in the majority, accounting for 23 of the total 44 studies with them. This is understandable in light of the fact that by-bidding direct sales of state-owned assets was proposed as a way of privatization in CEE and FSU countries, thereby dramat­ ically increasing FDI inflow in some cases, as symbolized by Hungary in the 1990s. Sub­ sequently, 11 papers employ general transition indicators; those that rely on liberalization indicators, enterprise reform indicators, and competition policy indicators are in the minor­ ity (from five to seven studies for each); interestingly, 19 studies deploy other transition indicators such as trade and forex systems, the efficiency of law institutions, infrastructure reform, and financial sector reform. This last point would suggest the breadth of researchers’ understanding of the economic transition or, alternatively, reflect that there is no clear consensus concerning the essence of the economic transition in the region. 9.3.2 Meta-synthesis Figure 9.4 shows the frequency distribution of the PCC and the t value of the transi­ tion-related variables; the Shapiro–Wilk test rejects the null hypothesis of normality at the 1% significance level for both. As Panel a of this figure shows, the PCC shows a sharp-pointed distribution with a mean of 0.197 and a median of 0.160. According to Cohen’s guidelines of PCC (1988), 25.7% (46 estimates) find no practical relation­ ship (|r| < 0.1) between transition progress and FDI performance in CEE and FSU countries, while 50.8% (91 estimates) and the remaining 23.5% (42 estimates) report a small effect (0.1 ≤ |r| ≤ 0.3) and a medium or large effect (0.3 < |r|), respectively. Meanwhile, Panel b of the figure tells us that the estimates of transition-related vari­ ables with respective absolute t values that are equal to or greater than 2.0 account for 54.7% (98 estimates) of the total. The estimation period of each study varies significantly: thus, we could expect that this difference has a noticeable influence on empirical results. In fact, Panel a of Figure 9.5 reveals that the effect size of the collected estimates has been decreasing over time, meaning that the correlation of economic transition and FDI would be weakened as market-oriented economic reforms continue. At the same time, Panel b of the figure shows that the collected estimates demonstrate a flat trend for the t value in chronological order; the coefficient of the average year of estimation period (yr) is estimated to be positive but statistically insignificant. To examine this point more strictly, we will test the influence of the estimation period on empirical results in the meta-regression analysis in the next subsection. 295

C H A P T E R 9

C H A P T E R

FOREIGN DIRECT INVESTMENT

9

Figure 9.5 Chronological order of partial correlation coefficients and t values of the collected estimates of determinants of FDI (K=179) Note: Figures in parentheses beneath the regression coefficients of the approximate straight line are robust standard errors. *** denotes statistical significance at the 1% level.

To consider the implications of the integration of empirical results in a more sys­ tematic way, we synthesized the collected estimates of the selected studies using the meta-synthesis methodology outlined in Chapter 1. Table 9.1 indicates the outcome of the synthesis of the collected estimates extracted from our sample. In addition to 296

0.055*** 0.053*** 0.148*** 0.926*** 0.225*** 0.039*** 0.069*** 0.098*** 0.037*** 0.232*** 0.137*** 0.135***

0.185***

0.122*** 0.208*** 0.260*** 0.163*** 0.050*** 0.013***

179 159 17 3 102 77 51 24 55 29 3 14

3

20 13 22 18 65 41

Notes:

a Null hypothesis: The synthesized effect size is zero.

b Null hypothesis: Effect sizes are homogeneous.

***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

All studies (a) Comparison in terms of data type Studies that employ panel data Studies that employ cross-sectional data Studies that employ time series data (b) Comparison in terms of model type Studies that adopt total FDI model Studies that adopt bilateral FDI model (c) Comparison in terms of the type of FDI variable Studies that use annual net FDI inflow Studies that use annual gross FDI inflow Studies that use cumulative gross FDI value or FDI (including fixed capital) stock Studies that use annual net or gross FDI inflow per capita Studies that use cumulative net FDI value per capita Studies that use annual net FDI inflow to GDP (including manufacturing value added) or annual gross FDI inflow to manufacturing output Studies that use other types of FDI variables (number of FDI projects, etc.) (d) Comparison in terms of the type of transition variable Studies that use general transition indicators Studies that use liberalization indicators Studies that use enterprise reform indicators Studies that use competition policy indicators Studies that use privatization indicators Studies that use other indicators

Number of Fixedestimates effect (K) model a

Table 9.1 Synthesis of collected estimates of determinants of FDI

0.224*** 0.257*** 0.312*** 0.153*** 0.138*** 0.085***

0.185**

0.260*** 0.174*** 0.186***

0.164*** 0.109*** 0.155***

0.263*** 0.090***

0.159*** 0.159*** 0.926***

0.166***

Randomeffects model a

445.870*** 50.030*** 57.664*** 73.376*** 430.853*** 223.408***

0.182

119.629*** 5.209* 74.537***

302.244*** 75.306*** 854.353***

729.976*** 497.709***

1463.511*** 25.080* 0.028

1588.308***

Test of homogeneity b

(a) Synthesis of PCCs

18.776*** 9.878*** 12.373*** 5.719*** 20.317*** 8.607***

3.580***

12.431*** 3.784*** 6.299***

18.321*** 8.409*** 19.617***

24.380*** 19.894***

29.533*** 7.777*** 9.431***

31.452***

Unweighted combination

65257 51090 363 96 22303 11185 6275 603 7767 1627 13 191

11

2586 456 1223 200 9850 1081

2.351 2.342 1.886 5.445 2.414 2.267 2.565 1.716 2.645 2.308 2.185 1.684

2.067

4.198 2.740 2.638 1.348 2.520 1.344

4.800*** 1.296* 2.969*** 1.290* 3.177*** 1.885**

0.895

2.416*** 0.617 1.057

2.861*** 1.363* 4.904***

5.040*** 3.228***

5.518*** 1.228 1.886**

5.774***

Failsafe N (fsN)

Median Weighted of t combination values

(b) Combination of t values

DETERMINANTS OF FDI 9.3

297

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

the overall synthesis results shown on the top line, we also report individual synthe­ sis results, focusing on differences in data types, model types, types of FDI variable, and types of transition variable, in light of the discussion in the previous subsection. As shown in Column a of the table, which reports the synthesis results of the PCC, the homogeneity test rejects the null hypothesis in almost every case; thus, the synthesized effect size, Rr , of the random-effects model is adopted as the reference value. The mag­ nitude of the synthesized effect size differs remarkably between subjects of comparison. More specifically, studies that conduct a time series data analysis tend to report a much larger positive effect on FDI performance than do those performing a panel or a crosssectional data analysis. With regard to model type, the total FDI model is highly likely to result in a greater influence of FDI determinants as compared to the bilateral FDI model. The type of FDI variable chosen seems to be essential for interpreting empirical results: studies using annual net or gross FDI inflow per capita tend to offer larger effect sizes than do others. In the case of transition-specific explanatory variables, their effect sizes are roughly classified into two groups—one for variables with comparatively larger effect sizes (indicators of general transition, liberalization, and enterprise reform), and the other for less powerful variables (competition policy, privatization, and other indicators). Remember that these results are simply compiled from the collected estimates of the ori­ ginal studies. In the next subsection, we will turn to this issue in a more rigorous way, so as to be more precise using multivariate meta-regression models. Column b of the table shows the results of the combined t value. A first inspection of both tables immediately reveals not only that the combined t value, Tw , weighted by the quality level of the study is substantially lower than the unweighted combined t value, Tu , but also that the former falls below the 10% level in terms of its statistical significance, in some cases. These results suggest that there may be a strongly nega­ tive correlation between the quality level of the study and the reported t value. At the same time, except for a few cases, the failsafe N (fsN) in the right column of the tables shows a sufficiently large value. This means that, even taking into consideration the presence of unpublished studies (working papers, discussion papers, conference papers, etc.) that have been omitted from our meta-analysis, the overall research impli­ cations obtained from the selected studies herein cannot be easily dismissed. 9.3.3 Meta-regression analysis Based on discussions in this section, one can foresee that the observed heteroge­ neous set of studies would largely affect their empirical results. In order to scrutinize this issue more carefully, we estimated meta-regression models that take either the PCC or the t value of a collected estimate as the dependent variable. Table 9.2 lists the names, definitions, and descriptive statistics of meta-independent variables to be introduced on the right-hand side of the regression model. As this table suggests, in our MRA, we quantitatively examined whether and to what extent empirical evidence from the pertinent literature is affected by differences in the composition of target coun­ tries in terms of both FDI donors and recipients, the estimation period, the data type, the presence or absence of controlling for individual and time effects,4 the estimator, the 298

DETERMINANTS OF FDI

9.3

Table 9.2 Name, definition, and descriptive statistics of meta-independent variables used in meta-regression analysis of heterogeneity among studies of determinants of FDI Descriptive statistics Variable name

Definition

Proportion of CEE 10 EU countries Proportion of other CEE countries Proportion of EU countries Proportion of nonEU countries First year of estimation Length of estimation Cross-section

Proportion of CEE 10 EU countries in the host target countries a Proportion of other CEE countries in the host target countries

Time series Individual Time FE RE SLS GMM Bilateral Log Annual gross inflow Cumulative gross value or stock Annual net or gross inflow per capita Cumulative net value per capita Annual net inflow to GDP etc.

Proportion of EU advanced countries in the home target countries b Proportion of non-EU advanced countries in the home target countries b First year of estimation period Years of estimation period 1 = if cross-sectional data is employed for analysis, 0 = otherwise 1 = if time series data is employed for analysis, 0 = otherwise 1 = if individual effects of the host target countries are controlled, 0 = otherwise 1 = if time effects during the estimation period are con­ trolled, 0 = otherwise 1 = if fixed-effects panel estimator is used for estima­ tion, 0 = otherwise 1 = if random-effects panel estimator is used for esti­ mation, 0 = otherwise 1 = if two-step least squares estimator is used for esti­ mation, 0 = otherwise 1 = if generalized method of moments estimator is used for estimation, 0 = otherwise 1 = if bilateral FDI model is used for analysis, 0 = otherwise 1 = if logarithmic value of the dependent variable is used for estimation, 0 = otherwise 1 = if FDI variable is measured in annual gross inflow, 0 = otherwise 1 = if FDI variable is measured in cumulative gross value or stock (including fixed capital), 0 = otherwise 1 = if FDI variable is measured in annual net or gross inflow per capita, 0 = otherwise 1 = if FDI variable is measured in cumulative net value per capita, 0 = otherwise 1 = if FDI variable is measured in annual net inflow to GDP (including manufacturing value added) or annual gross inflow to manufacturing output, 0 = otherwise

Mean

Median

S.D.

0.736

0.857

0.298

0.153

0.120

0.213

0.425

0.077

0.394

0.098

0.081

0.058

1993.508 1994 10.721

10.000

1.964 3.935

0.095

0

0.294

0.017

0

0.129

0.458

0.0

0.500

0.380

0

0.487

0.184

0

0.389

0.413

0

0.494

0.034

0

0.180

0.128

0

0.336

0.430

0

0.496

0.832

1

0.375

0.134

0

0.342

0.307

0

0.463

0.162

0

0.369

0.017

0

0.129

0.078

0

0.269

(Continued )

299

C H A P T E R 9

C H A P T E R

FOREIGN DIRECT INVESTMENT

9

Variable name

Definition

Other FDI variables Liberalization

1 = if another FDI variable is used, 0 = otherwise

Table 9.2 contd. Descriptive statistics

1 = if the liberalization indicator is used as the economic transition variable, 0 = otherwise Enterprise reform 1 = if the enterprise reform indicator is used as the economic transition variable, 0 = otherwise Competition policy 1 = if the competition policy indicator is used as the economic transition variable, 0 = otherwise Privatization 1 = if the privatization indicator is used as the economic transition variable, 0 = otherwise Other transition 1 = if another indicator is used as the economic transiindicators tion variable, 0 = otherwise √Degree of Root of degree of freedom of the estimated model freedom Quality level Ten-point scale of the quality level of the study c

Mean

Median

S.D.

0.017

0

0.129

0.073

0

0.260

0.123

0

0.329

0.101

0

0.302

0.363

0

0.482

0.229

0

0.421

18.881 4.654

11.314 19.383 5

2.839

Notes:

a CEE EU countries denote the 10 Central and Eastern European countries that joined the European Union either in 2004

or 2007. b For the total FDI model, all home countries are conveniently divided into four categories according to the country group classificaton of the UNCTAD Handbook of Statistics 2012: among 221 countries listed, 17 are classified as EU advanced countries, 18 as non-EU advanced countries, and the remaining 186 as emerging and developing countries, including the former socialist countries. c See Chapter 1 for more details.

model type, the form of dependent variable (exact numeric value versus logarithmic value), the type of FDI variable, the type of transition variable, and the degree of free­ dom, as well as the quality level of the study. Note that some meta-analysis studies of general FDI determinants have, thus far, demonstrated that the empirical evidence of ori­ ginal papers is highly dependent on what type of FDI variable is chosen.5 Table 9.3 reports the estimation results of the MRA of heterogeneity among the selected studies for transition-specific FDI determinants. Although the weighted least squares (WLS) models are sensitive to the choice of analytical weights, many variables are signifi­ cantly estimated uniformly. The coefficient of determination (R2), which indicates the explanatory power of a model, ranges from 0.220 (Model 7) to 0.828 (Model 4) if we set aside Model 14 with extremely low explanatory power due to the omission of several explanatory variables in the course of the fixed-effects estimation. This is of a sufficient level, as compared to previous meta-analysis studies of FDI performance. Based on the estimation results of two sets of MRA, we find that a number of coded characteristics of the selected studies exert a statistically significant influence on their empirical evidence. In other words, the empirical results of FDI determinants are highly likely to be affected as follows. First, the empirical evidence of original papers would be affected by differences in the composition of target countries. While studies with more non-EU CEE countries as FDI donors are conducive to larger effect sizes 300

[1]

Clusterrobust OLS

[2]

Cluster-robust WLS [Quality level]

Composition of host target countries (FSU) Proportion of CEE 10 EU 0.009 −0.056 countries Proportion of other CEE 0.244** 0.149 countries Composition of home target countries (Non-advanced countries) Proportion of EU −0.048 −0.095 Proportion of non-EU −0.748* −0.767*** Estimation period First year of estimation −0.011 −0.010 Length of estimation 0.000 0.001 Data type (Panel data) Cross-section −0.147** −0.120** Time series 0.689*** 0.681*** Control for individual and time effects (No control) Individual −0.058 −0.040 Time 0.133** 0.126**

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

−0.165* 0.133

−0.196 −0.343 0.004 −0.006 −0.034 0.590*** −0.061* 0.157***

0.286*

−0.032 −0.436* −0.012 −0.020** −0.194*** 0.632*** −0.056 0.055

[4] a

0.020

[3]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−0.053 0.109**

−0.171** 0.719***

−0.017* 0.001

0.030 −0.632*

0.233***

0.011

[5]

Multilevel mixed effects RML

Table 9.3 Meta-regression analysis of heterogeneity among studies of determinants of FDI

−0.052 0.106*

−0.173** 0.721***

−0.017 0.001

0.034 −0.629*

0.234***

0.011

[6] b

Random-effects panel GLS

(Continued )

−0.036 −0.082

−0.339** 0.758***

dropped dropped

dropped dropped

0.236***

0.003

[7] c

Fixed-effects panel LSDV

DETERMINANTS OF FDI 9.3

301

C H A P T E R 9

302

[1]

[2]

Cluster-robust WLS [Quality level]

Estimator (OLS) FE −0.104* −0.135** RE −0.128** −0.108** SLS −0.233*** −0.326*** GMM −0.116* −0.072 Model type (Total FDI model) Bilateral −0.088 0.003 Form of dependent variable (Exact numeric value) Log −0.041 −0.028 Type of FDI variable (Annual net inflow) Annual gross inflow −0.151** −0.095** −0.022 0.026 Cumulative gross value or stock Annual net or gross inflow 0.009 0.080 per capita Cumulative net value per 0.198* 0.161 capita Annual net inflow to GDP −0.141 −0.072 etc. Other FDI variables −0.042 −0.007 Type of transition variable (General transition indicators) Liberalization −0.018 0.050 Enterprise reform 0.005 0.085 Competition policy −0.153 −0.125

Meta-independent variable (Default)/Model

Clusterrobust OLS

−0.080*** 0.031 0.085* 0.010 −0.127 −0.004 −0.043 −0.069 −0.198***

−0.015 0.004 −0.121 −0.160** −0.110 −0.022 −0.120*

−0.050 −0.066 −0.225***

−0.084

−0.148

0.177*

−0.076

−0.192*** −0.051

−0.019

−0.090

0.009 −0.103** −0.071

−0.126

0.006

−0.086

−0.118** −0.130*** −0.209*** −0.123**

[5]

−0.055 −0.077 −0.197** −0.014

[4] a

Multilevel mixed effects RML

−0.053 −0.090 −0.085 −0.106**

[3]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−0.052 −0.070 −0.229**

−0.099 −0.136* −0.295***

dropped

dropped

−0.148 −0.087

dropped

dropped

−0.011*** −0.243***

dropped

−0.243***

−0.146** −0.141** −0.197*** −0.212**

[7] c

Fixed-effects panel LSDV

0.177*

−0.081

−0.195*** −0.053

−0.018

−0.127

−0.119** −0.130*** −0.208*** −0.124*

[6] b

Random-effects panel GLS

9

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

Table 9.3 contd.

C H A P T E R

FOREIGN DIRECT INVESTMENT

[8]

Clusterrobust OLS

177 0.558

[9]

Cluster-robust WLS [Quality level]

177 0.593

−0.004*** − 20.772

0.076 0.033

Composition of host target countries (FSU) Proportion of CEE 10 EU 1.033 0.480 countries Proportion of other CEE 3.655** 2.408 countries Composition of home target countries (Non-advanced countries) Proportion of EU −0.473 0.436 Proportion of non-EU −14.054** −11.667** Estimation period First year of estimation 0.014 −0.014 0.174** Length of estimation 0.142 Data type (Panel data) Cross-section −2.744*** −2.709*** Time series 3.851*** 3.965***

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

K R2

Privatization −0.001 Other transition indicators −0.062 Degree of freedom and research quality √Degree of freedom −0.003** Quality level 0.002 Intercept 21.990 177 0.828

−0.003*** 0.005 −6.637

−0.109* −0.163**

−2.201*** 3.067***

0.066 0.053

−0.130 −0.239 −4.806*** 1.030

0.503 −11.964**

−1.580 −7.353

1.475 −8.788*

−2.899*** 4.277***

−0.090 0.127

2.659***

3.564*

4.333

0.786

[12]

−0.216

[11] a

Multilevel mixed effects RML

177 −

−0.004** −0.002 34.191*

−0.045 −0.082

0.695

[10]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

177 0.638

−0.001 −0.001 25.009

−0.073** −0.084***

−2.945*** 4.329***

−0.095 0.125

0.591 −12.155**

2.577***

0.765

[13] d

Random-effects panel GLS

177 0.523

−0.004** −0.002 34.901

−0.049 −0.083

(Continued )

−5.967*** 3.639***

dropped dropped

dropped dropped

1.430***

0.144

[14] e

Fixed-effects panel LSDV

177 0.220

−0.001 dropped 0.665***

−0.138**

−0.107**

DETERMINANTS OF FDI 9.3

303

C H A P T E R 9

304

[8]

−2.494 −3.524**

−0.926 −0.637

1.433

−0.392

−1.823*

0.099 3.454***

−1.404*** 0.347

−1.952** −1.684

−1.118** 0.160

0.037

−1.056

2.471*

−0.681

0.311

−0.607

0.846

−0.198 −0.419 −1.973* −0.668

−1.352

−1.097 −1.033 −0.411 −2.782***

−1.945** −1.264** −3.630*** −1.733**

−1.119* 1.584**

[11] a

0.291

−1.134* −0.309

[10]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−0.640 0.977

[9]

Cluster-robust WLS [Quality level]

Control for individual and time effects (No control) Individual −1.083* Time 1.095 Estimator (OLS) FE −0.775 RE −1.064 SLS −2.120** GMM −1.879** Model type (Total FDI model) Bilateral 0.171 Form of dependent variable (Exact numeric value) Log −0.665 Type of FDI variable (Annual net inflow) Annual gross inflow −1.675** Cumulative gross value or 0.103 stock Annual net or gross inflow −0.710 per capita Cumulative net value per 3.609*** capita Annual net inflow to GDP −1.665 etc. Other FDI variables 0.028

Meta-independent variable (Default)/Model

Clusterrobust OLS

−0.414

−1.659

3.101***

−1.784*

−1.928*** −0.012

−0.209

−0.318

−1.161* −0.961 −1.807*** −2.045***

−0.868 0.788

[12]

Multilevel mixed effects RML

−0.391

−1.645

dropped

dropped

dropped

dropped

−1.884* 3.088**

−0.190*** −6.506***

dropped

−6.523***

−0.800** −1.083*** −1.470*** −3.627***

−1.454*** −1.360**

[14] e

Fixed-effects panel LSDV

−1.967** −0.044

−0.149

−0.361

−1.195* −0.956 −1.783*** −2.094***

−0.826 0.682

[13] d

Random-effects panel GLS

9

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

Table 9.3 contd.

C H A P T E R

FOREIGN DIRECT INVESTMENT

177 0.644

177 −

−0.043** 0.116 186.280

−0.041** 0.147 −125.489 177 0.630

−2.763** −3.271** −4.302*** −2.410** −3.364***

−2.093 −2.701** −3.328*** −2.427*** −3.565***

177 0.370

−0.041** 0.113 196.427

−2.892** −3.411** −4.448*** −2.548** −3.442***

177 0.011

0.083 dropped 11.924***

−4.063** −4.481*** −5.552*** −3.994*** −4.165***

b

a

Notes:

Excluding two estimates collected from Döhrn (2000) and Jensen (2002) that do not report the composition of host target countries.

Breusch-Pagan test: χ2=0.68, p=0.205

c Hausman test fails to meet the asymptotic assumptions.

d Breusch-Pagan test: χ2=3.73, p=0.027

e Hausman test fails to meet the asymptotic assumptions.

Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 9.2 for

definition and descriptive statistics of meta-independent variables.

