Regional Development And Economic Growth In China 9789814439855, 9789814439848

Citation preview

Series on

E conomic Development

7

and Growth Vol.

Regional Development and Economic Growth in China

8655_9789814439848_tp.indd 1

4/3/13 4:37 PM

Series on Economic Development and Growth

(ISSN: 1793-3668)

Series Editor: Linda Yueh (University of Oxford & London School of Economics and Political Science, UK) Advisory Board Members: Magnus Blomström (Stockholm School of Economics, Sweden) Stefan Dercon (University of Oxford, UK) John Knight (University of Oxford, UK) Ari Kokko (Copenhagen Business School, Denmark) Li Shi (Beijing Normal University, China) Jonathan Story (INSEAD, France)

Published Vol. 1 Globalisation and Economic Growth in China edited by Yang Yao & Linda Yueh Vol. 2 Elderly Entrepreneurship in an Aging U.S. Economy: It’s Never Too Late by Ting Zhang Vol. 3 Industrial Development in East Asia: A Comparative Look at Japan, Korea, Taiwan, and Singapore (With CD-ROM) by K. Ali Akkemik Vol. 4 Inclusive Value Chains: A Pathway Out of Poverty by Malcolm Harper Vol. 5 Narratives of Chinese Economic Reforms: How Does China Cross the River? edited by Xiaobo Zhang, Shenggen Fan & Arjan de Haan Vol. 6 Corporate Ownership and Control: Corporate Governance and Economic Development in Sri Lanka by Shalini Perera Vol. 7 Regional Development and Economic Growth in China edited by Yanrui Wu

Forthcoming titles Institutional Change and the Development of Industrial Clusters in China: Case Studies from the Textile and Clothing Industry by Jinmin Wang

Lixi - Regional Development and Economic.pmd1

3/4/2013, 1:36 PM

E

Series on conomic Development and Growth Vol.

7

Regional Development and Economic Growth in China Edited by

Yanrui Wu University of Western Australia, Australia

World Scientific NEW JERSEY

•

LONDON

8655_9789814439848_tp.indd 2

•

SINGAPORE

•

BEIJING

•

SHANGHAI

•

HONG KONG

•

TA I P E I

•

CHENNAI

4/3/13 4:37 PM

Published by World Scientific Publishing Co. Pte. Ltd. 5 Toh Tuck Link, Singapore 596224 USA office: 27 Warren Street, Suite 401-402, Hackensack, NJ 07601 UK office: 57 Shelton Street, Covent Garden, London WC2H 9HE

Library of Congress Cataloging-in-Publication Data Regional development and economic growth in China / edited by Yanrui Wu, University of Western Australia. pages cm. -- (Series on economic development and growth ; vol. 7) ISBN 978-9814439848 1. China--Economic policy--2000---Regional disparities. 2. Regional planning--China-Regional disparities. 3. Economic development--China--History--21st century. 4. Industrial productivity--China. 5. China--Economic conditions--2000– I. Wu, Yanrui. HC427.92.R446 2013 338.951--dc23 2012048839

British Library Cataloguing-in-Publication Data A catalogue record for this book is available from the British Library.

Copyright © 2013 by World Scientific Publishing Co. Pte. Ltd. All rights reserved. This book, or parts thereof, may not be reproduced in any form or by any means, electronic or mechanical, including photocopying, recording or any information storage and retrieval system now known or to be invented, without written permission from the Publisher.

For photocopying of material in this volume, please pay a copying fee through the Copyright Clearance Center, Inc., 222 Rosewood Drive, Danvers, MA 01923, USA. In this case permission to photocopy is not required from the publisher.

In-house Editor: DONG Lixi

Typeset by Stallion Press Email: [email protected]

Printed in Singapore

Lixi - Regional Development and Economic.pmd2

3/4/2013, 1:36 PM

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

Acknowledgments

This edited book contains a collection of papers written by a distinguished group of authors in Australia, China, Indonesia, Japan, and Vietnam. The papers were selected through a blind-refereed process. Some of the papers were presented at the ACESA Annual Conference on the Chinese Economy “China’s Growth and the World Economy”, July 7–8, 2011, UWA Business School, Perth, Australia. The conference was generously supported by AusAID (Canberra), the Ford Foundation (Beijing Office), UWA Business School and the UWA Vice-Chancellor’s Office. Others who helped the conference include Danni Figg, Jenny Hu, Aya Kelly, Ha Le, Eleni Stephanou, and Anna Wiechecki. The final publication of this volume is kindly supported by Michael Heng Siam Heng and Chen Kang. I also thank Rebecca Doran-Wu, Kristi Ng, David Silbert and Fei Yu for their excellent research assistance. Yanrui Wu Perth, Australia

v

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

Contents

Acknowledgments List of Contributors List of Tables List of Figures Chapter 1.

v ix xi xv

Regional Economies and Growth: An Introduction Yanrui Wu

1

Part I Regional Development Chapter 2. New Evidence of Regional Inequality Tsun Se Cheong

15

Chapter 3.

The Energy Tax and Regional Development Zhengning Pu and Yasuhisa Hayashiyama

47

Chapter 4.

Regional Distribution of the Creative Class and Its Determinants Jin Hong, Wentao Yu and Fengli Yang

Chapter 5.

Comparing Productivity Growth among the Regions Yanrui Wu

Part II Manufacturing Sector, FDI and Economic Growth Chapter 6. FDI and Economic Growth Chunlai Chen Chapter 7.

Chapter 8.

75 97

117

Manufacturing Sector FDI and Performance of Domestic Banks Sizhong Sun and Siqiwen Li

141

Patterns of Industrial Dynamics in the Manufacturing Sector Sizhong Sun

161

vii

March 5, 2013

viii

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

Regional Development and Economic Growth in China

Chapter 9.

Agglomeration and Export Performance of Manufacturing Firms Dahai Fu

Part III China and Neighboring Economies Chapter 10. Individual Country Approaches to Agriculture in the ASEAN–China FTA Ray Trewin, David Vanzetti, Nur Rakhman Setyoko, and Nguyen Ngoc Que Chapter 11.

Chapter 12.

Chapter 13.

Index

189

221

Environmental Regulation and Productivity Growth in APEC Economies Bing Wang and Yanrui Wu

253

Inflation Transmission in China’s Goods and Asset Markets Huawei Liu and Juan Yang

285

Inconsistency in the Assessment of China’s Domestic and Foreign Patents Fei Yu

303 327

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

List of Contributors Chunlai Chen, Australian National University, Canberra, Australia Tsun Se Cheong, University of Western Australia, Perth, Australia Dahai Fu, Central University of Economics and Finance, Beijing, China Yasuhisa Hayashiyama, Tohoku University, Japan Jin Hong, University of Science and Technology of China, Hefei, China Siqiwen Li, James Cook University, Townsville, Australia Huawei Liu, Peking University, Shenzhen, China Zhengning Pu, Tohoku University, Japan Nguyen Ngoc Que, Institute of Policy and Strategy for Agricultural and Rural Development, Hanoi, Vietnam Nur Rakhman Setyoko, Agency for Trade Research and Development, Jakarta, Indonesia Sizhong Sun, James Cook University, Townsville, Australia Ray Trewin, Australian National University, Canberra, Australia David Vanzetti, Australian National University, Canberra, Australia Bing Wang, Jinan University, Guangzhou, China Yanrui Wu, University of Western Australia, Perth, Australia Fengli Yang, University of Science and Technology of China, Hefei, China Juan Yang, Peking University, Shenzhen, China Fei Yu, University of Western Australia, Perth, Australia Wentao Yu, University of Science and Technology of China, Hefei, China

ix

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

List of Tables

1.1. 1.2. 1.3. 2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7. 2.8.

2.9.

3.1. 3.2. 3.3. 4.1. 4.2. 4.3.

Ranking of Chinese Regions in Terms of Per Capita Income . . . . . . . . . . . . . . . . . . . . Regional Shares (%), 2010 . . . . . . . . . . . . . . . . . Growth Rates (%) across the Regions . . . . . . . . . . . . Number of Counties and County-level Cities in Database . . . . . . . . . . . . . . . . . . . . . . . . . Changes in Inter-CU Inequalities for Different Spatial Groupings . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-CU Inequalities . . . . . . . Changes in Inter-County Inequalities for Different Spatial Groupings . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-County Inequalities . . . . . Changes in Inter-City Inequalities for Different Spatial Groupings . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-City Inequalities . . . . . . Inequality Changes and Percentage Changes for the National Inter-CU Inequality, Inter-County Inequality, and Inter-City Inequality . . . . . . . . . . . . Yearly Growth (%) of Gini Coefficient for the National Inter-CU Inequality, Inter-County Inequality, and Inter-City Inequality . . . . . . . . . . . . . . . . . . Primary Energy Composition . . . . . . . . . . . . . . . . Regional Division Code . . . . . . . . . . . . . . . . . . . Reclassified Commodity Sectors . . . . . . . . . . . . . . Descriptive Analysis of All Variables . . . . . . . . . . . . Correlation Between Creative Class and Other Variables . . . . . . . . . . . . . . . . . . . . . Ridge Parameters of Different Criteria for Each Variable Regression Coefficient . . . . . . . . . . . . . . . . . . .

xi

3 5 6 23 27 30 33 36 38 41

42

43 53 57 60 86 89 91

March 5, 2013

xii

6.2. 6.3. 6.4. 6.5. 7.1. 7.2. 7.3. 7.4. 7.5. 7.6. 7.7. 8.1. 8.2. 8.3. 8.4. 8.5. 8.6. 8.7. 8.8. 9.1.

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

Regional Development and Economic Growth in China

5.1. 5.2. 5.3. 5.4. A5.1. A5.2. 6.1.

11:55

Rates of Depreciation in the Chinese Economy . . . . . . . Estimation Results . . . . . . . . . . . . . . . . . . . . . . TFP Growth Rates . . . . . . . . . . . . . . . . . . . . . . TFP Growth Rates across the Regions . . . . . . . . . . . Technological Progress in China’s Three Regions . . . . . Technical Efficiency Changes in China’s Three Regions . . . . . . . . . . . . . . . . . . . . . . . . FDI Firms’ International Trade Performance, 1980–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . Variables of the Impact of FDI on China’s Provincial Economic Growth . . . . . . . . . . . . . . . . . . . . . . Regression Results of Production Function ofAll Provinces, 1987–2005 . . . . . . . . . . . . . . . . . . . . . . . . . . Economic and Technological Indicators by Region . . . . . Regression Results of Production Function by Regions, 1987–2005 (Fixed-effects Model) . . . . . . . . . . . . . . Summary Statistics of ROAs and ROEs of 12 Commercial Banks . . . . . . . . . . . . . . . . . . Summary Statistics . . . . . . . . . . . . . . . . . . . . . Regression Results of Fixed Effect Estimation . . . . . . . Regression Results of Random Effect Estimation . . . . . Regression Results of Pooled OLS Estimation . . . . . . . Regression Results Using Employment Share as Measure of FDI Presence . . . . . . . . . . . . . . . . . . . . . . . Regression Results Using Assets Share as Measure of FDI Presence . . . . . . . . . . . . . . . . . . . . . . . Number of Firms by Industry . . . . . . . . . . . . . . . . Descriptive Statistics of Normalized Firm Size by Years . . . . . . . . . . . . . . . . . . . . . . . . . . . Regression Results of Equations (8.2) and (8.3) . . . . . . Subbotin Estimation over Pooled Data . . . . . . . . . . . Sectoral Growth Process Regression Results . . . . . . . . Subbotin Estimation at the Disaggregate Level . . . . . . . Summary of the Estimate of the Subbotin Parameters . . . Determinants of Firm Growth Rate Distributions . . . . . . Export Patterns in the Data . . . . . . . . . . . . . . . . .

101 103 106 109 110 111 123 130 131 134 135 147 154 154 156 156 157 157 164 167 170 172 176 179 181 182 202

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

List of Tables

xiii

Industrial Distribution of Exporters in 1998 and 2007 . . . Regional Distribution of Exporters in 1998 and 2007 . . . Baseline Results for Export Participation: Pooled Probit Models . . . . . . . . . . . . . . . . . . . . . . . . 9.5. Robustness Check (I) for Export Participation . . . . . . . 9.6. Robustness Check (II) for Export Participation . . . . . . . 9.7. Robustness Check (III) for Export Participation . . . . . . 9.8. Baseline Results for Export Intensity: PPML Estimation . . . . . . . . . . . . . . . . . . . . . . 9.9. Robustness Check (I) for Export Intensity: Fractional Probit Models . . . . . . . . . . . . . . . . . . . . . . . . 9.10. Robustness Check (I) for Export Intensity: Heckit Model . . . . . . . . . . . . . . . . . . . . . . . . 10.1. Chinese, Indonesian and Vietnamese Food Trade Flows and Shares of Total Merchandise Trade . . . . . . . . . . . . . 10.2. Bound and Applied Simple Average Tariffs 2010 . . . . . . 10.3. Number of WTO Notifications and Measures in Force, and Number of Disputes . . . . . . . . . . . . . . . . . . 10.4. NRAs to all Agricultural Products, Indonesia, Vietnam and China, 1996 to 2005 . . . . . . . . . . . . . . . . . . 10.5(a). Base and Final Indonesian and Chinese Bilateral Tariffs . . . . . . . . . . . . . . . . . . . . . . . 10.5(b). Base and Final Vietnamese and Chinese Bilateral Tariffs . . . . . . . . . . . . . . . . . . . . . . . 10.6. Welfare Impacts . . . . . . . . . . . . . . . . . . . . . . . 10.7. Source of Welfare Gains . . . . . . . . . . . . . . . . . . . 10.8. Change in Exports . . . . . . . . . . . . . . . . . . . . . . 10.9. Change in Imports . . . . . . . . . . . . . . . . . . . . . . 11.1. Summary of Main Studies on Productivity Growth in APEC Economies . . . . . . . . . . . . . . . . . . . . . 11.2. Summary Statistics of the Sample . . . . . . . . . . . . . 11.3. Average Productivity Growth, 1980–2004 . . . . . . . . . 11.4. Countries Shifting the Frontiers . . . . . . . . . . . . . . . 11.5. Factors Associated with Changes in Productivity . . . . . .

203 204

9.2. 9.3. 9.4.

205 207 208 210 212 213 214 224 226 227 228 233 234 238 240 241 243 257 265 267 271 275

March 5, 2013

xiv

11:55

12.4. 12.5. 13.1.

Regional Development and Economic Growth. . . b1491-fm

Regional Development and Economic Growth in China

A11.1. A11.2. A11.3. 12.1. 12.2. 12.3.

9in x 6in

Average Indices and Changes under Scenario 1 . . . . . . . Average Indices and Changes under Scenario 2 . . . . . . . Average Indices and Changes under Scenario 3 . . . . . . . Augmented Dickey–Fuller Test for the Level Series . . . . Augmented Dickey–Fuller Test for the First Difference of Each Series . . . . . . . . . . . . . . . . . . . . . . . . One Step Schwarz Loss Criteria by Lags on the Number of Co-Integrating Vectors (R) and Model Specifications Fit over Period January 2005–December 2010 . . . . . . . . . The Correlation Matrix among the Variables . . . . . . . . Variance Decomposition of CPI (Unit %) . . . . . . . . . . Regression Results . . . . . . . . . . . . . . . . . . . . .

278 279 280 291 292

293 295 299 317

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

List of Figures

1.1. 1.2. 2.1. 2.2. 2.3. 2.4. 2.5. 2.6. 2.7. 2.8. 2.9. 2.10. 2.11. 2.12. 3.1. 3.2. 3.3. 3.4. 3.5.

Income Per Capita among China’s Regions in 2011 . . . . Regional Export Values over GRP (%) in 2011 . . . . . . . Inter-CU Inequalities for Different Spatial Groupings . . . Intra-Provincial Inter-CU Inequalities in 1997 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Intra-Provincial Inter-CU Inequalities in 2007 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-CU Inequalities in 1997 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . Inter-County Inequalities for Different Spatial Groupings . . . . . . . . . . . . . . . . . . . . . . Intra-Provincial Inter-County Inequalities in 1997 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Intra-Provincial Inter-County Inequalities in 2007 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-County Inequalities in 1997 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . Inter-City Inequalities for Different Spatial Groupings . . . . . . . . . . . . . . . . . . . . . . Intra-Provincial Inter-City Inequalities in 1997 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Intra-Provincial Inter-City Inequalities in 2007 (Gini Coefficients) . . . . . . . . . . . . . . . . . . . . . . Relative Intra-Provincial Inter-City Inequalities in 1997 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . Production Structure . . . . . . . . . . . . . . . . . . . . Capital–Energy Composite . . . . . . . . . . . . . . . . . Household Activities . . . . . . . . . . . . . . . . . . . . Government Activities . . . . . . . . . . . . . . . . . . . . Export Structure . . . . . . . . . . . . . . . . . . . . . . .

xv

2 4 26 28 29 31 32 34 35 37 37 39 40 42 52 53 54 55 56

March 5, 2013

xvi

3.6. 3.7. 3.8. 3.9. 3.10. 3.11. 3.12. 4.1. 4.2. 4.3. 5.1. 5.2. 5.3. 5.4. 5.5. 6.1. 6.2. 7.1. 7.2. 7.3. 7.4. 7.5. 8.1. 8.2. 8.3. 8.4.

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

Regional Development and Economic Growth in China

Import Structure . . . . . . . . . . . . . . . . . . . . . . . Energy Output Reduction of China . . . . . . . . . . . . . Industry Output Rate of Change . . . . . . . . . . . . . . Regional GDP Rate of Change . . . . . . . . . . . . . . . Regional Household Utility Rate of Change . . . . . . . . Regional Petroleum and Natural Gas Output Change . . . . Regional Coal Output Change . . . . . . . . . . . . . . . Creative Class over Total Population in 31 Provinces (2007) . . . . . . . . . . . . . . . . . . . Relative Growth Rate of Provincial Creative Class from 2000 to 2007 . . . . . . . . . . . . . . . . . . . . . . Ridge Trace of Six Variables Ridge Regression . . . . . . . Growth of China’s Total Employment 1979–2006 . . . . . China’s Best Practice Performers, 1978–2005 . . . . . . . Standard Deviation of Regional TFP Growth Rates, 1979–2005 . . . . . . . . . . . . . . . . . . . . . . . . . . Map of China . . . . . . . . . . . . . . . . . . . . . . . . Standard Deviation of TFP Growth Rates in the Western Regions . . . . . . . . . . . . . . . . . . . FDI Inflows as a Percentage of GFCF and Percent of Investment in Fixed Assets in China, 1994–2008 . . . . FDI Firms’ Manufacturing Employment in China, 1995–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . Return on Assets (ROA) of 12 Domestic Commercial Banks in China . . . . . . . . . . . . . . . . . . . . . . . . . . . Return on Equity (ROE) of Selected Banks . . . . . . . . . Average FDI Presence in the Manufacturing Sector 2000–2007 . . . . . . . . . . . . . . . . . . . . . . . . . . Standard Deviation of FDI Presence in the Manufacturing Sector 2000–2007 . . . . . . . . . . . . . . . . . . . . . . A Conceptual Framework . . . . . . . . . . . . . . . . . . The Growth Pattern of the Chinese Manufacturing Sector . . . . . . . . . . . . . . . . . . . . The Distribution of Firm Size by Year . . . . . . . . . . . The Distribution of Growth Rate by Year . . . . . . . . . . The Distribution of Firm Size in the Textile Sector . . . . .

56 61 62 63 64 65 66 78 80 88 102 105 107 108 109 121 122 146 147 148 148 149 165 168 171 173

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

List of Figures

8.5. 8.6. 10.1. 12.1. 12.2. 12.3. 12.4. 13.1. 13.2. 13.3. 13.4. 13.5. 13.6. 13.7. 13.8. 13.9. 13.10.

The Distribution of Firm Size in the Chemical Materials and Chemical Products Sector . . . . . . . . . . . . . . . The Distribution of Firm Size in the Non-metallic Mineral Products Sector . . . . . . . . . . . . . . . . . . . . . . . Indonesian Sugar Imports and Change in Welfare . . . . . The Growth Rate of Money Supply, CPI, GDP and the Interest Rate from 2006 to 2010 in China . . . . . Time Trend for the Housing Price and Shanghai Composite Stock Index from 2005 to 2010 . . . . . . . . . . . . . . . The Response of CPI, FP, HP, SP, GDP, ER to Impulse of M2 . . . . . . . . . . . . . . . . . . . . . . . . . . . . The Response of CPI to Impulses of Other Variables . . . . PATSTAT Coverage of the Chinese Patent Data . . . . . . Number of Patents by Source of Origin . . . . . . . . . . . Number of Patents in Patent Families (N), by Origins . . . Patents Granted and Rejected in the Chinese Patent Office . . . . . . . . . . . . . . . . . . . . . . . . . Granted Patents, by Origins . . . . . . . . . . . . . . . . . Rejected Patent Applications, by Origins . . . . . . . . . . Rejected Patent Applications, by Grant Lag (L) . . . . . . Granted Patents, by Grant Lag (L) . . . . . . . . . . . . . Transformed Results from the Logit Regression . . . . . . Transformed Results from the Count Regression . . . . . .

xvii

174 174 246 286 286 296 297 310 311 312 312 313 314 316 316 320 321

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-fm

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Chapter 1

Regional Economies and Growth: An Introduction Yanrui Wu

Regional development and growth is always considered as an important issue in a large country like China. It is vital to the overall national development and growth. In China, regional development and growth can lead to serious social and political implications given the country’s diverse ethnic mix. Having enjoyed uninterrupted high growth for over 30 years, Chinese policy makers are now facing the horrendous challenge to balance development and growth among the regions and promote a “harmonious society” in the country. This edited volume explores some of the critical issues associated with regional development and is hence timely. In this introductory chapter, some background discussion is presented first in Section 1.1. This is followed by an outline of the chapters in Section 1.2.

1.1. China’s Regional Economies and Development According to the administrative classification, mainland China is divided into 31 regions, namely, 22 provinces, 5 autonomous regions and 4 municipalities. The level of economic development among the regions is very diverse. In terms of gross regional product (GRP) per capita, for example, the three richest regions (i.e., Tianjin, Shanghai, and Beijing) in 2011 reached 84,337 RMB (or US$13,057), 82,560 RMB (or US$12,782) and 80,394 RMB (or US$12,447), respectively (Figure 1.1).1 In the same year, the three 1 The exchange rate used for the conversion is 6.459 Yuan or RMB per US$1, reported in

NBS (2012). RMB is short for Renminbi or the name of Chinese currency.

1

March 5, 2013

2

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Development and Economic Growth in China

Figure 1.1.

Income Per Capita among China’s Regions in 2011.

Sources: Author’s own work. The raw data are drawn from the NBS (2012) and converted into US dollars using the official exchange rate reported in NBS (2012).

poorest regions (i.e., Guizhou, Yunnan, and Gansu) were recorded with per capita income of 16,413 RMB (or US$2,541), 18,957 RMB (or US$2,935) and 19,517 RMB (or US$3,022), respectively. In general the coastal regions are better developed (four or five stars in Figure 1.1) than the interior ones. This pattern is observed in many countries of the world largely due to the advantage of being close to the seas and hence enjoying low costs of transportation. In the Chinese case, economic policies also play a role in affecting regional development and growth. Since the initiative of China’s economic reform in the late 1970s, the coastal regions have been offered special concessions in terms of international trade and foreign investment. Traditionally the coastal areas have also been better endowed with human capital and entrepreneurship. Thus geographic factors

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Economies and Growth: An Introduction

3

Table 1.1. Ranking of Chinese Regions in Terms of Per Capita Income. Regions Tianjin Shanghai Beijing Jiangsu Zhejiang Inner Mongolia Liaoning Guangdong Shandong Fujian Jilin Hubei Hebei Shaanxi Heilongjiang Ningxia Shanxi Hunan Xinjiang Henan Qinghai Sichuan Jiangxi Anhui Guangxi Gansu Yunnan Guizhou

1981

1994

2011

3 1 2 11 12 13 4 15 17 19 9 20 8 16 5 6 10 21 14 25 7 24 23 27 22 18 26 28

3 1 2 7 5 15 6 4 9 8 12 14 13 26 10 20 17 19 11 24 16 22 25 21 18 27 23 28

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28

Source: Author’s own work. Tibet, Chongqing and Hainan are excluded due to missing data.

and biased policies have changed the landscape of regional economies in China over the past three decades. There are clearly winners and losers over time as the change in the ranking positions of the regions shows. It is apparent in Table 1.1 that the winners are mainly coastal regions such

March 5, 2013

4

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Development and Economic Growth in China

as Zhejiang, Jiangsu, Fujian, Guangdong, and Shandong. There are three regions (i.e., Qinghai, Heilongjiang, and Ningxia) which recorded the large decline in their ranking position among the 28 regions. One of the main development policies in China’s growth over the past decades is economic openness. For example, the total value of China’s exports accounts for 35% of the country’s gross domestic product (GDP) which is far greater than those of other large economies in the world such as the US (13%) and Japan (18%).2 Many economists have provided evidence to confirm a positive contribution of openness to the economic growth (Edwards, 1998; Harrison, 1996; Frankel and Romer, 1999). There is however considerable regional variation in terms of openness in China. As shown in Figure 1.2, regional export value as a share of GRP ranges from about 1.3% in Guizhou to almost 70% in Guangdong in 2011. It is noted that, among the 10 regions with the smallest shares, nine of them are located in Western China (with the exception of Jilin, a Northeast province). 70

%

60 50 40 30 20 10

Qinghai Gansu Inner Mongolia Guizhou Jilin Shaanxi Hunan Shanxi Yunnan Guangxi Heilongjiang Henan Hainan Hubei Ningxia Sichuan Anhui Jiangxi Hebei Chongqing Tibet Beijing Xinjiang Liaoning Shandong Tianjin Fujian Jiangsu Zhejiang Shanghai Guangdong

0

Figure 1.2.

Regional Export Values over GRP (%) in 2011.

Source: The raw data are drawn from NBS (2012). 2 These are 2009 figures and are drawn from the World Development Indicators database (World Bank, 2011).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Economies and Growth: An Introduction

5

Table 1.2. Regional Shares (%), 2010. Indicators Population GRP Uni/college students Hospital beds Length of high ways Optical cables Length of railways

Coastal

Central

Western

40.6 56.8 43.7 45.4 30.1 26.9 26.3

29.7 22.4 31.2 26.2 30.8 25.8 30.2

29.7 20.7 25.0 28.4 39.1 47.3 43.6

Sources: Author’s own estimates. The raw data are drawn from NBS (2011).

The huge difference in human resource and infrastructure development across the mentioned regions is associated with the disparity in economic development and openness. In general there is a large gap between the coastal areas and the rest of the country (Table 1.2).3 China’s Western regions are well-endowed with resources and hence have a lot of potential for further growth. Since the late 1990s, Chinese government has adopted specific policies in order to reduce regional disparity. One of the main policy initiatives was the “Western development program” adopted in 1999. This was followed by specific policies to promote growth in the six “Central regions” and reinvigorate the economies of the three “Northeast provinces”, respectively. Table 1.3 demonstrates that since 1999 resources have been diverted towards the Western regions.4 This is clearly shown by the rapid growth in gross capital formation in the last decade (Table 1.3). Since the middle of the 2000s, further initiatives have been adopted to promote growth in the Central and Northeast regions.5 These new regional development policies 3 The coastal regions include Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang,

Fujian, Shandong, Guangdong, and Hainan (11 provinces). The Central area covers Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, and Shanxi (eight provinces). The rest of China, namely the 12 Western regions, include Inner Mongolia, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Tibet, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang. 4 See endnote 3. 5 The six “Central regions” include Anhui, Jiangxi, Henan, Hubei, Hunan, and Shanxi. The three “Northeast” provinces are Liaoning, Jilin, and Heilongjiang.

March 5, 2013

6

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Development and Economic Growth in China

Table 1.3. Growth Rates (%) across the Regions. Periods

Western

Central

Northeast

Rest of China

Average growth rates of gross capital formation 1995–1999 9.01 10.55 6.33 2000–2004 16.52 12.30 14.51 2005–2010 17.04 17.76 22.02

10.25 10.44 14.50

Average growth of GRP 1995–1999 9.48 10.90 2000–2004 10.62 10.41 2005–2010 13.19 13.01

11.29 11.70 12.87

9.12 10.16 13.37

Sources: Author’s own estimates. The raw data are drawn from the NBS (various years).

have contributed to fast growth in the Western, Central, and Northeastern regions. In the last five years, the current government has also introduced special policies to support agriculture and hence rural communities. Examples include the abolition of agricultural taxes and waive of school fees in rural areas. More recently resource-rich regions have been allowed to increase their shares of profits from resource development. Though the effects are yet to be assessed and in the case of the resource tax reform it is still experimental, it can be anticipated that a balanced development approach will be prevalent in the coming decades in China. Relevant chapters in this book provide important insights into regional development and can hence contribute to the current policy debates in China.

1.2. Outline of the Chapters This book is divided into three parts. Part I has four chapters (Chapters 2 to 5) with a main focus on regional development in China. In Chapter 2, Cheong presented a new evidence of regional inequality. Most of the existing studies of regional inequality are based on provincial level data. Although some researchers considered county-level data, their studies are still plagued by either the problem of limited coverage or a short time-span. Chapter

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Economies and Growth: An Introduction

7

2 contributes to the literature on intra-provincial regional inequality by using county-level data. This is the first time in the literature that the intraprovincial regional inequality of many inland provinces is measured for a long time span. In addition to the intra-provincial regional inequality amongst the county-level units (namely, counties and county-level cities) in each province, inequality among other spatial groupings is also explored. China’s economic development is highly dependent on the use of fossil fuels in particular coal. In 2008, 78% of China’s electricity was derived from burning coal. To change this situation, China has introduced some new initiatives to alter the country’s energy tax system. These initiatives include the implementation of a carbon tax and the reform of the resource taxes. The latter is currently being pilot-tested in the Western provinces which are well-endowed with energy resources. In Chapter 3, Pu and Hayashiyama used a spatial computable general equilibrium (SCGE) model to evaluate the economic effects of the new resource tax on different regions of China. Their findings show that carbon emissions in China can be reduced by the introduction of an ad valorem energy tax. The authors argue that the new policy may cause different effects among Chinese regions which vary in terms of resource endowments, income levels and so on. They also argue that the new policy may be more effective in the petroleum and natural gas sectors but less effective in the coal industry. In Chapter 4, Hong et al. examine the geographic distribution of creative class and the impacts of creative class on regional economic development. This is an under-documented topic about the Chinese economy. In the existing literature nearly all empirical studies on creative class are based on case studies of developed countries. This chapter is one of the few focussing on China. The author’s first analyze the distribution of creative class that is highly concentrated and uneven in China. They then attempt to offer explanations underlying the distribution pattern of creative class. The empirical analysis involves the use of Chinese provincial-level panel data. The results show that openness and amenities are important factors that affect the creative class distribution in China. In addition, ecological infrastructure, urbanization, the high-tech sector development and the number of patents and universities also affect the attraction of talents in different ways. Chapter 5 by Wu presents a comparative study of productivity growth among the Chinese regions. Wu particularly explored the impact of China’s

March 5, 2013

8

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Development and Economic Growth in China

Western development program which was initiated in 1999. The aim of the chapter is to compare economic performance in the Western regions with that in coastal China and shed light on the sustainability of the country’s growth in the near future. In particular, it provides a productivity perspective about economic growth in Western China. The findings may have policy implications for economic development among the regions as well as across the sectors within each region. In the second part of the book, four chapters (Chapters 6 to 9) address various issues in China’s manufacturing sector and regional economies. In Chapter 6, Chen presents an empirical study of the impacts of foreign direct investment on China’s economic growth. He first discusses the possible channels through which FDI may affect China’s economic growth. Then, using a panel dataset of China’s 30 provinces over the period from 1987 to 2005, Chen estimates an augmented growth model and analyses the direct effects (e.g., raising output and productivity through capital augmentation and technological progress) and spillover effects (e.g., improving production efficiency through diffusing technology and management skills to local economy) of FDI on China’s economic growth. He argues that FDI contributed to China’s economic growth both directly through capital augmentation and technological progress and indirectly through positive spillover effects on local firms. However, the impact of FDI on economic growth varies across China’s regions. Chen concludes that local economic and technology conditions, especially local absorptive capability, do influence the diffusion of spillovers from FDI to the local economy. Sun and Li in Chapter 7 examine the performance of domestic banks and its relationship with the presence of FDI in the manufacturing sector. The authors show that FDI in the manufacturing sector indeed affects the performance of domestic banks significantly. Their empirical findings imply that a higher average level of FDI’s presence is associated with better performance of domestic banks. They also showed that the level of variation in the FDI’s presence negatively affects domestic banking performance. The authors argue that the relationship between the performance of banks and foreign direct investment in the manufacturing sector has policy implications for reforms in China’s banking sector.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Economies and Growth: An Introduction

9

Sun in Chapter 8 investigates the industrial dynamics in Chinese manufacturing sector. His findings confirm the Gibrat’s Law and a typical tent shape of growth rate distribution, which are two common properties observed in many previous studies. Sun further connects the growth rate distribution with the average firm size in a sector and the industrial policy changes that occurred in China in 2006. A are found, Sun argued that bigger sectors tend to have growth rate distributions with a fatter tail but smaller dispersion and the policy regime shift also exerts a similar impact. These are defined as the “big sector effect” and “policy regime shift effect” in this chapter. In Chapter 9, Fu investigates how agglomeration of exporters as a key external force affects export participation and export intensity of firms while firm heterogeneity is controlled. The results show the presence of export spillovers which have an inverted-U shape relationship with export performance of firms. It is also found that export spillovers only benefit the firms within the same industry in a region. In 2002, China signed a free trade agreement (FTA) with ASEAN (ACFTA). The full implementation commenced in 2010. ACFTA allows ASEAN Member States (AMS) to negotiate tariff reductions independently. Trewin et al. argued that AMS are generally aware of the potential opportunities arising from their access to the Chinese market but individual AMS are concerned to differing degrees about Chinese imports. For example, Indonesia has expressed a desire to renegotiate its tariff reduction schedules to protect sensitive sectors. By contrast, Vietnam seems more accepting of the prospects. China has its own sensitive agricultural sub-sectors. The author’s proposed a global general equilibrium model, GTAP, which is used to compare the potential impacts of ACFTA on the agricultural sectors of China, Indonesia, and Vietnam. The modeling exercises aim to identify the differential impact of separate sensitive sectors. Their analytical results indicate that all countries would improve their trade and welfare if the agreement is implemented fully. They also point out that the extent of exemptions for sensitive products represents differing degrees of missed opportunities. At the sectoral level, their findings show that all countries can expect some output reductions in some agricultural sectors, but in general these reductions are relatively small.