177 0.426

K R2

177 0.450

0.000 0.077 270.740

−4.862*** −2.997** −3.945*** −3.736*** −3.879***

Type of transition variable (General transition indicators) Liberalization −1.778 −0.191 Enterprise reform −1.847 −0.485 Competition policy −2.815** −1.930* Privatization −1.286 0.225 Other transition indicators −2.747** −0.856 Degree of freedom and research quality √Degree of freedom −0.051** −0.051*** Quality level 0.176 − Intercept −22.209 31.225

DETERMINANTS OF FDI 9.3

305

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

and higher statistical significances, studies with more non-EU advanced countries as FDI suppliers report smaller effect sizes and lower statistical significances. Second, as suggested by the quantitative synthesis of the empirical results reported in Table 9.1, a notable result of the MRA herein is the large difference between the panel data and the time series data. Estimates of the time series data analysis: that is, single country studies, are larger by approximately 0.6–0.7 in terms of the PCC relative to the panel data analysis as a benchmark. At the same time, studies using cross-sectional data report statistically significant lower estimates for both PCCs and t values as compared to panel data studies. Although an overview of the original papers would tempt us to conclude that researchers were obliged to work with cross-sectional data during the early years of transition, mainly due to the unavailability and/or the incredibility of regionwide datasets,6 we examined whether the estimation period was associated with increased FDI performance and found no relationship between them in the MRA. This provided evidence that the effect is entirely attributable to differ­ ences in the data type. Third, the choice of estimator also greatly affects the estimation results. As com­ pared to the benchmark estimator: that is, OLS, more reflective estimators, such as FE, 2SLS, and GMM that pay more attention to possible biases in the estimates due to the individual effects of host target countries or to simultaneous causation between FDI performance and FDI determinants, tend to present a more conservative assessment of the effect size and statistical power. Since we can expect that there would be endogeneity between FDI performance and economic transition, this MRA result suggests that one must tackle the issue explicitly. Fourth, it seems that the choice of FDI variable type does not cause a large vari­ ance in the effect size or the statistical significance of the FDI variables. In other words, contrary to all expectations, the difference in the type of FDI variable does not give rise to large heterogeneity among the whole set of studies. Only studies using annual gross FDI inflow as the dependent variable are likely to report smaller effect sizes and lower statistical significances of economic transition. At the same time, the choice of transition variable type does not bring about a large significant difference in the PCC (Table 9.3, Panel a). This result seems to be consistent with the previous discussion, which pointed out the homogeneous population of transition variables, partly reflecting the fact that they are largely in reference to or compiled from EBRD transition indicators/sub-indicators. It is well known that there appears to be a strong positive correlation between those variables that are devised to indi­ cate the progress of economic reforms in CEE and FSU countries.7 However, the choice of transition variable seems to exert a certain influence on the statistical sig­ nificance: that is, the t value (Table 9.3, Panel b). As opposed to aggregated general transition indicators, functionally segmented transition variables act to reduce the statistical power of estimates. In addition to the above findings, Table 9.3 also suggests that the degree of free­ dom for estimates: that is, the number of samples, has a mild negative effect on the empirical evaluations of transition-specific FDI determinants. Accordingly, studies 306

DETERMINANTS OF FDI

9.3

with larger sample sizes, ceteris paribus, tend to assign lower values to transitional factors for stimulating foreign business, thus drawing conservative conclusions con­ cerning the causality between economic transition and FDI performance in CEE and FSU countries. Other meta-independent variables, such as the estimation period, con­ trol for individual and time effects, and, in all but a few cases, the form of depend­ ent variables is not statistically estimated at the 10% level of significance, reflecting the fact that these characteristics do not cause heterogeneity among individual stud­ ies under our meta-analysis. 9.3.4 Assessment of publication selection bias In aggregating the results of the relevant literature that examines the determinants of FDI in CEE and FSU countries, we must keep in mind that no empirical study is exempt from PSB. The objective of this final analytical section is to find the magnitude of PSB and attempt to grasp the true effect of the transition variables in question by removing the influence of PSB. Looking at the transition-related variables in Figure 9.6, if the true effect exists around zero, then the ratio of the positive-versus-negative estimates becomes 157:22, which strongly rejects the null hypothesis that the ratio is 50:50 (z = 10.090, p = 0.000); therefore, type I PSB is strongly suspected to be present in the existing litera­ ture. However, if the syn­ thesized value of 0.166 obtained from the randomeffects model reported in Table 9.1 is used as an approximate value of the true effect, the collected estimates herein have a ratio of 85:94; accord­ ingly, the null hypothesis is not rejected (z = −0.673, p = 0.749). In this case, we can see a relatively sym­ metric and triangular distri­ bution of the collected estimates in the figure; thus, the possibility of type I PSB is considered to be low. Next, looking at the Galbraith plot, we can confirm Figure 9.6 Funnel plot of partial correlation coefficients of collected esti­ that the presence of type II mates of determinants of FDI (K=179) PSB is highly likely in Note: Solid line indicates the synthesized value of 0.166 obtained from the this research field. For the random-effects model reported in Table 9.1 . 307

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

transition-specific variables in Figure 9.7, only 71 of the 179 estimates show t values within the range of ±1.96 or two-sided crit­ ical values at the 5% sig­ nificance level. This result strongly rejects the null hypothesis that the rate as a percentage of total estimations is 95% (z = 33.969, p = 0.000). Even based on the assumption that the synthesized effect size of 0.166 stands as the true effect, the correspond­ ing result also rejects the null hypothesis that estimates in which statistics Figure 9.7 Galbraith plot of t values collected estimates of determinants of |(k-th estimate – the true FDI (K=179) effect)/SEk| 　exceed the Note: Solid lines indicate the thresholds of two-sided critical values at the 5% critical value of 1.96 significance level ±1.96. account for 5% of all esti­ mates (z = 21.623, p = 0.000). All too often, empirical papers cling to more statistically significant results and, thus, are contaminated by type II PSB. This holds true for our case. Finally, we examined the two types of PSB and attempted to determine whether genuine empirical evidence is present by estimating the meta-regression models specially developed for this purpose. Table 9.4 summarizes the results. As Panels a and b of the table show, the null hypothesis that the intercept term β0 is equal to zero is rejected in four of five models; this supports the view that both types of PSB have thoroughly prevailed in the selected studies. At the same time, in Panel a, the null hypothesis that the coefficient of the inverse of standard error β1 is zero is rejected only in one of five models, meaning that genuine evidence would exist in the collected estimates in a very limited way, despite the fact that Panel c shows that the coefficient of the inverse of standard error β1 is statistically significantly different from zero in all five models. Detecting the true effect of the transitionrelated FDI determinant seems to be difficult due to the existence of strong PSB in this field. All things considered, we conclude that the empirical results reported in the per­ tinent literature that examined the FDI-inducement power of economic transition have provided limited empirical evidence to prove a nonzero FDI-inducing effect. Even if we assume a nonzero effect, its magnitude would be in the range of 0.0354 to 0.0498: this is really small, according to Cohen’s criteria. 308

DETERMINANTS OF FDI

9.3

Table 9.4 Meta-regression analysis of publication selection in the studies on deter­ minants of FDI (a) FAT (type I PSB)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 9

Estimator

OLS

Model Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0) K R2

Multilevel Clustermixed effects robust OLS RML [1]

2.5883***

−0.0123* 179 0.010

[2] 2.5883***

−0.0123 179 0.010

Cluster-robust random-effects panel GLS [4] a

[3] 2.4629***

−0.0051 179 −

Cluster-robust fixed-effects panel LSDV

2.5002***

−0.0069 179 0.010

[5] b −0.4579

0.1450 179 0.010

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v)

Estimator

OLS

Model Intercept (H0: β0=0) 1/SE K R2

Multilevel Clustermixed effects robust OLS RML [6]

2.8752*** −0.0145** 179 0.017

[7] 2.8752*** −0.0145 179 0.017

Cluster-robust random-effects panel GLS [9] c

[8] 2.8625*** −0.0118 179 −

2.8962*** −0.0136 179 0.017

Cluster-robust fixed-effects panel LSDV [10] d 1.1587 0.0742 179 0.017

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Estimator Model SE 1/SE (H0: β1=0) K R2

OLS

Multilevel Clustermixed effects robust OLS RML

[11]

[12]

14.8219*** 0.0354***

14.8219*** 0.0354**

179 0.392

179 0.392

Random-effects panel ML

[13] 9.3796*** 0.0498** 179 −

Population-averaged panel GEE

[14] 9.3796*** 0.0498*** 179 −

[15] 11.7769*** 0.0434** 179 −

Notes: a Breusch-Pagan test: χ2=29.70, p=0.000 b Hausman test: χ2=4.97, p=0.026 c Breusch-Pagan test: χ2=47.08, p=0.000 d Hausman test: χ2=2.15, p=0.139 Robust standard errors are used for hypothesis testing except for Model [14]. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

309

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

9.4 MACROECONOMIC IMPACTS OF FDI IN TRANSITION ECONOMIES This section aims to examine the impacts of FDI inflow on macroeconomic growth in transition economies. To this end, Subsection 9.4.1 gives an overview of selected studies for meta-analysis. Subsection 9.4.2 demonstrates the synthesis of collected estimates. Subsection 9.4.3 performs meta-regression analysis to explore the observed heterogen­ eity between studies. Then, Subsection 9.4.4 assesses publication selection bias in the extant literature. 9.4.1 Overview of selected studies for meta-analysis In accordance with the literature selection method described in Section 9.2, we selected a total of 31 studies that contain estimates suitable for meta-analysis of macroeconomic impacts of FDI in transition economies.8 In the Economic Survey of Europe 2001, the United Nations Economic Commission for Europe pointed out that “studies of the impact of FDI on GDP in the transition economies are lacking” (UNECE 2001, p. 204) given the background of the scarcity of time series data available for empirical analysis and poor investment results throughout the 1990s. However, this academic vacuum at the beginning of the new century has been largely filled by subsequent research efforts. According to our survey results, Barrell and Holland 2000 was a pioneering study that empirically examined the macroeconomic effect of FDI in transition economies. Barrell and Holland reported a positive and statistically significant correlation between FDI and the total value added per worker in the manufacturing sectors in Hungary, Poland, and the Czech Republic. Since its publication, empirical works in this study area have been published constantly, with Elmawazini et al. 2018 being the latest. However, the target countries for the 31 selected studies are heavily dis­ torted toward a handful of nations. In fact, the total number of country observations covered in these studies is 338, of which 57.7% (195 observations) deal with the 10 CEE countries that joined the EU in either 2004 or 2007. Meanwhile, other CEE countries and FSU states, excluding the three Baltic countries, accounted for only 16.7% (57 observations) and 23.4% (79 observations), respectively. Lyroudi et al. (2004) and Apergis et al. (2008) also included Mongolia in their target countries, and Acharya and Nuriev (2016) treated five transition economies in Asia as well. Empirical analysis in the 31 selected studies covers the 26 years from 1989 to 2014 as a whole. The average estimation period of collected estimates is 11.7 years (median: 11.5; standard deviation: 3.6). Twenty-three studies used panel data, while eight studies employed time series data. Twenty-five studies used GDP as the bench­ mark index for the macroeconomic variable to be introduced in the left-hand side of their respective regression models. The remaining six studies dealt with either the gross value added to the manufacturing industry, the gross industrial production, or the sectoral value added in order to measure macroeconomic growth in their target coun­ tries. As for the scale of economic growth, 10 studies adopted the level of output

310

MACROECONOMIC IMPACTS OF FDI

9.4

volume, 10 studies chose the change in output volume, eight studies used the level of productivity, and the remaining two selected the change in the productivity level. With regard to the FDI variable, which is to be introduced together with other vari­ ables in the right-hand side of the regression models, there is more variation in the types. In fact, the FDI to GDP ratio, which is the most widely used variable type, was adopted by ten studies. This type is followed by annual capital inflow (nine studies), cumulative investment value (seven studies), and cumulative investment per capita or worker (four studies). The FDI to the total value added ratio, the FDI to the gross fixed capital formation ratio, the growth rate, and other variables were adopted by one or two studies for their empirical analyses. From these 31 studies, we collected a total of 172 estimates (5.5 per study on average). According to our bold classification of the empirical results of these stud­ ies, 16 studies reported positive and statistically significant macroeconomic impacts of FDI, while Mencinger (2003) took a pessimistic view of the role of FDI in macroeconomic growth with negative and significant estimates of the FDI variable. The remaining 14 studies either detected no significant macroeconomic impact of FDI or reported that the FDI variable was not statistically robust. We conjecture that the above-mentioned differences in empirical methodologies resulted in such mixed results among the relevant studies. In the following subsections, we further explore this point using the meta-analytic techniques mentioned in Chapter 1. 9.4.2 Meta-synthesis Figure 9.8 illustrates a frequency distribution of the PCC and that of the t value using 172 estimates collected from the aforementioned 31 studies. As Panel a of this figure shows, the PCC shows a sharp-pointed distribution with a mean of 0.180 and a median of 0.173. Thus, the Shapiro-Wilk test rejects the null hypothesis of normality at the 1% significance level (V = 3.808, p = 0.001). According to Doucouliagos’ guidelines for the study field of FDI and economic growth (2011), 26.2% (45 estimates) found no practical relationship (|r| < 0.103) between FDI and macroeconomic growth in transition econ­ omies, while 29.7% (51 estimates), 25.6% (44 estimates), and the remaining 18.6% (32 estimates) reported a small effect (0.103 ≤ |r| ≤ 0.214), a medium effect (0.214 < |r| ≤ 0.338), and a large effect (0.338 < |r|), respectively. Meanwhile, as seen in Panel b of the same figure, the t value shows a skewed distribution toward the positive direction longwise with a mean of 2.311 and a median of 2.323. Accordingly, the Shapiro-Wilk normality test strongly rejects the null hypothesis (V = 11.985, p = 0.000) again. The estimates with respective absolute r values that are equal to or exceed the threshold of 1.96 account for 57.6% (99 estimates) of the total. Therefore, it can be said that the above 31 studies as a whole emphasize the presence of statistically significant and prac­ tically meaningful effects of FDI on macroeconomic growth in CEE and FSU countries. The estimation period of each study varied significantly, and it is possible that this difference might have had a certain influence on the empirical results. In Figure 9.9, however, we found that the collected estimates show a flat trend in chronological order. In fact, according to the approximately straight lines drawn in this figure, the 311

C H A P T E R 9

C H A P T E R

FOREIGN DIRECT INVESTMENT

9

Figure 9.8 Distribution of partial correlation coefficients and t values of the collected estimates of macroeco­ nomic impacts of FDI (K=172) Notes a

Shapiro–Wilk normality test: V=3.808, p=0.001

b

Shapiro–Wilk normality test: V=11.985, p=0.000

Figure 9.9 Chronological order of partial correlation coefficients and t values of the collected estimates of macroeconomic impacts of FDI (K=172) Note: Figures in parentheses beneath the regression coefficients of the approximate straight line are robust standard errors.

coefficient of the average year of estimation period (yr) is estimated to be positive but statistically insignificant for both cases of the PCC and t value. To examine this point more strictly, we tested the influence of estimation period on empirical results in the meta-regression analysis. 312

MACROECONOMIC IMPACTS OF FDI

9.4

Table 9.5 performs a synthesis of the collected estimates. In addition to the overall synthesis results shown on the top line, this table also reports results focusing on the differences in data types and benchmark indexes for and types of the macroeconomic variable, as well as the type of FDI variable in light of the discussion in the previous subsection. As shown in Column a of the table, which reports the synthesis results of the PCC, the homogeneity test rejects the null hypothesis in every case; thus, the synthesized effect size, Rr , of the random-effects model is adopted as the reference value. Here, the synthesized PCC of all studies is 0.186, with statistical significance at the 1% level. The presence of a statistically significant positive macroeconomic effect of FDI can be found in all conditions, with the only exception being a study in which the change in productivity level was adopted as a type of macroeconomic variable. However, the magnitude of the synthesized effect size remarkably differs between subjects of comparison. More specifically, studies that conducted a time series analysis tended to report a larger positive effect of FDI on macroeconomic growth than did those performing a panel data analysis (0.280 vs. 0.163). The same applies to the relationship between the output level and change indexes (0.268 vs. 0.191) and that between cumulative investment value and other types of FDI vari­ ables (0.337 vs. 0.116–0.182). Column b of Table 9.5 shows the results of the combination of the t values. Here, we can see that the combined t value, Tw , that is weighted according to the quality level of the study is substantially lower than the unconditionally combined t value, Tu . This result suggests that there may be a strongly negative correlation between the quality level of the study and the reported t value. Furthermore, the fsN in the right column of the same table shows a sufficiently large value, except for one case. This means that, even taking into consideration the presence of unpublished working papers that have been omitted from our meta-analysis, the overall research implications obtained from the 31 selected studies cannot easily be dismissed. 9.4.3 Meta-regression analysis As indicated in Table 9.5, the empirical evidence concerning the macroeconomic impact of FDI in transition economies is likely to be greatly affected by study condi­ tions and quality. In order to scrutinize this issue more rigidly, we estimated a metaregression model that takes either the PCC or the t value of a collected estimate as the dependent variable. Table 9.6 lists the names, definitions, and descriptive statis­ tics of meta-independent variables to be introduced on the right-hand side of the regression model. As this table shows, we examined whether and how empirical evi­ dence from the existing literature is affected by differences in the composition of target countries, the estimation period, the data type, the estimator, the benchmark index/type for the macroeconomic variable, the type of FDI variable, and the degree of freedom, as well as the quality level of the study. Table 9.7 reports the estimation results. As shown in the table, estimates of several meta-independent variables significantly vary with the choice of estimator. Thus, assuming that meta-independent variables that are statistically significant and have 313

C H A P T E R 9

314

Fixed-effect model a

Notes:

a Null hypothesis: The synthesized effect size is zero.

b Null hypothesis: Effect sizes are homogeneous.

*** and ** denote statistical significance at the 1% and 5% levels, respectively.

All studies 172 0.184*** (a) Comparison in terms of data type Studies that employ panel data 138 0.155*** Studies that employ time series data 34 0.406*** (b) Comparison in terms of the benchmark index of macroeconomic variable Studies that use GDP as the benchmark 115 0.198*** index of macroeconomic variable Studies that use a non-GDP index 57 0.168*** (c) Comparison in terms of the type of macroeconomic variable Studies that adopt the level of output 43 0.357*** volume Studies that adopt the change in output 42 0.201*** volume Studies that adopt the level of productivity 67 0.153*** Studies that adopt the change in product20 0.005 ivity level (e) Comparison in terms of FDI variable Studies that use FDI to GDP ratio 37 0.147*** Studies that use cumulative investment 32 0.342*** value Studies that use annual capital inflow 54 0.220*** Studies that use other types of FDI 49 0.106*** variable

Number of estimates (K) 1313.218*** 856.384*** 261.071*** 859.940*** 446.130*** 300.953*** 580.521*** 136.689*** 17.907

125.654*** 361.724*** 401.535*** 183.918***

0.186*** 0.163*** 0.280*** 0.184*** 0.188*** 0.268*** 0.191*** 0.179*** 0.005

0.141*** 0.337*** 0.182*** 0.116***

Randomeffects Test of homogeneity b model a

14.538*** 10.970***

11.044*** 25.918***

17.862*** 0.403

17.885***

20.359***

20.227***

22.819***

24.701*** 18.392***

30.303***

Unweighted combination

3.372*** 1.685**

2.246** 4.508***

2.687*** 0.081

4.481***

4.440***

3.081***

4.820***

4.447*** 3.847***

5.601***

Weighted combination

1.678 1.430

2.444 2.950

2.450 −0.250

2.649

2.413

2.500

1.798

2.300 2.412

2.323

Median of t values

(b) Combination of t values

9

(a) Synthesis of PCCs

Table 9.5 Synthesis of collected estimates of macroeconomic impacts of FDI

4164 2130

1631 7912

7832 −19

4923

6543

8561

22014

30977 4216

58194

Failsafe N (fsN)

C H A P T E R

FOREIGN DIRECT INVESTMENT

MACROECONOMIC IMPACTS OF FDI

9.4

Table 9.6 Name, definition, and descriptive statistics of meta-independent variables used in meta-regression analysis of heterogeneity among studies of macroeconomic impacts of FDI Descriptive statistics Variable name

Definition

Proportion of CEE 10 EU countries Proportion of other CEE countries Proportion of non CEE and FSU counries First year of estimation Length of estimation Time series data

Proportion of CEE 10 EU countries in target countries a Proportion of other CEE countries in target countries Proportion of non CEE and FSU countries in target countries

OLS Non-GDP index Changes Productivity Cumulative invest­ ment value Annual capital inflow FDI to gross value added FDI to gross fixed capital formation Cumulative FDI per capita Growth rate Other FDI variables √Degree of freedom Quality level

First year of estimation period Years of estimation period 1 = if time series data is employed for empirical analysis, 0 = otherwise 1 = if ordinary least squares estimator is used for estimation, 0 = otherwise 1 = if non-GDP index is used as macroeconomic variable, 0 = otherwise 1 = if macroeconomic variable is expressed in change rate, 0 = otherwise 1 = if macroeconomic variable is measured in productivity, 0 = otherwise 1 = if cumulative investment value is used as the type of FDI variable, 0 = otherwise 1 = if annual capital inflow is used as the type of FDI variable, 0 = otherwise 1 = if FDI to gross value added ratio is used as the type of FDI variable, 0 = otherwise 1 = if FDI to gross fixed capital formation ratio is used as the type of FDI variable, 0 = otherwise 1 = if cumulative FDI per capita (or worker) is used as the type of FDI variable, 0 = otherwise 1 = if growth rate is used as the type of FDI variable, 0 = otherwise 1 = if other FDI variable is used, 0 = otherwise Root of degree of freedom of the estimated model Ten-point scale of the quality level of the study b

Mean

Median

S.D.

0.674

0.857 0.385

0.180

0.100 0.294

0.008

0.000 0.034

1994.744

1995

2.806

11.686 0.198

11.5 0

3.638 0.399

0.250

0

0.434

0.331

0

0.472

0.360

0

0.482

0.506

1

0.501

0.186

0

0.390

0.314

0

0.465

0.087

0

0.283

0.006 0.140 0.012 0.041 11.752 4.953

0 0 0

0.076 0.348 0.108

0 0.198 11.045 5.279 5

2.181

Notes: a CEE EU countries denote the 10 Central and Eastern European countries that joined the European Union either in 2004 or 2007. b See Chapter 1 for more details.

the same sign in at least four of the seven models constitute statistically robust esti­ mation results, we indicate the following four points about factors that generate sys­ tematic differences among the empirical results regarding the macroeconomic impact of FDI in transition economies.

315

C H A P T E R 9

316

[1]

0.994

−0.020 −0.007 0.008 −0.113 −0.044 0.101 −0.192** −0.114 0.069 −0.023 −0.054

1.143*

−0.006 0.010 −0.026 −0.084 0.049 −0.191** −0.054 0.176* −0.012 0.013

0.090 0.380***

0.488***

−0.068 0.063

0.128

−0.104 −0.078

−0.041

−0.173** 0.082 0.142

0.020

−0.035

0.003 0.000

0.336

0.199**

0.065

[5]

Multilevel mixed effects RML

−0.018

0.101

−0.009 0.007

0.230*

0.225

0.200

0.163

[4] a

0.142

[3]

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

0.040

[2]

Cluster-robust WLS [Quality level]

Composition of target countries (FSU) Proportion of CEE 10 EU 0.124 countries Proportion of other CEE 0.266 countries Proportion of non CEE and 0.752 FSU countries Estimation period First year of estimation −0.008 Length of estimation 0.007 Data type (Panel data) Time series data −0.136 Estimator (non-OLS estimators) OLS −0.090 Benchmark index of macroeconomic variable (GDP) Non-GDP index 0.038 Type of macroeconomic variable Changes (Level) −0.185** Productivity (Output) −0.094 Type of FDI variable (FDI to GDP) Cumulative investment 0.195* value −0.009 Annual capital inflow FDI to gross value added 0.037

Meta-independent variable (Default)/Model

Clusterrobust OLS

−0.079 0.058

0.091

−0.085 −0.085

−0.051

0.036

−0.025

0.008 0.000

0.105

0.177**

0.056

[6] b

Random-effects panel GLS

−0.023*** dropped

0.068***

dropped dropped

dropped

0.035*

dropped

0.043*** 0.027***

−1.178***

0.109***

−0.010*

[7] c

Fixed-effects panel LSDV

9

Estimator (Analytical weight in parentheses)

(a) Dependent variable — PCC

Table 9.7 Meta-regression analysis of heterogeneity among studies of macroeconomic impacts of FDI

C H A P T E R

FOREIGN DIRECT INVESTMENT

[8]

Clusterrobust OLS

172 0.296

−0.004** −0.022 15.127

−0.006 − 12.583

−3.979 −0.127 0.137

−2.495 −0.063 0.093 −1.175 −0.103 1.312

11.363

−0.022 0.140 −0.416 −0.537 0.663

−0.691 −2.474**

0.025

0.122

2.737

1.626

0.574

[11] a

1.763

[10]

0.383

[9]

162 0.457

−0.008* −0.020 17.178

−0.038 0.489** 0.188

0.381**

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

172 0.370

−0.159** 0.595*** −0.033

−0.149** 0.673*** 0.077

172 0.417

0.033

0.126

Cluster-robust WLS [Quality level]

Composition of target countries (FSU countries) Proportion of CEE 10 EU 1.725 countries Proportion of other CEE 2.180 countries Proportion of non CEE and 6.415 FSU countries Estimation period First year of estimation −0.063 Length of estimation 0.076 Data type (Panel data)

Time series data −1.770 Estimator (non-OLS estimators)

OLS −0.981 Non-GDP index 0.755

Meta-independent variable (Default)/Model

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

K R2

FDI to gross fixed capital 0.124 formation

Cumulative FDI per capita −0.137* Growth rate 0.621*** Other FDI variables 0.009 Degree of freedom and research quality √Degree of freedom −0.007 Quality level −0.011 Intercept 16.884

0.163 0.044

−1.049

0.014 −0.011

2.033

1.892

1.031

[12]

Multilevel mixed effects RML

172 −

−0.006*** −0.005 −6.188

−0.136* 0.431** 0.082

0.113

0.447 −0.051

−0.999

0.065 −0.024

−0.154

1.846

0.849

[13] d

Random-effects panel GLS

172 0.154

−0.006*** −0.003 −15.515

−0.118 0.380* 0.121

0.083

(Continued )

0.704 dropped

dropped

0.407*** 0.216***

−1.400

2.197***

0.129

[14] e

Fixed-effects panel LSDV

172 0.001

−0.007*** dropped −86.232***

−0.105**

dropped

0.311***

−0.073***

MACROECONOMIC IMPACTS OF FDI 9.4

317

C H A P T E R 9

318

[8]

172 0.313

−0.504 −0.047 0.350 −1.949* 5.190** −1.049

−0.328 0.484 1.491 −1.192 6.047*** 0.494

172 0.429

172 0.364

0.030 −0.366 129.001

0.977

2.255*

0.024 − 45.703

−2.352** −1.972**

[10]

162 0.498

0.006 −0.369 252.693

0.385 3.738 1.877

0.320 5.591*** 6.027***

7.739***

1.643 0.764

[11] a

Cluster-robust Cluster-robust WLS [N] WLS [1/SE]

−2.098** −1.407*

[9]

Cluster-robust WLS [Quality level]

172 −

0.041 −0.121 −24.512

−1.420 3.700 0.102

−0.467 0.803 1.668

1.970

−1.001 −1.655

[12]

Multilevel mixed effects RML

172 0.193

0.037 −0.085 −126.833

−1.333 2.996 0.465

−0.656 0.711 1.140

1.288

−0.768 −1.788

[13] d

Random-effects panel GLS

172 0.042

0.012 dropped −813.363***

0.200** dropped 2.572***

0.429*** dropped 1.296***

−1.233***

dropped dropped

[14] e

Fixed-effects panel LSDV

a

Notes:

Excluding 10 estimates collected from Varamini and Kalash (2010) that report only t values of their estimation results.

b Breusch-Pagan test: χ2=1.87, p=0.086

c Hausman test: χ2=27.79, p=0.006 d Breusch-Pagan test: χ2=1.51, p=0.109 e Hausman test: χ2=17.79, p=0.087 Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coefficient at the 1%, 5%, and 10% levels, respectively. See Table 9.6 for definition and descriptive statistics of meta-independent variables.

K R2

Type of macroeconomic variable Changes (Level) −1.853* Productivity (Output) −1.755* Type of FDI variable (FDI to GDP) Cumulative investment 2.714* value Annual capital inflow 0.188 FDI to gross value added 0.580 FDI to gross fixed capital 1.562 formation Cumulative FDI per capita −0.995 Growth rate 5.493** Other FDI variables −0.164 Degree of freedom and research quality √Degree of freedom 0.032 Quality level −0.186 Intercept 127.138

Meta-independent variable (Default)/Model

Clusterrobust OLS

9

Estimator (Analytical weight in parentheses)

(b) Dependent variable — t value

Table 9.7 contd.