March 5, 2013

10

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Development and Economic Growth in China

Environmental regulation has become more and more important in policy making among the world economies. The answer to the question, “How has it affected productivity growth and hence economic growth?” is either controversial or yet to be explored in many cases. In Chapter 11, Wang and Wu present a case study of 17 Asian Pacific Economic Cooperation (APEC) economies. They adopted a directional distance function approach to estimate output-oriented Malmquist–Luenberger productivity indices. The latter are in turn decomposed into efficiency changes and technological progress. They considered three scenarios, namely no control on CO2 emissions (unregulated), maintaining current emission level, and a partial reduction of emissions. In general, it is found that the rates of productivity growth incorporating CO2 as an undesirable output are slightly higher than those estimated following the traditional method. Furthermore, the authors also investigate the causes. In Chapter 12, Liu and Yang modify the classical money demand function and build a vector error correction model (VECM) to investigate the interactive relationship among a series of variables including money supply, CPI, housing price, stock price, commodity price, GDP and exchange rate in China. This chapter has two goals, namely, (i) to explain the disproportional change in inflation relative to the change in money supply in China and (ii) investigate the dynamic interactions among good and asset markets for the inflation transmission. The basic conclusion in this chapter is that asset prices, good prices and economic fundamentals exhibit significant long-run co-integration relationship and that money supply is still the major factor affecting inflation in the long run, whilst asset markets play an important role to absorb the excessive liquidity and maintain money market equilibrium in the short run. The rising asset prices will accelerate and magnify the inflation progress over time. Thus the authors argue that both asset prices and good prices should be consistently targeted and monitored to better stabilize the inflation. Finally, Yu presents an interesting study of patent office behavior in Chapter 13. A country’s patent office could manipulate the procedures of patent examinations so as to protect domestic firms.Yu defines such practice as a kind of strategic patent policy. In this chapter, Yu exploits the possible strategic patent policy in the Chinese patent office by the analysis of patent data.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

Regional Economies and Growth: An Introduction

11

References Edwards, S., “Openness, Productivity and Growth: What Do We Really Know?” Economic Journal, 108: 383–398 (1998). Frankel, J. A. and Romer, D., “Does Trade Cause Growth?” American Economic Review, 89(3): 379–399 (1999). Harrison, A., “Openness and Growth: A Time-series, Cross-country Analysis for Developing Countries,” Journal of Development Economics, 48: 419–47 (1996). National Bureau of Statistics (NBS), China Statistical Yearbook 2011. Beijing: China Statistics Press (2011). National Bureau of Statistics (NBS), China Statistical Abstract 2012. Beijing: China Statistics Press (2012). National Bureau of Statistics (NBS), China Statistical Yearbook. Beijing: China Statistics Press (various years). World Bank, World Development Indictors. Washington, D.C.: The World Bank (2011). www.worldbank.org.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch01

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Part I

Regional Development

13

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

This page intentionally left blank

14

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Chapter 2

New Evidence of Regional Inequality Tsun Se Cheong

2.1. Introduction Much of the literature has been devoted to the topic of regional inequality in China. However, most of the studies on regional inequality are based on provincial-level data. Although some researchers employ county-level data in their studies, many of these studies are still plagued by either the problem of limited coverage or a short time-span. In-depth studies on intraprovincial regional inequalities are confined to the provinces in the Eastern zone, while intra-provincial regional inequalities in other economic zones are virtually unknown. The study in this chapter contributes to the literature by examining the intra-provincial regional inequalities, based on countylevel data, for provinces in all the economic zones. This is the first time in the literature that the intra-provincial regional inequalities of many inland provinces are measured and tracked for a long period of time. In addition to presenting the intra-provincial regional inequality amongst the county-level units (CUs) for each province, the inequalities in other spatial groupings are also shown. This chapter is organized as follows. Section 2.2 describes several different measures of inequality. Section 2.3 provides a discussion of the data used in the analysis. The results are presented in Section 2.4. The Subsection 2.4.1 shows the results for the complete dataset, which comprises both cities and counties. The dataset is then divided into the county subgroup and city subgroup so as to study the characteristics of these two subgroups individually. Section 2.4.2 shows the measurement results of the inequality amongst the counties (i.e., inter-county inequality), whereas Section 2.4.3 shows the measurement results of the inequality amongst

15

March 5, 2013

16

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

the cities (i.e., inter-city inequality). Section 2.4.4 provides a comparison between the complete dataset, the county-only dataset and the city-only dataset. This chapter concludes with Section 2.5.

2.2. Measurement of Inequality There are many different kinds of inequality measures — the most common being the Gini coefficient, Theil-T, and Theil-L indices. These three measurements are selected in this study because they all satisfy the Pigou–Dalton condition and the property of income-zero-homogeneity. The Pigou–Dalton principle suggests that a transfer of income from a rich person to a poor person should result in a decline in the inequality indicator, so long as the transfer does not reverse the ranking of the two in the income distribution. Income-zero-homogeneity refers to the value of the inequality measurement, which remains unchanged when there is a scale change of the whole income distribution. Most other conventional inequality measures do not fulfill these conditions (Bourguignon, 1979). Therefore, only the Gini coefficient, Theil-L, and Theil-T indices are used in this study. The Theil-T and Theil-L indices are often used in decomposition analysis because the two indices can be decomposed completely into the components of the subgroups (Bourguignon, 1979; Shorrocks, 1980, 1984). However, the Gini coefficient cannot satisfy the property of additive decomposability, therefore it cannot be decomposed completely into the components of the subgroups (Yao, 1999). Nevertheless, the Gini coefficient is still the most popular measure of inequality. The Gini coefficient is based on the Lorenz curve, which plots the cumulative share of income against the cumulative share of population from lowest to highest incomes. The 45◦ line of the Lorenz curve is the uniform (perfect) distribution line. The Gini coefficient is the ratio of the area that lies between the uniform distribution line and the Lorenz curve over the total area under the uniform distribution line. The value of the Gini coefficient ranges from zero to one. The value of zero corresponds to perfect income equality and “one” corresponds to perfect income inequality. It should be noted that the Gini coefficient is highly sensitive to inequality in the middle portion of the income distribution and it cannot capture small changes in the corners of the Lorenz curve.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

17

There are many different formulae for the calculation of the Gini coefficient, and one of the formulae for a population-weighted Gini coefficient is

|yi − yj | ni nj 2µ

j

i

NN

,

(2.1)

where yi and yj are the gross regional product (GRP) per capita in region i and region j respectively, ni and nj are the population in region i and region  j respectively, N is total population in all the regions, and µ = i yNi ni (Tsui, 1996). The formula of the unweighted Gini coefficient is

|Yi − Yj | i

j

2µR2

,

(2.2)

where Yi and Yj are the regional GRP in region i and region j respectively,  R is total number of regions, and µ = i YRi . Both Theil-L and Theil-T belong to the generalized entropy (GE) class of inequality measures (Shorrocks, 1980, 1984). The formula for the GE(c) inequality index is  
ni  yi c 1 GE(c) = − 1 , c = 0, c = 1, (2.3) c(1 − c) i N µ where c is a sensitivity parameter. The more negative c is, the more sensitive it is to income differences at the bottom of the income distribution. The more positive c is, the more sensitive it is to income differences at the top of the income distribution. It should be noted that GE(2) is half of the squared coefficient of variation; GE(1) is the Theil-T index, while GE(0) is the Theil-L index. The Theil-L index can also be referred to as the mean logarithmic deviation (MLD) and it is proved that this index is the only zerohomogeneous population-weighted measure. On the contrary, the Theil-T index can be simply referred to as the Theil index and it is the only zerohomogeneous income-weighted measure (Bourguignon, 1979). The interpretation is straightforward: the higher the Theil-L and Theil-T indices, the more unequal the income distribution. In addition, both Theil-L and Theil-T have unweighted versions.

March 5, 2013

18

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

The original formulae for Theil-L and Theil-T are: 1
µ Theil-L = ln , N i Ii Theil-T =

1
Ii Ii ln , N i µ µ

(2.4) (2.5)

where N is total population in all the regions, Ii is the income for the ith individual, µ is the mean of Ii , and i = 1, 2, . . . , N. After some manipulation, the formulae of the Theil-L and Theil-T can be obtained in terms of regional GRP and population. The formula for Theil-L is   ni
ni Theil-L = ln YNi , (2.6) N Y i where ni is the population in region i, N is the total population in all the regions, Yi is the regional GRP in region i, and Y is the total regional GRP for all regions. Similarly, the formula for Theil-T is   Yi
Yi Theil-T = (2.7) ln nYi . Y N i The formulae of the unweighted Theil-L and Theil-T respectively are:   1
1 R Theil-L = ln Yi , (2.8) R Y i   Yi
Yi (2.9) ln Y1 , Theil-T = Y R i where R is the total number of regions, Y is the total regional GRP of all the regions, and Yi is the regional GRP in region i. It can also be observed that the unweighted forms of Equations (2.8) and (2.9) can be derived from Equations (2.6) and (2.7) respectively by replacing ni /N by 1/R. According to Milanovic (2005), inequality can be examined from two different perspectives. The first approach emphasizes the inequality of the people, and hence population is incorporated into the formula. This

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

19

approach can be seen through Equations (2.1), (2.6), and (2.7). The second approach is the unweighted measure and so it can only show the disparity in output amongst the different regions without taking their population into consideration. This alternative can be seen through Equations (2.2), (2.8), and (2.9). It is deemed that the first approach is better because it shows the inequality amongst the people clearly. For this reason, population-weighted versions of the equations are used for inequality measurement in this study.

2.3. Data Issues Regional inequality can be studied using different indicators. Duncan and Tian (1999) note that it is important to distinguish between studies using livelihood and output indicators, as they can lead to different analysis results. Many inequality studies are based on expenditure, consumption, wage, earning, and household income data. These indicators are good measures for the livelihood, and economic well-being of the people. On the contrary, output per capita is deemed to be a good measure of economic development for a region. Because the focus of this study is on the regional inequality in economic development, regional outputs per capita are used in this study. GRP per capita is selected as the indicator of economic development in this research because it is the most frequently used indicator of output and is more comprehensive than other measures, for example, the gross value of industrial and agricultural output (GVIAO) and the national income (NI). This study is based on a dataset of real GRP per capita for the counties and county-level cities in China from 1997 to 2007. The year 1997 is selected as the beginning year because this is the year when Chongqing was separated from the Sichuan province. Chongqing was then upgraded to the administrative status of municipality. Thus, from this year onwards, individual data of Chongqing and Sichuan are available. There are three kinds of CUs, namely, the counties, the county-level cities and the city districts (or simply “districts”) in the prefectural level cities and municipalities. However, the data for the city district is unavailable for some provinces in the earlier years of the study period. Moreover, in some cases, only an aggregated value of the city districts, rather than the data of each individual district, is available. Therefore, this study is based on the data of the counties and county-level cities only. Many studies that examine

March 5, 2013

20

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

inequality amongst the CUs do not include city districts but are based on the data of the county-level cities and counties only.1 In this study, the four municipalities, namely, Beijing, Tianjin, Shanghai, and Chongqing are not included because most of the administrative regions in the municipalities are districts. The data is largely compiled from the Provincial Statistical Yearbook (State Statistical Bureau, 1998–2008a) for each province. However, where the data is unavailable, the data from the China Statistical Yearbook for Regional Economy (State Statistical Bureau, 2004–2008) and from the Provincial Yearbook (State Statistical Bureau, 1998–2008b) for each province is used. Some cities and counties are dropped from the dataset because of incomplete information in the sources. All the county-level GRP data are adjusted for inflation by converting them to 1997 constant prices. Since the deflator for each individual CU is unavailable, a provincial deflator is used for all the counties and county-level cities within a province. Therefore, the regional inequality within a province remains unchanged after the deflation process; however, the measures of inequality within other spatial groupings differ. Occasionally, there are some changes in the administrative divisions in China. Changes at the county-level may affect the GRP per capita of a region, which in turn can affect inequality levels. The common changes in administrative divisions include the change of the name of the county or city; the upgrade of the administrative status of a county to city or district; the upgrade of the administrative status of a city to district; and the change in boundaries of the CUs, for example, the transfer of control of a town or village from one CU to another. It should be noted that, except for the change in the name of the city and county, all the other changes in administrative divisions may affect the measurement of inequality. The official website for Administrative Divisions in China is checked to find out all the changes in administrative divisions that occurred during 1997–2007.2 First, all CUs are checked for any name change in the study period and then they are all renamed according to their latest names in 2007. 1 For example, see Veeck and Pannell (1989), Rozelle (1994), Lee (2000), Song et al. (2000),

Gustafsson and Li (2002), Wei and Kim (2002), Jones et al. (2003), Yu et al. (2007), Li and Xu (2008), Brajer et al. (2010), Zhou and Zou (2010), and Wu and Zhu (2011). 2 The web site is http://www.xzqh.org.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

21

Second, every CU is checked for any changes in boundary. The changes in boundary may affect the measurement of inequality. For example, the transfer of control of a town with a high GRP per capita from one rich county to a poor county may increase the GRP per capita of the latter, while significantly reducing the GRP per capita of the former; thereby giving a false impression of a decline in inequality. With the aim of keeping the data comparable, some modifications to the data are necessary to ensure that the changes in the GRP per capita can only be attributable to economic development rather than any administrative change. The approach suggested by Fan (1995) is adopted to tackle the problem of boundary changes amongst the CUs. If the cities and counties have any boundary changes in the study period, then they are aggregated so as to ensure comparability across time. However, the aggregation is strictly for boundary changes that involve cities and counties only, and no boundary change involving city districts is allowed. Accordingly, if there is a boundary change that involves city districts with any cities or counties, then these counties and cities are deleted from the database, in view of the fact that city districts are not included in this study. Similarly, if the counties or cities change their administrative status to city districts, then these cities and counties are deleted as well. The shortcoming of aggregation is the underestimation of inequality in the aggregated CUs, although according to the website of Administrative Divisions in China, the changes in administrative divisions are infrequent and few in number across the research period. After these adjustments, the data becomes comparable and the analysis can unveil the changes in GRP per capita due to economic development, rather than reflecting the impacts of changes in administrative divisions. In one recent work, Sakamoto and Fan (2010), who studied convergence at the county-level, also report their use of aggregation of some CUs into larger regions, so as to tackle the problem of data unavailability. Several counties were upgraded to cities within the research period. Accordingly, these CUs have different administrative statuses across the study period. In order to keep the data comparable in the study, the administrative statuses of all the cities and counties in the database are based on the latest classification, which is the administrative status in 2007. The same spatial entities should be used in the analysis for each year; otherwise,

March 5, 2013

22

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

the measurement of inequality will be distorted by the changes in spatial entities. For instance, a region may contribute a lot to overall inequality and hence the presence of this region would lead to a high level of inequality. If this region is omitted because of missing data in one year, then the calculated inequality indicator in this particular year may drop substantially. Therefore, it is necessary to use the same entities across the whole period of research. However, the data of some provinces are not available for some of the years; hence, the CUs in these provinces cannot be included in the study even though the data of these provinces are available in other years. The total number of counties and county-level cities in the database is shown in Table 2.1. The CUs can be grouped into larger spatial groupings in higher spatial levels for analysis. There are four spatial levels for the grouping of the CUs, namely, the national, inland-and-coastal, economic zonal, and the provincial levels. In this study, the coastal region is treated in the same way as the Eastern zone, whereas the inland region is defined to comprise all the provinces in the Central, Western and Northeastern zone. The definition of the inland and coastal regions used in this study is slightly different from the official definition. However, the results of the analyses in this study can be viewed as a study of the disparity between the Eastern zone and the other zones. As mentioned earlier, because the data of some provinces are not available in some of the years, the CUs within these provinces cannot be included in the measurement of inequality conducted at the national, inlandand-coastal, and economic zonal levels. Only the CUs in 22 provinces are included in these three levels of analyses, and they are3 :
Eastern zone: Hebei, Jiangsu, Zhejiang, Fujian, Guangdong, and Hainan. The municipalities of Beijing, Tianjin and Shanghai are excluded in this 3 The categorization of the zones is based on the 2006 China Statistical Yearbook (State Statistical Bureau, 2006). The dataset used in the study comprises 1,485 counties and countylevel cities every year. It is worth noting that, although the analyses conducted at the national, inland-and-coastal, and economic zonal levels are limited to the 22 provinces as mentioned above, the analysis conducted at the provincial level can be carried out for all provinces as long as the data of that province is available for that year. Therefore, the number of provinces reported in the results of the study of intra-provincial regional inequality (which is based on the measurement of inequality amongst the county-level units in each province) is larger than 22.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

23

Table 2.1. Number of Counties and County-level Cities in Database. Province Anhui Fujian Gansu Guangdong Guangxi Guizhou Hainan Hebei Heilongjiang Henan Hubei Hunan Inner Mongolia Jiangsu Jiangxi Jilin Liaoning Ningxia Qinghai Shaanxi Shandong Shanxi Sichuan Tibet Xinjiang Yunnan Zhejiang Total

Counties

County-level Cities

44 42 65 41 66 66 10 112 46 76 39 67 68 21 68 18 26 8 37 80 58 85 117 71 66 106 34

4 14 4 22 7 9 6 22 18 20 24 16 11 27 10 17 14 0 2 3 29 11 14 1 15 9 21

1,537

350

study. The province of Shandong is not included because of unavailability of data.
Central zone: Anhui, Jiangxi, Henan, and Hunan. The provinces of Shanxi and Hubei are not included because of unavailability of data.
Western zone: Inner Mongolia, Guangxi, Sichuan, Guizhou, Yunnan, Gansu, Qinghai, Ningxia, and Xinjiang. The municipality of Chongqing

March 5, 2013

24

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

is excluded in this study. The provinces of Shaanxi and Tibet are not included because of unavailability of data.
Northeastern zone: Liaoning, Jilin, and Heilongjiang. Population data is compiled from the Provincial Statistical Yearbook (State Statistical Bureau, 1998–2008a), Provincial Yearbook (State Statistical Bureau, 1998–2008b), and China Statistical Yearbook for Regional Economy (State Statistical Bureau, 2004–2008). It is a well-known fact that the provincial population data in China is based on the household registration system (hukou) population, whereas the number of temporary migrants is not taken into consideration. The temporary population would lead to a discrepancy between the actual and registered populations, but this is the only official data available from the government, and therefore virtually all studies in China, except the ones based on census or survey data, have to rely on it. The data of the actual population in 2000 for each CU is available in the County Data of 2000 Census (2000 ren kou pu cha fen xian zi liao) (State Statistical Bureau, 2003), though no such information is available for the census in 2005, and the census data for 2010 is not yet available. Therefore, for the time being, it is impossible to adjust the population data by interpolation because the only data available is for 2000. Further adjustment can be made when the data becomes available in the future. Another concern is the quality of output data in China. Many researchers such as Chow (1986), Wu (1997), Rawski (2001), and Holz (2006) have raised doubts about the reliability and accuracy of the GDP statistics. Despite this, just like in the case of the population data mentioned above, government publications are the only source of output data for the CUs. Therefore, all researchers need to rely on official data when studying inequality in output amongst the CUs. Recognizing these limitations, the data in official government publications still provides valuable information on both population and output. They are indispensable in studying regional inequality in China, although caution should be exercised in interpreting the results of all research that is based on these data.

2.4. Results and Discussions The terms “county-level city” and “city” are used interchangeably in this chapter. The dataset used in this research comprises the data of the counties

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

25

and county-level cities. Therefore, CU refers to both counties and countylevel cities. The term “inter-CU inequality”, refers to the inequality amongst all the counties and county-level cities within a region. The term “intraprovincial regional inequality”, or simply “intra-provincial inequality”, refers to the inequality amongst the CUs within a province (i.e., the inter-CU inequality within each province). The intra-provincial inequality can either be based on the complete dataset, which comprises both the counties and cities, or it can be based on the county-only dataset or the city-only dataset. The analyses in Section 2.4.1 are based on the complete dataset, which comprises both counties and cities. The complete dataset is then divided into the county-only dataset and the city-only dataset. The analyses in Section 2.4.2 are based on the county-only dataset, whereas those in Section 2.4.3 are based on the city-only dataset. Each of these sections is further divided into two subsections. The first subsection shows the inequalities in different spatial groupings in the national, inland-and-coastal, and economic zonal levels. The second subsection shows the measurement of inequality within each province. Essentially, this is the intra-provincial regional inequality for each province. In order to save space, only the intraprovincial inequalities in the beginning (1997) and the end (2007) of the study period are shown. It should be noted that the data in the first subsection is limited to the 22 provinces as listed in Section 2.3, while the provinces that are shown in the second subsection are not limited to the 22 provinces. Basically, any province whose data are available in 1997 and 2007 is shown in the second subsection. Finally, Section 2.4.4 provides a comparison of all datasets. Given that the Gini coefficient is the most popular measurement of inequality, the discussions in this section will be based mainly on the results of this coefficient. However, some sections will also present the results of Theil-T and Theil-L indices for comparison.

2.4.1. Inter-county-level-unit (inter-CU) inequalities 2.4.1.1. Inter-CU inequalities within various spatial groupings Figure 2.1 shows the substantial increase in the inter-CU inequalities within all spatial groupings from 1997 to 2007, and the presence of a considerable difference in the growth rates of inequality for the spatial groupings. The Gini coefficient in the inland region was slightly lower than that of the coastal region in 1997, but then the latter increased quickly, especially after 2004,

March 5, 2013

26

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

0.40 0.38 0.36

Gini

0.34 0.32 0.30 0.28 0.26 0.24

2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

0.20

1997

0.22

Year Nation

Inland Region

Figure 2.1.

Eastern Zone (Coastal Region)

Central Zone

Western Zone

North-Eastern Zone

Inter-CU Inequalities for Different Spatial Groupings.

Source: Author’s calculation. Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones.

whereas the former increased slowly. Accordingly, divergence in inequality can be observed for these two regions. Turning to the inequality in the economic zones, it can be observed that the Eastern zone had lower inequality than the Western zone in 1997. However, the increase in inequality in the Eastern zone was so huge that it eventually exceeded the inequality in the Western zone. The inequality in the Central zone was lower than the inequality in the Northeastern zone in the beginning of the research period, but at the end, the inequality in the Central zone was slightly higher than the Northeastern zone. In 2007, the Eastern zone had the highest inequality, followed by the Western zone. Both the Central zone and Northeastern zone had relatively low levels of inequality. Table 2.2 shows the changes in inequalities in different spatial groupings. The percentage change at the national level is 19.1% for the Gini coefficient while they are recorded as 54.7% for Theil-T and 39.9% for Theil-L. The percentage change in the coastal region is much higher than that for the inland region. The growth rates of inequality differ significantly for the economic zones. The Central zone had an enormous increase in inequality in that period, revealing an increase of 39.0% in the Gini coefficient, 96.9% for Theil-T and 93.3% for Theil-L. The Eastern zone had the second largest

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

27

Table 2.2. Changes in Inter-CU Inequalities for Different Spatial Groupings.

Inequality

1997

2007

Change from 1997 to 2007

% Change from 1997 to 2007

Nation

Gini Theil-T Theil-L

0.342 0.202 0.194

0.407 0.313 0.271

0.065 0.111 0.077

19.1 54.7 39.9

Inland Region

Gini Theil-T Theil-L

0.292 0.143 0.143

0.351 0.226 0.200

0.059 0.083 0.058

20.1 57.6 40.6

Eastern Zone (Coastal Region)

Gini Theil-T Theil-L

0.307 0.159 0.150

0.401 0.293 0.259

0.094 0.134 0.109

30.6 84.2 72.7

Central Zone

Gini Theil-T Theil-L

0.222 0.085 0.078

0.309 0.168 0.152

0.087 0.083 0.073

39.0 96.9 93.3

Western Zone

Gini Theil-T Theil-L

0.331 0.187 0.179

0.374 0.287 0.231

0.043 0.100 0.052

13.0 53.7 29.0

Northeastern Zone

Gini Theil-T Theil-L

0.244 0.098 0.100

0.302 0.149 0.155

0.058 0.051 0.055

23.7 51.9 55.4

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

increase, followed by the Northeastern zone. The Western zone had just a 13.0% increase in the Gini coefficient, 53.7% for Theil-T, and 29.0% for Theil-L.

2.4.1.2. Intra-provincial inter-CU inequalities The inter-CU Gini coefficients in the provinces in 1997 and 2007 are shown in Figures 2.2 and 2.3, respectively. It can be observed in Figure 2.3 that China can be roughly divided into the Northern portion and the Southern portion. Most of the provinces in the Northern portion had relatively high levels of inequality, while the provinces in the Southern portion had

March 5, 2013

28

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Figure 2.2.

Intra-Provincial Inter-CU Inequalities in 1997 (Gini Coefficients).

Note: The municipalities are not included in this study. The data of Shandong, Hubei and Tibet are not available. Source: Author’s calculation.

relatively low levels of inequality. Besides, it is shown that the inequality levels in Hainan were low in both 1997 and 2007. Table 2.3 shows the intra-provincial regional inequality for each province in 1997 and 2007. It is evident that the level of inequality increased in most provinces. It is thus difficult to observe their relative performance in inequality alleviation. To make this comparison easier, the inequality values are converted into relative inequality. The mean inequality for all provinces in 1997 is calculated first, and then the intra-provincial inequality in each province is divided by this mean, so as to compute the relative inequality for each province in 1997. The same procedure is repeated for 2007. The results of the relative Gini coefficients are plotted in Figure 2.4.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

Figure 2.3.

29

Intra-Provincial Inter-CU Inequalities in 2007 (Gini Coefficients).

Note: The municipalities are not included in this study. Source: Author’s calculation.

Figure 2.4 shows the severity and the evolution of intra-provincial regional inequality in each province very clearly. The horizontal axis is the relative Gini coefficient in 1997, while the vertical axis is the relative Gini coefficient in 2007. The figure is divided into four quadrants. The provinces in the first quadrant are those which had above-average levels of inequality in both 1997 and 2007. The provinces in the second quadrant are those which had below-average inequalities in 1997 but above-average inequalities in 2007. Those that had below-average inequalities in both 1997 and 2007 are situated in the third quadrant. The provinces in the fourth quadrant are those which had above-average inequalities in 1997 but below-average inequalities in 2007. The evolution of intra-provincial inequality for each

March 5, 2013

30

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Table 2.3. Relative Intra-Provincial Inter-CU Inequalities. Gini

Relative Gini

Province

1997

2007 Change 1997

Coastal Eastern

Fujian Guangdong Hainan Hebei Jiangsu Zhejiang

0.270 0.257 0.156 0.229 0.375 0.230

0.293 0.023 1.045 0.953 0.253 −0.004 0.994 0.823 0.144 −0.012 0.602 0.468 0.333 0.104 0.887 1.084 0.469 0.094 1.451 1.524 0.256 0.026 0.890 0.831

Inland

Central

Anhui Henan Hunan Jiangxi Shanxi

0.195 0.267 0.177 0.163 0.265

0.217 0.317 0.259 0.253 0.343

0.022 0.050 0.082 0.090 0.078

0.753 1.033 0.684 0.629 1.026

0.705 1.031 0.841 0.823 1.113

Western

Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan Xinjiang Yunnan

0.364 0.204 0.226 0.263 0.361 0.291 0.247 0.287 0.379 0.338

0.401 0.037 0.218 0.014 0.268 0.042 0.440 0.177 0.299 −0.062 0.372 0.081 0.410 0.163 0.281 −0.006 0.443 0.064 0.314 −0.024

1.407 0.788 0.875 1.018 1.394 1.123 0.955 1.108 1.463 1.306

1.303 0.708 0.871 1.431 0.971 1.209 1.333 0.913 1.439 1.020

0.181 0.168 0.316 0.259

0.301 0.120 0.217 0.049 0.283 −0.033 0.308 0.049

0.698 0.979 0.650 0.706 1.221 0.921

Northeastern Heilongjiang Jilin Liaoning Mean Gini

2007

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

province can be observed clearly by looking at the quadrant in which it is situated. It can be observed that five provinces in the first quadrant are from the Western zone. The provinces in the second quadrant are Hebei and

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

31

1.6 Jiangsu (E) Xinjiang (W)

Inner Mongolia (W)

1.4

Relative Inequality in 1997

Shaanxi (W)

Gansu (W) Qinghai (W)

1.2 Shanxi (C)

Hebei (E) 0.6 Heilongjiang (NE)

0.8

1.0

Jilin (NE)

Fujian (E)

Guizhou (W)

Hunan (C) Jiangxi (C)

Zhejiang (E) Guangxi (W)

0.8

Yunnan (W)

Henan (C)

1.0 0.4

1.2

Sichuan (W)

Liaoning (NE)

Ningxia (W) 1.4

1.6

Guangdong (E)

Anhui (C) 0.6

Hainan (E) 0.4 Relative Inequality in 2007

Figure 2.4.

Relative Intra-Provincial Inter-CU Inequalities in 1997 and 2007.

Note: The zones of the provinces are shown. E is Eastern zone, C is Central zone, W is Western zone, and NE is Northeastern zone. Source: Author’s calculation.

Shaanxi, which had below-average inequalities in 1997, but ended up with above-average inequalities in 2007. There are 10 provinces in the third quadrant, and these provinces had below-average inequalities in both years. The provinces within the Central zone are either in the first quadrant or in the third quadrant, which implies that there was some persistence for intraprovincial inequality in these provinces over the study period. The provinces of the Northeastern zone are all under the horizontal axis which means that they had below-average inequalities in 2007. The provinces in the fourth quadrant performed relatively well on inequality reduction because they had above-average levels of inequality in 1997 and below-average levels of inequality in 2007. There is no apparent trend for the provinces in the Eastern and Western zones as they appear in all four quadrants. However, there are some limitations in the usage of this figure alone. For example, inequality in Fujian rose from 0.270 to 0.293 in the study period, but the relative Gini coefficient is 1.045 in 1997 and 0.953 in 2007. Although the performance of Fujian was very good according to the calculated relative Gini coefficient, its inequality actually increased slightly in this period. For this reason, the changes in absolute level of inequality are also provided in Table 2.2 as a reference.

March 5, 2013

32

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China 0.36 0.34 0.32

Gini

0.30 0.28 0.26 0.24 0.22 0.20 2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

0.18

Year Nation

Inland Region

Figure 2.5.

Eastern Zone (Coastal Region)

Central Zone

Western Zone

North-Eastern Zone

Inter-County Inequalities for Different Spatial Groupings.

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

2.4.2. Inter-county inequalities 2.4.2.1. Inter-county inequalities within various spatial groupings Figure 2.5 shows that inter-county inequality increased in all spatial groupings from 1997 to 2007. The inter-county inequality within the inland region was higher than the coastal region. This finding is different from the one in the previous section, which is based on the data for both counties and cities. In 2007, the Western zone was found to have the highest intercounty inequality, followed by the Eastern zone, and then the Northeastern zone. Although the Central zone registered a large increase in inequality, it still had the lowest inter-county inequality among the zones in 2007. Table 2.4 shows that the national Gini coefficient of inter-county inequality increased by 18.1% from 1997 to 2007. The coastal region had a higher percentage of growth than the inland region, but the inter-county inequality in the inland region was still higher than the coastal region in 2007. The Central zone had the largest percentage increase amongst the four zones across the period. The increase was 39.4% for the Gini coefficient, 109.4% for Theil-T, and 98.7% for Theil-L. The Eastern zone had the second largest increase, followed by the Northeastern zone, while the Western zone had the smallest increase.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

33

Table 2.4. Changes in Inter-County Inequalities for Different Spatial Groupings.

Inequality

1997

2007

Change from 1997 to 2007

% Change from 1997 to 2007

Nation

Gini Theil-T Theil-L

0.283 0.132 0.134

0.334 0.203 0.183

0.051 0.072 0.048

18.1 54.3 35.7

Inland Region

Gini Theil-T Theil-L

0.262 0.113 0.116

0.318 0.196 0.167

0.056 0.083 0.051

21.5 73.3 44.3

Eastern Zone (Coastal Region)

Gini Theil-T Theil-L

0.229 0.086 0.084

0.290 0.143 0.133

0.061 0.057 0.049

26.6 66.8 57.8

Central Zone

Gini Theil-T Theil-L

0.191 0.059 0.057

0.266 0.124 0.113

0.075 0.065 0.056

39.4 109.4 98.7

Western Zone

Gini Theil-T Theil-L

0.304 0.154 0.150

0.355 0.269 0.209

0.051 0.115 0.059

16.8 75.0 39.2

Northeastern Zone

Gini Theil-T Theil-L

0.219 0.078 0.082

0.273 0.120 0.128

0.054 0.042 0.046

24.7 54.7 56.1

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

2.4.2.2. Intra-provincial inter-county inequalities The inter-county Gini coefficients for the provinces in 1997 and 2007 are shown in Figures 2.6 and 2.7, respectively. It can be observed from Figure 2.7 that the provinces, which had a high level of inter-county inequality in 2007, are all situated in the Northern and Northwestern part of China. All of them, except Xinjiang, share the border with Inner Mongolia. However, Ningxia registered a decline in inter-county inequality even though it shares a border with Inner Mongolia. In addition, it is found that the levels of inter-county inequalities in Jilin, Fujian and Hainan were low in both 1997 and 2007, while Hubei had a low level of inequality in 2007.

March 5, 2013

34

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Figure 2.6.

Intra-Provincial Inter-County Inequalities in 1997 (Gini Coefficients).

Note: The municipalities are not included in this study. The data of Shandong, Hubei and Tibet are not available. Source: Author’s calculation.

Table 2.5 shows the change in intra-provincial inter-county inequality for each province from 1997 to 2007. Figure 2.8 shows that many provinces in the first quadrant are from the Western zone; these provinces had aboveaverage inequalities in both 1997 and 2007. Many provinces of the Central zone are in the third quadrant, which means that these provinces had belowaverage inequalities in both years. The provinces in the Northeastern zone are all under the horizontal axis, and so these provinces had relatively low regional inequalities in 2007. Most of the provinces are in either the first or third quadrants, which implies that there was some persistence for intraprovincial regional inequality in many provinces over the study period. However, there is no apparent pattern for the provinces in the Eastern zone.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

Figure 2.7.

35

Intra-Provincial Inter-County Inequalities in 2007 (Gini Coefficients).

Note: The municipalities are not included in this study. Source: Author’s calculation.

2.4.3. Inter-city inequalities 2.4.3.1. Inter-city inequalities within various spatial groupings Figure 2.9 shows that inter-city inequality increased a lot in the study period for all spatial groupings.4 Although the inter-city inequality in the inland region increased in the research period, the increase was small compared to the coastal region. The coastal region had a dramatic increase in inequality and its Gini coefficient was higher than the national one in 2007. The Eastern zone had the highest level of inter-city inequality in 2007, followed by the 4 There is only one county-level city in Ningxia, and this city was removed from the dataset

because there was a boundary change in that city. Ningxia does not have any city and comprises counties only. Thus, Ningxia is excluded from the study in this section.

March 5, 2013

36

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Table 2.5. Relative Intra-Provincial Inter-County Inequalities. Gini

Relative Gini

Province

1997

2007 Change 1997

Coastal Eastern

Fujian Guangdong Hainan Hebei Jiangsu Zhejiang

0.177 0.227 0.135 0.194 0.189 0.264

0.184 0.239 0.136 0.287 0.241 0.280

0.007 0.778 0.681 0.012 1.002 0.887 0.001 0.594 0.502 0.093 0.854 1.064 0.052 0.832 0.892 0.016 1.164 1.037

Inland

Central

Anhui Henan Hunan Jiangxi Shanxi

0.170 0.199 0.184 0.163 0.264

0.200 0.260 0.253 0.244 0.324

0.030 0.061 0.069 0.081 0.060

0.748 0.876 0.811 0.720 1.163

0.743 0.965 0.938 0.905 1.199

Western

Gansu Guangxi Guizhou Inner Mongolia Ningxia Qinghai Shaanxi Sichuan Xinjiang Yunnan

0.338 0.196 0.191 0.240 0.361 0.236 0.246 0.239 0.341 0.279

0.363 0.025 0.223 0.027 0.216 0.025 0.447 0.207 0.299 −0.062 0.248 0.012 0.416 0.170 0.245 0.006 0.403 0.062 0.279 0.000

1.492 0.864 0.840 1.060 1.589 1.041 1.086 1.052 1.505 1.229

1.347 0.827 0.800 1.657 1.107 0.920 1.543 0.908 1.495 1.035

0.173 0.150 0.289 0.227

0.268 0.095 0.189 0.039 0.230 −0.059 0.270 0.043

0.763 0.993 0.662 0.702 1.274 0.852

Northeastern Heilongjiang Jilin Liaoning Mean Gini

2007

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

Western and Central zones, while the Northeastern zone had the lowest level of inequality. Surprisingly, the inter-city inequality in the Northeastern zone registered a decline from 2000 to 2004, however, the trend reversed and it increased from 2004 to 2007.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

37

Inner Mongolia (W) 1.6 Shaanxi (W) Xinjiang (W) 1.4

Relative Inequality in 1997

Gansu (W) 1.2 Hebei (E)

Heilongjiang (NE) 0.4

Shanxi (C)

1 Henan (C) 1 Qinghai (W) 0.8 Hunan (C) Sichuan (W) Jiangsu (E) Jiangxi (C) Guangxi (W) Guangdong (E) 0.8 Guizhou (W) Anhui (C)

0.6

Jilin (NE)

Ningxia (W)

Zhejiang (E) Yunnan (W) 1.2

1.4

1.6

Liaoning (NE)

Fujian (E) 0.6

Hainan (E) 0.4 Relative Inequality in 2007

Figure 2.8. Relative Intra-Provincial Inter-County Inequalities in 1997 and 2007. Note: The zones of the provinces are shown. E is Eastern zone, C is Central zone, W is Western zone, and NE is Northeastern zone. Source: Author’s calculation. 0.42 0.40 0.38 0.36 0.34 Gini

0.32 0.30 0.28 0.26 0.24 0.22 0.20 2007

2006

2005

2004

2003

2002

2001

2000

1999

1998

1997

0.18

Year

Nation

Inland Region

Eastern Zone (Coastal Region)

Central Zone

Western Zone

North-EasternZone

Figure 2.9. Inter-City Inequalities for Different Spatial Groupings. Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

Table 2.6 shows that the inter-city inequality in all spatial groupings increased considerably from 1997 to 2007. The percentage change for the national Gini coefficient was 27.4% within 11 years. The growth rate of the inter-city inequality in the coastal region was much higher than the inland

March 5, 2013

38

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Table 2.6. Changes in Inter-City Inequalities for Different Spatial Groupings.

Inequality

1997

2007

Change from 1997 to 2007

% Change from 1997 to 2007

Nation

Gini Theil-T Theil-L

0.314 0.164 0.159

0.400 0.285 0.261

0.086 0.121 0.102

27.4 74.1 64.5

Inland Region

Gini Theil-T Theil-L

0.260 0.111 0.108

0.316 0.169 0.160

0.057 0.058 0.052

21.9 52.6 48.0

Eastern Zone (Coastal Region)

Gini Theil-T Theil-L

0.298 0.144 0.144

0.415 0.296 0.290

0.116 0.152 0.146

39.0 105.4 101.1

Central Zone

Gini Theil-T Theil-L

0.238 0.093 0.088

0.309 0.153 0.152

0.071 0.060 0.064

30.0 64.7 72.7

Western Zone

Gini Theil-T Theil-L

0.298 0.153 0.149

0.341 0.226 0.190

0.043 0.073 0.041

14.4 47.6 27.6

Northeastern Zone

Gini Theil-T Theil-L

0.222 0.080 0.078

0.274 0.121 0.124

0.052 0.041 0.046

23.3 51.4 58.6

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

region. According to the measurements based on the Gini coefficient, the Eastern zone had a prominent 39.0% increase, while the Central zone had an increase of 30.0%. A moderate increase of 23.3% is observed in the Northeastern economic zone, while the Western zone only had a 14.4% increase.

2.4.3.2. Intra-provincial inter-city inequalities The inter-city Gini coefficients for the provinces in 1997 and 2007 are shown in Figures 2.10 and 2.11, respectively. It can be observed that Jiangsu had

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

Figure 2.10.

39

Intra-Provincial Inter-City Inequalities in 1997 (Gini Coefficients).

Note: The municipalities are not included in this study. The data of Shandong, Hubei, Tibet and Ningxia are not available. Source: Author’s calculation.

a very high level of inter-city inequality in 2007. The other provinces with high inter-city inequality are Xinjiang, Gansu, Inner Mongolia, and Hebei. They are all situated in the Northern part of China. It is surprising to find that although Zhejiang is in the Eastern zone, it had a relatively low level of inter-city inequality in 2007. Hainan had the lowest level of inter-city inequality in 2007. Jiangsu had the highest level of inter-city inequality in 2007, as shown in Table 2.7 and Figure 2.12. It can be observed that the provinces in the first quadrant are from both the Eastern and Western zones, and these provinces had above-average levels of inequality in both years. Provinces in the Northeastern zone are all under the horizontal axis, so all of them had

March 5, 2013

40

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Figure 2.11.

Intra-Provincial Inter-City Inequalities in 2007 (Gini Coefficients).

Note: The municipalities are not included in this study. The data of Ningxia and Tibet are not available. Source: Author’s calculation.

below-average levels of inequality in 2007. However, no apparent trend can be observed for the provinces in the Central zone.

2.4.4. Comparison of the inter-CU inequality, inter-county inequality and inter-city inequality It can be observed from Table 2.8 that inter-CU inequality was higher than that for inter-city inequality, while inter-city inequality was in turn higher than the inter-county inequality in both 1997 and 2007. The percentage increase in inter-city inequality was very large (27.4% as measured by the Gini coefficient), and it was much larger than the increase in inter-county inequality (18.1% as measured by the Gini coefficient). The evidence shows

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

41

Table 2.7. Relative Intra-Provincial Inter-City Inequalities. Gini

Relative Gini

Province

1997

2007 Change 1997

Coastal Eastern

Fujian Guangdong Hainan Hebei Jiangsu Zhejiang

0.240 0.266 0.084 0.173 0.347 0.146

0.276 0.036 1.193 1.094 0.264 −0.002 1.324 1.045 0.111 0.027 0.416 0.441 0.303 0.130 0.861 1.198 0.428 0.081 1.726 1.693 0.185 0.039 0.727 0.732

Inland

Central

Anhui Henan Hunan Jiangxi Shanxi

0.184 0.219 0.126 0.136 0.163

0.267 0.252 0.210 0.239 0.280

0.083 0.033 0.084 0.103 0.117

0.916 1.091 0.626 0.674 0.811

1.056 0.997 0.831 0.947 1.111

Western

Gansu Guangxi Guizhou Inner Mongolia Qinghai Shaanxi Sichuan Xinjiang Yunnan

0.231 0.228 0.115 0.281 0.178 0.090 0.212 0.338 0.322

0.342 0.111 0.163 −0.065 0.169 0.054 0.375 0.094 0.206 0.028 0.165 0.075 0.244 0.032 0.363 0.025 0.260 −0.062

1.149 1.136 0.571 1.401 0.884 0.447 1.054 1.683 1.601

1.353 0.645 0.670 1.486 0.815 0.654 0.967 1.439 1.031

0.142 0.146 0.255 0.201

0.240 0.098 0.228 0.082 0.236 −0.019 0.253 0.052

0.709 0.952 0.727 0.905 1.270 0.936

Northeastern Heilongjiang Jilin Liaoning Mean Gini

2007

Note: The coastal region is treated the same as the Eastern zone. The inland region includes the Central, Western, and Northeastern zones. Source: Author’s calculation.

that the inequality amongst the cities grew much faster than the inequality amongst the counties in the research period. Table 2.9 shows that there was a sharp increase of inter-CU inequality from 1997 to 1998. The annual increase in the period from 1998 to 2002 was moderate for every year, but there was another sharp increase from 2002

March 5, 2013

42

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China 1.9 Jiangsu (E) 1.7

1.5

Inner Mongolia (W) Xinjiang (W)

Gansu (W) Relative Inequality in 1997

1.3 Hebei (E) Shanxi (C) 1.1 Anhui (C) 1.3

0.5

Jiangxi (C)

0.7 Heilongjiang (NE) 1.9 Jilin (NE) Hunan (C)

Guizhou (W)

0.9

Fujian (E) Guangdong (E)

Henan (C) 1.1 Sichuan (W)

1.3 Liaoning (NE)

Yunnan (W) 1.5

1.7

Qinghai (W) Zhejiang (E)

0.7 Guangxi (W)

Shaanxi (W) 0.5 Hainan (E) 0.3 Relative Inequality in 2007

Figure 2.12.

Relative Intra-Provincial Inter-City Inequalities in 1997 and 2007.

Note: The zones of the provinces are shown. E is Eastern zone, C is Central zone, W is Western zone, and NE is Northeastern zone. Source: Author’s calculation.

Table 2.8. Inequality Changes and Percentage Changes for the National Inter-CU Inequality, Inter-County Inequality, and Inter-City Inequality.

1997

2007

Change from 1997 to 2007

Inter-CU Gini Inter-CU Theil-T Inter-CU Theil-L

0.342 0.202 0.194

0.407 0.313 0.271

0.065 0.111 0.077

19.1 54.7 39.9

Inter-County Gini Inter-County Theil-T Inter-County Theil-L

0.283 0.132 0.134

0.334 0.203 0.183

0.051 0.072 0.048

18.1 54.3 35.7

Inter-City Gini Inter-City Theil-T Inter-City Theil-L

0.314 0.164 0.159

0.400 0.285 0.261

0.086 0.121 0.102

27.4 74.1 64.5

Source: Author’s calculation.

% Change from 1997 to 2007

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

43

Table 2.9. Yearly Growth (%) of Gini Coefficient for the National Inter-CU Inequality, Inter-County Inequality, and Inter-City Inequality. Percent Change (%)

1997– 1998– 1999– 2000– 2001– 2002– 2003– 2004– 2005– 2006– 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Inter-CU Gini 3.025 1.419 2.383 2.103 1.414 3.658 0.602 0.067 1.987 1.025 Inter-County Gini 2.709 1.754 2.084 2.583 0.407 3.144 0.845 −0.323 2.015 1.631 Inter-City Gini 2.953 0.803 3.050 1.459 2.497 4.523 1.196 3.969 2.926 1.170 Source: Author’s calculation.

to 2003. However, the increase between 2003 and 2004 was small. National inter-CU inequality remained constant from 2004 to 2005. Surprisingly, the inter-county inequality registered a decline in this period, but the percentage increase in inter-city inequality was enormous, with a value of 4%. The results in this section show that large discrepancies exist for the growth rates of the levels of inequality in the city and county subgroups. It pinpoints the fact that much information will be lost if the cities and counties are combined together in calculation, and thus it is necessary to analyze city and county data separately.

2.5. Conclusions The results show that the inequality amongst the CUs has increased enormously in the study period. This is not only shown through the results derived from the national level study, but the analyses conducted at the inland-and-coastal level, the economic zonal level and the provincial level all show that regional inequality increased substantially in different spatial groupings over the period 1997–2007. The study based on inter-CU inequality shows that the coastal region had a higher level of inequality than the inland region. The zonal-level analysis showed that the Eastern zone had the highest inequality in 2007, followed by the Western zone, and then the Central zone, while the Northeastern zone had the lowest level of inequality. These findings are the same as the study which is based on the city-only dataset. However, the study based on inter-county inequality reports entirely different findings. For the study which is based on the county-only dataset, inequality in the inland region

March 5, 2013

44

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

is higher than that for the coastal region. Moreover, the study shows that the inequality in the Western zone was the highest, followed by the Eastern zone, and then the Northeastern zone, while the Central zone had the lowest level of inequality. Both the Eastern and Western zones had high levels of inequality no matter whether the measurement was based on the complete dataset, the county-only dataset, or the city-only dataset. As far as intra-provincial regional inequality for each province is concerned, Jiangsu is found to have the highest inter-city and inter-CU inequalities for the year 2007. Another observation is that most of the provinces that have high levels of inequality are in the Northern part of China and are situated close to each other. In addition, it can be concluded that the provinces have different patterns of inequality even if they are in the same economic zone. The results imply that there is a need to formulate province-specific policies in order to reduce inequality.

References Bourguignon, F., “Decomposable Income Inequality Measures,” Econometrica, 47: 901–920 (1979). Brajer, V., Mead, R. and Xiao, F., “Adjusting Chinese Income Inequality for Environmental Equity,” Environment and Development Economics, 15: 341–362 (2010). Chow, G. C., “Chinese Statistics,” The American Statistician, 40: 191–196 (1986). Duncan, R. and Tian, X., “China’s Inter-Provincial Disparities: An Explanation,” Communist and Post-Communist Studies, 32: 211–224 (1999). Fan, C. C., “Of Belts and Ladders: State Policy and Uneven Regional Development in Post-Mao China,” Annals of the Association of American Geographers, 85: 421–449 (1995). Gustafsson, B. and Li, S., “Income Inequality within and across Counties in Rural China 1988 and 1995,” Journal of Development Economics, 69: 179–204 (2002). Holz, C. A., “Why China’New GDP Data Matters,” Far Eastern Economic Review, 169: 54–57 (2006). Jones, D. C., Owen, A. L. and Li, C., “Growth and Regional Inequality in China During the Reform Era,” China Economic Review, 14: 186–200 (2003).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

New Evidence of Regional Inequality

45

Lee, J., “Changes in the Source of China’s Regional Inequality,” China Economic Review, 11: 232–245 (2000). Li, S. and Xu, Z., “The Trend of Regional Income Disparity in the People’s Republic of China,” ADBI Discussion Paper 85. Tokyo: Asian Development Bank Institute (2008). Milanovic, B., “Half a World: Regional Inequality in Five Great Federations,” World Bank Policy Research Working Paper 3699. World Bank (2005). Rawski, T. G., “What is Happening to China’s GDP Statistics?” China Economic Review, 12: 347–354 (2001). Rozelle, S., “Rural Industrialization and Increasing Inequality: Emerging Patterns in China’s Reforming Economy,” Journal of Comparative Economics, 19: 362–391 (1994). Sakamoto, H. and Fan, J., “Distribution Dynamics and Convergence among 75 Cities and Counties in Yangtze River Delta in China: 1990–2005,” Review of Urban and Regional Development Studies, 22: 39–54 (2010). Shorrocks, A. F., “The Class of Additively Decomposable Inequality Measures,” Econometrica, 3: 613–625 (1980). Shorrocks, A. F., “Inequality Decomposition by Population Subgroups,” Econometrica, 52: 1369–1385 (1984). Song, S., Chu, G. S.-F. and Cao, R., “Intercity Regional Disparity in China,” China Economic Review, 11: 246–261 (2000). State Statistical Bureau, Provincial Statistical Yearbook. Beijing: China Statistics Press (1998–2008a). State Statistical Bureau, Provincial Yearbook. Beijing: China Statistics Press (1998–2008b). State Statistical Bureau, County Data of 2000 Census. Beijing: China Statistics Press (2003). State Statistical Bureau, China Statistical Yearbook for Regional Economy. Beijing: China Statistics Press (2004–2008). State Statistical Bureau, China StatisticalYearbook. Beijing: China Statistics Press (2006). Tsui, K.-Y., “Economic Reform and Interprovincial Inequality in China,” Journal of Development Economics, 50: 353–368 (1996). Veeck, G. and Pannell, C. W., “Rural Economic Restructuring and Farm Household Income in Jiangsu, People’s Republic of China,” Annals of the Association of American Geographers, 79: 275–292 (1989).

March 5, 2013

46

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch02

Regional Development and Economic Growth in China

Wei, Y. D. and Kim, S., “Widening Inter-County Inequality in Jiangsu Province, China, 1950–95,” The Journal of Development Studies, 38: 142–164 (2002). Wu, H. X., “Measuring China’s GDP,” Briefing Paper Series No. 8. Canberra: East Asia Analytical Unit (1997). Wu, Y. and Zhu, J., “Corruption, Anti-Corruption, and Inter-County Income Disparity in China,” The Social Science Journal, 48(3): 435–448 (2011). Yao, S., “On the Decomposition of Gini Coefficients by Population Class and Income Source: A Spreadsheet Approach and Application,” Applied Economics, 31: 1249–1264 (1999). Yu, L., Luo, R. and Zhang, L., “Decomposing Income Inequality and Policy Implications in Rural China,” China & World Economy, 15: 44–58 (2007). Zhou, H. and Zou, W., “Income Distribution Dynamics of Urban Residents: The Case of China (1995–2004),” Frontiers of Economics in China, 5: 114–134 (2010).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Chapter 3

The Energy Tax and Regional Development Zhengning Pu and Yasuhisa Hayashiyama

3.1. Introduction The 200 years since the industrial age have been the most brilliant period in human history. This is based on various resources and the greatly enhanced use of energy, the wealth created for all of human society, and the rate of economic development. The progress of science and technology has reached a point that would have stunned earlier generations. This unparalleled growth and creation has also brought unparalleled destruction and consumption. Looking at current times, we can see that the consumption of oil, which is the primary class of human consumption of energy, has risen from 3,813 million tons of oil equivalent in 1965, to 11,164 million tons of oil equivalent in 2009. In just 44 years, the consumption of primary class energy in the world has increased nearly three times over. This wanton energy consumption has brought a series of environmental problems that poses a direct threat to the survival of humankind. For example, the extreme type of climate change which has putatively been caused by global warming is one consequence of mankind’s apparently unlimited need for energy consumption. Facing increasingly extreme environmental events, society has realized that more attention must be paid to protecting the environment instead of solely focusing on the development of our own species. Therefore, several international cooperation agreements such as the Kyoto Protocol have been instituted for the purpose of ensuring sustainable development for the whole world. However, the inconsistency of economic development throughout the world has complicated the efforts to implement further cooperative mechanisms. Several agreements have apparently reached an impasse. The United

47

March 5, 2013

48

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

Nations Climate Change Conference (COP15) held in Copenhagen in 2009 has come to a stalemate because of differences concerning the responsibility required amongst developing and developed countries. At COP16, which was held in Cancun in 2010, developed countries refused to continue with the second phase of the Kyoto Protocol commitments. Specific agreements have been delayed for a year to allow time for further discussions. In a long see-saw of policy, developed countries have been confronted with historically huge public debts at the same time in which they ask for limitations on the huge emissions of developing countries today. The two groups accuse each other of abrogating the main points of contention. In these controversies, China, as the world’s largest developing country, is usually pushed to the front of the stage. China, the second largest economy in the world and one of the world’s largest carbon emitters in recent times, has often been criticized due to its environmental policies being insufficiently strong and lacking in its responsibility for environmental protection. For example, a report from the International Energy Agency (IEA) in 2010 stated that1 “China has now overtaken the United States to become the world’s largest energy user”. A related news report added that “China’s demand today would be even higher still if the government had not made such progress in reducing the energy intensity (the energy input per dollar of output) of its economy”. In fact, despite accusations and warnings by international bodies, after 30 years of rapid economic development, China has severely damaged its own environment for the sake of greater consumption. The World Energy Report by BP (2010) showed that in 2009, China’s coal consumption accounted for more than 49% of the world’s total coal consumption. In terms of global petroleum use, China accounted for 10.6% of the total. This largescale energy consumption has caused shocking damage in China: in Shanxi,2 China’s leading coal output region, nearly 2 million square kilometers of land has been hollowed out for coal mining. The underground layers have been entirely emptied in these areas, which constitute one-eighth of the soil surface in this province. In addition, the use of multiple energy resources 1 The original sentence can be seen from IEA’s website report entitled “China overtakes the

United States to become world’s largest energy consumer”. 2 See report from The Beijing News December 14, 2010.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

49

creates large-scale emissions of carbon dioxide (CO2 ), sulfur dioxide, and other harmful substances. In the face of multiple accusations of environmental damage, China has taken a series of measures intended to change their situation. As been introduced in China Environment Bulletin 2009 (2010), a huge amount of capital investment has been devoted to the development of new energies. Simultaneously, China has shutdown many of its small, highly polluting companies: their so-called “backward production capacity”. In 2009,3 China shutdown 60.06 million small thermal power units, which is equivalent to saving 6,400 tons of raw coal. In 2010, China planned and completed 88% shutdowns of 1,539 small coal mines in an attempt to reduce the damage caused by mining and pollution. In addition to the regulation of market operation, China has sought to change its tax system to meet the country’s future demands for economic development. Those changes include attempts to introduce a carbon tax, to change the resource tax rate, and to start an energy tax pilot program. In terms of the above fiscal policy adjustments, the country has decided to change its energy resource tax from almost zero to a 5% ad valorem energy tax for all energy goods. This policy was first piloted in Western China in 2011. In this study, a spatial computable general equilibrium (SCGE) model has been used to evaluate the possible effect of the above pilot policy for China from multiple perspectives.

3.2. Literature Review China’s main energy users have paid very low taxes or resource occupation fees to obtain the right to use energy resources in China throughout this period. According to a report from the Institute of Economics of China, since January 1, 1994, China’s energy resource taxes were about 8–30 RMB/ton for petroleum, 2–15 RMB/thousand m3 for natural gas, and 0.3–5 RMB/ton for coal. Only in 2006 did China start a policy targeting crude oil, also known as the “Special income levy”, which requires 20–40% taxation of income greater than US$40 earned for every barrel of crude oil. Compared to the prices of these energy resources, China’s energy users paid a tax that was close to zero for their indiscriminate use of 3 Report see in Economic Observer Net 2010-10-25.

March 5, 2013

50

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

these valuable, non-renewable energy resources. Under pressure for their large-scale environmental pollution, China also sought to adjust their fiscal policy to address this pollution. To evaluate the efficiency of policy on energy goods, the computable general equilibrium (CGE) model has been a part of the evaluation. Borges and Goulder (1997) have discussed the effect of high growth energy prices on the economy by using the CGE model, while Fullerton et al. (1997) have analyzed early research about the use of the CGE model on taxation policies. In recent years, the CGE model in China has gradually become more common. Researchers have begun using CGE models to study the effects of changes in China’s fiscal policy towards energy resources on the nation’s economy. For example, He et al. (2002) conducted a study on the relation between the carbon tax and carbon emissions reduction, showing that if China reduced its carbon emissions to 10.5%, 15.5%, 20.5%, 24.5%, and 30.5%, this would alter China’s marginal abatement costs by 88.4 RMB/ton, 146.6 RMB/ton, 219.9 RMB/ton, 289.4 RMB/ton, and 418.2 RMB/ton, respectively. Wang et al. (2005) conducted a similar study in 2004, and established a one-country static CGE model to simulate the effect of an ad valorem tax on energy resources in China. Results showed that taking 2010 as the simulation period, China would suffer a loss of about 0–3.9% of its GDP when the nation’s carbon emission reduction rate is in the range of 0–40%. Moreover, the marginal abatement costs for the nation in this study were about 100 RMB/ton under a reduction rate of 10% and 470 RMB/ton under a reduction rate of 30%. Based on Jian and Saltzman’s model, Wei (2009) developed a CGE model including the environmental feedback. They chose five levels of ad valorem taxation on energy resources (10–50%) to investigate the effects this tax would have on China’s economy. According to their results, aside from the carbon emission reduction effect, an ad valorem tax on fossil fuels led to a worsening of several of the nation’s microeconomic indexes, such as GDP, household utility usage, and unemployment rate. However, their results also show that only when the ad valorem tax rate is over 20% it will cause all the microeconomic indexes of the country to be reduced by more than 1%. In 2009, Yang et al. (2009) established a model showing that an energy tax had only a slight impact on China’s economy (all effects were under about 1%); at the same time, energy taxes will effectively change

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

51

China’s energy structure, reducing the use of coal as a share of total energy use. In addition to the studies described above, Pang (2008), Xia et al. (2010) and others have undertaken related research into areas including fuel tax cutting. All the above-mentioned studies share one commonality — they all regarded the nation as a whole to elucidate the effects that are expected to occur as a result of the exogenous shock. There are some CGE models that took multiple regions into consideration, such as the GTAP model from Purdue University which is reported in Hertal (1997) and the LINKAGE model from the World Bank, which is detailed in Van der Mensbrugghe (2005). These models divide the world into different regions and use international trade as the link between them. Besides this kind of worldwide multi regional CGE model, Monash University also created a multi-region CGE model — MMRF-Green in which Australia is divided into several states and territories. The detail of such model could be found in Adams et al. (2000). This was used for a wide range of applications including the analysis of greenhouse problems. China is a huge country; with its large population and wide area, every policy that is instituted brings different effects on different regions and industries. These differences are impossible to observe using any single country model. For this reason, we use the SCGE model to augment the results provided by those earlier China studies.

3.3. Model Structure The single-country SCGE model used in this study is a static CGE model that incorporates the assumption of a perfectly competitive market and the minimization of production costs for producers. For final users, all regional final users’ savings would be re-invested into a single region’s regional economy. International trade follows a small-country assumption and the Armington assumption. CO2 emission in this model is assumed to occur mainly from energy resource consumption and in this SCGE model, total CO2 emissions in China was treated as an exogenous parameter which is in reference to the World Bank database.4 4 This dataset can be found in the World Bank database (www.worldbank.org).

March 5, 2013

52

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

Output for industry j of Region s

VAE

Labor

Composite Intermediate Input 1

Capital-Energy Composite

Figure 3.1.

…

Composite Intermediate Input 22

Local Region 1

…

Region 8

Production Structure.

This model is based on the one-country static CGE model presented by Hosoe et al. (2004) and partly refers to the CGE model by Ueda (2010). Details related to the model’s structure can be found in a discussion paper by Pu (2011). There were six nests included in the model structure: production agent activities, capital–energy substitution, household agent activities, regional government agent activities, exports, and imports. The structure nests are given in Figure 3.1. In Figure 3.1, the producer agent activity in the market is described. In this structure, the labor and capital–energy composite is combined under the constant elasticity of substitution (hereafter CES) function as in the VAE block. Different regions’ same intermediate commodity outputs are combined as a composite intermediate input also under the CES function.All intermediate inputs plus one region’s VAE input under the CES composition became the total input for industry j’s production in region r. In Figure 3.2, the structure of the Capital–Energy composite block of Figure 3.1 is described. For such a substitution structure, energy substitution has only been considered between primary energy resources, with no substitution between primary energy resources and secondary energy resources being considered. This composite was chosen as throughout China’s history, nearly 80% of China’s electricity has been derived from burning coal. This made the relationship between secondary energy resources (electricity) and primary energy resources (coal, etc) more likely to include demand and supply rather than substitution, in both the short term and medium term.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

53

Capital–Energy composite CES

Energy Composite

Capital

CES

Coal Composite

Figure 3.2.

Non-Coal Composite

Capital–Energy Composite.

Table 3.1. Primary Energy Composition. Item

1995

2000

2003

2004

2005

2006

2007

As Percentage of Primary Energy Production (%) (calorific value calculation) Coal 77.4 71.02 71.63 71.31 72.76 72.97 72.82 Petroleum 18.3 24.28 23.24 23.39 21.74 21.25 20.64 Natural Gas 1.88 2.46 2.7 2.72 2.9 3.18 3.67 Hydroelectric Power 1.84 2.08 2.11 2.26 2.3 2.31 2.57 Nuclear Power 0.13 0.16 0.32 0.32 0.3 0.29 0.3 As Percentage of Primary Energy Production (%) (coal equivalent calculation) Coal Petroleum Natural Gas Hydroelectric Power Nuclear Power

74.6 17.5 1.8 5.71 0.39

67.75 23.21 2.35 6.23 0.46

68.38 22.21 2.58 5.93 0.9

67.99 22.33 2.6 6.2 0.88

69.11 21 2.8 6.26 0.83

69.4 20.4 3 6.4 0.8

69.5 19.7 3.5 6.5 0.8

Source: China Energy Statistical Yearbook 2008.

Furthermore, the energy composite data from China’s Energy Statistical Yearbook (2008) presented in Table 3.1 shows that the two main energy consumption goods for China’s consumption were coal and petroleum and natural gas. All equations in this model can be found in the Appendix to this chapter. Equations (A3.1) to (A3.15) describe all production activities that are included in the CES function presented in Figures 3.1 and 3.2. As explained

March 5, 2013

54

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

in this paper, subscripts r and s stand for the regional divisions, i and j are divisions of industry or commodities and “e” and “ne” are used to differentiate energy goods from non-energy goods. For energy–capital substitution equations, EN e,s,j represents the composite intermediate input of the energy sector and PEN e,s,j is the price for it (every “P” variable was the price variable for the following explanation). Additionally, ENX r,e,s,j represents the intermediate inputs of the energy sector; ENC s,j represents the energy composite; and Ks,j and Ls,j stand for the capital input and labor input respectively, where KENC s,j means the composite good of capital and energy. Furthermore, VAE s,j signifies a composite good of labor, energy and capital; Zs,j denotes output from industry j in region s; XX r,ne,s,j denotes the intermediate input of non-energy sector in different regions; and Xne,s,j signifies the composite intermediate input of the non-energy sector. The activities of household final user agents are shown in Figure 3.3. As the structure shows, these final user agents earn their income based on the primary factors they own which include, in this model, only labor and capital. They will use part of the revenue for consumption activities and take the rest as savings. In this model, all savings from final users will be re-invested into the local regional economy. These activities are defined in Equations (A3.16)–(A3.18). For household final user agent activity equations, XXH r,i,s signifies the household consumption of good i from region r consumed by region s consumers, XH i,s denotes the composite household consumption, FK s and FL s represent the factor endowments of capital and labor respectively,

Capital Income

Labor Income

Household Saving (Investment)

Figure 3.3.

Household Consumption

Household Activities.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

Production Tax

Energy Tax

Government Saving (Investment)

Figure 3.4.

55

Direct Tax

Government Consumption

Government Activities.

SPs stands for the total saving of the private department and TDs represents the direct tax. Figure 3.4 is the activity description for the government agent. In such a structure, the regional government earns their income based on three kinds of taxation: a production tax, an energy tax, and a direct tax. A direct tax is levied towards the primary factor income of household agents. The government collects these taxes as income and spends them on consumption and investment as household agents. Government activities are defined in Equations (A3.19)–(A3.22). Equations (A3.23)–(A3.25) determine their investment activities. Among the equations, Equation (A3.19) shows the relationship between government consumption XGr,i,s , production tax TPs,j , direct tax TDs , energy tax TE s,j and government saving SGs . Equations (A3.20)–(A3.22) show the relationship for each tax with the respective tax rates (TDR as the direct tax rate, TPRs,j as the production tax rate, TER for the energy tax rate) and the tax base. Equations (A3.23)–(A3.25) describe the relationship between investment and saving. In Equation (A3.23) investment, INV r,i,s , is described as being related to private saving, SPs , the rate of currency exchange, EXR, foreign savings, FSs , and the government savings, SGs . Equations (A3.24) and (A3.25) describe the quantitative relationship between savings and the savings rate, including between private saving and the private saving rate, SPRs , and government savings and the government saving rate, SGRs . In the export structure, one region’s total output in one industry (see the top block of Figure 3.5) is divided to feed the export supply and domestic supply under a constant elasticity of transformation (and hereafter CET)

March 5, 2013

56

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

Region output CET

Export supply

Domestic supply

Figure 3.5.

World Market

Export Structure.

Domestic Supply

Import

CES Armington composite commodity

Intermediate Input

Private Consumers

Government Consumer

Figure 3.6.

Investment in Fixed assets

Stock Investment

Import Structure.

function. Equations (A3.26)–(A3.28) describe this CET relationship. In the CET expression, Zr,i stands for the total output, Er,i signifies the export supply and Dr,i represents the domestic supply. For the import structure represented in Figure 3.6, world market imports are combined with one region’s industry’s local supply (domestic supply block of Figure 3.5) in the CES function under the Armington assumption into the Armington composite commodity supply block. All the domestic demand for a model region (intermediate input, final user agent consumption, etc) is satisfied by the local Armington composite commodity supply block. Equations (A3.29)–(A3.31) constitute the expression for the CES function, where Qr,i represents the Armington composite commodity. Equation (A3.32) shows the relationship between the export price PE i and the world export price PWE i . Equation (A3.33) shows the relationship

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

57

between the import price, PM i , and the world import price, PWM i . As the last of these equations, Equation (A3.34) expresses the relationship through exports, foreign savings, and imports. Equations (A3.35)–(A3.38) represent the market clearing condition of the model. Equation (A3.35) portrays the commodity market clearing condition, while Equations (A3.36) and (A3.37) represent the balance in the labor and capital markets. In the capital market, the model is based on the assumption that all the capital in a country can be transferred freely amongst regions and industries. Equation (A3.38) shows how household utility UU s is calculated for this analysis. Equation (A3.39) is the objective function of the whole model, and describes the main target of this project: social welfare (SW).

3.4. Data Resource and Scenarios 3.4.1. Data resource The data used for this research is drawn from the 2000 China’s multiregional input–output matrix published by the Institute of Developing Economies (2003). This is the newest multi-regional input-output data of China besides Ichimura, S. and Huijiong, W’s (2004) data. The input– output matrix includes 8 regions and 30 commodity sectors. In this database, 31 Mainland China provinces and municipalities were divided into eight regions; the region division structure is shown in Table 3.2. Following this structure, the Western area of China is considered to include the regions of Table 3.2. Regional Division Code. Region Northeast North Municipalities North coast East coast South coast Central Northwest Southwest

Included Provinces and Cities Heilongjiang, Jilin, Liaoning Beijing, Tianjin Hebei Shandong Jiangsu, Shanghai, Zhejiang Fujian, Guangdong, Hainan Shanxi, Henan, Anhui, Hubei, Hunan, Jiangxi Inner Mongolia, Shanxi, Ningxia, Gansu, Qinghai, Xinjiang Sichuan, Chongqing, Yunnan, Guizhou, Guangxi, Tibet

March 5, 2013

58

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

the Northwest and Southwest. However, for energy resource production, it must be noted that the main coal energy producers are located in the Central and Northwest regions. Following this regional division, the East coast and South coast regions can be considered as the two most export-oriented economy areas in China. The Northeast region’s economy is heavily based on heavy industry. This regional division is based not only on geographical factors, but also on economics. Under this regional division, the eight divided regions have distinct economic characteristics. The Northeast region has long been an area of heavy industrial concentration. Its abundant mineral resources support and underpin this heavy industry. The nation’s largest oil field, Daqing, is located in this region, which also possesses a highly developed oil industry. In addition to its industrial advantages, this area is well-known as a crop production base. The North Municipalities region is special amongst the regional divisions. The region includes only two cities: Beijing and Tianjin. Although they might appear similar in area and economic scale, the political significance of these two cities, and the high-technology equipment manufacturing and financial services based within these cities, give sufficient reason for them to be regarded as independent area divisions. The Northern Coast area has rich natural resources and various industries including manufacturing, energy, steel, petrochemicals, and hightechnology industries. At the same time, this region has a large output of agricultural products, including cotton, edible oils, aquatic products, and vegetables. Its balanced economic structure gives the region a strong competitive advantage amongst the regional economies. The Eastern Coast and Southern Coast share certain similarities in their economic structures. Both areas have export-oriented structures, with textile products and toys being the Eastern Coast’s main export products, and textiles and light chemical products being the Southern Coast’s main export products. Benefiting from globalization due to their export-orientated model, these two regions have rapidly accumulated large amounts of wealth and have become the most economically developed areas in China. The Central area includes six provinces that even ancient Chinese generals designated as the Central Plains area. This region is a less well-off

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

59

and is the major supplier of labor in China. Every year, millions of workers move from this area to the coastal parts of the nation to find job opportunities. They are the major population of migrant workers nationwide. Aside from the labor supply, the Shanxi province in this area is also the major supplier of coal for the nation. The Northwest area includes several of the least-developed provinces in China, such as the Ningxia, Gansu, and Qinghai provinces. Although it may be the least developed region in the country, this region has many untapped mineral resources. These resource reserves endow this area with great economic development potential for the future. The Southwest area is the last region in this division. This region was called “the third line” in old China’s strategic planning. This area has a complete industrial system and could potentially achieve self-sufficiency. However, the industrial structure of this region emphasizes military– industrial production; this has at some level limited the economic development of the whole area. In addition to its industrial system, this area is known for its vast reserves of natural gas and rare earth minerals. For the classification of commodities, we use the GTAP-7 database (Burniaux and Truong, 2002) and reclassified the industry data. As Table 3.3 shows, these commodities were reclassified into 24 sectors.