C H A P T E R

FOREIGN DIRECT INVESTMENT

MACROECONOMIC IMPACTS OF FDI

9.4

First, the composition of target countries has a certain influence on the estimates collected from the relevant studies. Actually, the proportion of other CEE countries is estimated with a significant and positive sign in four of the seven models in Panel a of Table 9.7, suggesting that the inclusion of Croatia and non-EU CEE countries tends to produce empirical evidence with a larger effect size than those of CEE-10 EU members and FSU countries. Second, the selection of FDI variable types is an important factor in explaining the differences between the selected studies. Namely, more positive macroeconomic effects tend to be detected in estimations in which the cumulative value or growth rate of FDI is adopted as compared to those in which the FDI to GDP ratio is taken as an independ­ ent variable. Meanwhile, the estimates reported by studies where the cumulative FDI per capita is adopted are more negative concerning the size of the FDI impact. These results strongly indicate that the choice of FDI variable is critical for empirically assess­ ing the macroeconomic impacts of FDI into CEE and FSU countries. Third, the degree of freedom is also an influential factor in empirical evaluations of the macroeconomic impacts of FDI in the existing literature. The square root of the degree of freedom is estimated to be robust and negative in Panel a of Table 9.7. In other words, when other conditions remain the same, studies with larger sample sizes tend to give lower evaluations of the magnitude of FDI’s effect on growth. We surmise that more precise studies, in terms of empirical data, have a tendency to draw conservative conclu­ sions concerning the caus­ ality between FDI and macroeconomic growth in transition economies. Fourth, in contrast with the three factors above, the difference in the estimation period, data type, estimator, benchmark index/type of the macroeconomic vari­ able, and quality level of the study does not signifi­ cantly affect empirical results in the selected works when we control for a series of study conditions simultaneously. From Panel b of Table 9.7, we also found that no factor system­ atically influences the statis- Figure 9.10 tical significance of the Funnel plot of partial correlation coefficients of collected esti­ of macroeconomic impacts of FDI (K=172) collected estimates except mates Note: Solid line indicates the synthesized value of 0.186 obtained from the for the type of FDI variable. random-effects model reported in Table 9.5 . 319

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

Overall, the findings obtained from the meta-regression analysis are noteworthy for understanding the relationship between the study conditions and empirical evi­ dence in the existing literature on the macroeconomic impacts of FDI in transition economies. 9.4.4 Assessment of publication selection bias As a final step of meta-analysis, we tested for PSB and the presence of genuine empirical evidence of the growth-enhancing effect of FDI in the selected studies. First, we looked at a funnel plot of the collected estimates’ PCCs against the respect­ ive inverse of the standard errors in Figure 9.10. This figure shows the expected shape, which can be seen among studies of a given research subject without publication selec­ tion bias. In other words, we can see a relatively symmetrical and triangular distribution of the collected estimates in the figure if the synthesized value of 0.186 obtained from the random-effects model reported in Table 9.5 is used as an approximate value of the true effect. In fact, if the true effect is assumed to be close to the synthesized value, the collected esti­ mates are divided into a ratio of 79:93, with a value of 0.186 being the threshold; accordingly, the null hypothesis is not rejected (z = −1.068, p = 0.286). Thus, the possibil­ ity of type I PSB is con­ sidered to be low.9 Next, looking at the Galbraith plot in Figure 9.11, we can confirm that 73 of the 172 estimates show a t value that is within the range of ±1.96 or the two-sided critical Figure 9.11 values at the 5% signifiGalbraith plot of t values collected estimates of macroeconomic cance level. This result impacts of FDI (K=172)

Note: Solid lines indicate the thresholds of two-sided critical values at the 5% strongly rejects the null

significance level ±1.96.

hypothesis that the rate as a percentage of total estimations is 95% (z = 31.627, p = 0.000). Even with the assumption that the synthe­ sized effect size of 0.186 stands as the true effect, the corresponding result also rejects the null hypothesis that estimates in which the statistic |(the k-th estimatetrue effectÞ=SEk j exceeds the critical value of 1.96 account for 5% of all estimates 320

MACROECONOMIC IMPACTS OF FDI

9.4

Table 9.8 Meta-regression analysis of publication selection in the studies on macro­ economic impacts of FDI (a) FAT (type I PSB)-PET test (Equation: t=β0+β1(1/SE)+v)

C H A P T E R 9

Estimator

OLS

K R2

Cluster-robust random-effects panel GLS

Cluster-robust fixed-effects panel LSDV

[1]

[2]

[3]

[4] a

[5] b

0.2705

0.2705

0.9080

1.0609

0.7172

0.1653***

0.1653**

0.1624

0.1559

0.1291**

Model Intercept (FAT: H0: β0=0) 1/SE (PET: H0: β1=0)

Multilevel Clustermixed effects robust OLS RML

172 0.089

172 0.089

172 −

172 0.089

172 0.105

(b) Test of type II PSB (Equation: |t|=β0+β1(1/SE)+v)

Estimator

OLS

K R2

Cluster-robust random-effects panel GLS

Cluster-robust fixed-effects panel LSDV

[7]

[8]

[9] c

[10] d

1.1951***

1.1951*

1.3347

1.3900

0.7853

0.1240***

0.1240*

0.1680

0.1666

0.1572

Model Intercept (H0: β0=0) 1/SE

Multilevel Clustermixed effects robust OLS RML [6]

172 0.067

172 0.067

172 −

172 0.067

172 0.067

(c) PEESE approach (Equation: t=β0SE+β1(1/SE)+v)

Estimator Model SE 1/SE (H0: β1=0) K R2

OLS

Multilevel Clustermixed effects robust OLS RML

[11]

[12]

−0.7252 0.1879***

−0.7252 0.1879***

172 0.443

172 0.443

Random-effects panel ML

[13] 1.3113 0.2083** 172 −

Population-averaged panel GEE

[14] 1.3113 0.2083*** 172 −

[15] 0.7752 0.2104*** 172 −

Notes: a Breusch-Pagan test: χ2=100.27, p=0.000 b Hausman test: χ2=0.89, p=0.345 c Breusch-Pagan test: χ2=164.78, p=0.000 d Hausman test: χ2=0.26, p=0.611 Robust standard errors are used for hypothesis testing except for Model [14]. ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively.

321

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

(z = 19.382, p = 0.000). Therefore, the presence of type II PSB is likely in this research field. Finally, we examined the two types of PSB and the presence of genuine empirical evidence by estimating a set of meta-regression models specially developed for this purpose. Table 9.8 summarizes the results. As Panel a of the table shows, the null hypothesis that the intercept term β0 is equal to zero is not rejected in any of the five models, while three of the five models do not reject the same null hypotheses in Panel b. Therefore, we assert that both types I and II PSB are less likely in the lit­ erature despite the findings from the Galbraith plot mentioned above. Furthermore, in Panel a, the null hypothesis that the coefficient of the inverse of standard error β1 is zero is rejected in three of five models, meaning that genuine evidence does exist in the collected estimates. Moreover, Panel c shows that the coefficient of the inverse of standard error β1 is statistically significantly different from zero in all five models. Therefore, we can say that the true value of the macroeconomic impact of FDI should be in the range of 0.1879 to 0.2104. Judging from the above assessments, we conclude that the empirical results reported in previous literature that examined the macroeconomic impact of FDI in transition economies as a whole have successfully provided empirical evidence to prove a nonzero FDI effect and, according to the Doucouliagos’ criteria, its impact is positive but limited to a small size.

9.5 CONCLUSIONS In this chapter, we conducted a meta-analysis of the literature that empirically exam­ ined either the determinants of FDI in CEE and FSU countries or the causality between FDI and macroeconomic growth in the region over the past quarter century. The study of FDI in transition economies has made substantial progress in the second half of 1990s and the first decades of the new century. The related litera­ ture published during these years carried out various empirical analyses, reflecting the differences in their authors’ motivation, research aims, and theoretical ground­ ing. Nevertheless, the meta-analysis in this chapter has revealed that empirical results reported in the preceding studies indicate the close relationship between the progress of transition to a market economy and FDI and the positive effect of FDI on macroeconomic growth in the literature as a whole. In fact, the metasynthesis of estimates extracted from the selected studies shows that the synthe­ sized PCC for the study of the determinants of FDI and that of the macroeco­ nomic impacts of FDI are 0.166 and 0.186, respectively, and the combined t values, weighted by the quality level of the studies, reach as high as 5.774 and 5.601, respectively (see Tables 9.1 and 9.5). These synthesis results are notable in comparison with studies of the rest of the world. For instance, according to the meta-analysis of macroeconomic impacts of FDI by Doucouliagos et al. (2010), which covered most countries and regions of the world, the synthesized PCC of

322

NOTES

880 estimates collected from 108 studies is 0.12. If a comparison is allowed, we could say that CEE and FSU countries have benefited from FDI in terms of macroeconomic growth 1.55 times greater than the world average, indicating the high quality of foreign capital invested into the post-communist economies and the excellent absorption capacity of local firms and citizens in the former socialist bloc. Nevertheless, the results of the MRA in this chapter have unveiled that empirical evaluations in transition literature strongly depend on study conditions. Actually, we found that the composition of target countries, data type, control for time effects, choice of estimator, type of FDI, and transition variables are particularly important factors that explain the heterogeneity of the collected estimates in the study of FDI determinants, while the composition of target countries and type of FDI variable sys­ tematically influence the empirical results reported in the study of macroeconomic impacts of FDI. We also found that the degree of freedom greatly affects the empirical results in the selected studies. The fact that the square root of the degree of freedom is estimated to be robust and negative in both study areas implies that econometrical evaluations of the FDI-inducing effect of the transition process and the growthpromoting effect of FDI in transition economies may become more conservative in tandem with further improvements in the precision of empirical analyses. Furthermore, according to our assessment of the publication selection bias, studies of macroeconomic impacts of FDI contain genuine empirical evidence. In contrast, existing works have not yet proved the true effect in the study of determinants of FDI, due to the strong tendency of publication selection bias in the literature. It is likely that empirical evaluations of the effect of progress in transition on FDI might be revised downward in the future with the further accumulation of highly precise estimates. We hope that there will be more development and improvement in this research field so as to capture the true effect.

ACKNOWLEDGMENTS This chapter is an updated and combined version of Iwasaki and Tokunaga 2014 and Tokunaga and Iwasaki 2017. We thank Arun Agrawal, Josef C. Brada, Hristos Dou­ couliagos, Masaaki Kuboniwa, Tom D. Stanley, Taku Suzuki, and Miklós Szanyi for their helpful comments and suggestions on the earlier version of this paper.

NOTES

1 The final literature search was conducted in January 2019. 2 For details of the selected studies, see Iwasaki and Tokunaga 2019, app. 1, list a; app. 2. 3 See Stanley and Doucouliagos 2012, pp. 16–17.

323

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT

4 We include this in our MRA because controlling for unobserved host country het­ erogeneity and common time effects may reduce the variation of transition-related variables (Overesch and Wamser 2010). 5 See de Mooij and Ederveen 2003, 2008; Feld and Heckemeyer 2011. 6 As is clearly shown in Iwasaki and Tokunaga (2019, app. 2), studies that employ cross-sectional data are found mainly in the early original papers selected for our meta-analysis. 7 According to the IMF (2000, pp. 133–137), EBRD transition indicators and two alternatives (the liberalization index and the index of institutional quality) are highly correlated, which reflects the similarity of the concepts measured. 8 For details of the selected studies, see Iwasaki and Tokunaga 2019, app. 1, list b; app. 3. 9 In contrast, if we assume that the true effect exists around zero, the ratio of the posi­ tive vs. the negative estimates becomes 140:32, which strongly rejects the null hypothesis that the ratio is 50:50 (z = 8.235, p = 0.000). In this case, type I PSB is strongly suspected.

REFERENCES Acharya, Sanjaya, and Shamshimukhamed Nuriev (2016) Role of public investment in growth and poverty reduction in transition economies. Journal of Reviews on Global Economics, 5, pp. 310–326. Aghion, Philippe, and Peter W. Howitt (1997) Endogeneous Growth Theory. MIT Press: Cambridge, Mass. Apergis, Nicholas, Katerina Lyroudi, and Athanasios Vamvakidis (2008) The relationship between foreign direct investment and economic growth: Evidence from transition economies. Transition Studies Review, 15(1), pp. 37–51. Bandelj, Nina (2002) Embedded economies: Social relations as determinants of foreign direct investment in Central and Eastern Europe. Social Forces, 81(2), pp. 409–444. Bandelj, Nina (2008) From Communists to Foreign Capitalists: The Social Foundations of Foreign Direct Investment in Postsocialist Europe. Princeton University Press: Princeton, NJ, and Oxford. Bangert, David, and József Poór (1993) Foreign involvement in the Hungarian economy: Its impact on human resource management. International Journal of Human Resource Man­ agement, 4(4), pp. 817–840. Barrell, Ray, and Dawn Holland (2000) Foreign direct investment and enterprise restructur­ ing in Central Europe. Economics of Transition, 8(2), pp. 477–504. Borensztein, E., J. De Gregorio, and J.-W. Lee (1998) How does foreign direct investment affect economic growth? Journal of International Economics, 45(1), pp. 115–135. Botrić, Valerija, and Lorena Škuflić (2006) Main determinants of foreign direct investment in the Southeast European countries. Transition Studies Review, 13(2), pp. 359–377. Carlin, Wendy, and Michael Landesmann (1997) From theory into practice? Restructuring and dynamism in transition economies. Oxford Review of Economic Policy, 13(2), pp. 77–105. Cass, Fergus (2007) Attracting FDI to transition countries: The use of incentives and promo­ tion agencies. Transnational Corporations, 16(2), pp. 77–122.

324

REFERENCES Claessens, Stijn, Daniel Oks, and Rossana Polastri (2000) Capital flows to Central and East­ ern Europe and the Former Soviet Union. In Sebastian Edwards (ed.), Capital Flows and the Emerging Economies: Theory, Evidence, and Controversies. University of Chicago Press: Chicago, pp. 299–340. Cohen, J. (1988) Statistical Power Analysis for the Behavioral Sciences. Second edition, Lawrence Erlbaum Associates: Hillsdale, NJ. Dauti, Bardhyl (2015a) Determinants of foreign direct investment in South East European countries and new member states of European Union countries. Economic and Business Review, 17(1), pp. 93–115. Dauti, Bardhyl (2015b) Determinants of foreign direct investment in transition economies, with special reference to Macedonia: Evidence from gravity model. South East European Journal of Economics and Business, 10(2), pp. 7–28. de Mooij, Ruud A., and Sjef Ederveen (2003) Taxation and foreign direct investment: A synthesis of empirical research. International Tax and Public Finance, 10(6), pp. 673– 693. de Mooij, Ruud A., and Sjef Ederveen (2008) Corporate tax elasticities: A reader’s guide to empirical findings. Oxford Review of Economic Policy, 24(4), pp. 680–697. Deichmann, Joel I. (2013) Origins of foreign direct investment in Croatia: Application of an expanded gravity model. In Anastasios Karasavvoglou and Persefoni Polychronidou (eds.), Balkan and Eastern European Countries in the Midst of the Global Economic Crisis. Physica-Verlag: Heidelberg, pp. 3–21. Derado, Dražen (2013) Determinants of FDI in transition countries and estimation of the potential level of Croatian FDI. Financial Theory and Practice, 37(3), pp. 227–258. Djankov, Simeon, and Peter Murrell (2002) Enterprise restructuring in transition: A quantitative survey. Journal of Economic Literature, 40(3), pp. 739–792. Döhrn, Roland (2000) Foreign direct investment in Russia: An engine of structural adjust­ ment? In Paul J. J. Welfens and Evgeny Gavrilenkov (eds.), Restructuring, Stabilizing and Modernizing the New Russia: Economic and Institutional issues. Springer: Berlin and New York, pp. 127–148. Doucouliagos, Hristos (2011) How large is large? Preliminary and relative guidelines for interpreting partial correlations in economics. School Working Paper No. SWP 2011/5, School of Accounting, Economics and Finance, Faculty of Business and Law, Deakin University: Melbourne. Doucouliagos, Hristos, Janto Haman, and T. D. Stanley (2012) Pay for performance and cor­ porate governance reform. Industrial Relations, 51(3), pp. 670–703. Doucouliagos, Hristos, Sasi Iamsiraroj, and Mehm et Ali Ulubasoglu (2010) Foreign direct investment and economic growth: A real relationship or wishful thinking? School Working Paper No. SWP 2010/14, School of Accounting, Economics and Finance, Faculty of Business and Law, Deakin University: Melbourne. 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), pp. 285–306. Easterly, William (1993) How much do distortions affect growth? Journal of Monetary Eco­ nomics, 32(2), pp. 187–212. EBRD (European Bank for Reconstruction and Development) (1998) Transition Report 1998: Financial Sector in Transition. EBRD: London.

325

C H A P T E R 9

C H A P T E R 9

FOREIGN DIRECT INVESTMENT Elmawazini, Khaled, Elias G. Saleeby, Ahmed Ibn el Farouk, and Bashayer Al-Naser (2018) Tri­ partite decomposition of labor productivity growth, FDI and human development: Evidence from transition economies. Economic Change and Restructuring, 51(2), pp. 153–171. Estrin, S., and M. Uvalic (2014) FDI into transition economies: Are the Balkans different? Economics of Transition, 22(2), pp. 281–312. Feld, Lars P., and Jost H. Heckemeyer (2011) FDI and taxation: A meta-study. Journal of Economic Surveys, 25(2), pp. 233–272. Garibaldi, Pietro, Nada Mora, Ratna Sahay, and Jeromin Zettelmeyer (2001) What moves capital to transition economies? IMF Staff Papers, 48 (special issue), pp. 109–145. Grossman, Gene M., and Elhanan Helpman (1991) Innovation and Growth in the Global Economy. MIT Press: Cambridge, Mass. Hengel, Erin (2011) Determinants of FDI location in South East Europe (SEE). OECD Journal: General Papers, 2, pp. 91–104. IMF (International Monetary Fund) (2000) World Economic Outlook: Focus on Transition Economies. IMF: Washington DC. Iwasaki, Ichiro, and Masahiro Tokunaga (2014) Macroeconomic impacts of FDI in transition economies: A meta-analysis. World Development, 61, pp. 53–69. Iwasaki, Ichiro, and Masahiro Tokunaga (2019) The determinants and macroeconomic impacts of foreign direct investment in transition economies. CEI Working Paper No. 2019-8, Center for Economic Institutions, Institute of Economic Research of Hitotsubashi University: Kunitachi, Tokyo. Jensen, Camila (2006) Foreign direct investment and economic transition: Panacea or pain killer? Europe-Asia Studies, 58(6), pp. 881–902. Jensen, Nathan (2002) Economic reform, state capture, and international investment in tran­ sition economies. Journal of International Development, 14(7), pp. 973–977. Kosová, Renáta (2010) Do foreign firms crowd out domestic firms? Evidence from the Czech Republic. Review of Economics and Statistics, 92(4), pp. 861–881. Lankes, Hans-Peter, and A. J. Venables (1996) Foreign direct investment in economic tran­ sition: The changing pattern of investments. Economics of Transition, 4(2), pp. 331–347. Lansbury, Melanie, Nigel Pain, and Katerina Smidkova (1996) Foreign direct investment in Central Europe since 1990: An econometric study. National Institute Economic Review, 156(1), pp. 104–114. Lee, Hanhee (2015) Foreign direct investment in North Korea and the effect of special eco­ nomic zones: Learning from transition economies. Journal of Economic Development, 40 (2), pp. 35–56. Lyroudi, Katerina, John Papanastasiou, and Athanasios Vamvakidis (2004) Foreign direct investment and economic growth in transition economies. South-Eastern Europe Journal of Economics, 2(1), pp. 97–110. Mencinger, Jože (2003) Does foreign direct investment always enhance economic growth? Kyklos, 56(4), pp. 491–508. Mišun, Jan, and Vladimír Tomšík (2002) Does foreign direct investment crowd in or crowd out domestic investment? Eastern European Economics, 40(2), pp. 38–56. Myant, Martin, and Jan Drahokoupil (2012) Transition indicators of the European Bank for Reconstruction and Development: A doubtful guide to economic success. Competition and Change, 16(1), pp. 69–75. Overesch, Michael, and Georg Wamser (2010) The effects of company taxation in EU accession countries on German FDI. Economics of Transition, 18(3), pp. 429–457.

326

REFERENCES Selowsky, Marcelo, and Ricardo Martin (1997) Policy performance and output growth in the transition economies. American Economic Review (Papers and Proceedings), 87(2), pp. 349–353. Shukurov, Sobir (2016) Determinants of FDI in transition economies: The case of CIS countries. Journal of International and Global Economic Studies, 9(1), pp. 75–94. Sinn, Hans-Werner, and Alfons J. Weichenrieder (1997) Foreign direct investment, political resentment and the privatization process in Eastern Europe. Economic Policy, 12(24), pp. 177–210. Solow, Robert (1956) A contribution to the theory of economic growth. Quarterly Journal of Economics, 70(1), pp. 65–94. Stanley, T. D., and Hristos Doucouliagos (2012) Meta-Regression Analysis in Economics and Business. Routledge: London and New York. Stern, Nicholas (1997) The transition in Eastern Europe and the former Soviet Union: Some strategic lessons from the experience of 25 countries over six years. In Salvatore Zec­ chini (ed.), Lessons from the Economic Transition: Central and Eastern Europe in the 1990s. Kluwer Academic Publishers: Dordrecht, Boston, and London, pp. 35–57. Tokunaga, Masahiro, and Ichiro Iwasaki (2017) The determinants of foreign direct invest­ ment in transition economies: A meta-analysis. World Economy, 40(12), pp. 2771–2831. UNECE (United Nations Economic Commission for Europe) (2001) Economic Survey of Europe 2001 No. 1. United Nations: New York and Geneva.

327

C H A P T E R 9

10

Regime change and environmental reform Masahiro Tokunaga

10.1 A SHORT HISTORY OF ENVIRONMENTAL REFORM IN CEE COUNTRIES Central and Eastern European (CEE) countries would be the “black sheep” of a European society aiming for compatibility of economic growth and environmental protection. Historically speaking, it has been pointed out that the decades of centrally planned economic systems and iron-fisted dictatorships in these countries not only left both the national economy and the civil society in shambles but also left devas­ tating scars on the natural environment. It is still fresh in our minds that the border area of the former German Democratic Republic (GDR), Poland, and the Czech Republic was once called the “Black Triangle” and was considered a Europe-wide air pollution culprit before the process of radical transformation began in CEE soci­ ety. There is a commonly held view that Eastern bloc countries were responsible for serious environmental degradation and pollution across the region compared to advanced Western countries that had promoted industrial restructuring in a move toward a resource-saving and energy-efficient society after the oil shocks of the 1970s, which led to the amelioration of industrial pollution associated with economic growth. Some researchers who delved deeply into the environmental problems under the socialist regimes have attempted to convey the critical importance of the natural environment to readers by using a shocking and impressive phrase such as “ecocide” (Feshbach and Friendly Jr. 1992; McCuen and Swanson 1993). Hence, important political concerns for CEE countries after the revolutions of 1989 include not only the transition to market economies (economic reform) and the promotion of demo­ cratic political systems (political reform), but also the effective implementation of environmental policies necessary to solve various environmental issues (environmen­ tal reform). Moreover, the European Union (EU) encouraged the candidate CEE countries to handle these three reforms in parallel from the early phase of accession talks. EU–CEE environmental policy coordination dates back to the 1991 Associ­ ation Agreements signed with Poland, Hungary, and the former Czechoslovakia prior to their official accession negotiations. These agreements provided for the gradual elimination of trade barriers and CEE national legislation that mirrored that of the

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

EU (Caddy 1997a). Afterwards, the EU required that the candidate countries observe the environmental laws and regulations of the EU and fundamentally revise their own laws and regulations to ensure consistency. However, in looking back on a quarter-century history of social changes in the CEE countries, their environmental reforms traveled a bumpy road with many twists and turns just as it did for their economic and political reforms. Following are a couple of symbolic events after the revolutions of 1989. First, there was an emergence and expansion of new types of environmental issues observed in advanced economies: events not previously experienced in the former socialist societies occurred without warning in association with the transition to capitalism. This was true despite the fact that the economic crisis after regime change caused the erosion of heavy industries and the withering of “dirty industries” in the period of economic dislocation unintentionally led to a rapid reduction in environmental stress in many CEE countries. Although observers focus on discussing different aspects, those newly emerging ecological disturbances are broadly classi­ fied: (1) the emergence of automobile pollution and growing traffic problems due to the rapid development of motorization, (2) increasing pressure for estate develop­ ment and inappropriate use caused by land privatization or reinstitution mainly in agricultural and forest land, and (3) the aggravation of waste problems associated with the rise of westernized lifestyles and consumer behavior such as burgeoning household garbage and conflict over waste disposal and landfill sites. All of these issues were in place in the marketization process after regime change and spread in CEE societies, partly because developing and changing environmental authorities were not able to respond to them quickly. It has been frequently pointed out, there­ fore, that a laissez-faire economic reform based on market fundamentalism would have an adverse impact on the environment (Manser 1993; Scrieciu and Stringer 2008), and more than a few studies have persuasively demonstrated this risk in an empirical way (Gille 2004; Jorgenson et al. 2012; Křenová and Kindlmann 2014; Pryde 1995; Staddon 1999; Sumelius et al. 2005). In addition to this, many obser­ vers worried about the prospect of the CEE countries’ environmental reforms with the revival of once-frozen or withdrawn massive development programs (highway construction, dam building, nuclear power station projects, etc.) as a result of antiestablishment movements during the last stage of the socialist regimes. Second, an anti-environmentalism campaign was developed in the 1990s in some CEE countries: environmental non-governmental organizations (ENGOs) and envir­ onmental activists who enjoyed high public support for their anti-establishment orientation and had helped to achieve the revolutions of 1989 lost their esteem and popularity within just a few years of this key historical event. In many cases, accom­ plishing their biggest goal caused a rift among members, creating distrust of one another within and between environmental groups. After the first free elections failed to demonstrate strong popular support for green parties and their candidates, some distanced themselves from environmental groups and went back to work as usual; others started their careers as government officials who had to address all sorts of issues in the face of domestic economies that exacerbated many social problems. 330

A SHORT HISTORY OF ENVIRONMENTAL REFORM

10.1

The result was that environmental reform was low on the political agenda in the CEE region (Frankland 1995; Jancar-Webster 1993a, 1998). Most of all, in the territory of the former Czechoslovakia, where the environmental movement played such a prominent role in the 1989 regime change that it could be called the “green velvet revolution” and nonviolent and democratic movements realized many tangible positive outcomes such as newly built environmental authorities and a new range of environ­ mental laws (Podoba 1998), a vociferous anti-environmentalism campaign against ENGOs and green activists in the 1990s, which was reminiscent of political suppres­ sion under the old regimes, shocked those people who believed that the democratiza­ tion of politics and the infiltration of democracy into society should promote environmental reforms. After the secession and independence of the Czech and Slovak Republics, two political leaders, Václav Klaus (Czech Republic) and Vladimír Mečiar (Slovakia), implicitly achieved consensus on anti-environmentalism, in spite of their divergent views on politics and economy (Watzman 1992). On the one hand, in Slovakia, which had political friction with Hungary regarding the treatment of Hungarian minority groups on the borders, in addition to negotiations with the Czech Repub­ lic over secession and independence, increasing nationalistic feelings became tied to some infrastructure development projects as a measure of enhanced national prestige. For example, an anti-Slovakian label was applied to environmentalists and citizens peacefully protesting the expansion of the Mochovce nuclear power plant and the planned construction of the Gabčíkovo dam on the Danube.1 They were even severely criticized by nationalistic propaganda led by the Slovak media that spread a rumor that the protestors served as foreign agent provocateurs (Podoba 1998; Snajdr 2001). On the other hand, the Czech government seemed unable to allow for flexible policy responses to newly emerging environmental issues along with the social transformation of the 1990s when Václav Klaus, a leading kingmaker in the CEE countries, served as the prime minister. His administration not only aggressively developed an anti-environmentalism campaign aimed at green activists and ENGOs, but also intervened in Czech environmental policy by expunging the concept of “sustainable development” in their official documents. That cut them off from the international environmental policy commu­ nity (Fagin 2001; Fagin and Jehlička 1998; Jehlička 2001).2 Moreover, it was revealed in the mid-1990s that three hardline ENGOs had been included on a security services list of “subversive organizations” and were to be targeted for surveillance. This scandal also brought a sense of deep disappointment among people who had regarded widespread civic movements, including ENGOs, as a testament of democratization in the Czech Republic (Fagin 2002; Jehlička 2001; Sarre and Jehlička 2007). Although it sounds paradoxical, at the same time, the Klaus administration continued to adopt tough environmental regulations for clear­ ing up traditional industrial pollution (Slocock 1996): in fact, after controlling for other potential reduction factors in the country, tighter environmental protection policies proved to be the most important reason behind the dramatic reduction in air pollutant emissions (Earnhart and Lizal 2008). 331

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

Third, a major overhaul of domestic environmental laws and regulations was requested by the EU: as remarked above, this move dates back to the early 1990s, and major CEE actors voluntarily started to harmonize a range of statutes with those of the EC/EU. Europe-wide environmental policy coordination as part of market integration or single-market establishment was officially approved in the 1994 Euro­ pean Council meeting (Essen), and the 1995 White Paper for the prospect of future membership presented a general program of actions to be undertaken and identified key measures to be adopted in the environmental sector (Caddy 1997a). At that time in Hungary, which would later come into conflict with the EU over an environmental standard as detailed below, the government was surprised and then embarrassed by the first official EU document on Hungary’s application for membership. This was due to the fact that the commission had a more comprehensive view concerning the natural environment than the White Paper had initially suggested. It was made clear that meeting only the conditions as stipulated in the White Paper would be insuffi­ cient for EU accession (Kerekes and Kiss 1998). As is commonly known, all candi­ date countries are required to accept and implement the acquis communautaire, which comprises the whole body of EU law, including the treaties, regulations, dir­ ectives, decisions, and judgments of the European Court of Justice. Membership requirements related to environmental issues are often called environmental acquis (or green acquis), which calls for the harmonization and coordination of domestic and EU-wide environmental policies. Although an acceptance of the EU environ­ mental standard was initially welcomed as a whole, early in the mid-1990s, some researchers responded with sharp criticism as they saw a true picture of this process and began to understand what it really involved (Caddy 1997b; Jancar-Webstar 1998). Among other requests, the EU asked for the revision of approximately 450 provisions of the legal system in the environmental field for just a few years, which imposed tremendous costs and burdens on the candidate countries. This legal trans­ position process was managed by small groups of senior civil servants, selected experts, and minimally engaged national parliaments (Gorton et al. 2010). If there was any parliamentary discussion, it was often fast-forwarded for approval, and new bills went through without sufficient consultation with those stakeholders who were highly likely to bear the economic burdens (Börzel 2009a; Börzel and Buzogány 2010b; Börzel and Fagan 2015; Buzogány 2009a, 2009b, 2015; Guttenbrunner 2009; Slocock 1999). The European Commission even ignored calls for a more flexible approach from the World Bank, which was concerned that the environmental acquis placed excessive burdens on applicant countries (Gorton et al. 2010). As a result, the candidate countries had no choice but to “download” the relevant laws and regulations that the EU had “uploaded” in advance (Scrieciu and Stringer 2008). This caused resentment of the EU for demanding their top-down acceptance of the environmental acquis, non-compliance with which was ubiquitous even in Western European countries.3 Nevertheless, considering the negative history of the natural environment under the socialist regimes, in CEE countries, no one could reach the political capability and policy performance with which they could directly compete with the EU as a green superpower. 332

A SHORT HISTORY OF ENVIRONMENTAL REFORM

10.1

However, it was the waste issues in Hungary that challenged the legitimacy of the EU environmental governance: the revision of domestic laws and regulations in accordance with the EU environmental standard impeded an innovative environmen­ tal policy and exacerbated the case when a serious incident occurred in the country. Hungary’s waste history, which from 1949 to the late 1980s favored preventative waste reduction and reuse policies, was ahead of the times in the sense that the West relied largely on waste dumps and incinerators. However, the EU did not acknow­ ledge this progressiveness and effectiveness observed in the past and, in practice, encouraged the country to introduce remedial end-of-pipe technologies in the provi­ sion of environmental assistance. This not only created confusion in legislation and institution building in Hungary, and thus delay, but also established a practice that might lock in a retrospective path of development (Gille 2002, 2004). The 2010 red sludge spill accident revealed the contradiction between the two entities. In Ajka, a small town in western Hungary, a huge amount of red sludge, strongly alkaline residuum from the early stage of aluminum production, leaked from storage reser­ voirs and cascaded through local villages, killing ten people. This serious industrial accident became a hot international issue in the European arena, when the CEO of the aluminum company that was responsible for the spill accident expressed the offi­ cial view that the red sludge was not a harmful substance according to the EU cri­ teria in spite of significant loss of human life and tremendous ecological damage beyond Hungary’s borders. Although Hungarian experts had concluded that the red sludge was indeed hazardous in spite of the EU norms, based on their domestic stat­ utes, and the Hungarian government accepted this conclusion, they failed to form a consensus with the EU investigation team. Thus, Hungarian representatives raised this issue in the EU Parliament (Ieda 2011). Many ecological activists in the country shared their concerns that Hungarian environmental standards were actually lowered due to the EU accession, allowing the government to weaken the regulations (Hicks 2004), which suggests a divergence of realities and the philosophy of the EU that preaches the value of sustainable development. Among other things, the European­ ization of CEE countries has revealed the essential contradiction that the EU, in fact, compels candidate countries to introduce the environmental acquis in a non­ democratic way—accept these rules or be denied membership—so as to augment the democratization of environmental policies such as information disclosure and citizen participation (Bell 2004; Gorton et al. 2010). As described again in the next section, more than a few researchers expressed critical and skeptical views on European rule adoption in the environmental field. All of the examples mentioned above suggest that environmental changes are not linearly related to political and economic dynamics. This calls for input from mul­ tiple points of view to define the complex causal relationships among these social change factors.4 In light of these observations, and based on a systematic review of previous stud­ ies that investigated a number of issues concerning regime transformation and envir­ onmental reform in CEE countries, this chapter summarizes various views on the relationship between regime transformation and environmental reform, systematically 333

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

verifying the factors in the literature that contribute to conflicting views regarding this relationship. I decided to focus on CEE countries in this chapter, even though other countries from the former Soviet Union (FSU) area also recognize the neces­ sity of carrying through on environmental reforms in the face of the legacy of envir­ onmental degradation in the twentieth century. While these FSU countries together with CEE countries are often called transition countries, they have substantially dif­ ferent ecological issues to be solved and, thus, separate aims and directions of envir­ onmental reforms to be pursued. The FSU countries have a serious disaster area, represented by the territory radioactively contaminated by the Chernobyl accident, and the desertification of the seabed of the Aral Sea in Central Asia, both of which affect multiple countries and throw the region into ecological turmoil almost perman­ ently. However, the CEE countries, which are examined in this chapter, principally deal with industrial pollution and its related problems at the local level, except for a few cases in which military facilities and operations caused environmental and health damage in the Cold War years. Besides this, after the adoption of the 1986 Single European Act, which went into force in 1987, three important articles on the environment were introduced in the European Economic Community treaty, which implied that environmental protection became the primary objective and that the EC had competences in matters of the environment with its membership states (Naka­ nishi 2016).5 This modification means that the EU still has authority and responsibil­ ity for the environmental laws and regulations of CEE member states.6 It also means that the EU stands outside the legislative powers that FSU countries maintain within their boundaries, despite the fact that their environmental policies have been so greatly influenced by the EU’s standards and norms.

10.2 OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW As argued in Chapter 1 of this book, the goal of the systematic review in this chapter is to synthesize the research evidence of the literature, not ad hoc, but in a way that enables us to look up at a tower of research, a complete view of which we cannot grasp with a narrative review only. In this section, I will give a comprehensive review of the selected studies on the subject in the following systematic review. First, Subsection 10.2.1 introduces discussions of the relationship between multi-dimensional regime changes and various environmental reforms in the transition countries and briefly describes the aim of the systematic review. Next, Subsection 10.2.2 gives an explanation of the search strategy and selection criteria for the targeted original papers to be incorp­ orated into the systematic review. Finally, Subsection 10.2.3 examines research specifi­ cations such as research topics, targeted regions/countries, time and period of analysis, and methodology of each study, as well as personnel attributes, such as authors’ affili­ ations and disciplines, academic degree, and gender, and medium characteristics, such as publication year, academic fields, and quality level of the literature.

334

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

10.2.1 Purpose of the systematic review We see great variety in the discussions about how to understand the relationship between regime change and environmental reforms in transition countries. First, there is a large gap in authors’ evaluations, due to a difference in the composition of countries being studied. This is typically seen in studies that compare the dynamics of environmental reforms in CEE countries, including the Baltics, with those in the FSU region. It is evaluated, in most cases, that there has been a synergy among eco­ nomic reforms, democratic development, and environmental improvement in the former group; however, many countries in the latter group have been perceived as facing difficulties in establishing economic reform, stable societies, and environmen­ tal protection (Górz and Kurek 2001; Missfeldt and Villavicenco 2000; Zamparutti and Gillespie 2000). Based on this point, I decided to include those papers that cover these two country groups in a common framework in the sample of the sys­ tematic review, although one must handle this issue deliberately, as mentioned in the previous section. An assessment gap due to a difference in the countries being exam­ ined is seen in studies that deal with the development of environmental governance in CEE countries. Buzogány’s two papers (2009a, 2009b), for example, analyze the emergence and the effectiveness of environmental governance in Hungary and Romania, giving the former high marks for its achievement while at the same time looking skeptically at the latter. Second, even if authors engaged in a shared research subject and found common ground on the contents and dynamics, their final conclusions often show mixed results stemming from a conflict of opinions. To cite an example, we see a divergence of views on the activities of ENGOs in the CEE countries, which has radically changed before and after the revolutions of 1989. In many countries, onceradical and dissident grassroots environmental mobilizations were fading out in the context of the institutionalization of civil society groups. They showed clear signs of shifting toward more westernized, sophisticated, and new-generation movements with professionalism as non-state actors and collaborative relations with state actors. Although almost all scholars share the awareness that these ENGOs have become mainstream across post-socialist Europe, they engage in a heated debate over the characteristics of the new movements. Some respond negatively to the process of assimilating to act on Western European values and behaviors and the resulting loss of resistibility that the CEE environmental organizations demonstrated during the final phase of totalism. Others positively assess the development of cooperative state–society relations as a success story of civil society where ENGOs can be deeply committed to the planning, implementation, and evaluation of government environmental policies toward sustainable development societies with established democracies (Börzel and Buzogány 2010a; Carmin 2010; Carmin and Fagan 2010; Císař 2010; Fagan 2005, 2006, 2010; Fagin and Jehlička 1998; Gliński 1998; JancarWebster 1998; Snajdr 1998; Waller 1998). Third, differences in the awareness of regime change’s impact on the environment in CEE countries stand out even in the same authors, in cases where they have

335

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

another research topic or focus on a different period of analysis regarding the same topic. For instance, according to studies that examine the effects of political, struc­ tural, and economic changes on environmental quality by testing the environmental Kuznets curve (EKC), their econometric regression model provides a solid indication of the EKC validity for airborne pollutants, in the sense that there is already clear evi­ dence that emissions have been reduced during economic growth in these countries. At the same time, they find much less evidence for the EKC hypothesis when the selected environmental indicators are used with regard to surface water quality (Archibald et al. 2004, 2009). Another example is Buzogány’s work, where the author gives thought to the emergence of environmental governance in Romania with positive expectancy in a recent paper (2015), as compared with the harsh perspective on that in the early devel­ opment stage in the previous paper (2009b). In this respect, a most intriguing case seems to be the swaying views of Adam Fagan (Fagin), who has been investigating the development of ENGOs in the Czech Republic, among others, and other CEE countries for over the last two decades, and delivers as many as 14 works (including co-authored ones) to the sample for this systematic review. Fagan gave positive feedback regarding their activities in the first half of the 1990s (see Fagin 1994, for an example), but the initial euphoria turned to frustration, and the author started to criticize the policy process on environmentalism with a well-placed barb. Then, Fagan discussed the issues in a more moderate way than before and now shows a favorable attitude again toward environmental activism in the CEE countries through a recent comparative study of the development of multilevel environmental governance in non-EU post-socialist countries such as Bosnia–Herzegovina and Serbia (Fagan 2006, 2010; Fagan and Sircar 2010; Fagin 1999, 2001; Fagin and Jehlička 1998; Fagin and Tickle 2002). Reviewing his own past works critically, he admitted that there were a couple of fundamental problems with the earlier, somewhat negative analysis. The author thus explained this turnaround: Viewed from a different functional perspective, namely the extent to which new NGOs represent effective conduits for progressive change, including new forms of governance interaction, Europeanization and the reformulation of state power, the legacy of donor intervention and assistance is judged somewhat dif­ ferently. (Fagan and Sircar 2011, p. 302) The main purpose of the systematic review here is to quantitatively analyze those research and literature attributes that would exert an influence over the divergence of views and their changes over time. Much has been highlighted in Chapter 1, where we argue the merits of our systematic review as compared with normative and descriptive literature reviews. Not only can the tower of research be comprehended in an objective fashion, but one could also discern some plausible biases that would be attributable to the characteristics of each study, thus having a huge impact upon their conclusions. No one could assert that any great scientific work is exempt from unintentional biases, contrary to everyone’s best efforts. The essence of a systematic review can be ascribed to such a possibility that these biases will be grasped 336

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

objectively and quantitatively, and, also, researchers with a similar research task have an opportunity to identify their positions in advance through a comparative review with others. In this respect, we discover an evolvability or scalability of systematic-review-based comparative studies. 10.2.2 Procedure for selecting literature To search for past research works to target for a systematic review, some large-scale academic literature databases are commonly used as a first step. In this chapter, I chose the Web of Science, maintained by Clarivate Analytics, because it gives access to mul­ tiple databases that reference cross-disciplinary research. First, I put “transition*” and “environment*” as main keywords and then limited the scope of my searches by coun­ try/region and discipline, because thousands of papers turned up in the first round.7 However, the situation was the same, in that it brought a flood of works: worse still, most of the papers detected were not related to environmental reforms in the context of multi-dimensional social transition. This resulted because those two keywords inherently cover wide semantic domains that indicate diverse phenomena. Additionally, both are part of various academic terms: in natural science, “transition” means “a successive change in ecosystem,” to cite an example, and economists often use “environment” to refer to business terms such as “management environment” and “investment environment.” To address this issue, I replaced the wildcard “*” with some additional words and obtained two best keywords as a result of trial and error: that is “environmen­ tal transition” and “environmental reform.” Furthermore, I used “ecological moderniz(s)ation” as an additional keyword, the concept of which was established and has been advocated as one of the main targets of European countries by a group of environmental social scientists. From the 1980s onward, it has prevailed not only in several academic fields but also in industrial society and among Euro­ pean public authorities seeking a new strategic framework for economic growth that would be in harmony with the environment (see Tokunaga 2010). Among these publications, the relevant literature on CEE countries was chosen to make a base collection of studies under systematic review. I also searched the references in these studies and book chapters in leading collections by a method of referen­ cing references, or a snowballing approach. As a result, I collected 384 English articles8 that were published between 1989 and 2015 as potential candidates for the following literature survey.9 From these samples, articles that explicitly examined the impact of regime change on the environment in CEE countries in academic publications were selected for a total of 243 papers (see Figure 10.1).10 If a journal paper was included in a book with minor changes or in the case of replication,11 the first one published was chosen; otherwise, a later published work with an update and/or some revisions was preferentially considered. Note that I removed those studies that were (1) monographic publications and academic books with a limited number of authors; (2) unpublished research manuscripts, such as discussion papers and working paper 337

C H A P T E R 10

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Figure 10.1 Publication year and number of publications of the literature (left: number of selection, right: number of search) Note: Figures in parentheses denote the numbers of studies where non-CEE countries, mainly FSU countries and the former Yugoslavia, are studied.

series; and (3) included in reports published by international organizations and ENGOs. In this last case, whether we could put these works and purely academic papers in the same arena is disputable, despite the fact that researchers and experts write the reports in many cases. Although it seems far from an exhaustive collection of the relevant literature, reviewers are allowed to narrow the scope of works under examination in a clearly defined way. Where the literature is known to be vast, tighter exclusion criteria could be reasonably adopted, as in meta-analysis studies (see Stanley and Doucouliagos 2012, ch. 2). Furthermore, random—that is, unbiased—omission of samples seems inevitable no matter how hard one tries to avoid it. 10.2.3 Basic characteristics of selected studies for systematic review Figure 10.1 tells us that the studies selected for this systematic review were largely published from 1993 to 2004: they amount to almost three-quarters of the total. An increase and decrease in the number of studies are heavily influenced by the appear­ ance in academic environmental journals of special issues related to CEE countries12 and publications of academic collections on the CEE environment.13 In the following subsections, referring to Figure 10.2, I outline the four major attributes of the sample studies in which research topics, authors’ profiles, publication media, and any other information are coded in a certain way. 338

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

C H A P T E R

Basic characteristics of selected studies

Figure 10.2

10

339

340

Journal article 64%

Positive 38%

g

f

e

d

c

b

a

Negative 29%

Moderate 62%

Positive 9%

Env. studies 56%

Impact factor over 1.0 63%

Negative 62%

Moderate 34%

Positive 4%

Evaluation of environmental issues under socialism

Impact factor below 1.0 25%

No impact factor 12%

Research levelf

(d) Other attributes

Positive 21%

Moderate 63%

Negative 16%

Evaluation of environmental movements

(c) Media attributes

Development

studies

3%

Politics 13%

Economics 9%

Others Geography 6% 8% Sociology 5%

Research area e

Affirmative 29% Reserved 53%

Skeptical 18%

g

Appraisal of market economy

Reserved 39%

g

Affirmative 60%

Skeptical 1%

Appraisal of democracy

Broadly classified research topics are not strictly compatible with finely classified research topics due to a difference in coding.

The median is obtained by dividing the sum of the first year and the final year of analysis by two.

Quantitative analysis, statistical analysis, and descriptive analysis denote a study with an econometrical method, a study using statistical database, and a study with a qualitative method,

respectively. Classification is based on information regarding the names of authors’ affiliations, their academic degrees, and their past research activities. Classification is based on information from the 2012 Journal Citation Reports for academic journals and keywords (tags) available from search engines and publishers for books. This refers to the 2012 Journal Citation Reports for a total of 155 papers published in academic journals. This refers to authors’ normative value judgments on market principles (marketization) or democracy (democratization), not to their analytical assessment of the progress of economic or political reforms.

Notes:

Figure 10.2 contd.

Moderate 49%

Negative 13%

Early 2000s 27%

Late 1990s 32%

Evaluation of environmental institutions

Late 2000s 12%

Early Early 2010s 1990s 14% 15%

Publication year

10

Evaluation of EU accession/support

Book chapter 36%

Publication source

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

(A) RESEARCH ATTRIBUTES

First, the most studies selected focus on the ten CEE countries that joined the EU in 2004 and 2007. From a total of 243 papers, non-EU CEE countries, FSU countries excluding the Baltics, and non-CEE countries (China and Hong Kong, Vietnam, Spain, Portugal, and Greece) are included in the samples of 43, 34, and 3 papers, respectively. Single-country studies, in which a specific country is the focus, are a majority, amounting to 151 works; the remaining 92 works are multi-country stud­ ies, in which two or more countries are analyzed. The nations most frequently stud­ ied among the former group are Poland (37 papers), the Czech Republic (26), and Hungary (22); followed by Romania (16), Bulgaria (13), Slovakia (9), Lithuania (5), Bosnia–Herzegovina (5), the former GDR (4), and Estonia (3). These figures reveal that the most polluted countries even in Eastern Europe—Poland and the Czech Republic—capture the attention of researchers. However, the former GDR is found far less than expected, although it was part of the “Black Triangle,” with Poland and the Czech Republic. The worst legacy of environmental degradation across Eastern Europe has become an issue facing the West by the unification of Germany (Juergen­ smeyer et al. 1991). The policies to solve it are, thus, formulated both by the federal government of Germany and the EU (Boehmer-Christiansen 1992, 1998; Wilson and Wilson 2002). The former GDR, or Germany’s new Länder, has been seen as a special case, which would blur the facts for analysts; therefore, they showed less interest in this part of CEE countries. The least interest was shown in Southeastern Europe and the Western Balkan region in the 1990s. Only in the 2000s did we see some progress in research in this area. A large reorganization of chapters in the collection edited by Carter and Turn­ ock (1993, 2002), known as a representative academic work in the field of the envir­ onmental problems in post-socialist Europe, shows an early shift of research interest to those countries that were not included in analyses before. That is to say, a country that was part of Yugoslavia in their first edition of 1993 is investigated separately in the revised edition of 2002. Croatia, one of the former countries of Yugoslavia, clearly stated their intention to join the EU in the early 2000s. The EU itself started to be directly involved in the peace process and has, since 2000, assumed much broader influence in post-conflict Bosnia–Herzegovina, which is considered to have brought about a growing trend of research interest regarding the environmental affairs in this area. After the end of the worst civil war in postwar European history, these countries were facing a specific situation in which military operations during the war directly caused severe environmental degradation on both sides. This also led to an amelioration of the industrial pollution, due to an economic crisis at the same time (Clarke 2002a, 2002b). Furthermore, even after the international commu­ nity and local ENGOs started trying to restore the environmentally devastated area by reconstructing the region damaged by war, the weak governance capacity of state authorities, the nationalistic party politics in regions, and the ethnically divided and fractured communities coalesced to thwart the emergence of green politics in the region (Castán Broto et al. 2009; Fagan 2006, 2010; Fagan and Sircar 2010). At the