3.4.2. Scenarios China is willing to use fiscal policy to reduce energy resource consumption and carbon emissions. To do so, they changed their former energy resource policy to a 5% ad valorem energy tax on all energy goods. This policy would have been the first pilot in the Xinjiang province since 2011. In this study, we follow China’s lead and extend the 5% ad valorem energy tax to the whole nation as a scenario to elucidate regional differences that are expected to occur under a national energy tax policy. In this scenario, six indexes are useful for analysis. These indexes are: the petroleum and natural gas mining output reduction, the coal mining industry output reduction, the regional household utility rate of change, the industry output rate of change, the regional GDP rate of change, and the national GDP rate of change. Because this model

March 5, 2013

60

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

Table 3.3. Reclassified Commodity Sectors. Reclassified Commodity Sectors 1 Agriculture 2 Coal Mining 3 Oil and Gas Mining 4 Other Mining 5 Food Manufacturing 6 7

Textile Wearing Apparels

8 9

Sawmills and Wood Products Paper Products

10

Petroleum Processing and Coking Chemical Industry Non-metallic Mineral Products Metal Smelting and Pressing Metal Products Machinery Industry Transport Equipment Electrical, Machinery, and Equipment Electronic and Communication Equipment Other Manufacturing Industries

11 12 13 14 15 16 17 18 19

20

Electricity, Water and Gas Supply

21 22

Construction Transportation and Warehousing Commercial Services

23 24

China Multi-Regional I-O 30 Sectors Agriculture Coal Mining Oil and Gas Mining Metal ore Mining, Non-Metal ore Mining Food Manufacturing and Tobacco processing Textile Wearing apparel, leather, furs, down and related products Sawmills and furniture Paper and products, printing and recording medium reproduction Petroleum Processing and Coking Chemical Industry Non-metallic mineral products Metal smelting and pressing Metal products Machinery Industry Transport Equipment Electrical, Machinery, and Equipment Electronic and communication equipment manufacturing Measuring Instruments and Office Machinery, Machinery and equipment repair, Other manufacturing industries, Waste disposal Electricity, Steam, and Hot Water Production and Supply, Gas Production and Supply, Tap Water Production and Supply Construction Transportation and warehousing Wholesale and retail trade Services

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

61

incorporates energy resources as the major carbon emission factor of economic activities, energy production in this model is said to be produced by the coal mining sector and petroleum and natural gas mining sector. Because China’s energy self-sufficiency rate is higher than 95%, it can be assumed that China’s energy consumption products are produced mainly through its own energy production sector. Therefore, the first two indexes described above were used to judge the carbon emission control effects of this policy (more energy production reductions means greater reduction in carbon emissions). The other four indexes that followed are used to judge whether the scenario policy might or might not heavily affect the regional economy and household living conditions.

3.5. Analysis of the Simulation Results In the theoretical model, we assumed that capital was able to move freely throughout the nation, but under the consideration of stabilizing and ensuring the feasibility of the simulation results, we chose a model under which capital could not transfer freely throughout the regions as our simulation model. Results for the scenario simulation were the following. As Figure 3.7 shows, a nationwide 5% ad valorem energy tax will result in a total reduction of 596.4 million RMB of petroleum and natural gas mining output and 137.4 million RMB of coal mining output. We calculated this result using energy prices of about 166 RMB/ton for coal and 1,150 RMB/ton for the petroleum and natural gas composite.

Million RMB

-137.4

-596.4 Petroleum and natural gas mining

Figure 3.7.

coal mining

Energy Output Reduction of China.

March 5, 2013

62

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

Those energy prices were mainly referenced from data from the China Development and Reform Commission Consumer Division and the Brent crude oil price in 1997. The carbon emissions factor for this study was referred from the IPCC (2006) (0.755 for coal, 0.448 for natural gas, and 0.585 for petroleum), which implies a country total of about 686,496.8 tons in carbon emission reductions. The Marginal Abatement Cost (denoted hereinafter as MAC) for China was about 707 RMB/ton. Considering the rate of exchange between the RMB and US dollar during the period of simulation data, the MAC for this carbon emission reduction policy was about 85.2 USD/ton. At the same time, the nation’s total GDP decreased by only 0.0032%. The industry output rate of change in the national level for each industry is presented in Figure 3.8.

Services Commercial Transportation and Warehousing Construction Electricity, Water and Gas Supply Other Manufacturing Industries Electronic and Communication Equipment Electrical Machinery and Equipment Transport Equipment Machinery Industry Metal Products Metal Smelting and Pressing Non-metallic Mineral Products Chemical Industry Petroleum Processing and Coking Paper Products Sawmills and Wood Products Wearing Apparels Textile Food Manufacturing Other Mining Oil and Gas Mining Coal Mining Agriculture

%

-10

Figure 3.8.

-5

0

5

10

Industry Output Rate of Change.

15

20

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

63

As shown in Figure 3.8, 19 of the 24 industries showed a decreased output under the simulation scenario. The paper products industry decreased the most at over 7.5%, followed by the transport equipment industry, other manufacturing industries, and the oil and gas mining industry. Productivity decreased by about 5% in all of these industries. However, while most industries were in a recession during the scenario, the textile, non-metallic mineral products, electrical machinery and equipment, construction and services industries showed increased business activity. The increase in these industries might have occurred because these industries are less dependent on the two energy goods used in this study’s model structure. Although the national result shows that the simulation policy has a positive effect, regional level data show that under an identical simulation policy, the different regions’ reactions are entirely different. Figures 3.9 and 3.10 show the GDP rate of change and household utility rate of change in the eight regions for this simulation. Regarding the GDP change, although the rate of change was small, four of the eight regions showed a positive change in their GDP growth. However, three of the other four were negative. The regional differences show that for GDP change, relative developed areas such as the North municipalities, East coast and South coast show a decrease in GDP. At the same time, areas which support the nation’s energy requirements, like the Northeast (with the nation’s largest petroleum field), Central China and the Northwest (each

0.3 0.2 0.1 0 -0.1 -0.2

%

-0.3

Figure 3.9.

Regional GDP Rate of Change.

March 5, 2013

64

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China 0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 -0.05 -0.1 -0.15

%

Figure 3.10.

Regional Household Utility Rate of Change.

area includes one important coal supply province), show an increase in their GDP. We assume that the simulation scenario policy also benefits activities in these energy sectors at some level. In most regions, household utility changes in the same direction as the region’s GDP changes. However, the Northeast and the North Municipalities regions show a conflict between their GDP change and household utility change. This might have resulted due to industry differences. For example, the greatest increased industry under the simulation scenario for the Northeast region was the electrical machinery and equipment industry. For North Municipalities, the industry showing the largest increase was the textiles industry. Traditionally and in reality, these two areas are not wellknown for these industries. The simulation showed that increased industry activity might bring GDP growth for the region. However, because the main industry it benefited was not the major industry in the region, calculations of household utility will probably show a decrease when industries employing most of the region’s labor decrease in output. In these two regions, the above mentioned situation occurred coincidentally. In the Northeast, the most affected industry was the transport equipment industry, followed by the petroleum and natural gas industry. As described in Section 3.4.1 about

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

65

the region division, the major industries in which most people worked in the Northeast region are these two industries. For this reason, although the huge growth in other industries might lead to GDP growth for the region, the decrease in the region’s major industry is expected to engender losses in the region’s household utility. The same result in the North Municipalities can also be explained by this reason: with regional industry changes, a region’s most prominent industries like electronics and communication equipment and service showed the greatest reduction in output. Finally, with regards to the energy reduction component, different regions of China show different but interesting results. According to changes in petroleum and natural gas output as seen in Figure 3.11, every region showed a cut-back in petroleum and natural gas output. In this reduction, the Northeast region suffers the most, followed by the Northwest area. As opposed to the petroleum and natural gas output change, the regional coal output change shown in Figure 3.12 may easily confuse a first-time viewer: although five of eight regions show a significant decrease in their coal production, the North municipalities, East coast and Central regions showed an increased coal output. Especially for the Central area, this additional amount is remarkable. Regarding the percentage specifically, even in the Central area compared with the benchmark, its coal output increased only about 0.07%. As the previous data shows, national and regional industry output and the GDP did change significantly, which means that Million RMB

-17.50 -91.57

-3.92

-25.93

-36.59 -82.06

-99.33

-239.54

Figure 3.11.

Regional Petroleum and Natural Gas Output Change.

March 5, 2013

66

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China Million RMB 63.86 6.28

3.17

-2.16 -29.94

-9.29 -31.88

-85.13

Figure 3.12.

Regional Coal Output Change.

eventually some of the energy reduction in certain regions might be offset by growth in other regions. Since China’s energy usage is nearly 80% reliant on coal and Central China possesses the most important coal resource province, Shanxi, therefore what Figure 3.12 shows could be perfectly sensible.

3.6. Conclusion As it can be seen in the scenario simulation results described in Section 3.5, our SCGE model proves that an ad valorem energy tax can reduce carbon emissions in China, with an acceptably marginal abatement cost. However, the fiscal policy might have different effects in different regions. China’s light industry, service based, and energy resource based regions might benefit from this policy, although energy-hungry industries such as the heavy-industry-based areas may be strongly affected by this energy tax. In additional, in terms of the reduction in the output of coal mining, petroleum and natural gas, it should be noted that the energy tax seems to affect the petroleum and natural gas resources more than the coal sector. As with any CGE model, the SCGE model in this study also confronts the problems of missing data and disputes related to the elasticity selection. However, this model can present a useful forecast for China to elucidate the possible internal influences on its energy tax policy.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

67

Appendices to Chapter 3 A3.1 Subscripts Subscript Mark r, s i, j e ne

Explanation Region Industry Energy sector Non-energy sector

A3.2 Variables Variables Variable Name VAE s,j Ls,j KENC s,j Ks,j ENC s,j EN e,s,j ENX r,e,s,j Xne,s,j XX r,ne,s,j XXI r,i,s,j Zs,j XXH r,i,s XH i,s SPs SGs TDs TPs,j TE s,j XGr,i,s INV r,i,s

Explanation Composite good of Labor, Energy and Capital Labor input Composite good of capital and energy Capital input Energy composite Composite intermediate input of energy sector Intermediate input of energy sector Composite intermediate input of non-energy sector Intermediate input of non-energy sector Intermediate input Output Household Consumption Composite household Consumption Saving of private department Saving of government Direct Tax Production Tax Energy Tax Government Consumption Investment (Continued)

March 5, 2013

68

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

(Continued) Variables Variable Name

Explanation Export supply Import demand Domestic supply (demand) Armington composite good of import and Domestic demand Price of VAEs,j Price of Ls,j Price of KENCs,j Price of Ks,j Price of ENCs,j Price of energy composite Price of non-energy composite intermediate input Price of Composite household Consumption Price of output Price of export Price of import Price of domestic Price of Armington composite goods Foreign savings Utility Total social utility Factor endowment of capital Factor endowment of labor Export Price in world market Import price in world market Production Tax Rate Direct Tax Rate Energy Tax Rate Exchange rate

Er,i Mr,i Dr,i Qr,i PVAE s,j PL s,j PKENC s,j PK s,j PENC s,j PEN e,s,j PX ne,s,j PH i,s PZ s,j PE r,i PM r,i PDr,i PQr,i FS r UU s SW FK s FL s PWE i PWM i TPRs,j TDR TER EXR

A3.3 Equations
ENe,s,j = αENe,s,j

 r

 ρ1 ρ1 βENXr,e,i,s ENXr,e,s,j

1

,

(A3.1)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development


ENXr,e,s,j =

 1−ρ1

ρ1 PENe,s,j αENe,s,j βENXr,e,s,j

1

ENe,s,j ,

PQr,i


ENCs,j = αENCs,j



69

(A3.2)

 ρ1

2

ρ2 βENe,i,s ENe,s,j

,

(A3.3)

ENCs,j ,

(A3.4)

e


ENe,s,j =

 1−ρ1

ρ2 PENGs,j αENCs,j βENe,s,j

2

PENe,s,j

 1 ρ3 ρ3 ρ3 KENCs,j = αKENCs,j βENCs,j ENCs,j + βKs,j Ks,j ,
Ks,j =

 1−ρ1

ρ3 PKENCs,j αKENCs,j βKs,j

3

KENCs,j ,

PKs,j


ENCs,j =

ρ3 PKENCs,j αKENCs,j

3

KENCs,j ,

 1 ρ4 ρ4 VAEs,j = αVAEs,j βLs,j Lρs,j4 + βKENCs,j Ks,j , Ls,j =

ρ4 PVAEs,j αVAEs,j βLs,j

KENCs,j =

VAEs,j ,

(A3.9)

 1−ρ1

4

VAEs,j ,

(A3.10)

,

(A3.11)

Xne,s,j ,

(A3.12)

PKENCs,j


Xne,s,j = αXne,s,j

(A3.8)

4

ρ4 PVAEs,j αVAEs,j βKENCs,j



(A3.7)

 1−ρ1

PLs,j


(A3.6)

 1−ρ1

PENCs,j




(A3.5)

 ρ1 ρ5 βXr,ne,i,s XXr,ne,s,j

5

r


XXne,s,j =

ρ5 PXne,s,j αXne,s,j βXr,ne,s,j

PQr,i

 1−ρ1

5

VAEs,j = AVAEs,j Zs,j ,

(A3.13)

Xne,s,j = AXne,s,j Zs,j ,

(A3.14)

March 5, 2013

70

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

PZs,j = PVAEs,j AVAEs,j +



PXne,s,j AXne,s,j ,

(A3.15)

ne

 ρ1


XHi,s = αXHi,s



6

ρ6 βXXHr,i,s XXHr,i,s

,

(A3.16)

XHi,s ,

(A3.17)

r


XXHr,i,s =

ρ6 βXXHr,i,s PXHi,s αXHi,s PQr,i

 1−ρ1

6

βXHi,s (PLs FLs + PKs FKs − SPs − TDs ), PXHi,s     βGr,i,s  = TPs,j + TDs + TEs,j − SGs  , PQr,i j j

XHi,s =

XGr,i,s

(A3.18)

(A3.19)

TDs = TDR(PKs FKs + PLs FLs ),

(A3.20)

TPs,j = TPRs,j PZs,j Zs,j ,

(A3.21)

TEs,j = TER · PENCs,j ENCs,j ,

(A3.22)

INVr,i,s =

βINVr,i,s (SPs + EXR · FSs + SGs ), PQr,i

SPs = SPRs (PLs FLs + PKs FKs ),     TPs,j + TEs,j  , SGs = SGRs TDs + j

Er,i =
Dr,i =

(A3.24) (A3.25)

j

1

ρ7 ρ7  ρ7 Zr,i = αZr,i βEr,i Er,i + D1r,i Dr,i ,




(A3.23)

ρ7 βEr,i (1 + TPRr,i )PZr,i αZr,i PEi ρ7 βEr,i (1 + TPRr,i )PZr,i αZr,i PDi

(A3.26)

 1−ρ1

7

Zr,i ,

(A3.27)

Zr,i ,

(A3.28)

 1−ρ1

1

ρ8 ρ8  ρ8 Qr,i = αQr,i βMr,i Mr,i + βD2r,i Dr,i ,

7

(A3.29)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development


Mr,i =
Dr,i =

PQr,i αQρr,i8 βMr,i PEi

 1−ρ1

PQr,i αQρr,i8 βD2r,i PDi

8

Qr,i ,  1−ρ1

r

Qr,i ,

r

r

Qr,i =

 s

+

i

XXHr,i,s +

 s



s



(A3.33) (A3.34) XGr,i,s

s

XXr,i,s,j +



INVr,i,s ,

s

(A3.35) Ls,j = FLs ,

j



j

(A3.31) (A3.32)

PMi = EXR · PWMi ,   PWEi Er,i + FSr = PWMi Mr,i ,

i

(A3.30)

8

PEi = EXR · PWEi , 

71

Ks,j =



(A3.36)

FKs ,

(A3.37)

XHi,s ,

(A3.38)

UUs .

(A3.39)

s

j

UUs =

s

SW =

 s

References Adams, P.-D. and Parmenter, B.-R. et al., “Analysis of Greenhouse Policy using MMRF-GREEN,” Centre of Policy Studies, Monash University (2000). Burniaux, J.-M. and Truong, T. P., “GTAP-E: An Energy-Environmental Version of the GTAP Model,” GTAP Technical Paper No. 16 (2002). Borges, A.-M. and Goulder, L.-H., Decomposing the Impact of Higher Energy Prices on Long-Term Growth. Cambridge: Cambridge University Press (1997).

March 5, 2013

72

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

Regional Development and Economic Growth in China

BP, Statistical Review of World Energy 2010 (2010). China Statistics Press, China Energy Statistical Yearbook 2008. China: China Statistics Press (2009). Ministry of Environmental Protection of the People’s Republic of China, China Environment Bulletin 2009 (2010). Fullerton, D.-F. and Henderson, Y.-K. et al., A Comparison of Methodologies in Empirical General Equilibrium Models of Taxation. Cambridge: Cambridge University Press (1997). He, J. H., Shen, K. T. and Xu, S. L., “Carbon Taxation and CGE Model of Carbon Dioxide Mitigation,” Quantitative and Technical Economics, 19(10): 39–47 (2002). Hosoe, N., Gasawa, K. and Hashimoto, H., Textbook of Computable General Equilibrium Modeling. Japan: University of Tokyo Press (2004). Hertal, T. W., Global Trade Analysis: Modeling and Application. Cambridge: Cambridge University Press (1997). Institute of Developing Economies, Multi-Regional Input-Output Model for China 2000. Japan: Institute of Developing Economies, Japan External Trade Organization (2003). IPCC, “Introductions,” in 2006 IPCC Guidelines for National Greenhouse Gas Inventories. IPCC (2006). Ichimura, S. and Huijiong, W., Interregional Input-Output Analysis of the Chinese Economy. Tokyo: Sobunsha Press (2004). Pu, Z. N., “A Study on China’s Energy Tax Policy Using the SCGE Model,” Discussion Paper, Tohoku University (2011). Pang, J. et al., “Analysis on Economic Impact of China’s Fuel Tax by Using CGE Model,” Inquiry into Economic Issues, 11: 69–73 (2008). Ueda, T. (eds.), Regional and Urban Economic Analysis using Excel. San Antonio, Texas: Corona Publishing (2010). Van der Mensbrugghe, D., “Linkage Technical Reference Document: Version 6.0,” Development Prospects (DECPG), The World Bank (2005). Wang, C. and Chen, J. N. et al., “Impact Assessment of CO2 Mitigation on China Economy Based on a CGE Model,” Science and Technology, Journal of Tsinghua University, 45(12): 1621–1624 (2005). Wei, W. X., “An Analysis of China’s Energy and Environmental Policies Based on CGE Model,” Statistical Research, 3–13 (2009).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

The Energy Tax and Regional Development

73

Xia, C. W. et al., “A Dynamic CGE Research on Influence of China’s Energy Saving Emission Reduction in the Fuel Tax Levied,” On Economic Problems, 2: 64–69 (2010). Yang, L. et al., “Impact Assessment for Energy Taxation Policy Based on a Computable General Equilibrium Model,” China Population, Resource and Environment, 19(2): 24–29 (2009).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch03

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Chapter 4

Regional Distribution of the Creative Class and Its Determinants Jin Hong, Wentao Yu and Fengli Yang∗

4.1. Introduction There has been a growing concern over the role of human capital in regional economic growth. Human capital is created by people, especially those who are skilled and highly educated. Skilled and highly educated people not only have an ability to generate knowledge, but also are able to absorb it; this is why they are more productive. Firms therefore are more competitive if they are located in cities and regions with high levels of human capital. However, talent appears to be concentrated in cities, which are best equipped to attract, mobilize, and organize human capital for economic activity (Jacobs, 1961; Lucas, 1988; Glaeser, 1994). According to Robert Lucas (1988), human capital accumulation is a “social activity”. Highly educated and skilled people interact face-to-face and increase both their own knowledge and the knowledge of others (Jacobs, 1969). The need for face-to-face contact means that dense cities are an ideal “pool” for human capital accumulation. Generally speaking, the standard measure for human capital focuses on educational attainment, usually measuring the share of a population that possesses at least a bachelor’s degree. However, recent studies have shown that this measure only captures a part of a person’s overall capabilities, which reflect accumulated experience, creativity, intelligence, innovativeness, ∗ The authors want to acknowledge the support of the National Social Science Foundation of China (08&ZD043), National Natural Science Foundation of China (71172213), Ministry of Education, Humanities and Social Sciences project (09YJA630153) and the Chinese Academy of Sciences (CAS) Special Grant for Postgraduate Research, Innovation and Practice.

75

March 5, 2013

76

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

and entrepreneurial capabilities as well as the level of schooling. One research strand (Florida, 2002a, 2004; Asheim and Hansen, 2009; Boschma and Fritsch, 2009) suggests an alternative measure for human capital, based on the person’s occupation. The measure specifies a set of occupations that make up the “creative class”, including science, engineering, arts, culture, entertainment, and the knowledge-based professions of management, finance, law, healthcare, and education. Comparative studies show that the creative class measure outperforms the conventional human capital measure in accounting for regional development in Sweden (Mellander and Florida, 2007) and the Netherlands (Marlet and Van Woerkens, 2004). The concept of creative class emerging not only makes up for the deficiency of the new economics on human capital description, but also provides a new perspective for economic growth, under the bottleneck of population, resources, and environment. When it comes to the curriculum and teaching mode of China’s higher education system, the measurement of talent based upon academic degrees cannot exactly distinguish them in a region and correspondingly explain regional economic development as well. Therefore, research on the creative class in China will be particularly meaningful. When creative class theory is applied to China, the more important question that emerges is what are the factors that shape the distribution of talents or the creative classes in the first place? On this score, three different competing theories have been offered. The first argues that universities play a key role in creating initial advantages in human capital, which becomes cumulative and self-reinforcing over time (Glaesar et al., 2005). The second argues that amenities play a role in attracting and retaining highly-educated, high-skilled households (Glaeser et al., 2001; Shapiro, 2006; Li and Florida, 2006; Clark, 2003). The third theory argues that tolerance and openness to diversity are important (Florida, 2002a, 2002b, 2002c; Li and Florida, 2006). However, these empirical studies on this topic are conducted based upon cases of a developed “knowledge economy”. Is it different in a developing country? Or are there other factors that contribute towards the creative class geographic distribution? This chapter contributes to the literature by investigating the largest developing country, China. We conduct an empirical study into the factors influencing the distribution of the creative class’s through economic, technological, cultural, and ecological indicators. We find that urban culture

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

77

opportunity, public service, higher education, and ecological facilities play complementary roles in the distribution of talent and in regional development. In this chapter, we particularly explore the role of ecological indicators as environmental concerns are increasing in China. Methodologically, a ridge regression model is applied to the statistical analysis of the creative class case in order to avoid multicollinearity.

4.2. The Creative Class in China 4.2.1. Provincial distribution According to Richards Florida’s classification (2002a), the creative class is a category of people who are not necessarily highly educated but who are working in creative and innovative jobs. The creative class accounts for about 30% of the American labor force, which includes not only writers, designers, musicians, painters and artists, but also scientists, managers and people employed in computer, engineering, education, healthcare, legal and financial occupations. China is the biggest developing country and possesses a different industrial, societal, demographic structure. Importantly, the country has for a long time restricted internal migration through the HUKOU policy (inhabitant registration system). The central government holds enormous influence on the economic and social activities of the Chinese people, even after decades of decentralization. However, China’s economic development has become oriented toward higher human capital and knowledge-based industries since the late 1990s. A top national policy priority is to “build an innovative country”. Both national policy priority and Florida’s research legitimizes our interest in China’s distribution of talent. Because there is no appropriate occupational data for the creative class available in China’s official statistics, it is impossible to calculate the population of the creative class by following Florida’s exact classification and the corresponding measurement (Florida, 2002a). According to Qian’s research (2008), using China’s Zhuanye Jishu Renyuan (professionals and technical personnel) as a substitute to a large extent, represents the Florida-style creative class (Qian, 2008). Zhuanye Jishu Renyuan includes scientists and engineers, university professors, teachers, agricultural and sanitation specialists, aviators and navigators, economic and statistical specialists, accountant, translators, librarians, journalists, publishers,

March 5, 2013

78

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

25%

20%

15%

10%

6.6% 5%

Figure 4.1.

Tibet

Guizhou

Anhui

Hunan

Henan

Guangxi

Sichuan

Gansu

Yunnan

Jiangxi

Hebei

Chongqing

Shandong

Fujian

Qinghai

Hubei

Zhejiang

Hainan

Heilongjiang

Jiangsu

Ningxia

Guangdong

Shanxi

Mongolia

Shaanxi

Jilin

Liaoning

Tianjin

Xinjiang

Beijing

Shanghai

0%

Creative Class over Total Population in 31 Provinces (2007).

lawyers, artists, broadcasts, athletes, etc. However, this measurement does not contain another important and special talent, namely entrepreneurs who are highly regarded as creative people in China. Consequently, we add these entrepreneurs to the Zhuanye Jishu Renyuan as a Chinese creative class. The density of the creative class measures the proportion of the creative class amongst the local employment population. Figure 4.1 presents the density across 31 provinces in China in 2007, which basically reflect its distribution at regional level. In Figure 4.1, it can be seen that the entire density of creative class is still very low with a national average of 6.6%, and that China’s creative class distribution is highly concentrated and uneven. Particularly, it shows the following characteristics. First, Beijing, Shanghai and Tianjin, three of the four direct-controlled municipalities under the central government, have the most concentrated levels of the creative class. Beijing holds a 21.06% density of creative class out of the employment population. Second, among the 17 province-level regions whose densities are higher than the national average, 10 of these regions are located in the North. One interesting finding is that the Eastern regions (that are more developed) do not always attract more of the creative class than the Central and Western regions (that are less developed). The average density of the creative class is 7.89% in the East, 8.52% in the

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

79

Northeast, 5.18% in the Central region and 5.49% in the West.1 This phenomenon may imply that economic factors are not the only determinant for attracting the creative class. Although the economy is more developed in the Central region than in the West region, its creative class density (5.18%) is less than that in the West (5.49%). As Florida (2002a) and Qian (2008) note, the creative class like to flow to areas where the social culture is tolerant, open, and diversified. If a region has a high level of economic development but only a single dominant culture it will not attract the creative class. Once an area attracts more of the creative class, they become “creative talent magnets” that attract more talent and a high technology industry.

4.2.2. Distribution changes over time Besides the regional distribution of the creative class, its evolution over time is also very important. We calculate the relative growth rate of the creative class in different regions from 2000 to 2007, which shows the regional disparity compared to the other region’s density, using the following equation: V=

T2i − T1i T1i

(i = 1, 2 . . . 31).

(4.1)

T2i and T1i represent the creative class density for province i, in 2007 and 2000 respectively. We use this equation to analyze the increasing rate in the creative class from the year 2000 to 2007. According to the China Statistical Yearbook, we calculate its relative growth rate in each region as shown in Figure 4.2. Figure 4.2 shows the developing trend in the regional creative class in China. The creative class has increased significantly in size with an average growth rate of 36.3%. This growth is different when comparing the coastal 1 The division of the four areas was based on the traditional China economic spatial geographical factors, combining the regional innovative industrial development and the characteristics of industry structure. The Eastern region included Beijing, Tianjin, Hebei, Shandong, Shanghai, Jiangsu, Zhejiang, Fujian, Guangdong, Guangxi, Hainan, and 11 other provinces. The Central region includes Shanxi, Inner Mongolia, Anhui, Jiangxi, Henan, Hubei, Hunan, and seven other provinces. The West area includes Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Xinjiang, Ningxia, and nine other provinces and cities. The Northeast area includes Liaoning, Jilin, and Heilongjiang.

March 5, 2013

80

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

100%

80%

60%

40%

20%

Figure 4.2.

Jilin

Tianjin

Liaoning

Heilongjiang

Hebei

Mongolia

Anhui

Hunan

Gansu

Jiangxi

Qinghai

Xinjiang

Guangxi

Shandong

Henan

Guizhou

Hainan

Sichuan

Ningxia

Yunnan

Shanxi

Chongqing

Hubei

Fujian

Tibet

Shaanxi

Zhejiang

Jiangsu

Shanghai

Beijing -20%

Guangdong

0%

Relative Growth Rate of Provincial Creative Class from 2000 to 2007.

areas to the inland provinces at this time. The maximum growth rate calculated is 94.7% in Beijing, while the minimum value calculated is −10.9% in Jilin, from 2000 to 2007. The East with its higher economic growth rate also experiences a higher creative class growth rate, while in the West and Central it is similar to the provincial distribution. Creative class growth in the Central region is lower than that in the West of China, reflecting that economic factors are not the only determinant influencing the growth of the creative class. Among the 31 provinces, Beijing, Guangdong, Jiangsu, Shanghai and Zhejiang (all are coastal and developed provinces) rank as the top five fastest growing in terms of the creative class. However, there are some heterogeneous provinces. Tibet and Shannxi take the sixth and seventh growth rankings respectively, while Tianjin and Shandong located in the East only experienced a small increase. The Northeast, including Heilongjiang, Liaoning and Jilin, experience the least growth, with Jilin experiencing a negative relative growth rate. As the tourism industry is developing rapidly in Tibet and Shannxi, its education industry is the most improved in recent times. Shangdong and Tianjin have low levels of entrepreneurial activities, leading to a low creative class growth rate. Creative industries

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

81

in the Northeast have always been undeveloped, following on from their dependence on heavy industries since 1949. Economic development has not significantly impacted upon China’s creative class regional distribution and changing distribution over time as we previously took for granted. Consequently, to explore the other factors which impact upon the regional distribution need to be considered.

4.3. Variables and Data From current research, we see many scholars from developed countries paying more attention to both market and non-market factors to analyze factors influencing creative class distribution. Economists traditionally have focused on market factors such as economic development, job opportunities, productivity growth, etc. (Florida, 2002a; Hong et al., 2011). However, recently two different approaches have been utilized that highlight the role of non-market factors in attracting talent. One approach addresses the quality of life and examines the role of natural, cultural, or service amenities. Amenities, according to Kotkin (2000), can attract high-tech industries and workers. Glaeser et al. (2001) argue that the growth rate of cities depends more and more upon amenities that they can provide to attract residents possessing valuable human capital. Lloyd and Clark (2001) depict the city as an “entertainment machine”, implying that consumption and aesthetic innovation play a crucial role in urban economic growth. Shapiro (2006) found that “roughly 60% of the effect of college graduates on employment growth is due to productivity; the rest comes from the relationship between concentrations of skill and growth in the quality of life” (p. 324). The other approach suggests that diversity, openness and tolerance can attract talent to move into a region and then promote economic development in that region, although traditional wisdom favors specialization rather than diversity (Quigley, 1998; Ottaviano and Peri, 2005). Saxenian (1999) has found that approximately 1/3 of scientists and engineers, and 1/4 of new businesses in Silicon Valley have Chinese or Indian born founders. Regions characterized by social diversity and low entry barriers tend to attract talent from various backgrounds. In addition, the natural environment can have an impact on talent mobility. Glaeser and Kohlhase (2004) found that the development of new transport technologies encouraged individuals to vote

March 5, 2013

82

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

with their feet for warm and dry winters, which lead to the first economic stirrings of the modern Sunbelt since the 1960s in the United States. We add these two ecological variables together in our model. Each variable is explained as follows in the upcoming sections.

4.3.1. Dependent variable: Creative class As mentioned above, there are two alternative measures of talent or human capital. The first is the conventional measure that is based on educational attainment, measured as the percentage of a population with a bachelor’s degree or above. The second is an occupationally based measure of the creative class. Several studies have shown the efficacy of this occupationally based measure (Markusen, 2004; Marlet and Van Woerkens, 2004). Florida (2002a) suggested that the creative occupations or the creative class include a super-creative core of people in science and engineering, architecture and design, education, arts, music, and entertainment. Such people are responsible for creating new ideas, new technology, and/or new creative content. Such an occupation also includes a broader group of creative professionals in business and finance, law, healthcare and related fields, who “engage in complex problem solving that involves a great deal of independent judgment and requires high levels of education or human capital”. Although it is impossible to make the definition as exact as Florida’s measurement methodology (Florida, 2002c) due to lack of official statistical data, China’s Zhuanye Jishu Renyuan added entrepreneurs as one dimension to denote the Chinese creative class. In the empirical model, we use creative class density which is measured by the proportion of the creative class among the local employment population (not V index) to analyze the factors influencing their geographic distribution.

4.3.2. Economic variables Generally speaking, the more developed the regional economy, the greater the ability to attract talent. In recent years, many studies focusing on developed countries have found that the economic effect on talent distribution is increasingly weakening (Florida 2002b). However, China is a developing country whose per capita income is much lower than these developed countries, thus it is reasonable to assume that the regional

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

83

distribution of China’s creative class will be greatly influenced by economic factors. These factors are measured by GDP per capital which is from the China Statistical Yearbook (2004–2008). Second, wages are the remuneration for work. Most economists suggest that wages are thus a good proxy for regional labor productivity. This measure is the sum of regional wages divided by the regional employment population in the region. The data is taken from the China Statistical Yearbook (2004–2008). Third, the city (urbanization) index, measured by the proportion of the urban population in the total population from the 2004–2008 China Statistical Yearbook, is employed to examine whether urbanization can help attract talent. Talent generally likes to concentrate in cities to undertake innovative and entrepreneurial activities. Accordingly, this index of urbanization could be an important factor for the stock of talent. Jacobs (1961, 1969) illustrates the role of cities’ scale and diversity in the generation of new ideas, enforcing the important role of knowledge, culture, and communications in stimulating regional growth.

4.3.3. Technology variables The technological environment is a very important factor that impacts the distribution of talent. Generally speaking, regions with high creativity and levels of technological activities are more likely to attract a cluster of high-tech industries, where talent is more likely to move in. In China, the high-tech industries are officially defined as the electronic and telecommunications, computers and office equipment, pharmaceuticals, medical equipment and meters, and aircraft and spacecraft industries. Only 4.6% of the value-added by the Chinese high-tech industries is used on R&D expenditure, which is much lower than those in most developed countries. Data on the value-added for the high-tech industry is available from the China Statistical Yearbook (2004–2008). To better evaluate regional technology innovation, we assess the patents per capita that were officially approved in 2004 as a supplementary measure. In China, there are three types of patents granted: inventions, utility models and designs. Innovation can be measured either from the input side, such as R&D expenditures, or from the output side, in the form of patents. The output side is more reliable because high inputs do not necessarily lead to

March 5, 2013

84

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

higher output. The data for patents per capita is available from the China Statistical Yearbook (2004–2008).

4.3.4. Tolerance and related variables 4.3.4.1. Tolerance According to Florida et al. (2008b), tolerance is measured by combining the concentration of gay and lesbian households with the concentration of individuals employed in arts, design and related occupations. Unfortunately, statistical data on these concentrations are not available for China. Alternatively, we adopt the “Hukou index” as a proxy to represent regional openness or tolerance culture. The rules of Hukou (or the inhabitant registration system) are used by the central government to control internal migration. The system determines whether a person has rural or urban status. Those with a locally registered Hukou are permanent residents and receive local social, economic, and political benefits, such as education, social welfare, and voting rights. Those who live in a jurisdictional area without a local Hukou are “marginal” workers or visitors. The Hukou index of openness is defined as the proportion of the population without a locally registered Hukou. The higher the Hukou index, the more open a region is. The data is sourced from the China Statistical Yearbook (2004–2008).

4.3.4.2. Universities This variable is measured by the number of students who are in university in a region. University students are generally reluctant to seek a job in other places after graduation due to a well-established local network, inertia and the advantages in accessing local job information. In addition, institutional barriers (such as Chinese Hukou policy) further prevent the university graduates from moving elsewhere. Regions with more universities possess potential advantages in attracting talent, provided that they can provide enough jobs to retain these graduates. This data is sourced from the China Statistical Yearbook (2004–2008).

4.3.4.3. Culture industry and amenity We use employment in the culture industry to measure regional diversity. Regions that can offer a greater array of diverse services have higher

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

85

numbers of the creative class (Florida, 2002b). We also use employment in service industries that directly affect the convenience of living, which is a wellbeing measurement of the amenity level, whether in city or countryside. The service industries include the hotel and restaurant, environment and public-facility management, resident services, sanitation, social security, social welfare, culture, sports, and entertainment industries.

4.3.5. Ecology index Nowadays, few studies on the creative class make assessments of the weather and how environmental factors impact on the creative class distribution. As China’s economy grows, residents start to experience a higher quality of life, and more and more people want to live in elegant and aesthetic areas. However, this assumption is still debated both in academia and practice. We will test this debate through the use of empirical data. The amount of urban environmental infrastructure construction investment is used to measure this influence of this factor. Moreover, as it is assumed that more and more people pay attention to the quality of their life, we use the air quality index to denote another environmental factor. This is measured by the days of air quality that score two or more in the index. The data for this ecology index is taken from the China Environment StatisticalYearbook (2004–2008). These variables cover the 30 Chinese provinces (except Tibet and Taiwan as data is unavailable) from 2003 to 2007. Table 4.1 presents the definition, data sources and abbreviation for the previously discussed variables.

4.4. The Model Correlation and regression analysis are utilized to examine the relationships between the variables in the model. To avoid the problem of multico-linearity, we use the ridge regression model for factor analysis of the creative class distribution in this research. Ridge Regression is a variant of the ordinary Multiple Linear Regression, with the goal of circumventing the problem of the predictor’s co-linearity that can be found in the Least Squares (LS) method. Although it makes the parameters of the new model somewhat biased (whereas the parameters as calculated by the LS method

March 5, 2013

86

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

Table 4.1. Descriptive Analysis of All Variables. Dimension

Variable

Explanation

Dependent variable

CC

The number of Creative class/population

China Statistical Yearbook (2004–2008)

Economic

GDP

GDP per capita (the log of GDP per capita)

China Statistical Yearbook (2004–2008)

WAG

Average wage (the log of average wage)

China Statistical Yearbook (2004–2008)

CIT

Urban population/ population

China Statistical Yearbook (2004–2008)

HTI

High-tech value-added

Statistical Yearbook of China high-tech industry (2004–2008)

COL

Total patents

Statistical Yearbook of China high-tech industry (2004–2008)

HEC

University (university students/population)

China Statistical Yearbook (2004–2008)

PSC

Service amenities (employment in five service industries)

China Statistical Yearbook (2004–2008)

UCO

Cultural index (Number of cultural population)

China Statistical Yearbook (2004–2008)

OPE

Openness (non-Hukou residents/population)

China Statistical Yearbook (2004–2008)

Technology

Institutional and cultural

Data Source

(Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

87

Table 4.1. (Continued) Dimension

Variable

Explanation

Data Source

Ecological

EFL

Amount of Urban environmental infrastructure construction investment

China Environment Statistical Yearbook (2004–2008)

AIR

Days of Air quality equal to or above Grade 2

China Environment Statistical Yearbook (2004–2008)

are unbiased estimators of the true parameters), the variances of these new parameters are smaller than that of the LS parameters. They are so much smaller that their Mean Square Errors (MSE) may also be smaller than that of the parameters of the LS model. This is an illustration of the fact that a biased estimator may outperform an unbiased estimator, provided its variance is small enough. Moreover, the prediction errors of the Ridge Model will also turn out to be more accurate than that of the LS regression model when predictors exhibit near co-linearity (Hoerl et al., 1971; Marquardt, 1970).

4.4.1. Correlation analysis From the relationships depicted above, we can construct the following model to analyze the correlation between the variables: CC = c + β1 GDP + β2WAG + β3 CIT + β4 HTI + β5 COL + β6 HEC + β7 PSC + β8 UCO + β9 OPE + β10 EFL + β11AIR + ε.

(4.2)

The variable CC denotes the creative class distribution which is measured by the number of creative class divided by population; GDP, WAG, CIT denote economic indicators; HTI, COL denote Technology indicators; HEC, PSC, UCO, OPE denote institutional and cultural factors; EFL, AIR denote ecology indicators (all variables are in Table 4.1); c denotes intercept; ε denotes random error term; βi (i = 1, 2, . . . 11) denotes coefficients between the influence factors and CC.

March 5, 2013

88

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China Ridge Trace

0.7

HTI 0.6

COL HEC PSC

RR Coefficients

0.5

UCO

0.4

EFL

0.3 0.2 0.1 0 -0.1 0

0.1

0.2

0.3

0.4

0.5

0.6

0.7

0.8

0.9

1

K

Figure 4.3.

Ridge Trace of Six Variables Ridge Regression.

Before conducting regression analysis, we first test the correlation between the independent variables and the dependent variable, from which the valid independents were able to be extracted. The result of the correlation analysis is seen in Figure 4.3 (by use of SPSS 11.5). From Table 4.2, it can be seen that six variables are significant at the 0.01 level (HTI, COL, HEC, PSC, UCO, and EFL). These variables are used as the independent variables for the ridge regression analysis. According to the results, we can see that creative class distribution is strongly correlated with institutional and cultural factors such as university, service amenities and cultural indicators, whose coefficients are 0.87, 0.92, and 0.92, respectively. These results are consistent with many studies (Florida, 2002b, 2002c, 2006; Florida et al., 2008a, 2008b, 2007; Qian, 2008), showing that talent takes service amenities into account when choosing localities for living and working. It also proves that cultural elements play an important role in influencing the distribution of the creative class, as Florida elaborates in his book The Rise of the Creative Class. What surprises us is that the openness index is only just significant at 0.05 (with a coefficient of 0.17). After careful analysis, we concluded that one reason this may be is that our sample is so

March 5, 2013 11:55

1

Sig. (two-tailed) N

CIT

HTI

COL

HEC

PSC

UCO

OPE

EFL

AIR

0.114∗ 0.153 0.020 0.648∗∗ 0.706∗∗ 0.869∗∗ 0.924∗∗ 0.918∗∗ 0.174∗ 0.730∗∗ −0.078∗ 0.011

150

WAG

150

0.137 0.810 0.000 150

150

150

0.000

0.000

0.000

0.000

0.033

0.000

150

150

150

150

150

150

Note: ∗∗ indicates significance at the 0.01 level; ∗ indicates significance at the 0.05 level.

0.344 150

89

Regional Development and Economic Growth. . . b1491-ch04

Pearson Correlation

GDP

9in x 6in

CC

Regional Distribution of the Creative Class and Its Determinants

Table 4.2. Correlation Between Creative Class and Other Variables.

March 5, 2013

90

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

small that it cannot reflect all individuals. In fact, non-Hukou residents only account for a small portion of the population in China small portion of t as a result provincial data will ignore the role of HUKOU factors. Another reason may be that many cities are more and more open towards talent with special skills or high educated, in turn making HUKOU factors weaker. The technology index is positive and highly significant (at the 0.01 level) with a coefficient of 0.65, which is less than that found for culture. Environmental infrastructure investment is also positive and highly significant with a coefficient of 0.73. However, neither the average wage nor the city index appears highly significant to the stock of talent and the GDP index is only significant at the level of 0.05. The results for the economic index reflect that the creative class pays more attention to their human environment than the business environment. The air quality index has an inverse relationship with the distribution of the creative class at a significance of 0.05. We propose that perhaps it is not that Chinese talent does not pay attention to air quality, but rather that most of them live in cities where the air is seriously polluted.

4.4.2. Regression analysis Based on correlation analysis, we use a regression model to test the six variables’ importance in the distribution of creative class. SPSS11.5 is carried out through multiple linear regressions. CC = 3.633 − 0.012 HTI + 0.426 COL + 0.030 HEC + 0.089 PSC + 2.574 UCO + 0.437 EFL.

(4.3)

Although the R2 reaches 96.4%, the tolerances of the PSC and UCO variables are 0.031 and 0.025 respectively (all less than 0.1), and the variance inflation factors (VIF) are 32.451 and 39.643, respectively (all greater than 10), which means that there is serious multi-co-linearity in the model. Variable reduction method can certainly make the model undergo the test, but if we simply remove only some variables it may weaken the explanatory power of the model. Therefore, the ridge regression method is introduced in our statistical analysis. Ridge regression analysis is an improved least square regression method. The main principle of the ridge regression is that a parameter K(>0) will

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

91

be artificially added in the information matrix along the main diagonal. As a result, the Eigen value will be slightly larger in order to improve the stability of regression coefficient estimate. Thus multi-co-linearity amongst variables will be minimized. In this chapter, this method is used to estimate the parameters of the variables by programming. Regression coefficients of the different Ridge parameters are shown in Table 4.2. Figure 4.2 also shows the ridge trace. When the ridge parameter K reaches 0.4 or more, the variable coefficients estimated become stable and six curves of ridge traces are accordingly getting smoother as seen in Table 4.3 and Figure 4.3 (Golub et al., 1979). Therefore we select 0.4 as the ridge parameter K, and re-estimate the variables traces in Equation (4.4). According to Table 4.3, six independent Table 4.3. Ridge Parameters of Different Criteria for Each Variable Regression Coefficient. K 0.00 0.05 0.10 0.15 0.20 0.25 0.30 0.35 0.40 0.45 0.50 0.55 0.60 0.65 0.70 0.75 0.80 0.85 0.90 0.95 1.00

R2

HTI

0.96537 −0.04991 0.96221 0.00436 0.95890 0.02968 0.95611 0.04483 0.95355 0.05502 0.95108 0.06235 0.94861 0.06787 0.94612 0.07215 0.94357 0.07555 0.94096 0.07829 0.93829 0.08052 0.93555 0.08235 0.93275 0.08385 0.92988 0.08510 0.92695 0.08612 0.92396 0.08696 0.92093 0.08764 0.91784 0.08819 0.91470 0.08863 0.91153 0.08897 0.90831 0.08922

COL

HEC

PSC

UCO

EFL

0.32549 0.24381 0.20948 0.19006 0.17743 0.16847 0.16174 0.15646 0.15216 0.14857 0.14549 0.14281 0.14042 0.13827 0.13631 0.13451 0.13283 0.13126 0.12977 0.12836 0.12702

0.08216 0.13075 0.15033 0.16080 0.16698 0.17075 0.17303 0.17431 0.17491 0.17502 0.17477 0.17425 0.17353 0.17267 0.17168 0.17061 0.16948 0.16829 0.16707 0.16582 0.16455

0.04557 0.16632 0.20142 0.21513 0.22077 0.22269 0.22271 0.22167 0.22002 0.21802 0.21580 0.21348 0.21109 0.20869 0.20629 0.20390 0.20155 0.19924 0.19697 0.19474 0.19256

0.60054 0.44991 0.39146 0.35819 0.33561 0.31869 0.30520 0.29398 0.28437 0.27595 0.26846 0.26170 0.25554 0.24988 0.24464 0.23975 0.23518 0.23088 0.22682 0.22298 0.21933

0.14110 0.12440 0.12067 0.12002 0.12032 0.12090 0.12151 0.12204 0.12246 0.12275 0.12293 0.12300 0.12297 0.12284 0.12264 0.12237 0.12204 0.12166 0.12123 0.12076 0.12025

March 5, 2013

92

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

variables can explain variability of creative class of 94.3%, and all variables significantly contribute to the dependent at 0.01 (t value >2). CC = 0.075 HTI + 0.152 COL + 0.174 HEC + 0.220 PSC t = (5.7737)

(11.7940)

(12.9259)

(18.4279)

+ 0.284 UCO + 0.122 EFL (23.9399)

(4.4)

(8.3346)

R2 = 0.943 All variables is significant at 0.01.

4.4.3. The findings The regression results show that the high-tech index, university, service amenity, culture index and urban environmental index all contribute significantly to the distribution of creative class.

4.4.3.1. Service amenity and culture index Regression results show that each standard unit increase of the two indicators will result in an increase in the creative class distribution by 0.22 and 0.28 standard units respectively. Service amenity and culture index represent the region’s convenience and diversity. We argue that diversity within a region creates a more open-minded and tolerant cultural, social, and economic milieu, thus affecting both the geographic distribution of talent and the outcomes of regional development. Our findings here reinforce Florida’s (2002b, 2002c) claim that the creative class likes to concentrate in diversified and tolerant regions.

4.4.3.2. University index University presents a strong association with talent distribution, with a coefficient of 0.174. This is consistent with the results that have taken place in the Western context as shown by Berry and Glaeser (1968) and Mellander and Florida (2007). In China’s context, universities are producing a large number of talent possessing higher level of degrees and creativity. Although some institutional barriers such as the Hukou system restrict labor flow among regions, local universities play an important role in producing, attracting and reserving talent. Moreover, it is not uncommon that local employers

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

93

prefer local college graduates, even if talent from other localities may be more qualified.

4.4.3.3. Technology index The patent index (COL) is significantly positive in correlation with the distribution of talent (with a coefficient of 0.15) and its effect is lower than other variables (except HTI). Three types of patents in China are granted: inventions, utility models, and designs. In fact, only inventions may have more important an effect on the distribution of talent. At any rate, talent is distinguished based on its knowledge or ability to generate knowledge. The high-tech index, as another technology index, is considered as a measure of innovation capacity that is significantly associated with talent distribution. The high coefficient means denser high-tech industries and more innovative milieu, which attract the creative class to get jobs. However, its coefficient is only 0.08, which is the lowest value among all variables.

4.4.3.4. Urban environmental index The urban environmental index is significantly positively correlated with the distribution of talent, with coefficient of 0.12. This reflects the preference of the creative class to pay more attention to ecological factors. With promotion and development of the Chinese Government’s “Resourcesaving and Environment-friendly Society” strategy, people have become more and more concerned about environmental quality. Consequently, if a region can provide a higher urban environmental infrastructure construction investment, it will be able to attract and retain more talent.

4.5. Conclusion and Discussion According to our research, the geographic distribution of talent in China shows both similarities and differences in comparison to Florida’s creative class theory. Traditional economic factors, such as GDP, wage and urbanization, do not significantly affect the regional distribution of China’s creative class, which is greatly influenced by technological, human, and environmental factors. With our Chinese dataset we find that the single most important contributor to the distribution of talent is the presence of service amenity and the cultural index. Amenities operate on the consumption side

March 5, 2013

94

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

to attract talent to a region. Our results consolidate Glaeser’s amenities effect, rather than Florida’s openness effect. When we observe the different regional conditions of China, open culture is seen as lowering barriers to entry for residents into different demographic groups. The open culture can facilitate talent flowing inward and outward, which is considered to be able to increase the possibility of a region generating a pool of talent. However, we find that the openness index is just significant at 0.01 levels and thus we further explore this issue. The patent index and high-tech index are the significant factors in attracting talent in China; however its coefficient is lower than that found for amenity. It reflects China’s c tendency to be a knowledge-based or innovative economy, where the creative class is more likely to fulfill their self-realization value and ambition. In turn, talent is expected to play a central role in industrial innovation, entrepreneurial activity and social development. Non-market factors such as the university and urban environmental index also present significant and positive correlations with the creative class distribution. Consequently, we suggest that these factors do not operate in competition with one another, but rather tend to attract or affect different types of talent. They are thus said to play complementary roles in the geographic distribution of talent. We hope our findings can stimulate the recently increased debate and discussion about the nature and the way of creating and retaining Chinese talent, and that they can provide a new insight for prospecting further into Chinese transformation due to economic growth.

References Asheim, B. and Hansen, H., “Knowledge Bases, Talents, and Contexts: On the Usefulness of the Creative Class Approach in Sweden,” Economic Geography, 85: 425–442 (2009). Boschma, R. A. and Fritsch, M., “Creative Class and Regional Growth: Empirical Evidence from Eight European Countries,” Economic Geography, 85: 391–423 (2009). Clark, T. N., “The City as an Entertainment Machine,” Research in Urban Policy, 9: 103–140 (2003). Florida, R., The Rise of the Creative Class. New York: Basic Books (2002a).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Distribution of the Creative Class and Its Determinants

95

Florida, R., “The Economic Geography of Talent,” Annals of the Association of American Geographers, 92(4): 743–755 (2002b). Florida, R., “Bohemia and Economic Geography,” Journal of Economic Geography, 2: 55–71 (2002c). Florida, R., Cities and the Creative Class. New York: Routledge (2004). Florida, R., Gates, G., Knudsen, B. and Stolarick, K., “The University and the Creative Economy,” Working Paper (2006). http://www.creativeclass.org/rfcgdb/ articles/University_andthe_Creative_Economy.pdf (accessed on 3 July 2007). Florida, R., Mellander, C. and Qian, H., “Creative China the University, Tolerance, Talent in Chinese Regional Development,” Working Paper (2008a). http:// creativeclass.com/rfcgdb/articles/creative%20china.pdf. Florida, R., Mellander, C. and Tolarick, K., “Inside the Black Box of Regional Development Human Capital, the Creative Class and Tolerance,” Journal of Economic Geography, 8: 615–649 (2008b). Glaeser, E. L., “Cities, Information, and Economic Growth,” Cityscape, 1(1): 9–47 (1994). Glaeser, E. L., Kolko, J. and Saiz, A., “Consumer City,” Journal of Economic Geography, 1: 27–50 (2001). Glaeser, E. and Kohlhase, J., “Cities, Regions and the Decline of Transport Costs,” Paper in Regional Science, 83: 197–228 (2004). Golub, G. H., Heath, M. and Wahba, G., “Generalized Cross-Validation as a Method for Choosing a Good Ridge Parameter,” Tehnometrics, 21: 215–223 (1979). Hoerl, A. E. and Kennard, R. W., “Ridge Regression: Biased Estimates for Nonorthogonal Problems,” Technometrics, 12: 55–67 (1971). Hong, J., Yu, W. T. and Zhao, D. T., “Spatial Agglomeration of Creative Class and Differences of Regional Labour Productivity: Analysis Based on China’s Provincial Panel Data,” Journal of Finance and Economics, 37(7): 92–102 (2011). Jacobs, J., The Death and Life of Great American Cities. NewYork: Vintage Books (1961). Jacobs, J., The Economy of Cities. New York: Random House (1969). Kotkin, J., The New Geography: How the Digital Revolution is Reshaping the American Landscape. New York: Random House (2000). Li, T. and Florida, R., “Talent, Technological Innovation, and Economic Growth in China,” Working Paper (2006). http://www.creativeclass.org/rfcgdb/articles/ China%20report.pdf.

March 5, 2013

96

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch04

Regional Development and Economic Growth in China

Markusen, A., “Targeting Occupations in Regional and Community Economic Development,” Journal of the American Planning Association, 70(3): 253–268 (2004). Marlet, G. and Van Woerkens, C., “Skills and Creativity in a Cross-section of Dutch Cities,” Discussion Paper Series 04-29, Tjalling C. Koopmans Research Institute (2004). Marquardt, D. W., “Generalized Inverses, Ridge Regression, Biased Linear Estimation, and Nonlinear Estimation,” Technometrics, 12: 591–612 (1970). Mellander, C. and Florida, R., “The Creative Class or Human Capital? Explaining Regional Development in Sweden,” Working Paper (2007). http://www. creativeclass.com/rfcgdb/articles/The_Creative_Class_or_Human_Capital.pdf. Ottaviano, G. I. P. and Peri, G., “Cities and Culture,” Journal of Urban Economics, 58: 304–337 (2005). Qian, H., “Creativity and Regional Economic Performance: The Case of China,” The Annals of Regional Science, 45(1): 133–156 (2008). Quigley, J., “Urban Diversity and Economic Growth,” The Journal of Economic Perspectives, 12: 127–138 (1998). Saxenian, A., “Silicon Valley’s New Immigrant Entrepreneurs,” Working Paper, Public Policy Institute of California, Berkeley (1999). Shapiro, J. M., “Smart Cities: Quality of Life, Productivity, and the Growth Effects of Human Capital,” The Review of Economics and Statistics, 88(2): 324–335 (2006).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Chapter 5

Comparing Productivity Growth among the Regions Yanrui Wu

5.1. Introduction China has achieved impressive economic growth since the late 1970s when the economic reform program was implemented. This growth has lifted the standard of living for many Chinese citizens. However, the benefit of economic growth is not evenly distributed amongst the Chinese regions. Coastal China has become relatively more affluent than the interior areas. It was partly due to this imbalanced development that the Western development or “go-West” program was initiated by the Chinese government in 1999. The objective of this study is twofold. First this study attempts to provide an assessment of the economic performance of China’s Western regions in comparison with the rest of China. Second, this chapter sheds light on the question about whether China’s high growth is sustainable after three decades of uninterrupted growth. To answer this question, one has to explore the role of total factor productivity (TFP) in economic growth. To achieve these goals, the rest of the chapter begins with a brief review of productivity studies (Section 5.2). This is followed by descriptions of the analytical framework and data issues (Section 5.3). Empirical results using sector level statistics are then reported in Section 5.4. The final section (Section 5.5) concludes the chapter.

97

March 5, 2013

98

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

5.2. Productivity and Economic Growth: A Review The role of productivity in economic growth has been controversial and fiercely debated among economists.1 Research on this topic was boosted by the inquiries into the productivity slowdown in the USA in the 1960s and 1970s, in particular, relative to Japan.2 It is argued that the relative productivity slowdown in the USA and European countries may be due to the natural process of convergence as countries with a low level of productivity catch up to those with a high level (Abramovitz, 1986; Baumol, 1986). Though still unresolved, the debate reemerged in the 1990s with a focus on the East Asian economies. Krugman (1994) together withYoung (1994) and Kim and Lau (1994) raised serious doubts about the role of technological progress in the East Asian growth model and thus the sustainability of growth in those economies. As expected, their views have been questioned by other authors (Kawai, 1994; Oshima, 1995; Sarel, 1995). The Chinese economy is no exception. The role of productivity in growth at the macro as well as the micro level has been investigated intensively though a consensus is hardly reached.3 Earlier studies of the agricultural sector have focused especially on the impact of rural reforms on agricultural productivity and hence growth (McMillan et al., 1989; Lin, 1992; Fan, 1991; Zhang and Carter, 1997; Fan and Zhang, 2002).4 One common finding in these studies is that economic reforms at the early stage had a significant impact in stimulating productivity growth in Chinese agriculture. However, conclusions with regard to other areas are more controversial. While Borensztein and Ostry (1996) and Hu and Khan (1997) are more positive about the contribution of TFP to China’s growth, others only found a minor role played by TFP growth (Woo, 1998). At the firm level, authors are more pessimistic partly because of the dominance of state-owned enterprises (SOEs) in the 1980s and 1990s (Chen et al., 1988; Jefferson et al., 1996; Woo et al., 1994). 1 For a review, see Griliches (1994). 2 See, for example, Baumol (1986), De Long (1988), Dowrick and Nguyen (1989), and Wolff (1991, 1996). 3 For literature reviews, see Wu (1993, 2011), Wu and Yang (1999), and Fan and Zhang (2006). 4 There are also many studies which are related to the pre-reform period and hence ignored. Examples include Tang (1980), Wiens (1982), and Perkins and Yusuf (1984).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

99

Wu (2011) reviewed over 70 papers and found that the average estimated rate of TFP growth is 3.62%. There is however substantial variation among the studies. The objective of this study is to present a comprehensive study of the three economic sectors and the aggregate economy. For the first time, capital stock data are estimated for the three sectors, that is, agriculture, manufacturing and services. The derived database is then employed to examine the contribution of TFP to growth in the Chinese regions as well as among the three economic sectors during the period of 1978–2005.

5.3. Analytical Framework and Data Description 5.3.1. Analytical framework A variety of techniques have been developed to estimate productivity growth. There are advantages and disadvantages associated with each approach.5 This chapter employs a well-developed method which falls into the family of stochastic frontier analysis (SFA). According to the latter, TFP growth can be decomposed into two components, that is, technological ◦

◦

progress (TP) and technical efficiency change (TE). That is, ◦

◦

◦

TFP = TP + TE.

(5.1)

The computation of Equation (5.1) involves the estimation of production functions. Symbolically, assume that the production technology in logarithmic form can be modeled as follows: ln yit = αi + ln f(xit ; β) + uit ,

(5.2)

where yit and xit represent outputs and inputs, α is the individual effect, β is a vector of parameters to be estimated and uit is white noise. Given Equation (5.2), technical efficiency (TE), which is defined as the ratio of the observed output over the best practice output, can be derived using the following procedure: TEit = eεˆ it −ˆεt ,

(5.3)

where εˆ it = αˆ i + uˆ it and εˆ t = max(uˆ it ). 5 For a recent survey, see Kumbhakar and Lovell (2000), Coelli et al. (2005), and Greene

(2008).

March 5, 2013

100

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Furthermore, technological progress and technical efficiency change can be estimated using the following equations: ◦

TP = ∂y/∂t = ∂f/∂t = ft , ◦

TEit = TEit /TEi,t−1 − 1.

(5.4) (5.5)

5.3.2. Data issues To implement the procedure of estimating Equations (5.1) to (5.5), it is assumed that capital and labor are employed to produce one output, i.e., value-added. Production occurs among the regions and across the three economic sectors (agriculture, manufacturing, and services). In the agricultural sector, land as a production factor is also taken into consideration. As a result, data of these variables (output, labor, capital stock, and land) are required for the empirical exercises. In this study, output takes the values of real GDP at the regional level as well as in the three sectors. Labor is the total employment in the three sectors and among the 31 administrative regions in China. The main challenging task is to estimate capital stock for the Chinese regions and the three economic sectors. There are some estimates of capital stock at the national and regional levels (Wu, 2008a; Zhang, 2008). Hence the focus of this exercise is the construction of sectoral capital stock data which are obtained, for the first time, in this study. The main breakthrough here is the derivation and use of the rates of depreciation for each region as well as each sector (see the appendix for details). The latter are reported in Table 5.1. In general, the rate of depreciation is lowest in the agricultural sector and highest in the manufacturing sector. The weighted average rates for each region are consistent with those obtained in Wu (2008a).

5.4. Evidence at the Sector Level To examine productivity performance at the sector level, Equation (5.2) is estimated using data for the periods of 1978–1989 and 1990–2005, respectively. The point of division (in 1990) is determined for two reasons. First, there was a major revision of employment data in 1990 (see Figure 5.1). To ensure consistency, two datasets are constructed with each being internally consistent. Second, economic reforms and hence growth have

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

101

Table 5.1. Rates of Depreciation in the Chinese Economy.

Regions Beijing Tianjin Hebei Shanxi I-Mongolia Liaoning Jilin Heilongjiang Shanghai Jiangsu Zhejiang Anhui Fujian Jiangxi Shandong Henan Hubei Hunan Guangdong Guangxi Hainan Chongqing Sichuan Guizhou Yunnan Tibet Shaanxi Gansu Qinghai Ningxia Xinjiang Mean

Agriculture Manufacturing Services 1.4 1.0 1.6 1.2 1.6 1.6 1.6 1.6 0.6 2.3 2.3 1.6 1.6 1.6 2.7 1.6 1.6 1.6 2.3 2.5 1.6 1.5 1.5 1.3 0.8 0.6 1.8 1.8 0.6 1.8 1.9 1.6

Source: Author’s own work.

5.7 5.7 6.1 6.1 5.0 7.0 7.0 7.0 4.8 4.2 5.3 6.1 6.4 6.1 7.0 6.1 4.7 5.8 7.0 3.7 2.3 7.0 7.0 4.6 3.5 2.6 3.7 3.8 2.6 3.2 3.0 5.2

3.2 3.1 3.5 3.6 6.1 6.3 6.3 6.3 2.7 5.5 3.5 3.5 3.5 3.5 4.1 3.5 5.2 5.2 5.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.5 3.2 3.5 3.2 2.7 4.0

GDP Weighted Wu Average (2008b) 4.0 4.3 4.5 4.7 4.6 6.1 5.7 6.2 3.6 4.5 4.3 4.2 4.6 4.1 5.4 4.3 4.5 4.7 6.0 3.3 2.5 4.6 4.5 3.4 2.9 2.6 3.4 3.2 2.7 3.0 2.7 4.2

3.4 3.7 4.3 4.0 4.3 5.8 5.1 6.0 3.4 4.2 4.0 5.0 4.5 3.7 5.0 4.1 4.5 4.5 6.9 3.3 2.2 5.0 4.6 2.8 2.7 4.2 3.3 2.7 2.4 2.8 2.6 4.0

March 5, 2013

102

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China 18 16 14 12 %

10 8 6 4 2

Figure 5.1.

2005

2003

2001

1999

1997

1995

1993

1991

1989

1987

1985

1983

1981

1979

0

Growth of China’s Total Employment 1979–2006.

Source: National Bureau of Statistics (various issues).

accelerated since the early 1990s. Separate estimation of the models may be able to capture the structural changes that occurred between the two periods. According to the Lagrange and Hausman tests, fixed effect models are preferred to the random effect models in most cases. In some cases, for the sake of consistency, fixed effect results are accepted though Hausman tests cannot be conducted due to singular variance–covariance matrices. The estimation results are reported in Table 5.2. Most estimated coefficients are statistically significant at the level of 1%. Given the estimates in Table 5.2, Equation (5.3) can be used to compute technical efficiency scores for the economic sectors among the regions. The rates of technological progress and efficiency changes can then be estimated using Equations (5.4) and (5.5). The discussion of those results is broadly divided into three issues, i.e., the leaders, TFP growth, and regional variations.

5.4.1. The leaders The leaders (the frontier shifters) are identified and illustrated in Figure 5.2. During 1978–2005, Shanghai, which had been the best practice performer for most years (1978–2001), was replaced by Guangdong during 2002–2005. At the sector level, several best performers are observed. Shanghai (1978–1991) and Guangdong (1993–2005) have been the frontier

March 5, 2013

Agriculture

0.022 (2.52)∗∗ 0.276 (6.73)∗∗∗ 0.577 (5.56)∗∗∗ 0.002 (1.24) 0.003 (1.80)∗

0.078 (6.95)∗∗∗ 0.353 (7.26)∗∗∗ 0.181 (3.10)∗∗∗ −0.008 (−3.69)∗∗∗ 0.008 (2.77)∗∗∗

−0.018 (−2.03)∗∗ 0.104 (2.63)∗∗∗ 0.555 (10.13)∗∗∗ 0.019 (7.53)∗∗∗ −0.007 (−2.82)∗∗∗

372 0.99 1859.67∗∗∗ 21.25∗∗∗

372 0.99 1526.93∗∗∗ 35.04∗∗∗

372 0.99 1486.79∗∗∗ 48.00∗∗∗

0.086 (5.40)∗∗∗ 0.096 (1.46) 0.141 (2.06)∗∗ −0.004 (−1.77)∗ 0.005 (1.21) 0.061 (0.38) −0.006 (−1.27) 372 0.99 1646.77∗∗∗ n.a.

0.047 (9.24)∗∗∗ 0.286 (11.02)∗∗∗ 0.317 (4.95)∗∗∗ 0.004 (5.15)∗∗∗

0.098 (15.12)∗∗∗ 0.198 (6.77)∗∗∗ −0.031 (−0.66) 0.000 (0.20)

0.106 (16.64)∗∗∗ 0.526 (13.64)∗∗∗ 0.152 (3.17)∗∗∗ −0.005 (−2.90) ∗∗∗

0.058 (8.55)∗∗∗ 0.137 (6.55)∗∗∗ −0.238 (−5.77)∗∗∗ 0.003 (2.24)∗∗ (Continued )

103

Regional Development and Economic Growth. . . b1491-ch05

Manufacturing

9in x 6in

Services

Comparing Productivity Growth among the Regions

1978–1989 Time LogK LogL Time∗ LogK Time∗ LogL LogLD Time∗ LogLD Sample size Adjusted R2 LM values Hausman 1990–2005 Time LogK LogL Time∗ LogK

Economy-wide

11:55

Table 5.2. Estimation Results.

−0.002 (−2.74)∗∗∗

−0.001 (−0.86)

0.001 (0.60)

496 0.99 2671.71∗∗∗ n.a.

496 0.99 1476.64∗∗∗ 282.25∗∗∗

496 0.99 2123.10∗∗∗ n.a.

0.007 (4.23)∗∗∗ 0.592 (7.96)∗∗∗ −0.011 (−4.87)∗∗∗ 496 0.99 3223.37∗∗∗ n.a.

Sources: Author’s own work. ∗ , ∗∗ and ∗∗∗ indicate significance at the level of 10%, 5% and 1%. n.a. implies that Hausman statistics cannot be computed due to singular variance–covariance matrix.

Regional Development and Economic Growth. . . b1491-ch05

Agriculture

9in x 6in

Manufacturing

11:55

Services

Regional Development and Economic Growth in China

Time∗ LogL LogLD Time∗ LogLD Sample size Adjusted R2 LM values Hausman

Economy-wide

March 5, 2013

104

Table 5.2. (Continued )

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

Figure 5.2.

105

China’s Best Practice Performers, 1978–2005.

Source: Author’s own work.

setters in manufacturing with the exception of 1992 when Shandong appeared to be the best performer in that year. In the service sector, Shanghai (1978–1987) and Jiangsu (1988–2005) have been the best performers. In agriculture, Shandong has been the trend-setter for most years with the exception of three years, i.e., 1979, 1982, and 1990. Jiangsu was the best performer in 1979 and 1982 while Hubei was the best in 1990. As expected, the leaders are dominantly coastal regions, in particular Shanghai, Guangdong, Jiangsu, and Shandong. The latter have also been the frontrunners of economic liberalization in China over the past decades.