341

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

same time, the EU has exerted much greater influence on the formal compliance pro­ cess with the environment acquis in some EU candidate countries than they attempted in the CEE EU membership countries in the past, which has been strictly examined from the viewpoint of the validity and effectiveness of the EU environ­ mental standards (Fagan and Sircar 2015; Obradovic-Wochnik and Dodds 2015). Second, when we direct our attention to the research topics in each study, it turns out that there is great concern for the real state of affairs of various environmental damage, such as air and water pollution. Many researchers also focus on the environ­ mental policies and movements dedicated to solving these issues and the international cooperation and assistance supporting these activities through financial and human resources. Most of these studies dealing with the latter topics discuss the dynamics of domestic actors in CEE countries as well as a series of measures of the EU as the biggest outside donor. The EU’s financial and human assistance for the environmental problems of CEE countries had been given as part of PHARE (Poland and Hungary Action for Restructuring of the Economy) program, with the aim of restructuring domestic economies and moving to market-oriented institutions. When EU accession was on top of the political agenda, more political emphasis was then put on how to promptly follow obligatory procedures to join the EU rather than how to effectively put the environmental policies into practice. Researchers’ awareness of issues has changed with these movements, and they placed the relationship of the EU accession process and environmental reforms at the center of discussion, along with the impact of market economy transition and political democratization, in which they debate the results and meaning as well as the problems and lessons of EU accession in terms of ecological improvement (Börzel 2009b; Carmin and VanDeveer 2005). The association with the EU was questioned after new member states were in place in 2004 and 2007, and two issues are mainly the focus of attention in aca­ demia. The first issue is related to the EU’s posture toward non-EU CEE countries in Southeastern Europe and the Western Balkan region. Although incompatibility with the EU environmental acquis was a common problem for many of the CEE countries who joined the EU in 2004 and 2007, there was little room for negotiation between the parties. In the case of Croatia, a newcomer who became a member state in 2013, the entry requirements have been shown to be inconsistently applied and more stringent than they were for other member states (Kay 2014). As described in the previous section, the inflexible attitudes of the EU bureaucracy have been viewed with praise or censure, and, in fact, more than a few authors have expressed sharp and bitter criticism. The other issue in dispute is the introduction of Natura 2000, a unique EU-wide network of protected areas, which aims to maintain Euro­ pean biodiversity based on two directives: Bird Directive 79/409/EEC and Habitats Directive 92/43/EEC. To enlarge and coordinate the nature protection system at the pan-European level, the EU demands that member states consult in advance with various stakeholders (local inhabitants, municipalities, assemblies, land-owners, farmers, foresters, tourism agencies, ENGOs, etc.) and coordinate their incompatible interests, because the Natura 2000 project requires a change in the dominant model of ownership and access for protected areas. However, even in Western European 342

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

countries that have a great deal of experience and achievement in nature protection and conservation biology, the implementation of Nature 2000 was criticized for being a top-down approach that insufficiently engaged stakeholders, leading to con­ flicts, legitimacy crises, and active opposition to the program. In many cases, this contributed to delays in designating the sites. At the same time, the EU Commission took several western member states to the European Court of Justice, citing delays and failures in the development of the protected area network (Cent et al. 2014). Despite all of this confusion, the EU requested the new membership states to strictly implement Natura 2000 and comply with the legal regulations. On the one hand, this led to harsh criticism of the EU’s obstinacy and the ineffectiveness of policy in the domestic arena (Buzogány 2009b; Grodzinska-Jurczak and Cent 2011; Kay 2014; Knorn et al. 2012, 2013; Křenová and Kindlmann 2014; Mikulcak et al. 2013; Sotirov et al. 2015; Stringer and Paavola 2013; Švajda 2008). On the other, more than a few researchers appreciate the magnitude and vision of the Natura 2000 pro­ ject (Cent et al. 2007, 2013, 2014; Evans et al. 2013; Kluvánková-Oravská et al. 2009, 2013; Niedziałkowski et al. 2012). Third, as for the research period, except for 18 studies that are mainly concerned with movements under the old regimes before social transformation, many authors begin their analysis in 1989 or 1990, keeping in mind the wave of revolutions in CEE countries: they account for almost two-thirds of the total literature, or 153 of 243 papers. The average research period is 9.5 years: excluding the above 18 papers from our sample, it falls to 8.2 years for the remaining 225 papers. A distribution of the median of the research period indicates that over half (143 papers) of the total recorded the early 1990s; therefore, a majority of the sample studies discuss environ­ mental issues in the context of social transformation in CEE countries, keeping in mind a turbulent few years during the revolutionary period. At the same time, those studies in which the median is the late 1990s or later represent one-third (84 papers) of the total, in which Europeanization rather than transformation has been the main focus in many cases. They examine the results and lessons of environmental reforms associated with EU accession. As might be expected, the more recent a paper’s publication date, the later the research period. It is noteworthy that some recent studies do not necessarily start their analyses from the revolutions and regime changes in the CEE region, but some­ times begin empirical investigations much later. As Figure 10.3 clearly indicates, while most publications issued until the early 2000s start discussing environmental affairs around the year 1990, less than half of those published later (24 of 56 papers) show a period of analysis that ends with the 2000s or later and do not refer at all to the major events that occurred before that time, or just outline events with­ out any discussion. The other side of this seems to be that there is less interest in the environmental problems of older regimes. We can see a kind of stylized descrip­ tion pattern in the earlier studies: it begins with an overview and criticism of the environmental damage in the era of socialism and then examines changes in the environmental situation due to radical social transformation after the collapse of communism and a command economy in order to determine their significance and 343

C H A P T E R 10

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Figure 10.3 First year (○) and last year (+) of the analysis period of selected studies. Note: Vertical and horizontal lines denote first and last years of the analysis period of each study and the cumulative number of studies, respectively. They are placed in chronological order from left to right.

lessons as principal conclusions. However, this style of description has gradually dis­ appeared since the mid-2000s, and the number of those papers that touch upon past environmental issues has obviously decreased. As mentioned later in detail, authors’ views on the environmental problems in older regimes are manifested in no less than 80% of the literature up until the early 2000s (152 of 187 papers), but this figure has fallen to less than 40% since the late 2000s (22 of 56 papers). Finally, concerning analytical method, a majority use descriptive analyses, while a minority use statistical and quantitative analyses (for the differences, see Figure 10.2, n. c). At the same time, the number of empirical studies with some sort of estimators has steadily increased over the last decade. Whereas we see only five such papers (4.0%) of the total 126 publications in the 1990s, there are 17 such papers (14.5%) in the 117 samples published after 2000. (B) AUTHORS’ ATTRIBUTES

Next, we turn to the attributes of the authors of each paper. The following three points are worthy of remark. First, the total number of authors is 466, and their affiliation structure tells us that those who work for higher education institutions account for 85% of the total. Therefore, the majority of studies under systematic

344

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

review were written by professional researchers. The remaining 15% can be attrib­ uted to practitioners who serve in consulting agencies, including ENGOs, govern­ ment institutions in the CEE or Western countries, and international organizations such as the EU and the European Bank for Reconstruction and Development (EBRD); the others are two freelancers and one staff member of a political party. Then, the locations of authors’ affiliations suggest that environmental issues in CEE countries seem to be of strong interest to Western countries. Although we come across a few such cases occasionally in which a researcher from a CEE country teaches at an educational institution in the West,14 it is interesting that the academic significance of environmental studies of CEE countries has been acknowledged not only in Western Europe, which is geographically connected to the CEE countries, but also in the faraway USA, considering that any research project needs to acquire grants from their home institutions. The location where the author’s doctorate was earned was identified for a total of 304 authors: North America, Western Europe, and Eastern Europe each have approximately a 30% share, which suggests that a majority of the authors gain educational and work experience in western countries. Second, regarding researchers’ genders,15 there are 127 female authors, accounting for less than 30% of the total. As time has proceeded, the proportion has increased: spe­ cifically, female authors are included in 37 (29.4%) of 126 papers published in the 1990s, in 23 (37.7%) of 61 papers in the early 2000s, and in 32 (57.1%) of 56 papers in the late 2000s and later. A similar trend is also observable regarding the number of authors. We found that there are more singly authored than multi-authored papers in the whole sample (128 versus 115 papers), but the proportion of the latter has been rising over time: 47 (37.3%) of 126 papers in the 1990s, 30 (49.2%) of 61 papers in the early 2000s, and 38 (67.9%) of 56 papers in the late 2000s and later. The number of authors acknowledged in one paper has also been rising in recent years; the aver­ age number moved from 1.5 people in the 1990s and 1.7 people in the early 2000s to 2.7 people in the late 2000s and later. Whereas there was only one paper with 5 or more authors among 187 papers until the mid-2000s (Pickles et al. 2002), papers with more than 5 authors were found in four of 56 papers after that (Iojă et al. 2009; Knorn et al. 2012, 2013; Young et al. 2007). Many papers with multiple authors, to a maximum of 10 authors, combine the humanities and sciences to yield their research results. Natural scientists from the fields of ecology, pedology, and forestry as well as remote sensing experts contribute to the projects by analyzing the observa­ tion data from satellites. A majority of multi-authored papers (68 of 115 papers) were written by researchers who share their academic discipline, but interdisciplin­ ary works by authors with diverse academic backgrounds account for one-third of the total multi-authored papers (47 of 155 papers). In the latter case, a combination of environmental studies and economics is the largest group (20 of 47 papers), fol­ lowed by the pairing of environmental studies and sociology (eight papers), environ­ mental studies and geography (six papers), and economics and sociology (four papers). The rest include combinations of environmental studies and politics, politics and geography, politics and sociology, and sociology and geography in a few cases (no more than three papers for each). 345

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

Third, as for authors’ professional disciplines, judging from research topics in each study as well as authors’ careers (academic degrees and affiliations), interdisciplinary environmental studies rank first and constitute over one-third of the total, followed by economics, politics, and geography. While this indicates a diversity of authors’ aca­ demic backgrounds, their professional disciplines are significantly correlated with the research topics in each study, as is shown in Table 10.1. For instance, economics art­ icles show deep interest in the trends of air pollution and environmental policies. One major reason for the former theme seems to be a finely maintained database on air pollution substances and greenhouse gas emissions. Panel-formatted data are necessary for multiple-country and -year comparative studies, and, in many cases, indices of the environmental burden related to air pollution meet this precondition. As for the latter theme, there is interest in reviewing to what extent environmental policies have con­ tributed to a significant improvement in some environmental indices observed widely in the CEE region in the 1990s. It is unanimously agreed that the improvement has been achieved by closing production lines with outdated facilities in the transition to a market economy; however, econometrical analysts seem to have such a unique approach that they attempt to assess the presence and extent of the effectiveness of environmental policies, while controlling some beneficial effects of the economic depression in the early phase of transition (Archibald et al. 2004; Bluffstone 1999; Earnhart and Lizal 2008; Vukina et al. 1999). Regardless of authors’ academic back­ grounds, environmental policies have been a top concern for those studying environ­ mental issues in CEE countries. However, researchers other than economists do not

Table 10.1 Research topics (finely classified) of selected studies and authors’ academic disciplinesa, b Environmental studies General environmental issues Air pollution Water pollution Soil degradation Waste problems Transport problems Land use Environmental movements Environmental policies (public sector) Environmental management (private sector) Tourism development International cooperation Total

Politics Economics Sociology Geography Total

14 25 26 7 12 7 38 48

6 22 14 2 6 5 5 56

13 34 23 8 8 2 10 15

3 4 2 0 3 0 6 22

12 16 17 9 10 3 27 25

48 101 82 26 39 17 86 166

76

58

57

20

37

248

31 3 53

6 0 51

26 2 17

2 0 14

11 8 23

76 13 158

340

231

215

76

198

1060

Notes: a Multi-coding corresponding to finely classified research topics b Chi-square test for independence: χ2 = 149.168 and Cramér’s V: V = 0.265

346

OVERVIEWS OF SELECTED STUDIES FOR SYSTEMATIC REVIEW

10.2

seem to have the perspective of factor comparison, and more than a few papers assert­ ively conclude that the improvement in major environmental indices has been exclusively due to economic depression during the transition period (Baker 2002; Fagin and Jehlička 1998, Fagin, 2001). Turning to other fields, political and sociological papers are more likely to capture the relationship of regime transformation and environmental movements as a touchstone of democratization. Specifically, they discuss the rise and fall of ENGOs and their public participation in policy decision-making processes, the status of environ­ mental administration in the government structure of each country, and the tangible effects of environmental support from the EU (Baker and Jehlička 1998; Carmin and VanDeveer 2005; Fagan and Carmin 2011). Furthermore, geographical papers are mainly interested in the dynamics of rural and mountain areas; they deal with the impact of changes in land use in agricultural and forest lands and the progress of land privatization or reinstitution upon the natural environment (Dingsdale and Lóczy 2001; Drgona 1996; Górz and Kurek 2001; Iojă et al. 2009; Knorn et al. 2012; Sklenicka et al. 2014). Some argue for the development of ecotourism as an attempt to restore the devasted areas and revive impoverished rural economies (Mazurski 1999; Turnock 1999; Unwin 1996). (C) MEDIA ATTRIBUTES

Here we turn to the media attributes of where the selected studies for systematic review were published. Of a total of 243 papers, 155 appeared in professional journals and the remaining 88 in academic collections. According to the presence or absence and level of impact factor of the 155 journal articles based on information from the 2012 Journal Citation Reports used to quantify their research quality level, no less than 80%, or 133 articles, were published in journals with an impact factor; the remaining 22 articles were published in practical or enlightening magazines that do not have an impact factor, prob­ ably due to their publication aims and/or editorial policies. With regard to research area, reflecting the research contents of selected studies, more than half of the total papers appeared in literature on environmental studies, followed by politics, economics, geog­ raphy, sociology, and development studies. A few articles were published in some inter­ disciplinary journals, irrespective of the authors’ professional disciplines. As mentioned before, their publication years are not evenly distributed, and no less than 60% were published from the late 1990s to the early 2000s. It can be inferred that researchers have developed an interest in the environmental policies of CEE countries from a growing trend toward an overall evaluation of the results and future tasks in various sectors on the tenth anniversary of the revolutions of 1989, as well as growing concern over EU accession as a top political agenda in some candidate countries during this period. As described before, acceptance and implementation of the EU’s environmental acquis became an important matter for negotiation regarding EU accession.16 (D) OTHER ATTRIBUTES

I reviewed the other attributes that would be influential in an evaluation of environ­ mental reforms after regime transformation. First, in order to examine the assessment 347

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

of the role of the EU, which has often been considered to be a more important policy­ maker than domestic entities in CEE countries for improving environmental problems in the region, a total of 177 papers that clearly discuss this issue were selected. Their conclusions were classified using a three-grade evaluation (positive, moderate, and negative). Whereas less than one-third of the total studies positively appraised the role of the EU, its negative influences are emphasized in many papers. At the same time, a majority—almost half—espoused moderate views midway between these two obvious evaluations: while they eagerly anticipate the EU’s actions to improve the ecological disaster posed by the socialist regimes, they are more or less skeptical of the EU’s realistic policy capabilities for environmental amelioration.17 This ambivalence leads to mixed opinions regarding the powers and functions of the EU. Similarly, irrespective of the main subjects in each paper, a great number of cases deal with the quality and effectiveness of environmental policies and management and the results and lessons of environmental movement in CEE countries (see Table 10.1). I selected those papers that clearly described their observations about these two issues in a readable way, and their conclusions were codified as a three-grade evaluation, as described above. Almost 90% of the selected studies, or 218 papers, discussed the quality and effectiveness of environmental policies and management: they generally gave their negative comments, and no more than 10% of the literature highly appreciated the institutional changes from the standpoint of environmental reform. At the same time, no less than 60% of the selected studies, or 143 papers, refer to the results and lessons of environmental movement: here too, they are gener­ ally critical of the role of environmental movements, and studies that positively view non-profit private organizations such as civic movements and ENGOs as catalysts of environmental reform are in the minority. Finally, opting for 174 papers where the authors’ views on the environmental problems under the old regimes were clearly indicated, their judgments were classi­ fied as a three-point scale variable as before, since the evaluation of the environmen­ tal situation after transformation would be affected by the evaluation of the environmental issues before transformation. Although most studies judge sternly on this point, a few papers seem to suggest that we need to observe without prejudice the CEE’s environmental affairs, even in the socialist era. In many cases, these papers focus on the increase in environmental activism during the collapse of old rules and highly appreciate that environmental groups mobilized a great deal of public support for regime change (Gliński 2001; Hicks 2004; Rinkevicius 2000; Snajdr 2001). Moreover, in the CEE region, there still remain huge areas with tracts of virgin forests spared development.18 It is often stressed that they are better at forest management than their Western European counterparts in terms of preserving relic species and maintaining biodiversity (Andersson 2002; Ioras 2003). With regard to environmental policies as well, we occasionally come across a few such papers that support the progressive approach to preserving the environment, such as a society- and industry-wide recycling system (Gille 2000, 2004; Gorton et al. 2010; Jendrośka 1998). By and large, while studies that are in complete denial about all things environmental under the socialist regime have been gradually eclipsed, the 348

ASSESSMENT OF REGIME CHANGE

10.3

view that we should assess environmental issues during the socialist period based on their facts, as Pavlínek and Pickles (1999, 2004) insist on, seems to be prevailing. As discussed in the previous section, an evaluation of the environmental reforms in the CEE region would be related to how their economic and political reforms after the revolutions of 1989 should be seen. Thus, we cannot rule out the possibility that each author’s economic thought and/or political creed in essence exerts a certain influence upon their conclusions on the relationship of social transformation and environmental reform. As an example, according to Dryzek (2005, pt. 3), who ana­ lyzes practices and discourses of US environmental policies, expert-led administra­ tive rationalism, people-led democratic pragmatism, and market-driven economic rationalism still have a great deal of influence in the policy arena and are rivals with one another. Thus, based on the normative judgments in each study on the pros and cons of market principles/marketization or democracy/democratization in the aspect of environmental improvement, they were codified as three-point scale variables and tagged as positive, reserved, and skeptical appraisal for each category. A discussant like Václav Klaus, the former prime minister and president of the Czech Republic, who asserts without reservation that the penetration of market principles has a favorable effect on the environmental situation remains at around 30% of the total. The majority of researchers consider the existence of market failure and assume a cautious attitude toward the saturation of market forces. At the same time, in more than half of all cases, the democratization of political structure and the penetration of democracy into society are considered to contribute to ecological improvement through the establishment of a multiparty system and parliamentary politics, promotion of the decentralization of authority, and an enlargement of civil society. Nonetheless, more than a few authors take a reserved or skeptical position on the effect of democ­ racy upon the environment.19

10.3 ASSESSMENT OF REGIME CHANGE AND ENVIRONMENTAL REFORM IN CEE COUNTRIES In the following section, I examine whether and to what extent the basic characteris­ tics of selected studies exert an influence over the divergence of views on CEE environmental reforms after regime change. The main purpose here is to discern research backgrounds that would cause diverse assessment of the relationship between social transformation and environmental reform, considering the possibility that it might be reliant on the basic characteristics specified in each study or its “per­ sonality,” both in explicit and implicit ways. Accordingly, authors’ positions on the influence that regime change has exercised on environmental reform in CEE countries have been classified into the following four categories: fully support (full support without any reservation), conditionally support (partial support with reservation or under some conditions), difficult to sup­ port (little support as a whole with recognition of some positive results), and hardly

349

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

support (total nonsupport or denial of outcomes). A cross table shows a correlative relationship between these four-point scale evaluations and the basic characteristics in each paper, confirming that the total picture of the selected studies is almost evenly divided between the first two with positive assessment of social changes as related to environmental improvement and the last two with negative assessments to the contrary (see the last row in Table 10.2). In this table, the values of Cramér’s V are shown as an index for the strength of a relationship between the graduated evaluations and the basic characteristics, as well as the results of chi-square tests for independence based on the hypothesis that two variables are independent of each other. Some of the categories of the basic characteristics are aggregated or removed to avoid minimizing the estimates of expected frequency by reason of a limited number of samples and biased frequency distribution.20 The results suggest that the four-point scale evaluations of environmental reform are possibly related to some basic characteristics: the analytical method of each study, the place of the author’s doctorate, the publication year and source of each paper, and other attributes. Inter­ estingly, individual evaluations of EU accession/support, environmental policies and institutions, and environmental movements, as well as authors’ economic views from an environmental perspective, might affect their conclusions regarding the achieve­ ments of environmental reform, to a certain degree. Other characteristics such as the research topic of each paper, authors’ academic disciplines, and research area of publication media do not seem to be significantly related to evaluations of environ­ mental reform, as far as the cross table can tell. Then, multivariate ordinary probit regression models are employed to examine whether these basic characteristics of each study are correlated with the evaluations on environmental reform after regime change in CEE countries in a statistically significant way after controlling for them simultaneously. Table 10.3 lists the names and descriptive statistics of independent variables to be introduced, as well as simple correlation coeffi­ cients between each independent variable and dependent ordinal variable that could be arranged in descending order: 3 points (fully support), 2 points (conditionally support), 1 point (difficult to support), and 0 points (hardly support), with a mean of 1.5 and a median of 1. The independent variables consist of two continuous variables (median of the research period and the publication year), 3-point scale ordinal variables (EU accession/support evaluation, environmental institutions evaluation, environmental movements evaluation, environmental issues under socialism evaluation, market econ­ omy appraisal, and democracy appraisal), and other dummy variables with 0 or 1. The last column of the table demonstrates that some basic characteristics significantly influ­ ence the evaluations of CEE environmental reforms, although the results are somehow different from those of the cross-table analysis in Table 10.2. Table 10.4 indicates the estimation results by ordinary probit regression analysis. Because all attributes are not readable in one paper and, thus, are not able to be coded concurrently, other attributes among the basic characteristics are estimated separately in each panel so as to secure a certain number of samples. Also, a portion of the basic characteristics are removed from the analysis to cope with multicollinearity issues;21 the median of the research period and the publication year of the paper are estimated 350

ASSESSMENT OF REGIME CHANGE

10.3

Table 10.2 Cross table for four-point scale evaluations on environmental reform and basic characteristics (real number) a Four-point scale evaluations of environmental reform

Statistical test b

C H A P T E R 10

Fully support Research attributes (a) Number of targeted countries Multi countries 12 Specific country 13 Total 25 (b) Region of targeted countries c EU CEE countries 25 Non-EU CEE countries 3 FSU countries 4 Total 32 (c) Research topics (broadly classified) c Physical environment 16 Environmental 14 movements Environmental policy 17 and management Total 47 (d) Research topics (finely classified) c General environmental 2 issues Air pollution 9 Water pollution 3 Waste problems 3 Land use 6 Other issues 2 Environmental 7 movements Environmental policies 14 (public sector) Environmental manage­ 16 ment (private sector) International 3 cooperation Total 8 (e) Research period (median) 1970s and 1980s 3 Early 1990s 11 Late 1990s 5 2000s 6 Total 25 (f) Length of analysis Up to 5 years 3 From 6 to 10 years 12 Over 10 years 10 Total 25

Conditionally support

Difficult to Hardly support support

Upper: χ2 Lower: Total Cramér’s V

35 61 96

40 58 98

5 19 24

92 151 243

4.494 0.136

91 19 12 122

90 17 14 121

21 4 4 29

227 43 34 304

1.207 0.063

62 50

65 67

19 12

162 143

92

86

21

216

204

218

52

521

13

19

4

38

43 33 16 25 9 23

30 28 11 28 16 32

5 4 1 9 2 6

87 68 31 68 29 68

51

66

12

143

89

85

20

210

32

19

7

61

53

64

11

136

26.543 0.173

4 66 15 11 96

9 55 22 12 98

1 11 6 6 24

17 143 48 35 243

12.178 0.158

34 40 22 96

26 40 32 98

4 11 9 24

67 103 73 243

8.851 0.135

3.924 0.087

(Continued )

351

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM Table 10.2 contd. Four-point scale evaluations of environmental reform Fully support

10

(g) Analytical method Descriptive Statistical and quantitative Total Authors’ attributes (h) Affiliation c Higher education institutions Other organizations Total (i) Location of affiliation c North America Western Europe Eastern Europe Total (j) Place of doctorate c North America Western Europe Eastern Europe Total (k) Gender Female included Female not included Total (l) Number of authors Single author Multiple authors Total (m) Academic discipline c Environmental studies Politics Economics Sociology Geography Total Media attributes (n) Publication source Journal article Book chapter Total (o) Publication year Early 1990s Late 1990s Early 2000s Late 2000s and early 2010s Total

Conditionally support

Difficult to Hardly support support

Statistical test b

Upper: χ2 Lower: Total Cramér’s V

17 8

81 15

89 9

17 7

204 39

25

96

98

24

243

25

83

87

24

219

4 29

26 109

17 104

3 27

50 269

3.780 0.119

13 7 12 32

26 44 43 113

26 50 33 109

6 14 12 32

71 115 100 286

8.686 0.123

14 4 10 28

23 28 34 85

24 33 20 77

7 10 10 27

68 75 74 217

0.080* 0.161

11 14 25

31 65 96

41 57 98

9 15 24

92 151 243

2.326 0.098

8 17 25

51 45 96

57 41 98

12 12 24

128 115 243

5.548 0.151

9 8 8 7 3 35

33 21 33 10 15 112

29 32 23 9 19 112

10 6 3 4 7 30

81 67 67 30 44 289

14.588 0.159

20 5 25

53 43 96

61 37 98

21 3 24

155 88 243

11.847*** 0.156

0 6 9 10

19 35 23 19

15 31 29 23

2 7 4 11

36 79 65 63

25

96

98

24

243

11.227** 0.152

16.616* 0.185

(Continued ) 352

ASSESSMENT OF REGIME CHANGE

10.3

Table 10.2 contd. Four-point scale evaluations of environmental reform Fully support (p) Research area c Environmental studies Politics Economics Geography Others Total (q) Research level d Impact factor over 1.0 Impact factor below 1.0 No impact factor Total

Conditionally support

Difficult to Hardly support support

Statistical test b

Upper: χ2 Lower: Total Cramér’s V

16 7 5 2 5 35

80 14 12 9 14 129

77 17 12 13 19 138

19 7 2 2 9 39

192 45 31 26 47 341

10.500 0.124

11 9 2 22

40 13 7 60

42 19 12 73

18 3 1 22

111 44 22 177

8.259 0.153

17 47 11 75

2 7 6 15

67 87 23 177

41.637*** 0.343

1 46 40 87

0 4 17 21

20 134 64 218

116.265*** 0.516

5 46 15 66

0 9 3 12

30 90 23 143

36.210*** 0.356

32 40 72

5 11 16

66 108 174

2.841 0.090

Other attributes (r) Evaluation of EU accession/support Positive 16 32 Moderate 1 32 Negative 1 5 Total 18 69 (s) Evaluation of environmental institutions Positive 10 9 Moderate 7 77 Negative 0 7 Total 17 93 (t) Evaluation of environmental movements Positive 10 15 Moderate 4 31 Negative 0 5 Total 14 51 (u) Evaluation of environmental issues under socialism Positive and moderate 6 23 Negative 8 49 Total 14 72 (v) Appraisal of market economy Affirmative 12 26 Reserved 4 48 Skeptical 2 6 Total 18 80 (w) Appraisal of democracy Affirmative 10 29 Reserved and skeptical 1 20 Total 11 49

18 41 20 79

1 10 7 18

57 103 35 195

29.628*** 0.276

34 28 62

7 5 12

80 54 134

5.077 0.138

Grand total

98

24

243

25

96

Notes: a See Note 20 in the text. b ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively for chi-square test for independence. c Multi-coding corresponding to items in each category d Applicable only for journal articles

353

C H A P T E R 10

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Variable group and name

Table 10.3 Descriptive statistics of dependent and independent variables for ordinary probit regression analysis Variable type a

Dependent variable Four-point scale evaluations of environmental reform c Independent variable Research attributes Number of targeted countries Multi-country studies Region of targeted countries Non-EU CEE countries FSU countries Other countries Research topics Air pollution Water pollution Soil pollution Waste problems Transport problems Land use Environmental movements Environmental policies (public sector) Environmental management (private sector) Tourism development International cooperation Research period Median Analytical method Statistical Quantitative Authors’ attributes Affiliation Public authorities ENGO and consultancy International organizations Other organizations Location of affiliation North America Western Europe Other regions Place of doctorate North America Western Europe Other regions

Mean

S.D.