5.4.2. TFP growth rates The mean TFP growth rates together with the rates of technological progress and efficiency changes are presented in Table 5.3. In the past three decades, TFP has achieved considerable growth across all sectors in China. Thus the pessimistic view about the East Asian growth model does not hold in the case of the Chinese economy. It is also found that technological

March 5, 2013

106

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Table 5.3. TFP Growth Rates. 1979–1989 Sectors Economy-wide Agriculture Manufacturing Services

1991–2005

TE

TP

TFP

TE

TP

TFP

1.5 −0.4 3.5 1.3

4.7 5.3 4.0 7.3

6.3 4.9 7.5 8.7

−0.1 0.2 −0.6 −1.5

5.7 3.0 8.7 8.1

5.6 3.1 8.1 6.6

Source: Author’s own work.

progress was the major contributor to productivity growth in the Chinese economy. This is especially so during the period of 1991–2005, a conclusion supported by others (Wu, 2008b; Zheng et al., 2008). However, TFP growth is much slower in the agricultural sector than in the manufacturing and services sectors. It has also slowed down slightly in all three sectors since the early 1990s, particularly in the agricultural sector. Though technological progress dominates productivity growth, efficiency change tends to play a more important role during 1978–1989, which is the first phase of economic reforms especially in the manufacturing sector.

5.4.3. Regional issues There is also substantial variation among China’s 31 provinces and autonomous administrative regions. However, there is evidence of convergence in terms of TFP performance as demonstrated in Figure 5.3. During 1978–1989, the source of convergence was mainly from the agricultural sector while all sectors have shown the tendency of convergence during the second period, 1990–2005, particularly during the first half of the 1990s. Geographically, China can be divided into three regions — the coastal, middle, and Western regions (Figure 5.4). The Western region includes five autonomous regions, six provinces and one autonomous municipality which are all covered by the “Western development” program.6 Variations among these regions and particularly within the Western region can also be examined by analyzing the relevant mean rates of growth. In terms of 6 This classification is slightly different from the traditional, official grouping according to

which Guangxi belongs to the coastal region while Inner Mongolia is a middle region.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

107

0.14 0.12

Agriculture

0.10

Services

0.08

Manufacturing

0.06 0.04 0.02

Economy-wide 1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

0.00

Figure 5.3.

Standard Deviation of Regional TFP Growth Rates, 1979–2005.

Source: Author’s own work.

TFP performance, the Western regions lagged behind the coastal regions in the 1990s according to Table 5.4. However, there is evidence of catch-up during the period of 2000–2005 which coincided with the implementation of the “Western development” program. In particular, the manufacturing sector in the Western regions has on the average performed better than that in the coastal regions. This change has mainly been driven by technological progress (see Tables A5.1 and A5.2 in the Appendix). Within the Western regions, Chongqing and Sichuan have been found to be the leaders and there is also evidence of convergence (Figure 5.5).7

5.5. Remarks The role of productivity in China’s economic growth has attracted a lot of attention among academia and hence has been extensively investigated using data at the aggregate, regional and industry levels. This chapter presents detailed estimates of capital stock series for the three economic sectors 7A separate estimation using the Western regional data only shows that Chongqing and

Sichuan are the leaders among the 12 provinces and autonomous administrative regions. The estimation results are available upon request.

March 5, 2013

108

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Figure 5.4.

Map of China.

Source: Author’s own work.

among China’s 31 administrative regions. In particular these estimates are for the first time based on region and sector-specific rates of depreciation. It is found that China’s growth has been driven substantially by productivity, improvement in particular technological progress. This is a good indication of sustainability of growth in the near future. However, there is considerable regional variation in performance and a different degree of convergence among the regions and across the sectors. The “Western development” program initiated in 1999 has in general played a positive role in narrowing the gap between China’s Western and coastal regions.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

Table 5.4. TFP Growth Rates across the Regions. Regions 1978–1989 Coastal Middle Western 1990–2005 Coastal Middle Western 2000–2005 Coastal Middle Western

Aggregate

Agriculture

Manufacturing

Services

7.11 6.30 6.55

5.20 4.07 5.24

8.60 8.46 8.16

8.70 9.29 9.40

6.68 5.62 5.16

3.25 2.90 2.87

6.18 6.65 6.03

7.90 6.61 6.75

6.15 5.46 5.22

3.77 3.78 3.60

7.19 7.78 7.46

8.56 7.37 7.84

Source: Author’s own work.

0.18

Agriculture

0.16 0.14 Manufacturing 0.12 0.10 0.08 0.06 0.04 0.02 Services

Aggregate

1979 1980 1981 1982 1983 1984 1985 1986 1987 1988 1989 1990 1991 1992 1993 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005

0.00

Figure 5.5.

Standard Deviation of TFP Growth Rates in the Western Regions.

Source: Author’s own work.

109

March 5, 2013

110

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Appendices to Chapter 5 The derivation of China’s regional and sectoral capital stock follows the perpetual inventory method. That is, Kti = (1 − δi )Kt−1,i + Iti ,

(A5.1)

where Kti and Iti represent the stock of capital and realized investment at period t for the ith region or sector and δi is the rate of depreciation for the ith region or sector. The Chinese government has released capital formation data at the sector level for all regions from 1978 onwards. This data series is employed as the realized investment in Equation (A5.1). Thus, a capital stock series can be generated using Equation (A5.1) if the rate of depreciation and the initial value of capital stock are known. To derive the region and sector-specific rates of depreciation, a simulation process proposed by Wu (2008a) is adopted here. According to this approach, the estimated values of depreciation (using assumed rates of depreciation) converge with the actual value of depreciation reported by the National Bureau of Statistics (various issues). The derived rates of depreciation are reported in Table 5.1. To derive the initial value of capital stock, the following formula Table A5.1 Technological Progress in China’s Three Regions. Regions 1978–1989 Coastal Middle Western 1990–2005 Coastal Middle Western 2000–2005 Coastal Middle Western

Aggregate

Agriculture

Manufacturing

Services

5.31 5.21 4.92

5.66 5.69 5.86

4.82 4.72 3.77

7.89 8.29 8.30

6.31 5.99 5.84

3.14 2.74 3.02

7.57 7.89 8.19

9.11 9.11 9.19

6.52 6.17 6.02

3.27 2.86 3.11

7.34 7.73 8.00

9.09 9.10 9.19

Source: Author’s own work.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

111

Table A5.2 Technical Efficiency Changes in China’s Three Regions. Regions 1978–1989 Coastal Middle Western 1990–2005 Coastal Middle Western 1990–2005 Coastal Middle Western

Aggregate

Agriculture

Manufacturing

Services

1.80 1.09 1.63

−0.46 −1.62 −0.63

3.78 3.75 4.39

0.81 1.00 1.10

0.37 −0.37 −0.68

0.11 0.16 −0.15

−1.39 −1.25 −2.16

−.21 −2.50 −2.44

−0.37 −0.71 −0.80

0.51 0.93 0.49

−0.15 0.05 −0.54

−0.53 −1.73 −1.34

Source: Author’s own work.

is employed K0i = I0i /(gi + δi ),

(A5.2)

where K0i and I0i represent the stock of capital and realized investment in the initial period (say, 1978 for this study) for the ith region or sector and gi is the average rate of growth in investment or output for the ith region or sector during the initial three years.

References Abramovitz, M., “Catching Up, Forging Ahead, and Falling Behind,” Journal of Economic History, 46(2): 385–406 (1986). Baumol, W. J., “Productivity Growth, Convergence, and Welfare: What the LongRun Data Show,” American Economic Review, 76(5): 1072–1085 (1986). Borensztein, E. and Ostry, J. D., “Accounting for China’s Growth Performance,” American Economic Review (Papers and Proceedings), 86: 225–228 (1996). Chen, K., Wang, H. C., Zheng, Y. X., Jefferson, G. H. and Rawski, T. G., “Productivity Change in Chinese Industry, 1953–85,” Journal of Comparative Economics, 12: 570–591 (1988).

March 5, 2013

112

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Coelli, T., Rao, D. S. P., O’Donnell, C. J. and Battese, G. E., An Introduction to Efficiency and Productivity Analysis (2nd ed.). New York: Springer (2005). De Long, J. B., “Productivity Growth, Convergence and Welfare: Comment,” American Economic Review, 78(5): 1138–1154 (1988). Dowrick, S. and Nguyen, D., “OECD Comparative Economic Growth 1950–85: Catch-up and Convergence,” American Economic Review, 79(5): 1010–1130 (1989). Fan, S., “Effects of Technological Change and Institutional Reform on Production Growth in Chinese Agriculture,” American Journal of Agricultural Economics, 73(2): 266–275 (1991). Fan, S. and Zhang, X., “Production and Productivity Growth in Chinese Agriculture: New National and Regional Measures,” Economic Development and Cultural Change, 50(4): 819–838 (2002). Fan, S. and Zhang, X., “Production and Productivity Growth in Chinese Agriculture: New National and Regional Measures,” in Dong, X.-Y., Song, S. and Zhang, X. (eds.), China’s Agricultural Development: Challenges and Prospects, Chinese Economy Series. Aldershot, U.K. and Burlington, Vt.: Ashgate (2006). Greene, W. H., “The Econometric Approach to Efficiency Analysis,” in Fried, H. O., Lovell, C. A. K. and Schmidt, S. S. (eds.), The Measurement of Productive Efficiency and Productivity Growth. USA: Oxford University Press (2008). Griliches, Z., “Productivity, R&D, and the Data Constraint,” American Economic Review, 84(1): 1–23 (1994). Hu, Z. F. and Khan, M. S., “Why is China Growing So Fast?” IMF Staff Papers, 44: 103–131 (1997). Jefferson, G. H., Rawski, T. G. and Zheng, Y., “Chinese Industrial Productivity: Trends, Measurement Issues, and Recent Developments,” Journal of Comparative Economics, 23: 146–180 (1996). Kawai, H., “International Comparative Analysis of Economic Growth: Trade Liberalisation and Productivity,” Developing Economies, 17(4): 373–397 (1994). Kim, J. I. and Lau, L., “The Sources of Economic Growth in the East Asian Newly Industrialised Countries,” Journal of the Japanese and International Economies, 8: 235–271 (1994). Krugman, P., “The Myth of Asia’s Miracle,” Foreign Affairs, 73(6): 62–78 (1994).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Comparing Productivity Growth among the Regions

113

Kumbhakar, S. C. and Lovell, C. A. K., Stochastic Frontier Analysis. Cambridge: Cambridge University Press (2000). Lin, J.Y., “Rural Reforms and Agricultural Growth in China,” American Economic Review, 82(1): 34–51 (1992). McMillan, J., Whalley, J. and Zhu, L., “The Impact of China’s Economic Reforms on Agricultural Productivity Growth,” Journal of Political Economy, 97(4): 781–807 (1989). National Bureau of Statistics (various issues), Statistical Yearbook of China. Beijing: China Statistics Press. Oshima, M., “Trends in Productivity Growth in the Economic Transition of Asia and Long-Term Prospects for the 1990s,” Asian Economic Journal, 9(21): 89–111 (1995). Perkins, D. and Yusuf, S., Rural Development in China. Baltimore: The Johns Hopkings University Press (1984). Sarel, M., “Growth in East Asia: What We Can and What We Cannot Infer from it,” in Andersen, P., Dwyer, J. and Gruen, D. (eds.), Productivity and Growth, Proceedings of a Conference. Australia: Reserve Bank of Australia (1995). Tang, C., An Economic Study of Chinese Agriculture. New York: Garland Publishing, Inc (1980). Wiens, T., “Technological Changes,” in Barker, R., Sinha, R. and Rose, B. (eds.), The Chinese Agricultural Economy. Boulder: Westview Press, Inc (1982). Wolff, E. N., “Capital Formation and Productivity Convergence Over the Long Term,” American Economic Review, 81: 565–579 (1991). Wolff, E. N., “The Productivity Slowdown: The Culprit At Last? Follow-up on Hulten and Wolff,” American Economic Review, 86: 1239–1252 (1996). Woo, W. T., “Chinese Economic Growth: Sources and Prospects,” in Fouquin, M. and Lemoine, F. (eds.), The Chinese Economy. Paris: Economica Ltd (1998). Woo, W. T., Hai, W., Jin, Y. and Fan, G., “How Successful Has Chinese Enterprise Reform Been? Pitfalls in Opposite Biases and Focus,” Journal of Comparative Economics, 18: 410–437 (1994). Wu,Y., “Productive Efficiency in Chinese Industry: A Review,” Asian-Pacific Economic Literature, 7(2): 58–66 (1993). Wu,Y., Productivity, Efficiency and Economic Growth in China. London: Palgrave Macmillan (2008a). Wu, Y., “The Role of Productivity in China’s Growth: New Estimates,” Journal of Chinese Economic and Business Studies, 6(2): 141–156 (2008b).

March 5, 2013

114

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch05

Regional Development and Economic Growth in China

Wu,Y., “Total Factor Productivity Growth in China: A Review,” Journal of Chinese Economic and Business Studies, 9(2): 111–126 (2011). Wu, Y. and Yang, H., “Productivity and Growth: A Review,” in Kalirajan, K. P. and Wu, Y. R. (eds.), Productivity and Growth in Chinese Agriculture, pp. 29–51. London: Macmillan Press (1999). Young, A., “Lessons from the East Asian NICs: A Contrarian View,” European Economic Review, 110: 641–680 (1994). Zhang, B. and Carter, C. A., “Reforms, the Weather and Productivity Growth in China’s Rural Sector,” American Journal of Agricultural Economics, 79: 1266–1277 (1997). Zhang, J., “Estimating China’s Provincial Capital Stock (1952–2004) with Applications,” Journal of Chinese Economic and Business Studies, 6(2): 177–196 (2008). Zheng, J., Wang, Z. and Shi, J., “Industrial Productivity Performance in Chinese Regions (1987–2002): A Decomposition Approach,” Journal of Chinese Economic and Business Studies, 6(2): 1 (2008).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Part II

Manufacturing Sector, FDI and Economic Growth

115

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

This page intentionally left blank

116

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Chapter 6

FDI and Economic Growth Chunlai Chen

6.1. Introduction China achieved remarkable economic growth in the last three decades with an annual average real GDP growth rate of around 10%. China’s remarkable achievement in economic growth seems to owe much to the adoption of the overall market-oriented economic reform launched three decades ago, in which actively encouraging inward foreign direct investment (FDI) was one of the most important aspects. Foreign firms have been attracted by the huge domestic market and pool of relatively well-educated, low-cost labor, which has made China one of the most attractive destinations for FDI in the world. By the end of 2010, China had attracted a total of US$1,020 billion FDI inflows (at 2000 US dollar prices), making it the largest FDI recipient in the developing world. The huge amount of FDI inflows has contributed greatly to China’s economy in terms of capital formation, employment creation, expansion of exports, and economic growth. In 2008, FDI inflows accounted for 6% of China’s gross fixed capital formation, while FDI firms produced 33% of industrial output value, employed 33% of the manufacturing labor force, and created 55% of China’s total exports. The role of FDI in economic growth has been extensively studied in the literature. In the case of China, the positive effects of inward FDI on China’s economy have been found in recent studies, including, for example, Buckley et al. (2002), Chen et al. (1995), Dees (1998), Henley et al. (1999), Kueh (1992), Lardy (1995), Pomfret (1997), Ran et al. (2007), Tuan et al. (2009), Vu et al. (2008), Wei (1996), Whalley and Xin (2010), Yao and Wei (2007), and Zhang (2006). Existing studies provide useful insights and rich

117

March 5, 2013

118

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

empirical evidence on the role of FDI in economic growth, but the exact mechanism of how FDI contributes to the growth process of a developing economy has not been well studied. To understand how FDI contributes to developing countries’ economic growth, it is necessary to compare the different roles of FDI and domestic investment in the economic growth process. Domestic investment is a necessary condition for production growth and technical progress, but it may not enable a developing economy to take advantage of advanced technologies available in the developed world. FDI is different from domestic investment in three important aspects although both can be treated as a basic physical input in the production process. First, FDI accelerates the speed of adoption of general purpose technologies in the host countries. The general purpose technologies are technological inventions that affect the entire global economy. The most recent examples of general purpose technologies include the computer, Internet, and the mobile phone. Each general purpose technology is capable of raising the aggregate productivity of labor and capital, but it takes a considerable amount of time for all countries, especially for the developing countries, to explore its potential. The developed countries tend to be “fore runners” in the adoption of general purpose technologies and their experiences are useful for the developing economies through FDI (Yao and Wei, 2007). Second, according to Dunning’s “OLI” explanation for FDI (Dunning, 1993), for multinational enterprises (MNEs) to have a strong motive to undertake direct investment aboard, they must possess certain ownership advantage. It could be a product or production process, like a patent or blueprint. It could also be specific intangible assets or capabilities, like technology and information; managerial, marketing and entrepreneurial skills; organizational systems; access to intermediate or final goods markets. The ownership advantage confers some valuable market power or cost advantage on the firm, which is sufficient enough to outweigh the disadvantages of doing business abroad. In addition, MNEs must have an advantage of internalizing business activities, and at the same time, the foreign market must offer a location advantage that makes it profitable to produce the product in the foreign country rather than simply produce it at home and export it to the foreign market. Therefore, FDI is embedded with new technologies, know-how, management skills and other intangible proprietary assets and

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

119

information unavailable in the host countries. Such advanced technologies will be able to shift the host country’s production frontier to a new level so that the same amount of material inputs can lead to a higher level of output. Third, the ability of MNEs to combine the three advantages implies that they should be able to out-perform indigenous firms in production. However, through knowledge spillovers such as imitation (reverse engineering), human resource movement, training courses, vertical industrial linkages, technical assistance, and exposure to fierce competition, FDI can generate spillovers to increase the productivity of local firms, thus shifting the production frontier of a host country. China, with fast economic growth and large amounts of FDI inflows, provides an ideal case for an empirical analysis of the impact of FDI on economic growth. Therefore, in this chapter we examine empirically the impact of FDI on China’s economic growth. We first identify the possible channels through which FDI may affect China’s economic growth. Then, using a panel dataset containing China’s 30 provinces over the period 1987 to 2005, we estimate an augmented growth model in which direct effects (e.g., raising output and productivity through capital augmentation and technological progress) and spillover effects (e.g., increasing productivity through diffusing technology and management skills to local economy) of FDI on China’s economic growth are analyzed. The structure of the chapter is as follows. Section 6.2 presents an overview of the contributions of FDI to China’s economy in terms of capital formation, employment creation, and export promotion. Section 6.3 discusses the literature on the impact of FDI on China’s economic growth and raises the issues to be examined in this study. Section 6.4 sets out the framework of analysis and specifies the empirical model. Section 6.5 describes the variables and the data. Section 6.6 discusses the regression results. Finally, Section 6.7 provides the conclusion with policy implications.

6.2. The Contributions of FDI to China’s Economy In the literature, FDI is believed to have played some major roles in the development process of a host country’s economy, via capital formation, the creation of employment opportunities, promotion of international trade,

March 5, 2013

120

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

technology transfer and spillover effects to the domestic economy (e.g., Caves, 1996; Dunning, 1993; Markusen and Venables, 1999; UNCTAD, 1999, 2004). Over the past three decades, China has attracted huge amounts of FDI inflows and FDI firms have generated some important impacts on China’s economy.

6.2.1. Capital formation How important have FDI inflows been in China’s domestic capital formation? To answer this question we analyze the ratio of FDI inflows in China’s domestic gross fixed capital formation and the share of FDI in China’s total investment in fixed assets1 for the period 1994 to 2008.2 FDI has provided an important source of external finance to China’s economic development. As shown in Figure 6.1, FDI inflows reached 17.3% of China’s domestic gross fixed capital formation in 1994. However, since then the ratio of FDI inflows in China’s domestic gross fixed capital formation has declined, falling to 6% in 2008. The main reason for this declining trend is the much higher growth rate of China’s domestic gross fixed capital formation relative to that of FDI inflows into China. During the period 1994 to 2008, the annual growth rate of China’s domestic gross fixed capital formation was 14.7%, while that of FDI flows into China was 6.6%. The ratio of FDI inflows in China’s domestic gross fixed capital formation may overestimate the real contribution of FDI to China’s domestic capital formation. In fact, total FDI flows into China have not all been used for investment in fixed assets. There is evidence that investment in fixed assets made by FDI firms accounted for only a portion of the total FDI inflows into China each year. FDI firms’investment in fixed assets accounted for around 80% of the total FDI inflows into China in the late 1990s, and the figure for earlier years was much lower. The remainder (around 20%) 1 Gross fixed capital formation refers to the value of acquisitions minus those disposals of fixed assets during a given period. Gross capital formation equals gross fixed capital formation plus changes in inventories. Total investment in fixed assets refers to the volume of activities in construction and purchases of fixed assets and related fees, expressed in monetary terms during the reference period of the whole country. 2 China unified its dual exchange rate system in 1994. As a result, the exchange rate of RMB to US dollar depreciated sharply from 5.76 Yuan per US dollar in 1993 to 8.62 Yuan per US dollar in 1994. For consistency, data before 1994 were excluded from the analysis.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

121

20 18 16 14

(%)

12

As percentage of GFCF As percent of investment in fixed assets

10 8 6 4 2 0 1994 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Figure 6.1. FDI Inflows as a Percentage of GFCF and Percent of Investment in Fixed Assets in China, 1994–2008. Source: Calculated from National Bureau of Statistics of China (various issues (b)). Note: GFCF = gross fixed capital formation.

of FDI inflows may have been used by FDI firms as working capital and for inventory investment (Chen, 2002). To evaluate the contribution of FDI to China’s domestic capital formation, we use the share of FDI in China’s total investment in fixed assets. As shown in Figure 6.1, the share reached a peak of 9% in 1996. Since then it fell to around 3.5% or less after 2000. The above analysis suggests that FDI made an important contribution to China’s domestic capital formation during the 1990s. However, since 2000, the role of FDI in China’s domestic capital formation has been declining. Nevertheless, for a large and fast growing economy like China, with average annual GDP growth around 10% for the last three decades, FDI has provided an important supplementary source of finance to its domestic capital formation.

6.2.2. Creation of employment opportunities In developing countries, where capital is relatively scarce but labor is abundant, one of the most prominent contributions of FDI to the local

March 5, 2013

122

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

economy is the creation of employment opportunities. In general, FDI has direct and indirect employment effects in a host country. The direct employment refers to the total number of people employed within the FDI firms. The indirect employment effects refer to the employment opportunities indirectly generated by FDI firms’ activities in the host country. The indirect employment effects are difficult to measure, but country case studies conducted by the International Labor Organization (ILO) show that the indirect employment effects associated with inward direct investment may be as, if not more, important than the direct effects (Dunning, 1993). Because of the difficulties in measuring the indirect employment effects of FDI, we confine our analysis within the scope of the direct employment effect of FDI in China’s manufacturing sector. Figure 6.2 shows FDI firms’ employment in the manufacturing sector during 1995 to 2008 and indicates that FDI firms’ manufacturing employment increased significantly after 2001. While they employed 6.05 million workers or 8.9% of China’s manufacturing employment in 1995, the figures had increased to 25.45 million workers or 32.97% in 2008. In other words, by the end of 2008, FDI firms employed one-third of China’s manufacturing labor force.

35

30 Share of FDI firms in total manufacturing employment

25

Manufacturing employment by FDI firms

25

20

(%)

20 15 15 10

(million persons)

30

10 5

5 0 1995

1996

1997

Figure 6.2.

1998

1999

2000

2001

2002

2003

2004

2005

2006

2007

0 2008

FDI Firms’ Manufacturing Employment in China, 1995–2008.

Source: Data for 1995, 1999–2001, 2003, and 2005–2008 are calculated from National Bureau of Statistics of China (various issues (b)). Data for the other years are estimated by the author.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

123

Table 6.1. FDI Firms’ International Trade Performance, 1980–2008. Value of FDI Firms’ Trade (US$ Billion)

FDI Firms’ Trade as % of China’s Total Trade

Year

FDI Inflows (US$ Billion)

Total Trade

Exports

Total Imports Trade

1980 1985 1990 1995 2000 2001 2002 2003 2004 2005 2006 2007 2008

0.44 1.66 3.49 37.52 40.72 46.88 52.74 53.51 60.63 60.31 62.97 74.77 92.40

0.04 2.36 20.12 109.82 236.71 259.10 330.24 472.17 663.18 831.64 1,036.27 1,255.16 1,409.92

0.01 0.29 7.81 46.88 119.44 133.24 169.99 240.31 338.61 444.18 536.78 695.37 790.49

0.03 2.06 12.31 62.94 117.27 125.86 160.25 231.86 324.57 387.46 472.49 559.79 619.43

Exports

Imports

0.05 1.09 12.58 31.51 47.93 50.07 52.21 54.84 57.07 58.30 58.19 57.10 55.25

0.17 4.89 23.07 47.66 52.10 51.68 54.29 56.17 57.83 58.71 59.70 58.56 54.69

0.11 3.39 17.43 39.10 49.91 50.84 53.20 55.48 57.44 58.49 58.87 57.74 55.01

Source: National Bureau of Statistics of China (various issues (a), various issues (b)).

6.2.3. Contribution of FDI to China’s export expansion There is considerable evidence that FDI contributes to the growth of host countries’ exports. In the case of China, the most prominent contribution of FDI perhaps is expanding China’s exports. The direct way to measure the impact of FDI on China’s export growth is to examine the trade performance of FDI firms. Table 6.1 presents FDI inflows into China and FDI firms’export performance from 1980 to 2008. With the rapid increase in FDI inflows, FDI firms’ exports increased spectacularly from US$0.01 billion in 1980 to US$119.44 billion in 2000 and further to US$790.49 billion in 2008, with an annual growth rate of 47.82%. As a result, the importance of FDI firms in China’s exports has increased dramatically from only 0.05% in 1980 to 47.93% in 2000 and further to 58.30% in 2005, before falling slightly to 55.25% in 2008. The rapid increase in FDI inflows and their export performance partly reflects China’s policy in relation to FDI attraction deliberately biased toward export-oriented FDI. As a result, FDI firms have rapidly become a major exporter.

March 5, 2013

124

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

FDI firms have played an even more important role in China’s manufacturing exports. According to the National Annual Enterprise Census conducted by National Bureau of Statistics of China (NBS), during 2000 to 2003, on average, FDI firms’ export propensity (exports to sales ratio) was 42%, while that of domestic firms was only 10%. Among the 29 manufacturing industries, in 10 industries FDI firms’ export propensities exceeded 50%, including, for example, cultural, educational and sports goods (82%); leather and fur products (73%); furniture (71%); other manufacturing (71%); and instruments and meters (70%). In 18 industries, FDI firms have dominated the industries’ exports. For example, the share of FDI firms’ exports in the industry’s total exports was 92% in electronics and telecommunication equipment; 90% in instruments and meters; 87% in printing; 78% in plastic products; 76% in furniture; 75% in paper and paper products; and 72% in cultural, educational and sports goods. Overall, FDI firms accounted for 67% of China’s total manufacturing exports. The above analysis reveals that FDI has made great contributions to China’s economy in terms of capital formation, employment creation and export expansion. However, the impact of FDI on China’s economic growth is yet to be established. Therefore, in the following sections, we will investigate the impact of FDI on China’s economic growth.

6.3. FDI and Economic Growth in China: A Literature Review There has been an increasing body of literature on the impact of FDI on China’s economic growth. For example, Chen et al. (1995) find that FDI has been positively associated with the increase of total fixed investment and economic growth in China. Wei (1996) finds empirical evidence that FDI is positively associated with cross-city differences in growth rates in China. Dees (1998) finds evidence that FDI affects China’s growth through the diffusion of knowledge and ideas. Buckley et al. (2002) find that local economic and technological conditions in a host country influence the relationship of FDI with growth and FDI favors growth in the economically stronger provinces in China. Zhang (2006) finds that FDI promotes income growth and this positive growth

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

125

effect is stronger in the coastal than in the inland regions. Yao and Wei (2007) find that FDI has positive and significant impacts on China’s economic growth both as a mover of production efficiency and as a shifter of production frontier and the positive impact of FDI on economic growth is larger in the East than in the Central and West regions. More recently, Vu et al. (2008), using sectoral FDI inflow data, evaluate the sector-specific impact of FDI on growth in China. The study finds that FDI has a statistically significant positive effect on economic growth and the effects seem to be very different across economic sectors, with most of the beneficial impact concentrated in manufacturing industries. Tuan et al. (2009) investigate the role that inward FDI plays in the process of regional development and the channels through which economic growth would be affected by using city level panel data estimations for the period since China’s economic opening and reform. While the study finds that FDI exerted spillover effects and affected total factor productivity (TFP) growth of the recipients, major technology- and knowledge-related factors, including R&D and human capital, also played critical roles in TFP enhancement and regional growth. Whalley and Xin (2010) investigate the contribution of inward FDI to China’s recent rapid economic growth using a two-stage growth accounting approach. After decomposing the Chinese economy into FDI and non-FDI sectors, the study results indicate that foreign-invested enterprises may have contributed to over 40% of China’s economic growth in 2003 and 2004, and without this inward FDI, China’s overall GDP growth rate could have been around 3.4 percentage points lower. The literature provides empirical evidence that FDI brings into the host countries a package of resources, which promote economic growth. However, most of the previous studies only investigate the direct impact of FDI as a capital input to China’s economic growth; the roles of FDI on economic growth from technological progress and spillover effects are usually overlooked. A few studies examined the spillover effects of FDI on economic growth, but failed to investigate the roles of FDI on economic growth through capital augmentation and technological progress. Therefore, this study attempts to fill the gaps by examining empirically the impact of FDI on economic growth from three channels — capital augmentation, technological progress, and spillover effects.

March 5, 2013

126

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

6.4. Framework of Analysis and the Empirical Model We estimate the impact of FDI on China’s economic growth by specifying an aggregate production function as follows: β

β

β

Yit = Ait Lit1 DK it2 FK it3 ,

(6.1)

where Yit represents the real gross domestic product (GDP) of province i in year t; Ait is the TFP level of province i in year t; Lit is the total labor input of province i in year t; DK it is the domestic capital stock of province i in year t; FK it is the foreign capital stock of province i in year t. In this specification, FDI is treated as a separate factor of capital input (FK) along with domestic capital input (DK) and labor input (L) in the aggregate production function. Theoretically, because FDI brings into the host country a package of capital, technology, production know-how, management skills, marketing skills and information, competition and so on (Dunning, 1993), it is expected that FDI can increase the host country’s economic growth by a number of means. First, through capital augmentation in a recipient economy, FDI is expected to be growth enhancing by encouraging the incorporation of new inputs and technologies into the production function, thus expanding the production frontier of a host country. This positive effect of FDI is the contribution of FDI as a capital input to output growth, which can be expressed as ∂Y/∂FK > 0, implying that the higher the foreign capital input, the higher the output growth of the host economy. Second, FDI is believed to be a leading source of technology transfer and human capital augmentation in developing countries. Technological progress takes place through a process of capital deepening in the form of the introduction of new varieties of knowledge-based capital goods. It also proceeds via specific productivity-increasing labor training and skill acquisition promoted by MNEs. Therefore, FDI is expected to shift the production frontier of a host country over time resulting from technological progress. This positive effect of FDI is the contribution of technology progress to output growth, which can be expressed as ∂Y/∂FK = f(t) > 0, implying that the marginal product of FK is an increasing function of time. Third, through the knowledge spillovers such as learning by doing or learning by watching (demonstration effects), research and development, human resource movement, training courses, vertical industrial linkages,

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

127

technical assistance, and exposure to fierce competition, FDI increases the productivity of local firms, thus shifting the production frontier of a host country. As a result, the presence of FDI can improve the productivity of domestic firms and shift the production frontier of a host country to a new level. This positive effect of FDI is the spillover effect on domestic firms, which can be expressed as ∂Y/∂SFK > 0, implying that the higher the presence of FDI, the higher the spillover effects of FDI on local economic growth. With the above propositions, the TFP Ait can be defined as the following: Ait = Bit eg(t,t∗FK it ,SFK it ,Z) ,

(6.2)

where Ait is the TFP level of province i in year t; Bit is the residual TFP level of province i in year t; t is a time trend, which captures the Hicksneutral technological progress in province i in the absence of FDI or foreign technology; t ∗ FK it captures the additional technological progress that is attributed only to FDI; SFK it is the presence of FDI in province i in year t, which captures the spillover effects of FDI; Z is a set of other variables which can also improve productivity. One such variable is human capital (HK). It has been suggested in recent growth models as a determinant of growth (e.g., Barro and Salai-Martin, 1995; Levin and Raut, 1997). In particular, these models predict a positive impact of human capital on economic growth. Incorporating Equation (6.2) into the aggregate production function, Equation (6.1), by taking the natural logarithm of the variables of labor (L), domestic capital (DK) and foreign capital (FK) in the production function, adding the variable of human capital (HK) and a regional dummy variable (EAST ) to control for the impact of human capital and regional difference on economic growth, and re-arranging the items on the right-hand side, with the addition of a constant term (β0 ) and an error term (εit ), we obtain the following empirical regression equation: lnYit = β0 + β1 lnLit + β2 lnDKit + β3 lnFKit + β4 SFKit−1 + β5 t + β6 t ∗ lnFKit + β7 HKit + β8 EASTi + εit ,

(6.3)

where i(i = 1, 2, . . . , 30) and t(t = 1987, . . . , 2005) denote province i and year t; L and DK are labor and domestic capital stock respectively; FK is foreign capital stock which captures the effect of FDI on economic growth

March 5, 2013

128

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

through capital augmentation; SFK is the presence of FDI (share of foreign capital stock to total capital stock), which captures the spillover effects of FDI to economic growth through increasing productivity of domestic firms; t is a time trend, which captures the Hicks-neutral technological progress in the absence of FDI or foreign technology; the interaction term t ∗ lnFK captures the additional technological progress to economic growth that is attributed only to FDI; HK is human capital and EAST is a dummy variable for the Eastern region provinces. In this specification, if β3 , β4 and β6 are positive and statistically significant, then FDI has generated positive effects to provincial economic growth through capital augmentation (β3 ), technological progress (β6 ) and spillover effects (β4 ). Equation (6.3) is the form of an augmented production function model that we will use to estimate the impact of FDI on China’s economic growth. The following section will describe the variables and the data.

6.5. Variable Specification and the Data The data for provincial gross domestic product (Y ) and provincial total capital stock are from Wu (2009). Wu uses the conventional perpetual inventory method by employing the recently released national accounts figures to derive a capital stock series for China’s 31 provinces and three economic sectors (i.e., agriculture, manufacturing, and services) for the period 1977 to 2006. This is one of the most comprehensive datasets of capital stock series for China’s 31 provinces and three economic sectors. The data for FDI stock (FK) is calculated in several steps. First, the US dollar value of annual FDI inflows is converted into RMB value by using the annual average official exchange rate. Second, the RMB value of annual FDI inflows are deflated into the real value at 1978 prices by using China’s national Consumer Price Index (CPI). Third, a 5% depreciation rate is assumed for foreign capital (FDI). Finally, FDI stock is accumulated for each year-end measured in billionYuan at 1978 prices. The hypothesis is that provinces with a larger FDI stock will have higher expansion in production and higher technological progress over time, thus shifting the production frontier and accelerating provincial economic growth. The domestic capital stock (DK) of each province is obtained by deducting the FDI stock (FK) from the total capital stock. Labor (L) is the

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

129

total number of employed persons in each province measured in million persons. The presence of FDI is measured as the share of FDI stock in total capital stock of a province (SFK) to capture the spillover effects of FDI on local economic growth. It is reasonable to assume that FDI inflows and spillover effects from FDI have a time lag, so the value of a one year lag of SFK is used in the model. The use of the lagged value of SFK can also avoid the endogeneity problem in the regression. The hypothesis is that provinces with a higher share of FDI stock in total capital stock will have higher spillover effects from FDI to the local economy, thus increasing productivity and efficiency and promoting provincial economic growth. In this study, the human capital (HK) is measured as the ratio of the number of university students to total population of each province. We expect the human capital to be positively related to the economic growth of the host province. The Eastern region dummy variable (EAST ) takes the value of one for the provinces in the Eastern region and zero otherwise. The dependent and independent variables and the data sources are summarized in Table 6.2.