Correlation Median Maximum Minimum coefficient b

O

1.5

0.810

1

3

0

−

D

0.379 0.486

0

1

0

0.082

D D D

0.177 0.382 0.140 0.348 0.012 0.111

0 0 0

1 1 1

0 0 0

−0.008 −0.016 −0.069

D D D D D D D

0.358 0.280 0.086 0.128 0.066 0.280 0.588

0.480 0.450 0.282 0.334 0.249 0.450 0.493

0 0 0 0 0 0 1

1 1 1 1 1 1 1

0 0 0 0 0 0 0

0.131** 0.010 0.045 0.083 0.020 −0.070 −0.050

D

0.864 0.343

1

1

0

−0.066

D

0.251 0.435

0

1

0

0.004

D D

0.053 0.225 0.560 0.497

0 1

1 1

0 0

−0.102 −0.105

5.406

1994

2009

1972

−0.041

D D

0.070 0.256 0.091 0.288

0 0

1 1

0 0

−0.031 0.106

D D D

0.070 0.256 0.086 0.282 0.037 0.189

0 0 0

1 1 1

0 0 0

0.029 0.045 0.013

D

0.012 0.111

0

1

0

0.069

D D D

0.292 0.456 0.473 0.500 0.012 0.111

0 0 0

1 1 1

0 0 0

0.116* −0.140** 0.023

D D D

0.280 0.450 0.309 0.463 0.025 0.156

0 0 0

1 1 1

0 0 0

0.112* −0.129** −0.066

C

1994

(Continued ) 354

ASSESSMENT OF REGIME CHANGE

10.3

Table 10.3 contd. Variable group and name Gender Female included Number of authors Multiple authors Academic discipline Politics Economics Sociology Geography Media attributes Publication source Book chapter Publication year Year Research area Development studies Politics Economics Sociology Geography Other areas Other attributes Evaluation of EU accession/support d Evaluation of environmental institutions d Evaluation of environmental movements d Evaluation of environmental issues under socialism d Appraisal of market economy e Appraisal of democracy e

Variable type a

Mean

S.D.

Correlation Median Maximum Minimum coefficient b

C H A P T E R 10

D

0.379 0.486

0

1

0

−0.023

D

0.473 0.500

0

1

0

0.095

D D D D

0.337 0.412 0.132 0.325

0.590 0.774 0.362 0.846

0 0 0 0

1 1 1 1

0 0 0 0

−0.027 0.150** 0.055 −0.131**

D

0.362 0.482

0

1

0

0.062

6.232

2000

2015

1991

−0.017

C

2001

D D D D D D

0.041 0.185 0.128 0.066 0.107 0.086

0.199 0.389 0.334 0.249 0.310 0.282

0 0 0 0 0 0

1 1 1 1 1 1

0 0 0 0 0 0

0.076 −0.021 0.068 −0.042 −0.034 −0.173***

O

1.6

1.154

2

3

0

0.182***

O

1.6

0.781

2

3

0

0.354***

O

1.2

1.113

1

3

0

0.091

O

1.0

0.803

1

3

0

−0.038

O

1.7

1.039

2

3

0

O

1.4

1.344

2

3

0

0.173*** −0.052

Notes: a C: continuous variable, D: dummy variable, O: ordinal variable b ***, **, and * denote statistical significance at the 1%, 5%, and 10% levels, respectively. c Fully support: 3; Conditionally support: 2; Difficult to support: 1; Hardly support: 0 d Positive evaluation: 3; Moderate evaluation: 2; Negative evaluation: 1 e Affirmative appraisal: 3; Reserved appraisal: 2; Skeptical appraisal: 1

separately, as are estimations of authors’ academic disciplines and the research areas of publication media. The Huber–White sandwich estimator is used to estimate robust standard errors. Akaike’s information criterion (AIC) and Bayesian information criter­ ion (BIC) are used to determine desirable models for analysis. It follows from the estimation results shown in Table 10.4 that: first, among the research attributes, the number of targeted countries, region of targeted countries,

355

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Independent variable (default)/

model

Table 10.4 Estimation results of ordinary probit regression analysis Dependent variable

Four-point scale evaluations of environmental reform

[1]

[2]

Number of targeted countries (Single-country studies) Multi-country studies 0.556** 0.498** Region of targeted countries (EU CEE countries) Non-EU CEE countries 0.045 0.003 FSU countries 0.108 0.062 Other countries 0.089 −0.505 Research topics (General environmental issues) Air pollution 0.642** 0.647** Water pollution −0.508* −0.419 Soil degradation 0.315 0.144 Waste problems 0.193 0.264 Transport problems 0.113 −0.010 Land use 0.203 0.000 Environmental movements −0.175 0.042 Environmental policies (public −0.446 −0.365 sector) Environmental management −0.036 −0.225 (private sector) Tourism development −0.774* −0.898** International cooperation 0.104 −0.103 Research period Median 0.008 Analytical method (Descriptive) Statistical −1.437*** −1.348*** Quantitative −0.405 −0.225 Affiliation (Higher education institutions) Public authorities −0.202 ENGO and consultancy 0.371 International organizations −0.123 Other organizations 1.747*** Location of affiliation (Eastern Europe) North America 0.105 Western Europe −0.276 Other regions 2.521** Place of doctorate (Eastern Europe) North America Western Europe Other regions Gender (Female not included) Female included −0.323 Number of authors (Single author) Multiple authors 0.063 Academic discipline (Environmental studies) Politics 0.194 Economics 0.131 Sociology 0.854*** Geography −0.032

0.007 0.343 −0.153 1.886***

[3]

[4]

[5]

[6]

0.381

0.304

0.250 −0.932*** 0.726

0.268 −0.904*** 0.661

−0.103 0.163 −0.698*

−0.228 0.069 −1.262***

0.260 −0.045 −0.205 0.229 −0.237 0.392 0.074 −0.502

0.246 0.000 −0.425 0.379 −0.354 0.199 0.139 −0.352

0.068 0.277 0.339 0.254 −0.144 0.427 0.680 −0.500

0.098 0.002 0.167 0.693 −0.326 0.176 0.341 −0.540

−0.100

−0.235

0.050

−0.013

−0.697** −0.308

−0.666** −0.483**

−0.003

0.624**

0.125 −0.428

0.722**

0.021 −0.598*

0.011

−0.607* −0.202

−0.487 −0.427

−1.415** 0.720

−1.108* 0.085

−0.024 −0.323 −0.490 1.828***

0.106 −0.345 −0.503 2.009***

1.218** 0.161 −0.279 1.528***

1.189** 0.085 −0.236 1.644***

0.010 −0.046 0.132 0.226 0.011 −0.535

0.268 −0.176 −2.304*** 0.123 0.054 −0.793**

0.559** −0.168 −1.252***

−0.147

−0.089

−0.049

−0.217

−0.294

0.109

0.214

0.244

0.394

0.374

0.003 0.288** 0.535 −0.122

0.144 0.388* 0.615* −0.066

(Continued ) 356

ASSESSMENT OF REGIME CHANGE

10.3

Table 10.4 contd. Dependent variable

Four-point scale evaluations of environmental reform

Independent variable (default)/

model

10 [1]

[2]

Publication source (Journal article) Book chapter −0.157 Publication year Year Research area (Environmental studies) Development studies Politics Economics Sociology Geography Other areas Evaluation of EU accession/ support

Evaluation of environmental institutions

Evaluation of environmental movements

N Log pseudolikelihood Pseudo R2 AIC BIC Wald test (χ2) a

−0.330

0.988***

[3]

0.468**

[4]

0.156

0.018

0.009

1.234** −0.214 0.136 −0.029 −0.052 −0.687*

1.003** −0.310 0.127 0.583 −0.855** −0.856**

[5]

[6]

0.383

0.053 0.052** 0.817 −0.544 0.410 0.531 −0.785** −0.847**

0.863***

1.627***

1.731***

0.956***

1.066*** 177 −167.504 0.193 405.007 516.173 –

Dependent variable Independent variable (default)/ model

C H A P T E R

177 218 218 143 143 −168.394 −182.553 −180.544 −129.305 −125.741 0.189 0.275 0.283 0.221 0.242 412.789 437.106 437.089 330.611 327.481 533.482 558.948 565.700 437.273 440.069 111.920*** 154.860*** 173.140*** 154.380*** 110.350***

Four-point scale evaluations of environmental reform [7]

[8]

Number of targeted countries (Single-country studies) Multi-country studies 0.600** 0.628** Region of targeted countries (EU CEE countries) Non-EU CEE countries 0.163 0.104 FSU countries −0.434 −0.434 Other countries −0.626 −1.067 Research topics (General environmental issues) Air pollution 0.695** 0.876*** Water pollution −0.387 −0.512* Soil degradation 0.116 0.069 Waste problems 0.165 0.290 Transport problems 0.052 0.064 Land use 0.080 −0.111 Environmental movements −0.040 −0.006 Environmental policies (public −0.205 −0.086 sector)

[9]

0.427*

[10]

[11]

[12]

0.373

0.638**

0.538*

0.259 −0.546* −0.920

0.151 −0.328 −1.074**

0.583* −0.545 −0.277

0.200 −0.436 −0.916

0.352 −0.093 −0.099 0.201 0.096 −0.156 0.110 −0.531

0.362 −0.140 −0.019 0.169 0.111 −0.257 0.053 −0.362

0.702** −0.255 0.928* −0.690 0.110 0.095 −0.197 −0.134

0.805** −0.557** 0.772 −0.409 0.017 −0.269 0.166 0.200

(Continued )

357

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Independent variable (default)/ model

Table 10.4 contd. Dependent variable

Environmental management (private sector) Tourism development International cooperation Research period Median Analytical method (Descriptive) Statistical Quantitative

Four-point scale evaluations of environmental reform [7]

[9]

[10]

[11]

[12]

0.193

0.126

−0.010

0.049

0.373

0.241

−0.228 −0.461*

−0.348 −0.572**

−0.151 −0.094

−0.075 −0.172

−0.065 −0.243

0.151 −0.299

0.006 0.036 1.197*

Affiliation (Higher education institutions) Public authorities −0.433 ENGO and consultancy 0.353 International organizations −0.362 Other organizations 0.996*** Location of affiliation (Eastern Europe) North America −0.006 Western Europe −0.327 Other regions −1.035* Place of doctorate (Eastern Europe) North America Western Europe Other regions Gender (Female not included) Female included 0.039 Number of authors (Single author) Multiple authors 0.312 Academic discipline (Environmental studies) Politics 0.248 Economics −0.014 Sociology 0.292 Geography 0.026 Publication source (Journal article) Book chapter 0.315 Publication year Year Research area (Environmental studies) Development studies Politics Economics Sociology Geography Other areas Evaluation of environmental issues under socialism Appraisal of market economy Appraisal of democracy

[8]

0.050

0.005

0.001

−0.416 0.663

−0.823** −0.158

−0.764** −0.280

−0.735 2.299***

−1.020 1.567**

−0.413 0.274 −0.407 1.953***

−0.414 −0.375 −0.745* 1.088***

−0.201 −0.281 −0.683 1.204***

0.776 0.506 0.451 1.374***

0.853* 0.250 0.120 1.102**

−0.029 −0.215 −0.814* 0.174 −0.212 −0.398 0.231 0.262

−0.251 −0.671** −1.033* 0.153 −0.151 −0.201

−0.097 0.620***

−0.050 0.591***

−0.257 0.138 0.120 −0.160 0.178

0.430**

0.216 −0.195 −0.549 −0.286

−0.123

0.390

0.344

0.422 −0.049 0.711** 0.167 0.340

−0.009

−0.258

0.027

0.008

0.036

1.084* −0.293 0.200 0.311 0.046 −0.583

0.377 0.005 0.132 −0.353 0.253 −0.541

1.933** −0.486 0.105 0.776* −0.577 −1.050**

0.058 0.659***

0.673*** 0.218

0.170

(Continued )

358

ASSESSMENT OF REGIME CHANGE

10.3

Table 10.4 contd. Dependent variable Independent variable (default)/ model N Log pseudolikelihood Pseudo R2 AIC BIC Wald test (χ2) a

Four-point scale evaluations of environmental reform [7]

174 −178.258 0.111 422.516 526.764 –

[8]

174 −174.806 0.128 421.613 535.339 –

[9]

[10]

[11]

195 195 134 −197.160 −198.718 −130.438 0.137 0.130 0.150 466.319 473.437 330.877 584.147 597.811 432.301 75.930*** 140.460*** –

[12]

134 −126.883 0.174 327.766 434.986 –

Notes:

a Null hypothesis: all of the coefficients are equal to zero.

Robust standard errors are used for hypothesis testing. ***, **, and * denote statistical significance of the regression coeffi­ cient at the 1%, 5%, and 10% levels, respectively. See Table 10.3 for definition and descriptive statistics of variables used

in estimation.

and research topics (finely classified) influence the evaluations of environmental reform. As compared to single-country studies where a specific country is discussed comprehensively, multi-country studies are likely to produce positive evaluations of environmental reforms. At the same time, multi-country studies that incorporate nonEU Southeastern European and the FSU countries into their samples demonstrate less-positive evaluations. Many scholars believe that these countries have lagged far behind CEE countries concerning environmental reform (Castán Broto et al. 2009; Clarke 2002a, 2002b; Fagan 2006, 2010; Fagan and Sircar 2010; Mol 2009; Pryde 1995; Tokunaga 2010; Turnock 2002). It has been established that there is a large gap between these two country groups in the process of mandatory implementation of environmental reforms in CEE EU member states. Turning to the research topics (finely classified), it is clear that air pollution and tourism development are polariz­ ing subjects. Whereas research on the former considers that regime change has con­ tributed remarkably to the improvement of air pollution relative to general environmental issues, research that focuses on the latter seems to be very cautious about the impact of regime change upon the natural environment. On the one hand it was pointed out early on that heavy industry, which used to be a core business sector of socialist economies, recorded a dramatic reduction in air pollutant emis­ sions regardless of country and region, since operations were greatly reduced in association with industrial restructuring in the early phase of economic transition. On the other, many scholars share concerns about an adverse effect on the natural environment from privatization—land privatization, inter alia—being characterized as the pillar of economic reform. It is fully understandable that their studies are motivated by a sense of vigilance regarding the destruction of the landscape and nature possibly caused by real estate development in urban areas and recreational development in rural areas. As for the rest, a relatively negative opinion has been expressed in the study of international cooperation. This point seems to be often

359

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

reflected in such an unflattering comment that overseas financial and human aid mainly from the EU have not sufficiently contributed to improving the natural envir­ onment in CEE countries. The differences in analytical methods variously impact upon the conclusions as well: compared to a descriptive analysis mainly with a case study approach, a bird’s-eye survey of the dynamics of environmental issues relying on a statistical database is generally inclined to understand the impact of regime change in a negative way. Second, it is hard to say that authors’ attributes and media attributes are not such influential factors as to determine the conclusions. It looks as if some variables being classified as other organizations for authors’ affiliations and other regions for location of affiliation and place of doctorate significantly influence the evaluations of environmental reform after regime change; however, this result should not be over­ rated, because the number of sample papers is very small (three, four, and seven papers for each category, respectively). Although the proportions of papers with a female author and multi-authored papers demonstrate an upward trend, as men­ tioned in the previous section, it is hardly related to a difference in the above evalu­ ations. In contrast, authors’ academic disciplines and the research area of publication media seem to influence the conclusions to some extent. The following serves as an example: compared with an interdisciplinary expert on environmental studies, a researcher who majors in sociology is likely to offer a positive assessment of the impact of regime change on the natural environment. There are many cases where sociologists tackle the question of environmental movements; therefore, authors’ aca­ demic backgrounds can possibly make a difference to their conclusions on this research theme. Another example: while the literature on development studies includes papers that are more optimistic about the achievements of environmental reform, papers published in the media of geography and other research areas (mainly interdisciplinary magazines and specialty journals for the judicial community) are more often written by outspoken critics. However, these estimation results regarding authors’ attributes and media attributes are substantially inconsistent with those obtained from the cross-table analysis (see Table 10.2), which is generally seen as lacking sufficient explanations.22 Third, as regards the other attributes among the basic characteristics, an indi­ vidual evaluation of EU accession/support, environmental policies and institutions, and environmental movements has a high possibility of closely relating to its evaluations of environmental reform, in the same way as the cross-table analysis suggested; at the same time, how to evaluate the environmental problems before regime change does not seem to be interrelated with their subjective judgments on the environmental problems after that, because the estimation results regarding an evaluation of the environmental issues under socialism are not statistically sig­ nificant. Many studies on the enforcement of environmental laws and regulations and the reorganization of environmental authorities in the government sector, the implementation of environment and management policies in the business sector, and the development of environmental movements by ENGOs and other civic organizations touch upon the EU’s initiatives for environmental assistance. Quite 360

FUTURE RESEARCH PROSPECTS AND TASKS

10.4

importantly, their individual evaluations correlate strongly with one another.23 These results suggest that authors’ recognition of CEE environmental reforms is greatly contingent on how they see the environmental effect of EU accession and support. In fact, the environmental impact of the EU accession process is not smaller than its political and economic impacts, and, therefore, the verification of the effects on environmental change has been one of the main research themes for experts in CEE regional studies. This is reflected by the fact that, in the selected studies, many papers tackle this issue head on. Innovative environmental policies and an ecological modernization approach have featured prominently when we see the EU as a green superpower. This makes it all the more important to inquire into the coherence and inconsistencies between their discourses and reality.24 Fourth, according to the estimation results of market economy appraisal as norma­ tive judgments on the pros and cons of market principles/marketization, those who see environmental reform and economic reform as inextricably linked together and support market-led environmental policies on the basis of economic rationality tend to appreciate the achievements of environmental reform to date. As a matter of fact, some researchers insist that an effective market economy mechanism should improve the natural environment; at the same time, they express their outright displeasure over unrealized goals and the incomplete operation of market economy principles. A typical case involves authors who consider that an inflexible bureaucratic policy­ making style and cost-ineffective direct regulations of the EU stand as the greatest obstacles to environmental reform in CEE countries (Archibald et al. 2004; 2009; Żylicz 1994, 1995). At the same time, as suggested by the insignificant estimation results of democracy appraisal, it seems that few scholars consider democratic polit­ ical reform and/or the enlargement of civil society with democracy as a vehicle for the development of environmental reform. All things considered, in light of the empirical results reported in this systematic review, I conclude that the heterogeneity observed in the pertinent literature that exam­ ined the relationship between regime change and environmental reform in CEE coun­ tries could be attributable to the following basic characteristics: number of targeted countries, region of targeted countries, research topics, and the analytical method of each paper. This heterogeneity depends as much on the effects of environmental sup­ port from the EU and the effectiveness of domestic environmental institutions and movements that partially resulted from EU accession as well as normative judgments on the impact of economic reform and marketization upon environmental affairs.

10.4 FUTURE RESEARCH PROSPECTS AND TASKS

In this final section, I will expand the argument thus far and remark on a couple of points for further research related to environmental reform in CEE countries. To begin with, Figure 10.1 indicates declining research interest in this theme in and beyond Europe after successive enlargements of the EU. In particular, CEE countries

361

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM

that achieved their goals of EU accession in 2004 and 2007 have lost their distinct­ iveness as transition countries and are now being perceived as the EU hinterland (Börzel 2009b). It is symbolic that, at the end of 2007, the Czech Republic was pushed aside in the Transition Report published annually by the EBRD. At the same time, the research interests of those scholars and experts who have been engaged in environmental issues and reforms specific to the transition period are moving toward non-EU Southeastern European countries, which are often compared with EU CEE countries that were beset with similar problems earlier (Fagan and Sircar 2010; Gorton et al. 2010).25 Next, as might be expected from the estimation results in Table 10.4, if we add relevant studies focused on the FSU countries to our samples for another system­ atic review of this research field, it would cause a significant decrease in the evalu­ ation of environmental reform in transition. In doing so, we should not overlook the possibility that conclusions could be biased by the use of specific language that is also a basic characteristic of the literature. This problem has already happened: a large difference in experts’ views between English and non-English (especially Russian) literature surfaced with regard to the health and environmental impact of the 1986 Chernobyl accident. Against the 2006 official report and recommenda­ tions, which were based on about 350 articles written almost exclusively in English and edited by the World Health Organization and the International Atomic Energy Agency as 20-year anniversary project, in the following year some researchers from the afflicted areas of the FSU region (mainly Ukraine, Belarus, and Russia) published a survey report that summarized the discussion in almost a thousand research papers written in local Slavic languages and drew a definite conclusion on the still-catastrophic outcome of the accident, with harsh criticism for the optimis­ tic view of the former group.26 This impressive case tells us that it is necessary to control for differences in working languages when environmental studies of the FSU region are included in the samples for future systematic review, considering the fact that numerous academic and professional works have been written and published in the Russian language, one of the global languages, especially in academia. Finally, we need to pay attention to the existence of biases mainly caused by dif­ ferences in authors’ attributes. Although it does not seem to be an influential factor that would radically change the conclusions of this chapter, some systematic surveys on other research topics have revealed that authors’ attributes exercise a certain influence over their final conclusions.27 While these authors’ attributes, unlike other attributes, are not fully readable and literally interpretable in the original literature only, we can partially cope with this hardship by performing an online search of authors’ profiles and, if possible, developing a questionnaire about authors and/or interviewing some noted researchers. What information cannot be accessed directly in the original literature that should be collected and to what extent this additional information could be used remain matters for debate. How to operate a follow-up investigation should be considered to further develop systematic review surveys.

362

NOTES

ACKNOWLEDGMENTS

This chapter is a revised version of Tokunaga 2016. I thank Yu Hasumi and Ichiro Iwa­ saki for their helpful comments and suggestions on the earlier version of this paper.

C H A P T E R 10

NOTES

1 In the original plan launched in September 1977, two huge waterworks were scheduled to be built transnationally on the Danube: the Gabčíkovo Dam on the Slovakian side and the Nagymaros Dam on the Hungarian side. These works became the target of criticism from both Slovak and Hungarian environmentalists in the mid-1980s. Although the new Hungarian government unilaterally with­ drew in 1989 from the construction project of the Nagymaros Dam due to eco­ logical and financial concerns, this decision made without consultation with the Czechoslovakian side provoked a major backlash in the Slovak Parliament and became an international dispute closely tied to the historical Slovak–Hungarian ethnic conflict. The then-European Community (EC) failed to serve as a mediator, and both parties then fought in the International Court of Justice (Fitzmaurice 1996, ch. 7; Fleischer 1993). After a five-year inquiry, the Hague Court finally supported Slovakia’s argument, and it ended up building the dam only within the country’s boundaries. 2 Slocock (1996) analyzed in depth the environmental discourse and environmental policy discussions inside the Klaus administration in the early 1990s. See Klaus 2008 for his criticism of environmentalists, focusing on the global warming issue. He still insists that “sustainable development” is neither an appropriate nor correct concept, saying “it is not a neutral term. It is … an empty, undefined and undefinable, more or less leftist ideological concept” (Klaus lecture in Russia, January 15, 2014; see http://www.klaus.cz/clanky/3504). 3 The EU noted that for all member states as of the beginning of new millennium, the transposition deficit for the environmental directives was much higher than for overall directives (Bell 2004). Furthermore, among all complaints on an offense against the EU law that the Commission has received until 2010, the environmental and its derivative sector counted the highest number by sector and amounted to about 20% of the total complaints (Usui 2013, ch. 3). 4 A collection of works titled Dilemmas of Transition: The Environment, Democ­ racy and Economic Reform in East Central Europe (Baker and Jehlička 1998) accurately describes the circumstances at the time. 5 Environmental legal authority was given to the then-EC, mainly because in the 1980s there was growing concern about an occurrence of “eco-dumping” with the accession of Greece, Spain, and Portugal, who had lagged far behind the ori­ ginal EC member states regarding environmental performance (Hakogi 2002). 6 According to Article 4(2)(e) of the Treaty for the Functioning of the European Union, the environmental policy is one of the areas where the EU and member 363

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

7 8 9 10 11

12

13

14 15

16

17

states have shared competences, i.e., as a matter of principle, both the EU and member states may legislate and adopt legally binding acts in that field. At the same time, Article 2(2) of the Treaty clarifies that the right of the member states to exer­ cise their legislative powers only exists to the extent that the EU has not exercised its competence; even if they maintain or introduce more stringent protective meas­ ures according to a special rule in Article 193, these measures must be compatible with the Treaties, and the Commission shall be notified (Proelss 2016). * denotes truncation of a retrieval term. I searched only the category of article and rejected other types of works, such as proceeding papers, book reviews, etc. The final literature search was conducted in June 2017. For a list of sample papers, see Tokunaga 2019, app. To cite some examples, all papers published in special issues 7(1), 13(1), and 19(5) of Environmental Politics are included in Baker and Jehlička 1998, Carmin and VanDeveer 2005, and Fagan and Carmin 2011, respectively. For the special issues on the CEE countries, see Environmental Politics, 7(1) (1998); Geographical Journal, 165(2) (1999); Environment and Planning B: Planning and Design, 27(3) (2000); Environment and Planning A, 33(4) (2001); Environmental Politics, 13(1) (2004); Land Use Policy, 22(3) (2005); Environmental Politics, 19(5) (2010); Environmental Conservation, 40(2) (2013); Environment and Planning C: Government and Policy, 33(5) (2015). The following are considered to be especially noteworthy: Carter and Turnock 1993, 2002, Jancar-Webster 1993b, Vari and Tamas 1993, Carraro et al. 1994, DeBardele­ ben and Hannigan 1995, Bluffstone and Larson 1997, Klarer and Moldan 1997, Clark and Cole 1998, Tickle and Welsh 1998, Turnock 2001, Auer 2004, and Börzel 2009b. Petr Pavlínek (Czech Republic), Zbigniew Bochniarz (Poland), and Zsuzsa Gille (Hungary) are some examples. They all teach at US universities. I acknowledge that it is arguable to code gender as a binary choice. Nevertheless, it is classified as one of the authors’ attributes in this study, considering the results that gender has significantly influenced upon the conclusions of some studies (e.g. see Stanley and Jarrell 1998). It is undoubtedly a fact that candidate countries faced more difficulties in accept­ ing the environmental acquis because the EU’s environmental policies also radic­ ally changed during the negotiation period. However, as far as I have seen, only a limited number of studies examined issues of EU enlargement clearly related to the dynamics of the EU’s environmental policies, most of which focus on the controversy over Soviet atomic power stations located in the regions bordering Western European countries, such as the Temelin nuclear power plant in the Czech Republic (see e.g. Hakogi 2002; Axelrod 2004). A dispute over the discretion of the EU and its member states due to a shared internal competence in the field of environmental policy, myriad violations of EU environmental laws and regulations, their bureaucratic and rigid policy­ making processes, and other issues (Usui 2013, ch. 3; Proelss 2016).

364

REFERENCES

18 The largest virgin forest area in Europe is found in the Carpathian Mountains, one of the mountain ranges belonging to the Alpine–Himalayan orogenic belt. The range stretches from the border of the Czech Republic, Slovakia, and Poland in the northwest through Ukraine to Romania in the southeast. 19 It is no doubt difficult to classify and code a way of democracy/democratization appraisal, as compared to market/marketization appraisal. Therefore, an approach like this seems to be problematic. In particular, it is worth considering Pick­ vance’s contention (1997) that democratization and decentralization affect the environment in totally different ways, although these two political processes are often identified as interchangeable. 20 The cross table was adjusted so that each cell has an estimated expected frequency of 1.0 or more; yet, at the same time, the number of cells with an estimated expected frequency below 5.0 is reduced to about one-third of the total number of cells. 21 Research topics (broadly classified) and length of analysis are strongly correlated to research topics (finely classified) and the median of the research period or the publication year of the paper, respectively; thus, the first two variables are removed from the following analysis. 22 The estimation results regarding authors’ attributes and media attributes are also unstable in the sense that they are highly dependent on the coding approach for and the number of samples of the selected studies. 23 Ordinal scores for evaluations of EU accession/support, environmental institu­ tions, and environmental movements are positively correlated at the 1% level of statistical significance. 24 I owe much to Usui (2013, chs. 5, 6) for this discussion. 25 This tendency is true of foreign direct investment (FDI) studies of transition economies (see Chapter 9 of this book). 26 Yablokov et al. 2009 is an English version of the original report in Russian. 27 See the other chapters in this book.

REFERENCES Andersson, Magnus (2002) Environmental policy in Poland. In Helmut Weidner and Martin Jänicke (eds.), Capacity Building in National Environmental Policy: A Comparative Study of 17 Countries. Springer: Berlin, pp. 347–373. Archibald, Sandra O., Luana E. Banu, and Zbigniew Bochniarz (2004) Market liberalisation and sustainability in transition: Turning points and trends in Central and Eastern Europe. Environmental Politics, 13(1), pp. 266–289. Archibald, Sandra O., Zbigniew Bochniarz, Masahiko Gemma, and Tanja Srebotnjak (2009) Transition and sustainability: Empirical analysis of environmental Kuznets curve for water pollution in 25 countries in Central and Eastern Europe and the Commonwealth of Inde­ pendent States. Environmental Policy and Governance, 19(2), pp. 73–98. Auer, Matthew R. (ed.) (2004) Restoring Cursed Earth: Appraising Environmental Policy Reforms in Eastern Europe and Russia. Rowman & Littlefield: Lanham.

365

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM Axelrod, Regina (2004) Nuclear power and EU enlargement: The case of Temelín. Environ­ mental Politics, 13(1), pp. 153–172. Baker, Susan (2002) Environmental protection capacity in post-communist Bulgaria. In Helmut Weidner and Martin Jänicke (eds.), Capacity Building in National Environmental Policy: A Comparative Study of 17 Countries. Springer: Berlin, pp. 97–122. Baker, Susan, and Petr Jehlič ka (1998) (eds.) Dilemmas of Transition: The Environment, Democracy and Economic Reform in East Central Europe. Frank Cass: London and Portland. Bell, Ruth Greenspan (2004) Further up the learning curve: NGOs from transition to Brussels. Environmental Politics, 13(1), pp. 194–215. Bluffstone, Randall (1999) Are the costs of pollution abatement lower in Central and Eastern Europe? Evidence from Lithuania. Environment and Development Economics, 4, pp 449–470. Bluffstone, Randall, and Bruce A. Larson (1997) (eds.) Controlling Pollution in Transition Economies: Theories and Methods. Edward Elgar: Cheltenham and Lyme. Boehmer-Christiansen, S. A. (1992) Taken to the cleaners: The fate of the East German energy sector since 1990. Environmental Politics, 1(2), pp. 196–228. Boehmer-Christiansen, Sonja (1998) Environment-friendly deindustrialization: Impacts of unification on East Germany. In Andrew Tickle and Ian Welsh (eds.), Environment and Society in Eastern Europe. Addison Wesley Longman: Harlow, pp. 67–96. Börzel, Tanja A. (2009a) After accession: Escaping the low capacity trap? In Tanja A. Börzel (ed.), Coping with Accession to the European Union: New Modes of Environmental Gov­ ernance. Palgrave Macmillan: Basingstoke, pp. 192–210. Börzel, Tanja A. (2009b) (ed.) Coping with Accession to the European Union: New Modes of Environmental Governance. Palgrave Macmillan: Basingstoke. Börzel, Tanja, and Aron Buzogány (2010a) Environmental organisations and the European­ isation of public policy in Central and Eastern Europe: The case of biodiversity governance. Environmental Politics, 19(5), pp. 708–735. Börzel, Tanja, and Aron Buzogány (2010b) Governing EU accession in transition countries: The role of non-state actors. Acta Politica, 45(1/2), pp. 158–182. Börzel, Tanja, and Adam Fagan (2015) Environmental governance in South East Europe/ Western Balkans: Reassessing the transformative power of Europe. Environment and Planning C: Government and Policy, 33(5), pp. 885–900. Buzogány, Aron (2009a) Hungary: The tricky path of building environmental governance. In Tanja A. Börzel (ed.), Coping with Accession to the European Union: New Modes of Environmental Governance. Palgrave Macmillan: Basingstoke, pp. 123–147. Buzogány, Aron (2009b) Romania: Environmental governance: Form without substance. In Tanja A. Börzel (ed.), Coping with Accession to the European Union: New Modes of Environmental Governance. Palgrave Macmillan: Basingstoke, pp. 169–191. Buzogány, Aron (2015) Building governance on fragile grounds: Lessons from Romania. Environment and Planning C: Government and Policy, 33(5), pp. 901–918. Caddy, Joanne (1997a) Harmonization and asymmetry: Environmental policy co-ordination between the European Union and Central Europe. Journal of European Public Policy, 4 (3), pp. 318–336. Caddy, Joanne (1997b) Hollow harmonisation? Closing the implementation gap in Central European environmental policy. European Environment, 7(3), pp. 73–79.

366

REFERENCES Carmin, JoAnn (2010) NGO capacity and environmental governance in Central and Eastern Europe. Acta Politica, 45(1–2), pp. 183–202. Carmin, JoAnn, and Adam Fagan (2010) Environmental mobilisation and organisations in post-socialist Europe and the Former Soviet Union. Environmental Politics, 19(5), pp. 689–707. Carmin, JoAnn, and Stacy D. VanDeveer (2005) EU Enlargement and the Environment: Insti­ tutional Change and Environmental Policy in Central and Eastern Europe. Routledge: London and New York. Carraro, C., A. Haurie, and G. Zaccour (1994) (eds.) Environmental Management in a Transition to Market Economy: A Challenge to Governments and Business. Éditions Technip: Paris. Carter, F. W., and David Turnock (1993) (eds.) Environmental Problems in Eastern Europe. Routledge: London and New York. Carter, F. W., and David Turnock (2002) (eds.) Environmental Problems of East Central Europe. Second edition, Routledge: London and New York. Castán Broto, Vanesa, Claudia Carter, and Lucia Elghali (2009) The governance of coal ash pollution in post-socialist times: Power and expectations. Environmental Politics, 18(2), pp. 279–286. Cent, Joanna, Małgorzata Grodzińska-Jurczak, and Agata Pietrzyk-Kaszyńska (2014) Emer­ ging multilevel environmental governance: A case of public participation in Poland. Jour­ nal for Nature Conservation, 22(2), pp. 93–102. Cent, Joanna, Cordula Mertens, and Krzysztof Niedziałkowski (2013) Roles and impacts of non-governmental organizations in Natura 2000 implementation in Hungary and Poland. Environmental Conservation, 40(2), pp. 119–128. Cent, Joanna, Hanna Kobierska, Małgorzata Grodzińska-Jurczak, and Sandra Bell (2007) Who is responsible for Natura 2000 in Poland? A potential role of NGOs in establishing the programme. International Journal of Environment and Sustainable Development, 6 (4), pp. 422–435. Císař, Ondřej (2010) Externally sponsored contention: The channelling of environmental movement organisations in the Czech Republic after the fall of communism. Environmen­ tal Politics, 19(5), pp. 736–755. Clark, John, and Daniel H. Cole (1998) (eds.) Environmental Protection in Transition: Eco­ nomic, Legal and Socio-Political Perspectives on Poland. Ashgate: Aldershot. Clarke, Richard (2002a) Bosnia and Hercegovina. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe. Second edition, Routledge: London and New York, pp. 283–304. Clarke, Richard (2002b) Yugoslavia. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe. Second edition, Routledge: London and New York, pp. 396–416. DeBardeleben, Joan, and John Hannigan (1995) (ed.) Environmental Security and Quality after Communism: Eastern Europe and the Soviet Successor States. Westview Press: Boulder, Col. Dingsdale, Alan, and Dénes Lóczy (2001) The environmental challenge of societal transition in East Central Europe. In David Turnock (ed.), East Central Europe and the Former Soviet Union: Environment and Society. Arnold and Oxford University Press: London and New York, pp. 187–199. Drgona, Vladimir (1996) Ecological problems arising from intensive agriculture in western Slovakia. GeoJournal, 38(2), pp. 213–218.

367

C H A P T E R 10

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Evans, Douglas, András Demeter, Peter Gajdoš, and Ľuboš Halada (2013) Adapting envir­ onmental conservation legislation for an enlarged European Union: Experience from the Habitats Directive. Environmental Conservation, 40(2), pp. 97–107.

Dryzek, John S. (2005) The Politics of the Earth: Environmental Discourses. Second edition, Oxford University Press: Oxford. Earnhart, Dietrich, and Lubomir Lizal (2008) Pollution reductions in the Czech Republic. Post-Communist Economies, 20(2), pp. 231–252.

Fagan, Adam (2005) Taking stock of civil-society development in post-communist Europe: Evidence from the Czech Republic. Democratization, 12(4), pp. 528–547. Fagan, Adam (2006) Neither “north” nor “south”: The environment and civil society in postconflict Bosnia–Herzegovina. Environmental Politics, 15(5), pp. 787–802. Fagan, Adam (2010) The new kids on the block: Building environmental governance in the Western Balkans. Acta Politica, 45(1/2), pp. 203–228. Fagan, Adam, and JoAnn Carmin (2011) Green Activism in Post-Socialist Europe and the Former Soviet Union. Routledge: London and New York. Fagan, Adam, and Indraneel Sircar (2010) Environmental politics in the Western Balkans: River basin management and non-governmental organisation (NGO) activity in Herzegovina. Environmental Politics, 19(5), pp. 808–830. Fagan, Adam, and Indraneel Sircar (2011) Foreign donor assistance and environmental cap­ acity building: Evidence from Serbia and Bosnia–Herzegovina. European Political Sci­ ence Review, 3(2), pp. 301–320. Fagan, Adam, and Indraneel Sircar (2015) Europeanisation and multi-level environmental governance in a post-conflict context: The gradual development of environmental impact assessment processes in Bosnia–Herzegovina. Environment and Planning C: Govern­ ment and Policy, 33(5), pp. 919–934. Fagin, Adam (1994) Environment and transition in the Czech Republic. Environmental Polit­ ics, 3(3), pp. 479–494. Fagin, Adam (1999) The development of civil society in the Czech Republic: The environ­ mental sector as a measure of associational activity. Journal of European Area Studies, 7 (1), pp. 91–108. Fagin, Adam (2001) Environmental capacity building in the Czech Republic. Environment and Planning A, 33(4), pp. 589–607. Fagin, Adam (2002) The Czech Republic. In Helmut Weidner and Martin Jänicke (eds.), Cap­ acity Building in National Environmental Policy: A Comparative Study of 17 Countries. Springer: Berlin, pp. 177–200. Fagin, Adam, and Petr Jehlič ka (1998) Sustainable development in the Czech Republic: A doomed process? Environmental Politics, 7(1), pp. 113–128. Fagin, Adam, and Andrew Tickle (2002) Environmental movements, nation states and glo­ balisation. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe Second edition, Routledge: London and New York, pp. 40–55. Feshbach, Murray, and Alfred Friendly Jr. (1992) Ecocide in the USSR: Health and Nature under Siege. Basic Books: New York. Fitzmaurice, John (1996) Damming the Danube: Gabč ikovo and Post-Communist Politics in Europe. Westview Press: Boulder, Col. Fleischer, Tamás (1993) Jaws on the Danube: Water management, regime change and the movement against the middle Danube hydroelectric dam. International Journal of Urban and Regional Research, 17(3), pp. 429–443.

368

REFERENCES Frankland, Erich G. (1995) Green revolutions? The role of green parties in Eastern Europe’s transition, 1989–1994. East European Quarterly, 29(3), pp. 315–345. Gille, Zsuzsa (2002) Social and spatial inequalities in Hungarian environmental politics: A historical perspective. In Peter Evans (ed.), Livable Cities? Urban Struggles for Liveli­ hood and Sustainability. University of California Press: Berkeley, London, pp. 132–161. Gille, Zsuzsa (2004) Europeanising Hungarian waste policies: Progress or regression? Envir­ onmental Politics, 13(1), pp. 114–134. Gliński, Piotr (1998) Polish greens and politics: A social movement in a time of transformation. In John Clark and Daniel H. Cole (eds.), Environmental Protection in Transition: Economic, Legal and Socio-Political Perspectives on Poland. Ashgate: Aldershot, pp. 129–153. Gliński, Piotr (2001) The ecologocial movement as the element of the civil society. In Helena Flam (ed.), Pink, Purple, Green: Women’s, Religious, Environmental and Gay/Les­ bian Movements in Central Europe Today. East European Monographs: Boulder, Col. pp. 112–119. Gorton, Matthew, Philip Lowe, Steve Quarrie, and Vlade Zarić (2010) European rule adop­ tion in Central and Eastern Europe: A comparative analysis of agricultural water manage­ ment in Serbia. Environmental Politics, 19(4), pp. 578–598. Górz, Bronislaw, and Włodzimierz Kurek (2001) Sustainable agriculture. In David Turnock (ed.), East Central Europe and the Former Soviet Union: Environment and Society. Arnold and Oxford University Press: London and New York, pp. 200–206. Grodzinska-Jurczak, Malgorzata, and Joanna Cent (2011) Expansion of nature conservation areas: Problems with Natura 2000 implementation in Poland? Environmental Manage­ ment, 47(1), pp. 11–27. Guttenbrunner, Sonja (2009) Poland: When environmental governance meets politics. In Tanja A. Börzel (ed.), Coping with Accession to the European Union: New Modes of Environmental Governance. Palgrave Macmillan: Basingstoke, pp. 148–168. Hakogi, Masumi (2002) Environmental problems of Central and Eastern Europe and EU accession. Review of World Economy, 8, pp. 23–32 (Japanese). Hicks, Barbara (2004) Setting agendas and shaping activism: EU influence on Central and Eastern European environmental movements. Environmental Politics, 13(1), pp. 216–233. Ieda, Osamu (2011) Red sludge spill in Ajka, Hungary in 2010: Disputes on the dual criteria of waste management between the EU and Hungary. Japan Border Review, 2, pp. 149–172 (Japanese). Iojă, Cristian Ioan, Maria Pătroescu, Laurenţiu Rozylowicz, Viorel D. Popescu, Mircea Vergheleţ, Mihai Iancu Zotta, and Mihaela Felciuc (2009) The efficacy of Roma­ nia’s protected areas network in conserving biodiversity. Biological Conservation, 143 (11), pp. 2468–2476. Ioras, Florin (2003) Trends in Romanian biodiversity conservation policy. Biodiversity and Conservation, 12(1), pp. 9–23. Jancar-Webster, Barbara (1993a) Eastern Europe and the former Soviet Union. In Sheldon Kamieniecki (ed.), Environmental Politics in the International Arena: Movements, Parties, Organizations, and Policy. State University of New York Press: Albany, pp. 199–221. Jancar-Webster, Barbara (1993b) (ed.) Environmental Action in Eastern Europe: Responses to Crisis. M. E. Sharpe: Armonk, NY. Jancar-Webster, Barbara (1998) Environmental movement and social change in the transi­ tion countries. Environmental Politics, 7(1), pp. 69–90.

369

C H A P T E R 10

C H A P T E R

REGIME CHANGE AND ENVIRONMENTAL REFORM

10

Jendrośka, Jerzy (1998) Environmental law in Poland, 1989–1996: An assessment of past reforms and future prospects. In John Clark and Daniel H. Cole (eds.), Environmental Protection in Transition: Economic, Legal and Socio-Political Perspectives on Poland. Ashgate: Aldershot, pp. 81–115.

Jehlič ka, Petr (2001) The new subversives: Czech environmentalists after 1989. In Helena Flam (ed.), Pink, Purple, Green: Women’s, Religious, Environmental and Gay/Les­ bian Movements in Central Europe Today, East European Monographs: Boulder, Col., pp. 81–94.

Jorgenson, Andrew K., Brett Clark, and Vincentas R. Giedraitis (2012) The temporal (in)sta­ bility of the carbon dioxide emissions/economic development relationship in Central and Eastern European nations. Society and Natural Resources: An International Journal, 25 (11), pp. 1182–1192. Juergensmeyer, Julian, Michael Kulesza, and Ewa Gmurzynska (1991) Environmental pro­ tection in post-socialist Eastern Europe: The Polish example. Hastings International and Comparative Law Review, 14(4), pp. 831–848. Kay, Kelly (2014) Europeanization through biodiversity conservation: Croatia’s bid for EU accession and the Natura 2000 designation process. Geoforum, 54, pp. 80–90. Kerekes, Sándor, and Károly Kiss (1998) Hungary’s accession to the EU: Environmental requirements and strategies. European Environment, 8(5), pp. 161–170. Klarer, Jürg, and Bedř ich Moldan (eds.) (1998) The Environmental Challenge for Central European Economies in Transition. Wiley: Chichester. Klaus, Václav (2008) Blue Planet in Green Shackles: What is Endangered: Climate or Free­ dom? Competitive Enterprise Institute: Washington DC. Kluvánková-Oravská, Tatiana, Veronika Chobotová, and Eva Smolková (2013) The chal­ lenges of policy convergence: The Europeanization of biodiversity governance in an enlarging EU. Environment and Planning C: Government and Policy, 31(3), pp. 401–413. Kluvánková-Oravská, Tatiana, Veronika Chobotová, Ilona Banaszak, Lenka Slavikova, and Sonja Trifunovova (2009) From government to governance for biodiversity: The perspec­ tive of Central and Eastern European transition countries. Environmental Policy and Gov­ ernance, 19(3), pp. 186–196. Knorn, Jan, Tobias Kuemmerle, Volker C. Radeloff, William S. Keeton, Vladimir Gancz, Iovu-Adrian Biriş, Miroslav Svoboda, Patrick Griffiths, Adrian Hagatis, and Patrick Hostert (2013) Continued loss of temperate old-growth forests in the Romanian Carpathians despite an increasing protected area network. Environmental Conservation, 40(2), pp. 182–193. Knorn, Jan, Tobias Kuemmerle, Volker C. Radeloff, Alina Szabo, Marcel Mindrescu, William S. Keeton, Ioan Abrudan, Patrick Griffiths, Vladimir Gancz, and Patrick Hostert (2012) Forest restitution and protected area effectiveness in post-socialist Romania. Biological Conservation, 146(1), pp. 204–212. Kř enová, Zdenka, and Pavel Kindlmann (2014) Natura 2000: Solution for Eastern Europe or just a good start? The Šumava National Park as a test case. Biological Conservation, 186, pp. 268–275. McCuen, Gary E., and Ronald P. Swanson (1993) Toxic Nightmare: Ecocide in the USSR and Eastern Europe. Gary E. McCuen Publications: Hudson, Wis. Manser, Roger (1993) Failed Transitions: The Eastern European Economy and Environment since the Fall of Communism. The New York Press: New York. Mazurski, Krzysztof R. (1999) Environmental problems in the Sudetes, Poland. GeoJournal, 46(3), pp. 271–277.

370

REFERENCES Mikulcak, Friederike, Jens Newig, Andra I. Milcu, Tibor Hartel, and Joern Fischer (2013) Integrating rural development and biodiversity conservation in Central Romania. Environ­ mental Conservation, 40(2), pp. 129–137. Missfeldt, Fanny, and Arturo Villavicenco (2000) The economies in transition as part of the climate regime: Recent developments. Environment and Planning B: Planning and Design, 27(3), pp. 379–392. Mol, Arthur P. J. (2009) Environmental deinstitutionalization in Russia. Journal of Environ­ mental Policy and Planning, 11(3), pp. 223–241. Nakanishi, Yumiko (2016) Introduction: The impact of the international and European Union environmental law on Japanese basic environmental law. In Yumiko Nakanishi (ed.), Contemporary Issues in Environmental Law: The EU and Japan. Springer: Tokyo, pp. 1–13. Niedziałkowski, Krzysztof, Jouni Paavola, and Bogumiła Jędrzejewska (2012) Governance of biodiversity in Poland before and after the accession to the EU: The tale of two roads. Environmental Conservation, 40(2), pp. 108–118. Obradovic-Wochnik, Jelena, and Anneliese Dodds (2015) Environmental governance in a contested state: The influence of European Union and other external actors on energy sector regulation in Kosovo. Environment and Planning C: Government and Policy, 33(5), pp. 935–949. Pavlínek, Petr, and John Pickles (1999) Environmental change and post-communist trans­ formations in the Czech Republic and Slovakia. Post-Soviet Geography and Economics, 40(5), pp. 354–382. Pavlínek, Petr, and John Pickles (2004) Environmental pasts/environmental futures in post-socialist Europe. Environmental Politics, 13(1), pp. 237–265. Pickles, John, Mariana Nikolova, Caedman Staddon, Stefan Velev, Zoya Mateeva, and Anton Popov (2002) Bulgaria. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe. Second edition, Routledge: London and New York, pp. 305–329. Pickvance, Christopher G. (1997) Decentralization and democracy in Eastern Europe: A sceptical approach. Environment and Planning C: Government and Policy, 15(2), pp. 129–142. Podoba, Juraj (1998) Rejecting green velvet: Transition, environment and nationalism in Slovakia. Environmental Politics, 7(1), pp. 129–144. Proelss, Alexander (2016) The scope of the EU’s competences on the field of the environ­ ment. In Yumiko Nakanishi (ed.), Contemporary Issues in Environmental Law: The EU and Japan. Springer: Tokyo, pp. 15–28. Pryde, Philip. R. (1995) The environmental implications of republic sovereignty. In Philip. R. Pryde (ed.), Environmental Resources and Constraints in the Former Soviet Republics. Westview Press: Boulder, Col., pp. 1–23. Rinkevicius, Leonardas (2000) Ecological modernisation as cultural politics: Transformations of civic environmental activism in Lithuania. Environmental Politics, 9(1), pp. 171–202. Sarre, Philip, and Petr Jehlič ka (2007) Environmental movements in space-time: The Czech and Slovak Republics from Stalinism to post-socialism. Transactions of the Institute of British Geographers, 32(3), pp. 346–362. Scrieciu, S. Şerban, and Lindsay C. Stringer (2008) The transformation of post-communist societies in Central and Eastern Europe and the Former Soviet Union: An economic and ecological sustainability perspective. European Environment, 18(3), pp. 168–185.