6.6. Regression Results and Explanations 6.6.1. The impact of FDI on economic growth: All provinces The data used in this study are a panel dataset at the provincial level, containing 30 of China’s provinces over the period 1987 to 2005.3 We first estimate Equation (6.3) under the random-effects model in order to test the variable EAST. We then estimate Equation (6.3) under the fixed-effects model to eliminate the province-specific and time-invariant factors, which may affect economic growth. The Hausman test prefers the fixed-effects model. The two models performed very well. All of the independent variables have the expected signs and are statistically significant at the 1% level and the models have high explanatory power. The regression results show that labor input (L) and domestic capital stock (DK) are positive and statistically significant at the 1% level, indicating the significant contributions of labor and domestic capital inputs to provincial economic growth. 3 Tibet is excluded from the sample due to lack of data. Data for FDI inflows into provinces

are not available after 2005.

March 5, 2013

130

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

Table 6.2. Variables of the Impact of FDI on China’s Provincial Economic Growth. Variable Name

Specification of Variables

Sources

Dependent variable Yit

Gross domestic product of province i Wu (2009) in year t. Billion Yuan at 1978 prices.

Independent variables Lit

Total number of employed persons of Various issues of province i in year t. Million persons. National Bureau of Statistics of China, China Statistical Yearbook.

DKit

Domestic capital stock of province i in Calculated from Wu year t. Billion Yuan at 1978 prices. (2009) and various issues of National Bureau of Statistics of China, China Statistical Yearbook.

FK it

FDI stock of province i in year t. Billion Yuan at 1978 prices.

Calculated from various issues of National Bureau of Statistics of China, China Statistical Yearbook.

SFK it−1

Share of FDI stock in total capital stock of province i in year t−1. Per cent.

Same as above.

HK it

Human capital of province i in year t measured as the ratio of the number of university students to total population. Per cent.

Same as above.

EAST i

Eastern region dummy, one for the provinces in the Eastern region, zero for other provinces.

The regression results are reported in Table 6.3. The coefficients of the variables of our main interest, namely lnFK, SFK and t ∗ lnFK, are positive and statistically significant in both regressions. These results provide strong support for the three propositions presented in Section 6.4.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

131

Table 6.3. Regression Results of Production Function ofAll Provinces, 1987–2005. Variables

Random-Effects Model

Fixed-Effects Model

Constant

0.1810 (1.67)∗ 0.4836 (12.92)∗∗∗ 0.3488 (10.73)∗∗∗ 0.0224 (3.63)∗∗∗ 0.0198 (9.80)∗∗∗ 0.0794 (3.96)∗∗∗ 0.0435 (14.23)∗∗∗ 0.0024 (5.20)∗∗∗ 0.3188 (4.78)∗∗∗

1.1890 (7.30)∗∗∗ 0.2472 (4.88)∗∗∗ 0.2814 (8.78)∗∗∗ 0.0184 (2.96)∗∗∗ 0.0193 (10.54)∗∗∗ 0.0474 (2.83)∗∗∗ 0.0553 (16.12)∗∗∗ 0.0026 (5.30)∗∗∗

560 30 0.95 Wald chi2 = 35,812.94∗∗∗

560 30 0.86 F-statistics = 4,819.06∗∗∗

lnL lnDK lnFK SFK HK T T*lnFK EAST No. of observations No. of groups R2 Overall

Note: Standard errors are adjusted for heteroscedasticity and clustering on group. t-statistics are in parentheses. ∗ Statistically significant at 0.10 level (two-tail test). ∗∗ Statistically significant at 0.05 level (two-tail test). ∗∗∗ Statistically significant at 0.01 level (two-tail test).

First, the variable of foreign capital stock (FK) is positive and statistically significant at the 1% level in both models, which provides strong support that FDI as a factor of capital input directly contributes to provincial economic growth. The estimation results show that a 10% increase in foreign capital stock will raise provincial GDP by about 0.18% to 0.22%. This implies that provinces with higher FDI inflows will have a higher economic growth contributed to directly by the increase in foreign capital input associated with high productivity of FDI firms.

March 5, 2013

132

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

Second, the variable measuring the spillover effects of FDI, or the share of FDI stock in total capital stock (SFK), is positive and statistically significant at the 1% level in the two regressions. This is consistent with the hypothesis that FDI has positive spillover effects on provincial economic growth through improving the productivity and efficiency of domestic firms. Thus, the regression results have provided strong empirical evidence to support the hypotheses that FDI inflows into China together with a package of knowledge-based intangible assets have produced positive spillover effects on China’s provincial economic growth. The coefficient of the share of FDI stock in total capital stock is around 0.019. It means that, holding the labor and capital inputs and other variables constant, a 10 percentage point increase in the share of FDI stock in total capital stock will raise provincial GDP by 0.19%. This implies that provinces with a higher share of FDI stock in total capital stock will have higher spillover effects from FDI to the local economy, thus improving productivity and efficiency and raising provincial economic growth. Third, the interaction term of a time trend and the FDI stock (t ∗ lnFK) is positive and statistically significant at the 1% level in both regressions, which supports the proposition that FDI is a shifter of the domestic production frontier. The regression results show that over time, FDI helps the domestic economy to move continuously onto a higher steady state technology. This change in the domestic production frontier caused by FDI is an additional enforcement of the Hicks-neutral technological progress represented by the coefficient on a time trend, which is positive and statistically significant at the 1% level in both regressions. The variable of human capital (HK) is positive and statistically significant at the 1% level in both models, which provides empirical evidence that human capital contributes to economic growth. The coefficient of the human capital is around 0.05–0.08, which means that a 10 percentage point increase in the ratio of university students to total population will raise provincial GDP by around 0.5–0.8%. This implies that provinces with better education and higher human capital will have higher economic growth. Finally, the EAST is positive and statistically significant in the randomeffects model, implying that provinces in the Eastern region have a higher economic growth than the provinces in the Central and Western regions.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

133

6.6.2. The impact of FDI on economic growth: By regions China is a large country with enormous contrasts in geographical and economic conditions between provinces. The degree of economic development is substantially different across the provinces of China and the geographical distribution of FDI is characterized by its concentration in the coastal areas. While an overall positive impact of FDI on economic growth is revealed in the above empirical analysis, China’s large absolute size and economic diversity may mean that the impact of FDI on economic growth may be different between regions. To investigate the impact of FDI on economic growth by region, we divide the sample of 30 provinces into two sub-samples, namely the Eastern region provinces4 and the Central and Western region’s provinces.5 Between the two regions, as Table 6.4 shows, the Eastern region is more economically and technologically developed, has more interactions with the rest of the world, better infrastructure and has attracted an overwhelming share of total accumulative FDI inflows into China. To investigate the impact of FDI on economic growth by region, we also use Equation (6.3) to conduct the empirical analysis. To eliminate the province-specific and time-invariant factors which may affect economic growth, we run the regression under the fixed-effects model with panel data. The panel dataset for the Eastern region contains 11 provinces and the panel dataset for the Central and Western regions contain 19 provinces over the period 1987 to 2005. The fixed-effects regression results for the two models are reported in Table 6.5. The regression results reveal a number of interesting findings. First, the three FDI variables, lnFK, SFK and t ∗ lnFK, are all positive and statistically significant at the 1% level in the Eastern region, which provide strong empirical evidence that FDI shifts the local production frontier through capital augmentation, technological progress and spillover effects to raise economic growth in the Eastern region. 4 The Eastern region provinces include Beijing, Tianjin, Hebei, Liaoning, Shanghai, Jiangsu, Zhejiang, Fujian, Shandong, Guangdong, and Hainan. 5 The Central and Western region’s provinces include Shanxi, Inner Mongolia, Jilin, Heilongjiang, Anhui, Jiangxi, Henan, Hubei, Hunan, Guangxi, Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Gansu, Qinghai, Ningxia, and Xinjiang.

March 5, 2013

134

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

Table 6.4. Economic and Technological Indicators by Region.

Economic and Technological Indicators Economic growth rate (year 1986–2005) (%) Per capital GDP (year 2005) (Yuan/person) (1978 = 100) Accumulative FDI inflows (end 2005) (US$ billion current prices) Human capital (year 2005) (university students/population) (%) Level of R&D (year 2005) (patent applications/10,000 persons) Exports to GDP ratio (year 2005) (%) Transportation intensity index (year 2005) (km per 100 km2 ) Level of telecommunications (year 2005) (phone sets/100 persons)

Eastern Region

Central and Western Regions

12.80 5,736

10.48 2,273

530.55

83.49

1.73

1.05

7

1

41.73 80

6.89 33

91

47

Note: The measure of the transportation intensity index is the ratio of the sum of the length of highways, railways and interior transport waterways divided by the land size of the corresponding province. Sources: Calculated from National Bureau of Statistics of China (various issues (b)).

Second, for the Central and Western regions, the variables of lnFK and t ∗ lnFK are positive and statistically significant at the 1% and 5% levels respectively, but the variable of SFK, though positive, is not statistically significant even at the 10% level. The regression results suggest that FDI shifts the local production frontier through capital augmentation and technological progress but may not improve productivity and efficiency of local firms through spillover effects in the Central and Western regions. One possible explanation is that the volume of FDI in the Central and Western regions is still small and it may take time for FDI to become an important and integrated part of local production and to generate spillover effects on local firms.6 This finding has important policy implication on regional economic development. 6 By 2008, the Central and Western regions attracted only 13.74% of the total accumulative

FDI inflows into China.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

135

Table 6.5. Regression Results of Production Function by Regions, 1987–2005 (Fixed-effects Model). Variables Constant lnL lnDK lnFK SFK HK T T∗ lnFK No. of observations No. of groups R2 Overall F -statistics

Eastern Coastal Region 1.4279 (5.70)∗∗∗ 0.4849 (6.71)∗∗∗ 0.1612 (3.76)∗∗∗ 0.0644 (5.47)∗∗∗ 0.0118 (5.75)∗∗∗ 0.0334 (1.08) 0.0389 (5.39)∗∗∗ 0.0072 (6.24)∗∗∗ 209 11 0.88 2731.42∗∗∗

Central and Western Regions 1.6902 (9.17)∗∗∗ −0.0176 (−0.36) 0.2893 (5.96)∗∗∗ 0.0188 (3.11)∗∗∗ 0.0099 (1.24) 0.0657 (2.82)∗∗∗ 0.0597 (11.88)∗∗∗ 0.0014 (2.33)∗∗ 351 19 0.65 4533.20∗∗∗

Note: Standard errors are adjusted for heteroscedasticity and clustering on group. t-statistics are in parentheses. ∗ Statistically significant at 0.10 level (two-tail test). ∗∗ Statistically significant at 0.05 level (two-tail test). ∗∗∗ Statistically significant at 0.01 level (two-tail test).

Third, comparing the two regions, although FDI has a positive and statistically significant shifting effect on the production frontier through capital augmentation and technological progress in both regions, the regression results do show that the coefficients of the variables of lnFK and t ∗ lnFK are higher in the Eastern region, at 0.0644 and 0.0072, than those in the Central and Western regions, which are 0.0188 and 0.0014, respectively. This implies that FDI has a larger impact on shifting the local production frontier to accelerate economic growth in the Eastern region than in the Central and Western regions.

March 5, 2013

136

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

Another interesting finding is that the variable of human capital (HK) is positive and statistically significant in the Central and Western regions. However, though positive it is not statistically significant in the Eastern region. This suggests that in the Central and Western regions, which are economically and technologically backward compared to the Eastern region, increases in investment in human capital make a positive contribution to economic growth. In contrast, for the Eastern region provinces, which already have a relatively high level of human capital, an increase in the abundance of human capital may no longer be critical for economic growth. Finally, there is significant difference between the Eastern region and the Central and Western regions in the values of estimated labor (L) and domestic capital (DK) elasticities. At the national level, there is no significant difference between labor elasticity and domestic capital elasticity in the fixed-effects model. In the Eastern region, both variables of labor and domestic capital are positive and statistically significant, however, labor elasticity is three times the domestic capital elasticity, implying that the marginal product of labor is not only much higher than that of the marginal product of domestic capital in the Eastern region but is also substantially more than that in the rest of the country. This also suggests that labor may be a constraint on economic growth in the Eastern region. In contrast, in the Central and Western regions, the variable of domestic capital is positive and statistically significant but the variable of labor is negative and statistically insignificant. This implies that the marginal product of domestic capital is much higher than that of labor due to the relative abundance of labor supply and scarcity of capital in the Central and Western regions. Referring to the differences in economic and technological conditions between the Eastern region and the Central and Western regions and the regression results, we may argue that, given the level of FDI stock, provinces with higher level of economic development (per capita GDP), higher level of R&D activities, higher level of infrastructure, and higher level of economic interactions with the rest of the world, will facilitate and enhance the role of FDI to local economic growth through capital augmentation, technological progress, and spillover effects. Thus, this study also provides empirical evidence to suggest that local economic and technology conditions, especially local absorptive capability, play an important role in influencing the diffusion of knowledge spillovers from FDI to the local economy.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

137

6.7. Conclusion The main purpose of this chapter is to investigate empirically the contribution of FDI to China’s economic growth through the channels of capital augmentation, technological progress and spillover effects to the shift of the local production frontier. Based on theoretical foundations, an augmented empirical growth model is specified, and a panel dataset containing 30 of China’s provinces over the period 1987 to 2005 is used under the panel regression models. The study has provided the following main findings. First, the regression results of all provinces provide strong evidence that FDI contributes to China’s economic growth both directly through capital augmentation and technological progress and indirectly through spillover effects to shift the local production frontier. This implies that provinces with higher FDI inflows will have higher economic growth contributed to directly by the increase in foreign capital input associated with technological progress and indirectly by the diffusion of knowledge spillovers from FDI to the local economy. Second, the study finds that the impact of FDI on economic growth is different between China’s regions. The regression results show that the contributions of FDI to economic growth are higher in the Eastern region than those in the Central and Western regions. In the Eastern region, FDI shifts the local production frontier not only through capital augmentation and technological progress but also through spillover effects to contribute to economic growth. In contrast, in the Central and Western regions, FDI is found to contribute to economic growth through capital augmentation and technological progress but not through spillover effects due to the lack of knowledge spillovers from FDI. Referring to the differences in economic and technology conditions between the regions and the regression results, this finding provides empirical evidence to suggest that local economic and technology conditions, especially local absorptive capability, do matter in influencing the diffusion of knowledge spillovers from FDI to the local economy. Third, the regional difference in the role of FDI on economic growth deserves attention for both policy making and academic research. The real problem is not because FDI causes the widening gap between the Eastern region and the Central and Western regions, but because FDI has played a much larger and significant role in the former than in the latter. As a

March 5, 2013

138

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

result, policy should be designed to encourage FDI flows into the Central and Western regions. To achieve the full potential of FDI, conditions have to be created, such as investment in education and infrastructure. However, other policies such as inter-regional migration and cross-regional investments are also important in reducing regional disparity in income and production. Finally, the study finds that although the empirical regression results show a positive and statistically significant impact of FDI on economic growth, the magnitude of the contribution from the technological progress and spillover effects of FDI to China’s economic growth was still very small. This implies that China still has a lot of benefits to gain from FDI. Therefore, apart from improving local economic and technology conditions to attract more FDI inflows, China should encourage contact, information exchange, production and technological cooperation, joint R&D activities, industrial linkages between domestic firms and FDI firms, in order to enhance and accelerate the technological progress and the diffusion of positive spillovers from FDI to China’s economy.

References Barro, R. and Sala-i-Martin, X., Economic Growth. New York: McGraw-Hill (1995). Buckley, P., Clegg, J., Wan, C. and Cross, A., “FDI, Regional Differences and Economic Growth: Panel Data Evidence From China,” Transnational Corporations, 2(1): 1–28 (2002). Caves, R., Multinational Enterprise and Economic Analysis (2nd edn.). Cambridge: Cambridge University Press (1996). Chen, C., Chang, L. and Zhang, Y., “The Role of Foreign Direct Investment in China’s Post-1978 Economic Development,” World Development, 23(4): 691–703 (1995). Chen, C., “Foreign Direct Investment: Prospects and Policies,” in Pigott, C. (ed.), China in the World Economy: The Domestic Policy Challenges, pp. 321–358. Paris: OECD (2002). Dees, S., “Foreign Direct Investment in China: Determinants and Effects,” Economics of Planning, 31(2): 175–194 (1998).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

FDI and Economic Growth

139

Dunning, J., Multinational Enterprises and the Global Economy. Wokingham, England: Addison-Wesley (1993). Henley, J., Kirkpatrick, C. and Wilde, G., “Foreign Direct Investment in China: Recent Trends and Current Issues,” The World Economy, 22(2): 223–243 (1999). Kueh, Y., “Foreign Investment and Economic Change in China,” The China Quarterly, 131: 637–690 (1992). Lardy, N., “The Role of Foreign Trade and Investment in China’s Economic Transformation,” The China Quarterly, 144: 1065–1082 (1995). Levin, A. and Raut, L., “Complementarities Between Exports and Human Capital in Economic Growth: Evidence from the Semi-industrialized Countries,” Economic Development and Cultural Change, 46(1): 155–174 (1997). Markusen, J. and Venables, A., “Foreign Direct Investment as a Catalyst for Industrial Development,” European Economic Review, 43(2): 335–356 (1999). National Bureau of Statistics of China (NBS), China Foreign Economic Statistical Yearbook. Beijing: China Statistics Press (1994, 1996, 1998, 1999–2005). National Bureau of Statistics of China (NBS), China Statistical Yearbook. Beijing: China Statistics Press (1994–2010). Pomfret, R., “Growth and Transition: Why has China’s Performance been so Different?” Journal of Comparative Economics, 25: 422–440 (1997). Ran, J., Voon, J. and Li, G., “How does FDI Affect China? Evidence from Industries and Provinces,” Journal of Comparative Economics, 35(4): 774–799 (2007). Tuan, C., Ng, L. and Zhao, B., “China’s Post-Economic Reform Growth: The Role of FDI and Productivity Progress,” Journal of Asian Economics, 20(3): 280–293 (2009). United Nations Conference on Trade and Development (UNCTAD), World Investment Report 1999: Foreign Direct Investment and the Challenge of Development. New York and Geneva: United Nations Publication (1999). United Nations Conference on Trade and Development (UNCTAD), World Investment Report 2004: The Shift Towards Services. New York and Geneva: United Nations Publication (2004). Vu, T., Gangnes, B. and Noy, I., “Is Foreign Direct Investment Good for Growth? Evidence from Sectoral Analysis of China and Vietnam,” Journal of the Asia Pacific Economy, 13(4): 542–562 (2008).

March 5, 2013

140

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch06

Regional Development and Economic Growth in China

Wei, S., “Foreign Direct Investment in China: Sources and Consequences,” in Ito, T. and Krueger, A. (eds.), Financial Deregulation and Integration in East Asia, Vol. 5, pp. 77–105. Chicago: University of Chicago Press (1996). Whalley, J. and Xin, X., “China’s FDI and Non-FDI Economies and the Sustainability of Future High Chinese Growth,” China Economic Review, 21(1): 123–135 (2010). Wu, Y., “China’s Capital Stock Series by Region and Sector,” Business School, University of Western Australia, Discussion Paper No. 09.02 (2009). http:// www.business.uwa.edu.au/__data/assets/pdf_file/0009/260487/09_02_Wu. pdf (accessed 30 August 2011). Yao, S. and Wei, K., “Economic Growth in the Presence of FDI: The Perspective of Newly Industrialising Economies,” Journal of Comparative Economics, 35(1): 211–234 (2007). Zhang, K., “Foreign Direct Investment and Economic Growth in China: A Panel Data Study for 1992–2004,” Paper Prepared for the Conference of WTO, China and Asian Economies, June 24–26, 2006, University of International Business and Economics, Beijing, China (2006). http://faculty.washington. edu/karyiu/confer/beijing06/papers/zhang.pdf (accessed 30 August 2011).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Chapter 7

Manufacturing Sector FDI and Performance of Domestic Banks Sizhong Sun and Siqiwen Li

7.1. Introduction Since the market reform in the late 1970s, China’s economy has been growing quickly, to which the manufacturing sector has made significant contribution. One characteristic of development in China’s manufacturing sector is the importance of foreign direct investment (FDI). FDI in the manufacturing sector has been found to affect domestic firms’ productivity (see, for example, Buckley et al., 2002, 2007a, 2007b; Li et al., 2001; Liu, 2002, 2008; Sun, 2011; Xu and Sheng, 2012a, 2012b) and exports (see for example Sun, 2009, 2010; Swenson, 2008) in China. In the same period, China’s banking sector, a major funding source for firm production and exporting activities, has also experienced significant development. For example, from 2005 to 2010, the four major commercial banks in China have all been listed in the stock market. Even though researchers have investigated the impacts of FDI on economic growth, domestic productivity, and domestic exports extensively, little is known on whether FDI in the manufacturing sector affects banking sector performance. Understanding of whether the manufacturing sector FDI has impacts on domestic banking performance has two implications. First, it allows us to better understand the contribution of FDI to the domestic economy. Second, it sheds light on a better knowledge of banking performance. This chapter intends to fill in this gap, through examining the impacts of FDI in the manufacturing sector on domestic banking performance in China.

141

March 5, 2013

142

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

The rest of this chapter is organized into six sections. Section 7.2 discusses China’s financial reform and financial market opening up. Section 7.3 provides an overview of China’s banking sector. Section 7.4 reports the presence of FDI in the manufacturing sector. In Section 7.5, we discuss how the presence of FDI can affect banking performance. Section 7.6 presents the empirical findings. Section 7.7 concludes the chapter.

7.2. China’s Financial Reform and the Opening Up of Market Historically banks were highly regulated in contrast to other non-bank financial institutions (NBFIs): domestically via regulation of the quantity, type and pricing of banking services; and externally through a managed exchange rate system and the regulation of foreign exchange and foreign entry. Limited foreign bank entry was one aspect of the controls over the financial sector in both developed and developing economies until financial deregulation in the 1980s. After a series of financial crises and with the poor performance of the banking sector (namely decreased market share compared to their NBFIs competitors), there was a major redirection of policy, from the early 1980s onwards, and the financial system was deregulated (Calomiris, 2000; Edwards and Mishkin, 1995; Kent and Debelle, 1999). China’s market became one of the major destinations of international capital inflows, both in terms of securities investment and FDI after China adopted the opening-up policy and financial reform in the late 1970s. FDI in the manufacturing sector of China’s market often requires an efficient path to serve the transfer of capital which is offered by the financial sector, in particular banks of the host country (Ali and Guo, 2005). Positive spillovers from the prosperity of foreign investment potentially promote further growth of domestic businesses in China. Sufficient capital financing from international capital markets therefore play a vital role in supporting such business growth. Financial system reform beginning from the 1980s in China opened up the financial market gradually to foreign financial institutions following the

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

143

trend of global financial deregulation. This breakup of monopoly banking systems promoted further FDI growth in China as the consequence of increasing foreign bank entry. Foreign bank entry in China occurred in five ways: branch networking; subsidiaries which acted as legal entities; joint venture with Chinese partners; representative offices which were common in the early stages of financial reform; and stake holding of Chinese commercial banks which appeared to be growing in recent years (CBRC, 2004). By December 2003, the ceiling on foreign ownership in domestic banks was raised from 15% to 20% for a single bank and to 25% overall. By the end of 2001, over 200 foreign banks/branches and 16 foreign insurance companies were allowed to operate only in a few cities (CBRC, 2005). Within the several years after making WTO commitments, foreign bank operated entities increased to 312 despite several merger cases (CBRC, 2004). A further opening-up of the domestic financial market has been observed, however performance of local banks seems insufficient (Garcia-Herrero et al., 2006). Obviously, this raises direct challenge to domestic Chinese banks located in these geographic areas. In the aftermath of theAsian Financial Crisis, foreign bank entry in China experienced a short period of slow down — several banks retreated from the Chinese market (CBRC, 2007). Foreign institutional investor’s access to China’s financial market was also limited to high quality institutions. China took further steps to liberalize the financial market including: a lower minimum requirement on foreign banks’ operating capital which appropriately eases their liquidity constraints of operating business in China; and they were allowed, like their local counterparties, to engage in the trading of financial derivatives, offshore wealth management and insurance agency businesses (CBRC, 2007). With China’s entry into the WTO, obstacles to foreign banks were loosened significantly as shown by the larger foreign bank presence in more regions: by the end of 2004 foreign-funded financial institutions were permitted to engage in Renminbi business in an additional five cities (Kunming, Xiamen, Xi’an, Beijing, and Shenyang) under supervision of the China Banking Regulatory Commission (CBRC). This dramatically improved the ability and authority of foreign banks to offer financial services to foreign investors whose investment decisions could be largely influenced

March 5, 2013

144

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

by the availability of bank services and convenience to access these facilities.

7.3. Overview of the Banking Sector Performance in China In classic corporate finance literature, the performance, in particular financial performance of an enterprise is measured by financial ratios. Profitability and capital ratios are two pillars of bank performance indicators used in business performance analysis. Return on Equity (ROE) and Return on Assets (ROA) are major profitability ratios while the liquidity ratio, capital adequacy ratios and efficiency ratios, including cost-toincome ratio, quantify other financial aspects of a business in the banking sector. Profitability and efficiency ratios — ROE, ROA and the cost-to-income ratio measure how effectively a firm uses resources to generate profit. ROE and ROA share the same numerator — net profit, and an increase in the net profit improves both ROE and ROA. However, ROE and ROA focus on different aspects of profitability. ROE measures the amount of profit generated by the firm by using equity, while in contrast ROA focuses on operating profitability using total assets as the denominator regardless of whether investments are financed by equity or debt (namely liability). Debt financing such as using liability magnifies (or “levers”) the effect of operating profitability (ROA) on overall profitability (ROE), which is described as the “financial leverage” process. Financial risk associated with aggressive usage of financial leverage is expected to be buffered by holding the proper amount of capital. Under Basel Accords, banks are required to hold sufficient regulatory capital against risks. In banks’ performance evaluation, capital adequacy ratios including the core capital adequacy ratio therefore serve as the indicator of banks’ risk-taking and risk management performance. Core capital (or Tier-1 capital) includes more liquid capital stock, surplus and undivided profits in contrast to subordinated capital. Capital ratios are applied as complementary to ROA and ROE, as they are both non-risk adjusted profitability ratios. This is particularly the case in the financial sector due to the risktaking activities that financial institutions, such as banks, are involved in.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

145

Figure 7.1 shows the trend of ROA movements of 12 domestic commercial banks within the past eight years.1 First, six banks have been slowly increasing their ROA over the eight years. The ROA of banks 7, 8, 9 and 10 slightly fluctuate which was mostly due to internal operation and business policy changes that frequently occurred with regional commercial banks. The last two banks have relatively stable ROAs over the observation period. Figure 7.2 presents the trend of ROE movements of four selected banks that are publically listed at an earlier time and with a larger market capitalization in contrast to others. Figure 7.2 illustrates that the ROE of these banks remain constant, although Bank 10’s ROE has some fluctuations along the curve. Table 7.1 provides summary statistics of the ROA and ROE of these 12 commercial banks. The ROA and ROE of these banks have substantial variations with the highest ROA being 1.46, the lowest ROA being 0.14; and the highest ROE being 28.58, and the lowest ROE being 3.74.

7.4. Manufacturing Sector FDI in China The FDI inflow in the manufacturing sector has been very active in the past few decades in China. From 2000 to 2007, on average 34.35% of outputs in the four-digit industries are produced by FDI firms, and in the four-digit industries 21.86% of firms are FDI invested. Figure 7.3 presents the average FDI presence in the manufacturing sector from 2000 to 2007. Two features emerge from Figure 7.3. First, the presence of FDI in the Chinese manufacturing sector appears to be quite stable and significant from 2000 to 2007. Second, FDI firms are more productive than their domestic counterparts, with around 22% of FDI firms accounting for about 34% of industry outputs. Figure 7.4 reports the variance in the level of FDI presence across the four-digit industries in the manufacturing sector. The standard deviation of FDI presence across the four-digit industries is relatively low, around 63% of the mean of FDI presence. Besides, the standard deviation exhibits 1 Financial ratio data used in this study are collected from annual financial reports of each

bank; however financial ratio data of a few regional commercial banks at certain years were not available in financial reports at the time of data collection.

March 5, 2013

146

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

Figure 7.1.

Return on Assets (ROA) of 12 Domestic Commercial Banks in China.

Source: Authors’ compilation.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

Figure 7.2.

147

Return on Equity (ROE) of Selected Banks.

Source: Authors’ compilation.

Table 7.1. Summary Statistics of ROAs and ROEs of 12 Commercial Banks. ROA

ROE

Mean S.D. Maximum Minimum Mean S.D. Maximum Minimum Bank 1 Bank 2 Bank 3 Bank 4 Bank 5 Bank 6 Bank 7 Bank 8 Bank 9 Bank 10 Bank 11 Bank 12 Average

0.80 1.05 0.92 0.57 0.85 0.78 0.52 1.10 0.91 0.90 0.86 1.37 0.88

0.42 0.32 0.19 0.27 0.30 0.18 0.08 0.19 0.15 0.36 0.07 0.13 0.22

1.32 1.32 1.14 0.86 1.2 1.09 0.64 1.26 1.13 1.46 0.99 1.55 1.16

Source: Authors’ compilation.

0.14 0.42 0.66 0.24 0.35 0.56 0.41 0.7 0.69 0.51 0.8 1.25 0.56

18.79 20.90 14.04 14.30 17.55 17.26 13.38 23.37 23.45 19.49 21.51 13.29 18.11

3.02 3.29 2.55 8.53 2.86 2.26 2.73 3.49 9.44 5.13 1.39 0.92 3.80

22.79 25.86 18.87 24.58 20.87 21.61 18.25 26.21 35.6 28.58 22.49 14.89 23.38

15.37 15 11.22 3.74 13.68 14.42 11 17.51 13.57 12.57 20.53 12.67 13.44

March 5, 2013

148

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

0.4 0.35 0.3 0.25 0.2 0.15 0.1 0.05 0 2000

2001

2002

2003

2004

2005

2006

2007

Average share of FDI firms' output in four digit industries Average share of number of FDI firms in four digit industries

Figure 7.3. Average FDI Presence in the Manufacturing Sector 2000–2007. Source: Authors’ compilation.

0.25 0.2 0.15 0.1 0.05 0 2000

2001

2002

2003

2004

2005

2006

2007

Standard deviation of the share of FDI firms' output in four digit industries Standard deviation of the share of number of FDI firm in four digit industries

Figure 7.4.

Standard Deviation of FDI Presence in the Manufacturing Sector 2000–2007.

Source: Authors’ compilation.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

149

a declining trend across time, albeit quite moderate. A declining standard deviation of FDI presence across the four-digit industries implies that FDI activities converge across the four-digit industries.

7.5. How Does the Manufacturing Sector FDI Affect Banking Performance? Figure 7.5 presents a conceptual framework that illustrates the way the manufacturing sector FDI affects domestic banking performance. Via foreign banks, the foreign investment flows into foreign firms and generates impacts on domestic banks through three channels, namely the impact due to spillovers to domestic firms (Channel 1 in Figure 7.5), direct business activities between foreign firms and domestic banks (Channel 2 in Figure 7.5), and the competition from foreign banks to domestic banks (Channel 3 in Figure 7.5). First, on Channel 1, a number of previous studies have found that the presence of FDI can affect domestic firms in the manufacturing sector, and researchers have identified three channels through which foreign firms can affect domestic firms in the manufacturing sector, that is the backward and forward linkages, labor mobility, and demonstration and competition effects (Blomstrom and Kokko, 1998). Domestic firms can be either (or both) local suppliers and customers to FDI firms, which help domestic firms to improve their performance,

Foreign Investment

Foreign Banks

1

2

3

Domestic Banks

Foreign Firms

1

1

Domestic Firms

Figure 7.5. A Conceptual Framework.

March 5, 2013

150

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

namely positive spillovers taking place. FDI firms can help domestic firms, which are their local suppliers, to set up production facilities, provide technical assistance and information, help in purchasing raw materials and intermediate inputs, provide training and help in management and organization, and assist domestic firms to diversify by finding additional customers (Lall, 1980). The movement of employees trained by FDI firms to domestic firms is another channel through which spillovers can occur. Employees in FDI firms are usually well trained, through activities such as skills training, on-the-job training and even overseas education and training at the parent company. If these well-trained employees move to domestic firms or set up their own businesses, they will naturally carry the skill they obtained in FDI firms over to domestic firms. FDI firms can also act as a learning model for domestic firms. Domestic firms can learn from, and imitate, FDI firms in adopting similar production techniques and improving their efficiency, through observing the business activities of FDI firms. Even if domestic firms are not inclined to imitate FDI firms, increased competition from FDI firms can force domestic firms to do so. In the short run, such competition may generate negative impacts on domestic firms, as some of them are forced out of the market (Aitken and Harrison, 1999; Harrison, 1994), but in the long run it may have a positive impact (Barrios et al., 2005), since weak domestic firms will already be crowded out of the market and existing firms adopt more advanced technology and management know-how to compete. Therefore foreign firms in the manufacturing sector can positively affect domestic firms. Subsequently, a large body of empirical research is devoted to finding the spillovers (impacts) from FDI to domestic firms. In China, many of these studies find positive spillovers from FDI. Using the third industrial census data in 1995, Li et al. (2001) find positive productivity spillovers, where the magnitude of spillovers depends on the types of domestic firm ownership and different sources of FDI. Based on the same dataset, positive productivity spillovers are further confirmed by Buckley et al. (2002, 2007a), and Chuang and Hsu (2004). Sun (2011) investigates the FDI productivity spillovers, using a simultaneous equation model estimated over a firm level dataset in 2003, and finds significantly positive productivity spillovers. Xu and Sheng (2012b) explore the backward

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

151

and forward linkage of FDI productivity spillovers using a comprehensive firm-level dataset from 2000 to 2003 in China’s manufacturing sector. They find significant and positive spillovers through the forward linkage. Besides, they also find that the productivity spillovers are likely to be regional (Xu and Sheng, 2012a). Positive productivity spillovers also exist at a regional level. Liu (2002) find positive productivity spillovers of FDI in the manufacturing sector of Shenzhen City, in the form of a positive and significant relationship between FDI in the manufacturing sector and the level and growth rate of productivity in component industries. Later, Liu (2008) further confirms that FDI raises the long-term rates of productivity growth of domestic firms, although it lowers their short-term productivity levels, using firm-level data in the Chinese manufacturing sector. In addition to the productivity spillovers, foreign firms can affect other aspects of domestic firms’ performance. Sun (2009, 2010) explores whether foreign firms affect domestic firms’exporting behavior. He finds that an increase in the presence of foreign firms exerts a significant impact on both the likelihood of participating in exporting and the exporting intensity by domestic firms. Such positive spillovers from FDI in the manufacturing sector are expected to be transmitted to the banking sector. Commercial banks are the main funding source for firm investment, particularly in China. Positive impacts from FDI on domestic firms in the manufacturing sector lead domestic firms to expand their production capacity and other business activities such as exporting, which in turn require increased financing from the commercial banks and thus increases the performance of domestic banks, ceteris paribus. Second, FDI firms in the manufacturing sector themselves can directly affect the performance of domestic banks, through their financing activities with domestic banks (Channel 2 in Figure 7.5). For the green field FDI, new plants and production lines need to be built; new machines need to be purchased; new employees need to be recruited and trained; and new sales/distribution networks needs to be established. All these activities involve substantial investment, which may be larger than the amount of foreign investment. Therefore, the newly established foreign firms are likely to finance these investments through borrowing from domestic banks. It is also likely that existing firms with foreign investment will

March 5, 2013

152

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

resort to domestic banks for further financing. After firms are acquired by foreign investment, the new management may like to renovate, upgrade, or expand their existing production facilities, purchase new machines, establish new production lines, and increase investment in human capital. All these business activities will increase their financing activities with domestic banks and subsequently boost the performance of domestic banks. Third, the inflow of foreign investment to the manufacturing sector via foreign banks also increases the contacts and competition between foreign and domestic banks, which in turn affects the performance of domestic banks (Channel 3 in Figure 7.5). The increased exposure to foreign banks, due to the inflow of FDI in the manufacturing sector, can generate two impacts on domestic banks. On the one hand, domestic banks can learn and imitate the business activities of foreign banks, either through their business activities with foreign banks or employee movement from foreign banks to domestic banks, and thus improve their performance. On the other hand, as the financial reform progresses and the financial market opens up, foreign banks are gradually granted licenses to conduct businesses that are previously exclusive to domestic banks. For example in 1996 and 1998 qualified foreign banks were allowed to conduct Chinese Yuan business in Pudong of Shanghai and Shenzhen respectively, and later in 2006 qualified foreign banks were granted further access to regions other than Pudong and Shenzhen. Such opening-up clearly increases the competition between domestic and foreign banks, which forces domestic banks to improve their efficiency and subsequently their performance. Although in the short run such competition can hurt domestic banks, in the long run it will generate positive impacts on domestic banks. In summary, the FDI in the manufacturing sector can generate impacts on the performance of domestic banks conceptually through three channels, namely its spillover effect to domestic firms and subsequently domestic banks, its financing activities with domestic banks, and the increased exposure of domestic banks to foreign banks due to the inflow of FDI. In the next section, we will estimate the impact of manufacturing sector FDI on the performance of domestic banks. Ideally we would like to capture the three channels separately. However, lack of data prevents us from doing so. Instead, we regress the bank’s performance against the distribution of FDI

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

153

in the manufacturing sector (both the mean and standard deviation) and measure the impact by examining the magnitude of point estimates.