371

C H A P T E R 10

C H A P T E R 10

REGIME CHANGE AND ENVIRONMENTAL REFORM Sklenicka, Petr, Vratislava Janovska, Miroslav Salek, Josef Vlasak, and Kristina Molnarova (2014) The farmland rental paradox: Extreme land ownership fragmentation as a new form of land degradation. Land Use Policy, 38, pp. 587–593. Slocock, Brian (1996) The paradoxes of environmental policy in Eastern Europe: The dynamics of policy-making in the Czech Republic. Environmental Politics, 5(3), pp. 501–521. Slocock, Brian (1999) “Whatever happened to the environment?” Environmental issues in the eastern enlargement of the European Union. In Karen Henderson (ed.), Back to Europe: Central and Eastern Europe and the European Union. UCL Press: London, pp. 151–167. Snajdr, Edward K. (1998) The children of the greens: New ecological activism in post-socialist Slovakia. Problems of Post-Communism, 45(1), pp. 54–62. Snajdr, Edward (2001) Grassroots and global visions: Slovakia’s post-socialist environmen­ tal movement. In Helena Flam (ed.), Pink, Purple, Green: Women’s, Religious, Environ­ mental and Gay/Lesbian Movements in Central Europe Today. East European Monographs: Boulder, Col., pp. 95–103. Sotirov, Metodi, Marko Lovric, and Georg Winkel (2015) Symbolic transformation of envir­ onmental governance: Implementation of EU biodiversity policy in Bulgaria and Croatia between Europeanization and domestic politics. Environment and Planning C: Govern­ ment and Policy, 33(5), pp. 986–1004. Staddon, Caedmon (1999) Localities, natural resources and transition in Eastern Europe. Geographical Journal, 165(2), pp. 200–208. Stanley, T. D., and Hristos Doucouliagos (2012) Meta-Regression Analysis in Economics and Business. Routledge: London and New York. Stanley, T. D., and Stephen B. Jarrell (1998) Gender wage discrimination bias? A meta-regression analysis. Journal of Human Resources, 33(4), pp. 947–973. Stringer, Lindsay C., and Jouni Paavola (2013) Participation in environmental conservation and protected area management in Romania: A review of three case studies. Environ­ mental Conservation, 40(2), pp. 138–146. Sumelius, John, Stefan Bäckman, and Timo Sipiläinen (2005) Agri-environmental problems in Central and Eastern European countries before and during the transition. Sociologia Ruralis, 45(3), pp. 153–170. Švajda, Juraj (2008) Participatory conservation in a post-communist context: The Tatra National Park and Biosphere Reserve, Slovakia. International Journal of Biodiversity Sci­ ence and Management, 4(4), pp. 200–208. Tickle, Andrew, and Ian Welsh (eds.) (1998) Environment and Society in Eastern Europe. Addison Wesley Longman: Harlow. Tokunaga, Masahiro (2010) Environmental governance in Russia: The “closed” pathway to ecological modernization. Environment and Planning A, 42(7), pp. 1686–1704. Tokunaga, Masahiro (2016) Regime change and environmental reform: A systematic review of research on Central and Eastern Europe. Economic Review, 67(4),pp. 339–353 (Japanese). Tokunaga, Masahiro (2019) Regime change and environmental reform: A systematic review of research on Central and Eastern Europe. CEI Working Paper No. 2019-10, Center for Economic Institutions, Institute of Economic Research of Hitotsubashi University: Kunita­ chi, Tokyo. Turnock, David (1999) Sustainable rural tourism in the Romanian Carpathians. Geographical Journal, 165(2), pp. 192–199.

372

REFERENCES Turnock, David (2001) (ed.) East Central Europe and the Former Soviet Union: Environment and Society. Arnold and Oxford University Press: London and New York. Turnock, David (2002) The Soviet Union and the Successor States. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe. Second edition, Routledge: London and New York, pp. 92–116. Unwin, Tim (1996) Tourist development in Estonia: Images, sustainability, and integrated rural development. Tourism Management, 17(4), pp. 265–276. Usui, Yoichiro (2013) Normative Politics in the EU Environmental Governance. Nakanishiya Publication: Kyoto (Japanese). Vari, Anna, and Pal Tamas (eds.) (1993) Environment and Democratic Transition: Policy and Politics in Central and Eastern Europe. Kluwer Academic Publishers: Dordrecht and Boston, Mass. Vukina, Tomislav, John C. Beghin, and Ebru G. Solakoglu (1999) Transition to markets and the environment: Effects of the change in the composition of manufacturing output. Environment and Development Economics, 4, pp. 582–598. Waller, Michael (1998) Geopolitics and the environment in Eastern Europe. Environmental Politics, 7(1), pp. 29–52. Watzman, Nancy (1992) Velvet revolution no longer green. New Scientist, June 20. Wilson, Olivia J., and Geoff A. Wilson (2002) East Germany. In F. W. Carter and David Turnock (eds.), Environmental Problems of East Central Europe. Second edition, Routledge: London and New York, pp. 139–156. Yablokov, Alexey V., Vassily B. Nesterenko, and Alexey V. Nesterenko (2009) Chernobyl: Consequences of the Catastrophe for People and the Environment. Blackwell Publishers: Boston, Mass. Young, Juliette, Caspian Richards, Anke Fischer, Lubos Halada, Tiiu Kull, Antoni Kuzniar, Urmas Tartes, Yordan Uzunov, and Allan Watt (2007) Conflicts between biodiversity con­ servation and human activities in the Central and Eastern European countries. Ambio: A Journal of the Human Environment, Research and Management, 36(7), pp. 545–550. Zamparutti, Tony, and Brendan Gillespie (2000) Environment in the transition towards market economies: An overview of trends in Central and Eastern Europe and the new independent states of the former Soviet Union. Environment and Planning B: Planning and Design, 27(3), pp. 331–347. Żylicz, Tomasz (1994) In Poland, it’s time for economics. Environmental Impact Assessment Review, 14(2–3), pp. 79–94. Żylicz, Tomasz (1995) Cost-effectiveness of air pollution abatement in Poland. Environmen­ tal and Resource Economics, 5(2), pp. 131–149.

373

C H A P T E R 10

Index

absorption capacity 290, 323 acquis communautaire 19, 332 adverse incentive effect 290 adverse selection effect 190 AIC see Akaike’s information criterion Akaike’s information criterion 48, 53, 56, 355, 357, 359 American-style HRM 247, 250–251, 253, 257 analysis of variance 68, 73 ANOVA see analysis of variance anti-environmentalism 330–331 Arrow, Kenneth J. 34, 41, 47, 191 asset stripping 191 auction system 181 bank loan penetration 77 Bayesian information criterion 48, 53, 56, 355, 357, 359 Bayesian model averaging 21n6 Berlin Wall 1, 25, 27, 37, 239 BIC see Bayesian information criterion big-bang approach 25, 37, 39 bilateral FDI model 294, 297–299 biodiversity 342, 348 Bosnian War 19, 82 Brewster, Chris 240, 247, 249–255 Bribery 16, 145, 168 business culture 234n6, 249 capital flight 40, 294 capital liberalization 272–273

374

capitalist market economy 12–14, 16, 25, 57–58, 80 central bank independence 22, 82 centralized system 13–14 chi-square tests for independence 350 CIS see Commonwealth of Independent State civil war 19, 75, 82–83, 341 Clarivate Analytics 11, 337 CMEA see Council of Mutual Economic Assistance coefficient of determination 300 Cohen’s criteria/guidelines 113n7, 295, 308 Cold War 1, 25, 32, 57, 67, 334 collectivism 248–249 collectivist value 248 combination of t values 86, 89, 98, 133, 135–137, 195–196, 268, 270, 272–274, 297, 314 COMECON see Council for Mutual Economic Assistance common decision-making system 185 Commonwealth of Independent State 81, 265 Communist Party 14, 20n1, 169, 187, 246 comparative economic systems 184 conditional radicalism 32, 34–38, 43–44, 46, 48, 51, 60n10 contextual approach/paradigm/perspective 240, 247, 250 convergence/converge 42, 153, 248, 251, 253, 257n10, 283n9 corporate finance 181, 184, 230

INDEX corporate governance 181, 250 corruption 16, 58, 145–160, 162–172, 173n2, 173n4, 173n7, 173n10, 189 Corruption Perception Index 146, 148 Council for Mutual Economic Assistance 18, 263, 266, 283n1, 283n3 Council of Ministers (of USSR) 13 CPI see Corruption Perception Index Cramér’s coefficient of association (Cramér’s V) 42–46, 61n15, 346, 350–353 Croatian War 82 culture 16–17, 150, 153, 157, 160, 166, 171–172, 173n8, 234n6, 239, 241, 244, 247–249 culture value 241 democracy 32–33, 37, 78, 146, 150–151, 159–160, 164, 171, 173n9, 331, 340, 349–350, 353, 355, 358, 361, 365n19 democracy indicator 78 democratization 78, 90, 96–98, 101–102, 112n3, 151–152, 158, 164, 172, 331, 333, 340, 342, 347, 349, 365n19 dependency ratio 131–132, 138 dichotomy 27, 54, 57–58, 184 direct sales to strategic investors 181, 189–190, 208, 217, 231–232 disinflation 22 disorganization 39 disorganization hypothesis 75 distance 15, 18, 34, 53, 249, 269–275, 279, 281–282, 330 divergence 17, 253–254, 256, 257n10, 333, 335–336, 349 diversity 17, 76, 179, 181, 184, 188, 251, 256, 265, 346 EBRD liberalization index 272–273 EBRD see European Bank for Reconstruction and Development eclectic gradualism 15, 26, 34–36, 38, 40–41, 43–44, 46, 48, 51, 54–56 ecological modernization 361 EconLit 27, 60n7, 84, 125–126, 142n1, 192, 240, 266, 291 economic risk index 272–274, 284n16 economies of shortage 75 education level 15–16, 70, 75, 112n6, 134, 138, 141 employee stock ownership plan 185 endogeneity 114n14, 158, 306 endogenous growth theory 290

ENGO see environmental non-governmental organization environmental acquis 19, 332–333, 342, 347, 364n16 environmental governance 333, 335–336 environmental non-governmental organization 330–331, 335–336, 338–339, 341–342, 347–348, 354, 356, 358, 360 environmental reform 19, 329–331, 333–335, 337, 342–343, 347–362 environmentalism 330–331 ESOP see employee stock ownership plan ethnic conflict 75, 83, 363n1 EU accession 293, 332–333, 340, 342–343, 347, 350, 353, 355, 357, 360–362, 365n23 EU agreement 273 EU candidate 273, 342 EU Commission 343 EU enlargement 188, 364n16 EU factor variable 18, 272, 274, 282 EU member countries/states 85, 148–149, 154, 170, 180, 200, 242–243, 252, 254–255, 285, 287, 359 EU membership 154, 188, 243, 251, 272, 283n9, 342 EU participation 242, 245, 252, 256 EU see European Union European Bank for Reconstruction and Development 69, 72, 74, 78–79, 81, 151, 153–154, 179–180, 182, 188, 257n6, 272–273, 288–289, 293, 295, 306, 324n7, 345, 362 European Court of Justice 19, 332, 343 European Union 17, 19, 76, 85, 91, 148, 154, 170, 188, 242, 244–245, 251–252, 255–256, 265, 269, 272–273, 293–294, 301, 303, 310, 329–334, 341–343, 345, 347–348, 350, 360–362, 363n3, 363n6, 364n16, 364n17 Europeanization 333, 336, 343 evolutionary economics 80 excess liquidity 75 FAT see funnel asymmetry test FAT-PET-PEESE approach 21n10 FDI see foreign direct investment financial deepening 273 financial depth 77, 272, 274, 284n16 Fisher’s z-transformation 20n4 fixed-effects panel estimator 114, 199, 299 foreign direct investment 8, 18–19, 21n13, 112n6, 158, 166, 272–273, 283n9, 285–323, 365n25

375

INDEX foreign investor 18, 183–184, 186–190, 193–198, 201–202, 204, 206–213, 215–231, 288 frequentist model averaging 21n6 funnel asymmetry test 9–10, 21n10, 105, 107–108, 111, 139–141, 225–226, 229, 275–280, 309, 321 funnel plot 7–9, 103–104, 111, 133, 137–138, 222–223, 225, 275, 307, 319–320 Galbraith plot 7–8, 10, 103, 106, 111, 224–225, 307–308, 320, 322 GDP see gross domestic product genuine evidence/effect 7, 9–11, 107, 111, 133, 137–142, 222, 225, 231, 275–282, 291, 308, 320, 322–323 gigantism 76 Gini coefficient 121 goodness-of-fit test 103 Gorbachev administration 82 governance 150, 152–153, 156, 158, 162, 168, 171, 173n7, 180–181, 250, 289, 333, 335–336, 341 gradualism 14–15, 25–27, 32–45, 48, 51–59, 60n11, 61n18, 96 gravity model 18, 264, 267–272, 274, 282, 283n13, 284n14, 294 greasing-the-wheels hypothesis 16, 146, 152, 166, 171–172 gross domestic product 18, 67–69, 71–72, 74–77, 84–85, 91, 93, 95–100, 129–130, 164, 182, 191, 200, 217, 234, 265–267, 269–275, 278, 281–282, 283n9, 289, 294, 297, 299, 302, 304, 310–311, 314–319 hard budget constraints 186 Hofstede, Geert 249–255 homogeneity test 5–6, 97, 196, 298, 313 household size 16, 131–132, 134–136, 138–140 HRM see human resource management Huber-White sandwich estimator 48, 53, 56, 355 human resource management 17, 239–257 hyperinflation 75, 82, 107 IDEAS database 11–12 IMF see International Monetary Fund impact factor 11, 31, 133, 272, 340, 347, 353 individualism 248–249 individualist value 248 inflation 15, 27, 69–70, 75, 81–83, 85–90, 94, 96, 104–111, 113n8, 114n14

376

informal economy/sector 152, 158, 162, 167, 170–171 informal institutions 41 information asymmetry 188, 190, 233 in-house labor market 185 insider inefficiency hypothesis 185 institutional economics 40, 80 institutional gradualism 33, 35–36, 38–40, 56, 60n11 institutional investor 186–187, 193, 195, 198, 206–207, 214, 217, 222, 229 International Monetary Fund 29–30, 43, 47–49, 52, 55, 60n2, 294, 324n7 intra-EU division of labor 265 investment fund 189 investment privatization fund 189 Jacobs, Jane 145 J-curve phenomenon 107 J-curved growth path 15, 67, 70–71, 84, 87, 96, 107, 111 kernel density estimation 86, 88 Klaus, Václav 32, 37, 331, 349, 363n2 Kornai, János 15, 33, 71, 75, 77, 81, 232 Kosovo War 82 Kruskal-Wallis test 68, 73 legacy of socialism 15, 70, 75, 80–81, 83, 87, 111 Lenin, Vladimir Illich 13 liberalization 27, 37, 39–40, 45, 48, 50, 52, 55, 60n2, 78–79, 82, 90, 96–98, 101–102, 151–153, 157, 160, 166, 170–171, 270–273, 289, 293, 295, 297–298, 300, 302, 305, 324n7 logistics 271 management and employee buyouts 165, 181–183, 185, 189–191, 200, 206, 214–217 management know-how 186 management/managerial turnover 22, 249–250 market capitalization 90, 96–100 market failure 349 market fundamentalism 330 Marx, Karl 11–12 Marxist society 145 mass privatization 232 MEBOs see management and employee buyouts meta fixed-effect model 5, 21n5, 86 meta random-effects model 5–6, 86

INDEX meta-analysis 3–5, 7, 11–12, 14–19, 20n3, 61n19, 70–71, 75, 78, 83–85, 96, 103, 107, 112n1, 113n6, 113n11, 125, 130–131, 133, 140, 181–183, 187–188, 192, 194, 222, 231–233, 233n1, 233n3, 234n9, 234n11, 263–264, 268–270, 272–273, 283n11, 283n13, 291–293, 298, 300, 307, 310, 313, 320, 322, 324n6, 338 meta-regression analysis 6–7, 21n6, 70–71, 88, 107, 137, 197, 201, 208, 230, 233n4, 268, 291–292, 295, 298, 300, 306, 310, 312–313, 320, 323, 324n4 meta-synthesis 3–4, 6, 15, 70, 86–88, 97, 113n8, 183, 194, 196, 208, 230, 272, 282, 295–296, 311 method of moments estimator 6, 91, 299 missed studies problem 2 misuse of power 150 model uncertainty 21n6 model-averaging approach 21n6 modern rationalism 290 monetary overhang 75 MRA see meta-regression analysis multilevel mixed-effects model/estimator 7, 10, 92, 94, 99, 100, 114n11, 209–210, 212–213, 215–216, 218–221 multinational enterprise 18, 241, 256n2, 257n12, 285, 290–291, 294 multinomial logit choice model 54 narrative review 2–4, 20n3, 334 Natura 2000 342–343 natural resources 158, 165 neoclassical growth theory 288 neoclassical model 41 nepotism 16, 145 neutralism 34–36, 41–44, 46–48, 51 North, Douglass C. 34, 40–41, 47 Nye, Joseph S. 16, 145 objectivity 3 OECD see Organization for Economic Cooperation and Development OLS see ordinary least squares ordered logit model 61n16 ordered probit model 48 ordinary least squares model/estimator 7, 10, 21n7, 91–94, 99–102, 108–110, 138–141, 199, 202–207, 209–210, 212–213, 215–216, 218–221, 226–228, 253, 255, 275–280, 301–304, 306, 309, 315–318, 321

Organization for Economic Cooperation and Development 149, 294 outsourcing 290 ownership concentration 8–10, 22 partial correlation coefficient 4–5, 7, 20n4, 86–89, 92, 96–99, 101, 103, 105, 133, 135–136, 195–197, 201–202, 206, 209, 212, 215, 217–218, 220, 222–223, 272–274, 281, 293, 295–298, 301–302, 306, 311–314, 316, 320–322 path dependency 22, 26, 80 path-dependent 244–245 PCC see partial correlation coefficient PEESE see precision effect estimate with standard error PET see precision effect test physical-good-focused production system 76 political governance 156, 158 political risk 272–274, 284n16 political systems 19, 150, 164, 171, 173n9, 329 post-Keynesian 41 poverty 15–16, 33, 119–143, 151, 153 poverty headcount 120–125 poverty line 119–120, 134, 138 precision 3, 7–9, 222, 323 precision effect estimate with standard error 10, 21, 107, 110–111, 139–141, 225, 228–229, 275–280, 309, 321 precision effect test 9–11, 21, 107–108, 111, 139–141, 225–226, 229, 275–280, 309, 321 privatization 16–17, 21–22, 33, 39, 59, 60n2, 78, 90, 133–134, 151, 157, 160, 165–166, 170–171, 179–234, 241, 249, 289, 293, 295, 297–298, 300, 330, 347, 359 privatization check 189 privatization coupon 189 privatization debate 184 production technology 186, 290 PSB see publication selection bias publication selection bias 4, 7–11, 15, 21n7, 21n10, 56, 70–71, 103, 107, 109, 111, 133, 137–138, 142, 183, 194, 222, 224–225, 227, 229, 231, 268, 275–280, 282, 291–292, 307–310, 320–323, 324n9 Q statistic 5 radicalism 14–15, 25–27, 32–39, 41–44, 46, 48–54, 56–59, 60n10, 61n18, 96 random-effects panel estimator 7, 91, 114, 199, 299

377

INDEX red executives 185, 249 regime change/transformation 19, 329–331, 333–335, 337, 343, 347–350, 359–361 regional conflict 15, 69–70, 75, 82–83, 85–86, 88–90, 92, 94, 96, 104–111, 113n8 regression analysis of regression analyses 6 religion 153–154, 160, 169, 171, 173n8 rent-seeking 145, 164–165 replicability 3 risk of poverty 131–132, 138 Rosenthal’s failsafe N 6, 86, 89, 98, 133, 135–137, 195, 272–274, 297–298, 314 rule of law 33, 39, 61n20, 78–80, 90, 96–98, 101–102, 146, 157, 159, 166, 168, 173n7 rural residence 131, 135–136, 138 ruralization of poverty 137 sampling error 5–6 scientific socialism 12 selective reporting 7 Shapiro-Wilk normality test 86–87, 296, 311–312 shock therapy 25, 27, 37, 61n17 slow-paced gradualism 15, 26, 33, 35–36, 38–39, 42, 48, 51, 54–56 socialist legacy 17, 69–70, 80–81, 85–90, 92, 94, 96, 104–111, 113n8, 243–245, 249–252, 256 socialist market economy 1, 20 socialist personnel management 17, 239–240, 244–246, 251, 253–254, 257n4 socialist planned economy 1, 12–14, 16–17, 25, 59, 75, 81, 243 Southeastern European countries 257n6, 271, 362 soviet model of management 240 soviet-type personnel management 246, 251 soviet-type planned system 13 state capture 164, 168, 188 state monopolization of trade 263 step-by-step gradualism 15, 26, 33, 35–36, 38, 48, 51, 54–56 Stiglitz, Joseph E. 34, 40–41, 47, 79, 113n9, 184, 232 strategic investor 181, 189–191, 208, 217, 231–232 structural change 15, 69–70, 75–77, 79, 83, 85–92, 94, 96–100, 103–111, 113n8, 264, 266, 273–275, 282 structural change variable 18, 272, 274–275, 282, 284n16

378

structural reform 1, 25–26, 33, 37–40, 57, 61n19, 91, 97–98, 101–102, 113n11, 187, 191, 273–274, 282 structural reform variable 18, 272, 274–275, 281–282, 284n16 sudden poverty 120–121, 125, 129 survey article 2–3, 20n2 sustainable development 19, 331, 333, 335, 363n2 systematic review 3–4, 14, 16–17, 19, 20n3, 127, 130, 147, 170, 181–182, 189, 333–338, 347, 361–362 t value 4–9, 86–89, 94, 96–98, 100, 102–103, 105, 133, 135–137, 195–197, 201, 204, 207, 210, 213, 216–217, 219, 221–222, 224–225, 268, 270, 272–274, 295–298, 303–304, 306, 308, 311–314, 317–318, 320, 322 tacit knowledge 186 third-way thinkers 57–59 Thomson Reuters 11 total FDI model 293, 297–300, 302, 304 trade openness 76, 90, 96, 98–100, 150, 271–274, 284n16 trade-related indicator 267 transaction cost 146, 151–152 transcendentalist 34, 59 transformation factor 18 transformation policy 15, 69–71, 75, 77–80, 83, 85–92, 94, 96–98, 101–104, 106–111, 113n8, 113n11 transformational recession 15, 33, 67, 71–72, 75, 77, 79–80 transition index 272–273 transition indicator 78–79, 133, 288, 293, 295, 297, 300, 302–303, 305–306, 324n6 transition strategy debate 14, 25–27, 29, 31–32, 34–35, 40–42, 54, 56–57, 59, 59n1, 60n7 Transparency International 16, 146–148, 155, 172n1 type I publication selection bias 8, 10, 21n7, 103, 105, 108, 111, 139–141, 224–226, 229, 231, 275–280, 307, 309, 320–321, 324n9 type II publication selection bias 8–10, 21n10, 103, 105, 107, 109, 111, 139–141, 224–225, 227, 229, 231, 275–276, 307–309, 321–322 UNCTAD see United Nations Conference on Trade and Development UNECE see United Nations Economic Commissions for Europe

INDEX unemployment 33, 131 United Nations Conference on Trade and Development 286–287, 289, 294, 300 United Nations Economic Commissions for Europe 264–265, 290, 310 universal radicalism 32, 35–38, 43–44, 46, 48, 51 universalist approach/paradigm 247, 250–251 urban residence 16, 131–132 voucher privatization 17, 21n13, 151, 183, 200, 211–214, 231–232 voucher system 181, 189–191, 208, 211, 214, 231–232

Washington Consensus 25, 27, 59n2, 232 Web of Science 84, 154, 192, 240, 291, 337 weighted least squares estimator/model 7, 11, 21n7, 21n8, 92, 94, 99–102, 202, 204,

206–207, 209–210, 212–213, 215–216,

218–221, 300–304, 316–318

Western/western-style HRM 17, 239, 241, 245–248, 257n3 WLS see weighted least squares model World Bank 29–30, 43, 47–49, 52, 55, 60n2, 78, 122, 124, 127, 130, 173n7, 271, 294, 332 World Development Indicators 122, 124, 130, 294 World Economic Outlook Database 294 Xiaoping, Den 20n1

379