7.6. The Empirical Evidence In this section, we empirically test the impact of the manufacturing sector FDI on domestic banking performance. We set up the following econometric model: bpit = β0 + β1 TotalAssetsit + FDI
it−3 β2 + αi + εit , where bp denotes bank performance and is a set of financial ratios as discussed in the following; TotalAssets is the bank’s total assets that measures bank size; FDI is a set of measurements of the distribution of FDI in the manufacturing sector, and includes the mean and standard deviation of the FDI presence in the 280 four-digit industries in the manufacturing sector, where the FDI presence is calculated as the share of FDI firms’ outputs in the four-digit industry; α is the individual fixed effect, and ε is an i.i.d. normal error term. We use financial ratios to measure bank performance. In our analysis, we use the ROA, ROE, and capital adequacy ratio to measure bank performance. We collect the financial ratios of 12 commercial banks in China from 2000 to 2010, and calculate the mean and standard deviation of FDI presence in the four-digit industries in China’s manufacturing sector from 2000 to 2007, where data come from the National Bureau of Statistics. For the measurement of manufacturing sector FDI, we use a three year lag in the regression, which on the one hand is due to the constraint of data availability and on the other hand allows for the possibility that spillover transmission takes time to occur. Table 7.2 reports the summary statistics of the data used in the regression. It can be observed from Table 7.2 that there exist significant variations in all variables. For example the mean of foreign presence in the manufacturing sector is as high as 34%, with an associated standard deviation of 0.01. The standard deviation of foreign presence in the manufacturing sector is 0.22, indicating that there exists a substantial difference in the level of foreign presence across different industries in the manufacturing sector.

March 5, 2013

154

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

Table 7.2. Summary Statistics. Variable ROA ROE Capital adequacy ratio Total assets Foreign presence (mean) Foreign presence (standard deviation)

Obs

Mean

Std. Dev.

Min

Max

89 80 85 92 96 96

0.87 17.97 10.52 3.08 0.34 0.22

0.33 5.68 2.83 3.25 0.01 0.01

0.13 3.74 2.3 0.20 0.33 0.21

1.55 35.6 24.12 13.46 0.36 0.23

Sources: National Bureau of Statistics, 2000–2007, and the financial reports from various commercial banks.

Table 7.3. Regression Results of Fixed Effect Estimation.

ROA

Total Assets FDI (1) FDI (2) Constant Obs. R2 F

Capital Adequacy Ratio

ROE

Coef.

S.E.

Coef.

S.E.

Coef.

0.03 6.65∗ −20.22∗∗ 2.84∗ 79 0.15 6.61

0.03 3.46 8.84 1.37

0.001 201.14∗∗ −426.62∗∗ 40.15 79 0.15 6.6

0.59 83.39 162.91 34.90

−0.26 60.87∗∗∗ −191.61∗∗ 31.49 83 0.004 5.57

S.E. 0.15 18.73 71.63 13.98

Note: FDI (1) and FDI (2) denotes the mean and standard deviation of FDI presence in the manufacturing sector; standard errors are robust; ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% level, respectively. Source: The authors’ regression results.

We then apply the fixed effect estimator in the regression. Table 7.3 reports the regression results. Table 7.3 indicates that FDI in the manufacturing sector appears to exert a significant effect on domestic bank performance, measured either by ROA and ROE or the capital adequacy ratio, as the coefficients of both the average FDI presence and standard deviation of FDI presence are significant at the 5% level. In addition, the

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

155

coefficient of the average FDI presence is significantly positive indicating that on average a higher level of FDI presence in the manufacturing sector generates a bigger impact on domestic bank performance, namely a positive effect from the FDI in the manufacturing sector to bank sector. Such positive impact can occur through the three channels discussed above. In contrast, the coefficient of standard deviation of FDI presence in the manufacturing sector is significantly negative, which suggests that the variation in the FDI inflow into the manufacturing sector is not good to domestic bank performance. A higher standard deviation of foreign presence in the manufacturing sector implies that FDI flows into industries in an uneven way, i.e. some industries have a high level of foreign presence while the others have a low foreign presence. If this high variation in the foreign presence across industries reflects the associated business risk in the manufacturing sector, then it is not surprising that it will negatively affect the performance of domestic banks. Using the fixed effect estimation, we discover that the distribution of FDI presence in the manufacturing sector indeed significantly affects domestic banks’ performance, with a positive impact from the mean and a negative impact from the standard deviation. Is this pattern robust to other estimation methods? To examine the robustness, we apply the random effect estimator and ordinary least square (OLS) estimator, and Tables 7.4 and 7.5 reports the results. It appears that this pattern continues to hold in both the random effect and pooled OLS estimations, except in the pooled OLS estimation of the capital adequacy ratio, where the standard deviation of FDI presence is insignificant at the 5% level. Up to now, the FDI presence in the manufacturing sector has been measured as the share of foreign firms’ outputs in a four-digit industry. Will different measures of FDI presence in the manufacturing sector lead to different findings? To investigate this point, we re-estimate the empirical model using two alternative measures of FDI presence, namely the shares of foreign firms’ employees and assets in a four-digit industry, using the fixed effect estimator. Tables 7.6 and 7.7 present the regression results. Compared the point estimate of Tables 7.6 and 7.7 with that of Table 7.3, the coefficient of average FDI presence is still significantly positive, but the coefficient of standard deviation of FDI presence is now insignificant at the 5% level. Therefore, the impact of mean FDI presence continues to hold in alternative

March 5, 2013

156

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

Table 7.4. Regression Results of Random Effect Estimation.

ROA Coef. Total Assets 0.03∗∗ FDI (1) 7.08∗∗ FDI (2) −21.72∗∗∗ Constant 3.04∗∗∗ Obs. 86 R2 0.26 Wald(chi2) 39.74

Capital Adequacy Ratio

ROE S.E.

Coef.

S.E.

0.01 0.055 3.36 195.69∗∗ 5.47 −412.67∗∗∗ 0.58 38.80∗ 79 0.15 23.01

Coef.

0.18 0.11 84.53 52.07∗∗∗ 114.78 −121.48∗∗∗ 22.72 18.50∗∗ 83 0.16 14.85

S.E. 0.09 17.85 42.32 8.82

Note: FDI (1) and FDI (2) denotes the mean and standard deviation of FDI presence in the manufacturing sector; standard errors are robust; ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% level respectively. Source: The authors’ regression results.

Table 7.5. Regression Results of Pooled OLS Estimation.

ROA

Total Assets FDI (1) FDI (2) Constant Obs. R2 F

Capital Adequacy Ratio

ROE

Coef.

S.E.

Coef.

S.E.

Coef.

0.02 8.15∗∗ −22.13∗∗∗ 2.75∗∗∗ 86 0.26 16.37

0.02 3.63 6.08 0.75

0.023 181.50∗ −405.57∗∗ 42.32∗ 79 0.15 7.53

0.27 86.99 132.88 22.78

0.30∗∗ 50.73∗∗ −86.54 10.72 83 0.19 5.5

S.E. 0.12 19.76 54.85 10.23

Note: FDI (1) and FDI (2) denotes the mean and standard deviation of FDI presence in the manufacturing sector; standard errors are robust; ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% level respectively. Source: The authors’ regression results.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

157

Table 7.6. Regression Results Using Employment Share as Measure of FDI Presence. ROA

Total Assets FDI (1) FDI (2) Constant Obs. R2 F

Capital Adequacy Ratio

ROE

Coef.

S.E.

Coef.

S.E.

Coef.

0.03 4.90∗ 10.73 −2.74 86 0.28 11.12

0.03 2.41 13.07 2.46

−0.320 130.19∗∗∗ −54.46 −6.75 79 0.14 7.41

0.57 38.55 336.73 68.59

−0.39∗∗ 58.28∗∗∗ 3.03 −5.16 83 0.001 7.98

S.E. 0.17 15.44 121.19 22.83

Note: FDI (1) and FDI (2) denotes the mean and standard deviation of FDI presence in the manufacturing sector; standard errors are robust; ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% level, respectively. Source: The authors’ regression results.

Table 7.7. Regression Results Using Assets Share as Measure of FDI Presence.

ROA

Total Assets FDI (1) FDI (2) Constant Obs. R2 F

Capital Adequacy Ratio

ROE

Coef.

S.E.

Coef.

S.E.

Coef.

0.03 8.16∗∗ −4.09 −1.22 86 0.26 11.44

0.02 3.28 6.53 1.71

−0.319 216.98∗ −108.71 −34.78 79 0.14 6.87

0.55 108.13 169.29 67.32

−0.31∗ 101.87∗∗ −9.33 −22.47 83 0.001 6.68

S.E. 0.14 33.30 65.65 22.86

Note: FDI (1) and FDI (2) denotes the mean and standard deviation of FDI presence in the manufacturing sector; standard errors are robust; ∗∗∗ , ∗∗ , and ∗ denote significance at the 1%, 5%, and 10% level respectively. Source: The authors’ regression results.

March 5, 2013

158

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

measures of FDI presence, while the impact of standard deviation of FDI presence is not robust to alternative FDI measures.

7.7. Conclusions This chapter explores the impacts of FDI in the manufacturing sector on domestic banking performance, measured in terms of the ROA, ROE, and capital adequacy ratio. Significant impacts from the manufacturing sector FDI are observed. The average FDI presence in the four-digit industries in the manufacturing sector is shown to exert a positive effect on banking performance, while in contrast the standard deviation of FDI presence across these four-digit industries has a negative effect on banking performance. A higher level of FDI presence in the manufacturing sector can positively affect banking performance through either increased exposure to foreign banks due to the inflow of FDI, FDI firms’financing activities with domestic banks, or positive spillovers from FDI to other domestic firms in the manufacturing sectors.

References Aitken, B. and Harrison, A., “Do Domestic Firms Benefit from Direct Foreign Investment? Evidence from Venezuela,” American Economic Review, 89(3): 605–618 (1999). Ali, S. and Guo, W., “Determinants of FDI in China,” Journal of Global Business and Technology, 1(2): 21–33 (2005). Barrios, S., Gorg, H. and Strobl, E., “Foreign Direct Investment, Competition and Industrial Development in the Host Country,” European Economic Review, 49(7): 1761–1784 (2005). Blomstrom, M. and Kokko, A., “Multinational Corporations and Spillovers,” Journal of Economic Survey, 12(2): 247–277 (1998). Buckley, P. J., Clegg, J. and Wang, C., “The Impact of Inward FDI on the Performance of Chinese Manufacturing Firms,” Journal of International Business Studies, 33(4): 637–655 (2002). Buckley, P. J., Clegg, J. and Wang, C., “Is the Relationship Between Inward FDI and Spillover Effects Linear? An Empirical Examination of the Case of China,” Journal of International Business Studies, 38(3): 447–459 (2007a).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Manufacturing Sector FDI and Performance of Domestic Banks

159

Buckley, P. J., Clegg, J. and Wang, C., “The Impact of Foriegn Ownership, Local Ownership and Industry Characteristics on Spillover Benefits from Foreign Direct Investment in China,” International Business Review, 16(2): 142–158 (2007b). Calomiris, C., U.S Bank Deregulation in Historical Perspective. New York: Cambridge University Press (2000). CBRC, “Public Notice of the CBRC on Further Opening Up China’s Banking Industry,” China Banking Regulatory Commission Report (2004). CBRC, “Latest Developments in China’s Banking Reform, Opening-up and Supervision,” China Banking Regulatory Commission Report (2005). CBRC, “Report on the Opening-up of the Chinese Banking Sector,” China Banking Regulatory Commission Report on January 25 (2007). Chuang, Y. and Hsu, P., “FDI, Trade, and Spillover Efficiency: Evidence from China’s Manufacturing Sector,” Applied Economics, 36: 1103–1115 (2004). Edwards, F. and Mishkin, F., “The Decline of Traditional Banking: Implications for Financial StabilityAnd Regulatory Policy,” NBER Working Paper No. 4933 (1995). Garcia-Herrero, A., Gavila, S. and Santabarbara, D., “China’s Banking Reform: An Assessment of its Evolution and Possible Impact,” CESifo Economic Studies, 52(2): 304–363 (2006). Harrison, A., “Productivity, Imperfect Competition and Trade Reform,” Journal of International Economics, 36(1/2): 53–73 (1994). Kent, C. and Debelle, G., “Trends in the Australian Banking System: Implications for Financial System Stability and Monetary Policy,” Reserve Bank of Australia Research Discussion Paper No. 5 (1999). Lall, S., “Vertical Interfirm Linkage in LDCs:An Empirical Study,” Oxford Bulletin of Economics and Statistics, 42: 203–226 (1980). Li, X., Liu, X. and Parker, D., “Foreign Direct Investment and Productivity Spillovers in the Chinese Manufacturing Sector,” Economic System, 25: 305–321 (2001). Liu, Z., “Foreign Direct Investment and Technology Spillover: Evidence from China,” Journal of Comparative Economics, 30: 579–602 (2002). Liu, Z., “Foreign Direct Investment and Technology Spillovers: Theory and Evidence,” Journal of Development Economics, 85(1/2): 176–193 (2008). Sun, S., “How Does FDI Affect Domestic Firms’ Exports? Industrial Evidence,” World Economy, 32(8): 1203–1222 (2009).

March 5, 2013

160

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch07

Regional Development and Economic Growth in China

Sun, S., “Heterogeneity of FDI Export Spillovers and Its Policy Implications: The Experience of China,” Asian Economic Journal, 24(4): 289–303 (2010). Sun, S., “Foreign Direct Investment and Technology Spillovers in China’s Manufacturing Sector,” Chinese Economy, 44(2): 25–42 (2011). Swenson, D. L., “Multinationals and the Creation of Chinese Trade Linkages,” Canadian Journal of Economics, 41(2): 596–618 (2008). Xu, X. and Sheng, Y., “Are FDI Spillovers Regional? Firm-level Evidence from China,” Journal of Asian Economics, 23(3): 244–258 (2012a). Xu, X. and Sheng, Y., “Productivity Spillovers from Foreign Direct Investment: Firm-Level Evidence from China,” World Development, 40(1): 62–74 (2012b).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Chapter 8

Patterns of Industrial Dynamics in the Manufacturing Sector Sizhong Sun

8.1. Introduction One strand of research in the domain of industrial dynamics is to characterize the distribution of firm size and the relation between firm size and growth rate. As the outcome of underlying growth dynamics arising from the firm entry, growing, shrinking, and exit, Gibrat (1931) observed that firm size is logarithmically normally distributed in the French manufacturing sector, and formulated the well-known “Law of Proportionate Effect”, namely the firm growth rate does not depend on its size. Subsequently a large number of literature has been focused on this area, for example Hart and Prais (1956), Simon and Bonini (1958), Hart (1962), Mansfield (1962), Steindl (1965), Quandt (1966), Silberman (1967), Ijiri and Simon (1977), Clarke (1979), Evans (1987), Hall (1987), Dunne et al. (1988), Stanley et al. (1995), Wilson and Williams (2000), Lotti et al. (2001), Axtell (2001), and Goddard et al. (2006). De Wit (2005) summarizes the steady-state firm size distributions that result from different firm dynamics models, and Sutton (1997) and Coad (2007) provide reviews on the empirical insights on the shape of firm size distribution. However, the research does not yield a consensus result, and Sutton (1997) concludes that there may not exist a general density function that describes all empirical densities well. An extension of the research on firm size distribution is to explore the distributional properties of the growth rate. Stanley et al. (1996) and Amaral et al. (1997) find that the distribution of the logarithm of growth rates displays an exponential tent-shape that closely resembles the Laplace distribution in the US manufacturing sector. More recently, Bottazzi and his

161

March 5, 2013

162

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

colleagues have pioneered to apply the Subbotin family of distributions, a more flexible class of distribution where both the Gaussian and Laplace distributions are special cases. In the US manufacturing sector (Bottazzi and Secchi, 2003), Italian manufacturing sector (Bottazzi et al., 2007), and the worldwide pharmaceutical industry (Bottazzi et al., 2001; Bottazzi and Secchi, 2005), a growth rate distribution that is close to the Laplace distribution has been found. Nevertheless, the Laplace behavior of growth rates does not appear to be universal. Bottazzi et al. (2009) find a distribution of growth rates that has a fatter tail than the Laplace distribution in the French manufacturing sector. Many of the previous studies explore the distributions of firm size and growth rate by pooling firms that operate in different sectors together. Bottazzi and Secchi (2003) argue that pooling heterogeneous firms together (aggregation) can introduce the statistical regularities that are merely the outcome of the pooling procedure and may at the same time cover up the specific characteristics of the firm dynamics in different sectors. Given the Bottazzi and Secchi (2003) critique, it is appropriate to analyze the distributions of firm size and growth rate not only over the pooled firms but also across different sectors where firms are more homogenous. In this chapter, we follow the analytical framework of Bottazzi and Secchi (2003), Bottazzi et al. (2007), and Bottazzi et al. (2009) to explore the patterns of industrial dynamics in the Chinese manufacturing sector. Specifically, we conduct a set of parametric and non-parametric statistical analyses of the firm growth dynamics, namely the size distribution, firm growth process, and growth rate distribution, in the Chinese manufacturing sector. This procedure is conducted over both the pooled (aggregate) firms and by sectors. Furthermore, since the data used cover a period during which the Chinese government changed its industrial policy regime, we are able to explore the impact of the shift of industrial policy regime on the distributions of firm size and growth rate. Therefore, the contribution of this chapter is twofold. First, utilizing an extensive firm level dataset, we provide evidence of distributions of firm size and growth rate from China, a large country where the manufacturing sector has contributed to around 40% of its fast economic growth. A number of previous studies have investigated China’s economic growth, for example, among others, Wu (2008a, 2008b). However to the best of our knowledge, this has not been done in China previously, possibly

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

163

due to a lack of a comprehensive dataset. Second, we relate the characteristics of growth rate distribution to economic factors such as the average firm size and industrial policy regime shift, and examine the role of the shift in the firm growth process and distribution of growth rate, which provides an insight into the firm dynamics from a policy perspective. The rest of the chapter is organized into five sections. In Section 8.2, we describe the data. Section 8.3 presents a brief overview of the Chinese manufacturing sector and discusses the five-year planning procedure where the Chinese government regularly makes a five-year plan for future economic development. Section 8.3 lays down background knowledge for subsequent empirical exercises. The statistical analysis of the distributions of firm size and growth rate at the aggregate level is presented in Section 8.4, and we discuss the results from the similar analysis which is conducted at the sectoral level in Section 8.5. Section 8.6 discusses and summarizes the findings.

8.2. The Data We draw upon a comprehensive firm-level dataset collected annually by the Chinese National Bureau of Statistics (NBS) that covers over 85% of China’s total industrial output. Data from NBS have been used to study various aspects of Chinese industrial economy, for instance, Hu et al. (2005) in R&D and technology transfer, Jefferson et al. (2008) in productivity growth, Sun (2009) in export spillovers of foreign direct investment, Wu (2011) in innovation and economic growth in China, and Chen and Wu (2012) in regional economic growth in China’s Pan Pearl River Delta area. The data used in this study cover seven years from 2001 to 2007 and are constructed according to the following procedure. First, we exclude firms that employed less than eight workers as they may not have reliable accounting systems (Jefferson et al., 2008), and that reported negative net values of fixed assets, wages, and total assets, which may occur due to firm misreporting or errors in inputting records. Second, as in Bottazzi et al. (2009), we only consider continuing firms over this period. Therefore we are left with 22,774 firms over the seven years.1 Table 8.1 presents 1 The proportion of these firms in total firms ranges from 8% of all firms in 2007 to 18% in

2000.

March 5, 2013

164

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Table 8.1. Number of Firms by Industry. Code 13 14 15 16 17 18 19 20

21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37

Sector Name Agro-food processing industry Food manufacturing Beverage manufacturing Tobacco manufacturing Textiles Textiles and garments, shoes, hat manufacturing Leather, fur, feathers (velvet) and its products industry Wood processing and wood, bamboo, rattan, palm and grass products industry Furniture manufacturing Paper and paper products industry Printing and record medium reproduction Cultural, educational, and sporting goods manufacturing Petroleum processing, coking and nuclear fuel processing industry Chemical materials and chemical products manufacturing Pharmaceutical industry Chemical fiber manufacturing industry Rubber products industry Plastics industry Non-metallic mineral products industry Ferrous metal smelting and rolling processing industry Non-ferrous metal smelting and rolling processing industry Fabricated metal products General equipment manufacturing Special equipment manufacturing Transport equipment manufacturing

Source: NBS, Beijing, 2001–2007.

Number of Firms

4 d20

1,225 562 512 53 2,127 1,355

0.25 0.26 0.28 0.34 0.25 0.26

632

0.25

286

0.29

259 964 657

0.26 0.25 0.27

449

0.24

159

0.26

2,322

0.24

868 92 315 1,203 2,653 415

0.29 0.35 0.26 0.25 0.23 0.24

316

0.28

1,198 1,894 857 1,401

0.24 0.25 0.29 0.25

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

165

the list of industries and the number of firms in each. We can observe from Table 8.1 that there exists a significant variation in the number of firms across different industries. The non-metallic mineral products industry has the biggest number of firms (2,653), which is almost 29 times that of the chemical fiber manufacturing industry. The significant variation in the number of firms also hints at the necessity of investigating the distributions of firm size and growth rate at a disaggregated (sectoral) level. We measure the firm size by the number of employees. One advantage of measuring the firm size as the number of employees is that it is naturally immune to inflation, and we can therefore avoid any possible distortion introduced by deflating the monetary value of the firm size.

8.3. The Chinese Manufacturing Sector and the Five-year Planning Procedure In this section, we present a brief picture of the Chinese manufacturing sector and discuss China’s five-year planning procedure, which lays down background knowledge for later empirical exercises. Figure 8.1 presents the growth pattern of the Chinese manufacturing sector from 1979 to 2006. Three characteristics can be observed from

Figure 8.1. The Growth Pattern of the Chinese Manufacturing Sector. Note: The growth rate is calculated using the value of output in constant price. Source: China Statistical Yearbook, http://www.stats.gov.cn/tjsj/ndsj/2008/indexeh.htm, China Statistics Press (2008).

March 5, 2013

166

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Figure 8.1. First, since 1979 the Chinese manufacturing sector has maintained a positive growth trend with the average growth rate as high as 11.6%. Nevertheless, there appears to be significant variation in the growth rate across the period. The lowest growth rate is 1.7% in 1981, and it reaches a peak of 21.2% in 1992. Second, there exists a structural break around 1995. Even though the average growth rate appears to fluctuate around 10% both before and after 1995, it is clear that before 1995 there exists much bigger fluctuation, while in contrast after 1995 the trend is more stable, suggesting that the manufacturing sector may have entered its stationary growth process since 1995. Third, despite the growth rate exhibiting significant fluctuation, the contribution of the manufacturing sector to GDP has been remarkably constant since 1979, with an average of around 40%. Since 1953, the Chinese government has maintained a tradition of establishing a five-year plan, which sets the directions for national development in the next five years. The five-year plans are key indicators of the direction and change in development philosophy (Fan, 2006), and therefore we expect them to have a significant impact on industrial dynamics. The period from 2001 to 2005 is the 10th five-year plan, and since 2006, China has started to implement its 11th five-year plan. Regarding industrial policy, the 10th five-year plan aims to (1) speed up industrial reorganization and transformation; (2) develop high-tech industries; (3) optimize the enterprise organization structure; (4) promote the restructuring of old industrial bases; and (5) promote industrialization, utilizing information technology. In contrast, the 11th five-year plan has changed its goals to (1) enhance the electronic manufacturing industry; (2) cultivate bio-industries; (3) promote the aerospace industries; and (4) develop new materials manufacturing industries. Since our data cover both the 10th and 11th five year plans, we are able to examine the impact of the policy regime shift on firm dynamics.

8.4. Aggregate Picture In this section, we analyze firm dynamics over pooled data across all sectors from three aspects: First we examine the distributions of firm size across seven years, studying their stationarity and shape; second we explore the growth process of firm size, namely the autoregressive structure of firm size, where we explicitly consider the role of shifts in the industrial policy

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

167

regime; and third, we investigate the growth rate distributions across the seven years.

8.4.1. Size distribution We first normalize the logarithm of the number of employees to eliminate the common trend, according to the following formula: si (t) = log(Si (t)) −

N 1
log(Si (t)), N i=1

(8.1)

where S denotes the non-normalized firm size, s denotes the normalized firm size, and N is the total number of firms. Then we calculate the standard deviation, skewness, and kurtosis of the firm size distribution by years (see Table 8.2), and provide the kernel density estimate of firm size in the seven years (see Figure 8.2). Three properties can be characterized from Table 8.2 and Figure 8.2. First, the distribution is remarkably stationary from 2001 to 2007. The standard deviation, skewness, and kurtosis exhibit little variation across the seven years, which is further confirmed by Figure 8.2. Second, the positive values of skewness indicate that the distributions are right skewed, consistent with the finding of many previous studies, for example Hart and Prais (1956), Ijiri and Simon (1977), Bottazzi et al. (2007), and Bottazzi et al. (2009). Third, the positive and big values of kurtosis indicate that the distributions have fatter tails than those of Gaussian distributions. Table 8.2. Descriptive Statistics of Normalized Firm Size by Years. Year

Std. Dev.

Skewness

Kurtosis

2001 2002 2003 2004 2005 2006 2007

1.11 1.11 1.11 1.11 1.12 1.13 1.14

0.60 0.58 0.57 0.54 0.56 0.57 0.55

3.92 3.86 3.79 3.73 3.77 3.71 3.68

Source: Author’s own compilation.

March 5, 2013

168

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

0

.1

Density .2

.3

.4

Kernel density estimate

-4

-2

0 2 normalized log firm size 2002

4

2004

6 2006

kernel = epanechnikov, bandwidth = 0.1280

Figure 8.2. The Distribution of Firm Size by Year. Source: Author’s own estimation.

8.4.2. Growth process To investigate the growth process of firm size, we set up an AR(1) model where the error term is first order autocorrelated, as follows: si,t = β1 si,t−1 + β2 d115i,t + εi,t ,

(8.2)

εi,t = ρεi,t−1 + ui,t ,

(8.3)

where the subscript i and t denotes firm and year respectively; s denotes the normalized firm size; d115 is a dummy variable that takes a value of one if a firm is in the 11th five-year plan period, namely 2006 and 2007, and it therefore captures the shift of the industrial policy regime; ε is the first order autocorrelated error term; and u is an independently and identically distributed error term. The three coefficients, β1 , β2 , and ρ, are the interest of the regression. If β1 = 1, then the growth rate is independent of firm size, namely Gibrat’s Law (Law of Proportionate Effect) holds. If β1 < 1, smaller firms grow faster than bigger firms, and if β1 > 1, bigger firms grow faster than smaller firms. The coefficient β2 captures the impact of the regime shift, with a positive value indicating that the 11th five-year plan promotes firm growth. A significant estimate of the coefficient ρ indicates the existence of

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

169

autocorrelation, which, if not controlled, leads to a biased and inconsistent estimate of the coefficients β1 and β2 . Therefore, Gibrat’s Law cannot be properly tested without considering the autocorrelation in the error term (Chesher, 1979). Equations (8.2) and (8.3) can be transformed into: si,t = γ1 si,t−1 + γ2 si,t−2 + γ3 d115i,t + γ4 d115i,t−1 + ui,t ,

(8.4)

where γ1 = β1 + ρ, γ2 = −β1 ρ, γ3 = β2 , and γ4 = −β2 ρ. We estimate Equation (8.4) over the pooled data, using the ordinary least square (OLS) estimator with robust standard errors computed to account for the possible heteroskedasticity in the error term u. From the estimate (8.4),   of Equation  we can compute the β1 , β2 , and ρ, as follows: β1 = 21 γ1 + γ12 + 4γ2 ,    β2 = γ3 , and ρ = 21 γ1 − γ12 + 4γ2 . Linearize them at the estimated γ1 and γ2 using Taylor’s expansion, and we can obtain their standard errors, as follows:   2  1 4  + γˆ 1 (γˆ 2 + 4γˆ 2 )− 21 σ 2 + σ2 1 γ1  2 2 γˆ 1 + 4γˆ 2 γ2  

 − 1 σβ1 =  4 21 + γˆ 1 γˆ 12 + 4γˆ 2 2   + σγ1γ2 ,   γˆ 12 + 4γˆ 2 σβ2 = σγ3 , and    1  1 2 4 2  − γˆ 1 γˆ 2 + 4γˆ 2 − 2 σ 2 + σγ2 1 γ1  2 2 γ ˆ + 4 γ ˆ 2  1 

 2 − 21 σρ =  1 4 2 + γˆ 1 γˆ 1 + 4γˆ 2   − σγ1γ2 ,   2 γˆ 1 + 4γˆ 2 where σ denotes the standard error; γˆ 1 and γˆ 2 are the estimated γ1 and γ2 from Equation (8.4). Table 8.3 reports the regression results. The estimated coefficient β1 is 0.9733 with a standard error of 0.0036, and therefore we cannot reject the

March 5, 2013

170

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Table 8.3. Regression Results of Equations (8.2) and (8.3).

β1 β2 ρ

Coefficient

Robust Std. Err.

t

0.9733 −1.11 × 10−7 −0.1550

0.0036 0.0019 0.0097

267.8406 −5.8 × 10−5 −16.0163

Source: Author’s own estimation.

null hypothesis that β1 = 1. Thus Gibrat’s Law holds in the Chinese manufacturing sector. The shift in the industrial policy regime does not appear to significantly affect the growth process as its estimated coefficient is insignificant. The estimate of the ρ coefficient is significantly negative, indicating that Chinese firms experience a negative growth rate autocorrelation with a magnitude of 16%.

8.4.3. Growth rate distribution Following Bottazzi et al. (2007), the growth rate is defined as the first difference of the normalized logarithm of the firm size: gi,t = si,t − si,t−1 .

(8.5)

Figure 8.3 presents the kernel density estimate of the distributions of growth rate in 2002, 2004, and 2006. A characteristic tent shape can be observed from Figure 8.2, confirming the finding of Bottazzi et al. (2007) and Bottazzi et al. (2009). Then we applied the Subbotin family of densities (Subbotin, 1923) to estimate the parameters of the distributions. The Subbotin family of densities is flexible with both the Gaussian and Laplace distributions as its special cases, and has been previously applied by Bottazzi and his colleagues to study the distributions of firm growth rate in the US (Bottazzi and Secchi, 2003), Italy (Bottazzi et al., 2007), and France (Bottazzi et al., 2009). The Subbotin family of densities has the following functional form: b 1 − 1 x−µ f(x) = (8.6) 1 e b| a | , 1 2ab b  b + 1 where (·) is the Gamma function. The distribution is characterized by three parameters: the positioning parameter (mean) µ, which due to

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

171

0

2

Density 4

6

8

Kernel density estimate

-5

0 g 2002

5 2004

2006

kernel = epanechnikov, bandwidth = 0.0145

Figure 8.3. The Distribution of Growth Rate by Year. Source: Author’s own estimation.

the normalization in Equation (8.1) is equal to zero in our case, the shape parameter b, and the dispersion parameter a. The smaller the shape parameter b is, the fatter the tail of the distribution becomes. If b = 1, the distribution reduces to the Laplace distribution, and if b = 2, a Gaussian distribution is obtained. As b → ∞, the distribution becomes a uniform distribution. The values of the shape and dispersion parameters are estimated using the maximum likelihood estimation procedure described in Bottazzi and Secchi (2006). Table 8.4 reports the Subbotin estimation results. We can observe a robust pattern regarding the shape parameter b from Table 8.4, namely they are all significantly less than one, which indicates that the growth rate distribution has a fatter tail than that of the Laplace distribution. This pattern differs from the finding of a close-to-Laplace distribution of growth rate by Amaral et al. (1997), Bottazzi and Secchi (2003) in the US, and Bottazzi et al. (2007) in Italy, but is consistent with the finding of Bottazzi et al. (2009) in France. Compared with that of Bottazzi et al. (2009), the estimate of the shape parameter b is smaller here. Therefore, Chinese firms are more likely

March 5, 2013

172

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Table 8.4. Subbotin Estimation over Pooled Data. Shape Parameter

Dispersion Parameter

Year

b

Std. Err.

a

Std. Err.

2001 2002 2003 2004 2005 2006 2007

0.5802 0.5605 0.6133 0.44 0.5409 0.4914 0.5338

0.0065 0.0062 0.0069 0.0047 0.0060 0.0053 0.0059

0.1295 0.1295 0.1339 0.1264 0.1284 0.1027 0.1164

0.0013 0.0013 0.0013 0.0014 0.0013 0.0011 0.0012

Source: Author’s own estimation.

to experience significantly positive or negative growth than firms in the US, Italy, and France, which may owe to the impact of distinct institution in China. Later in the sectoral analysis, we will explore the impact of a shift in the industrial policy regime on the estimated shape and dispersion parameters.

8.5. Sectoral Analysis Given the Bottazzi and Secchi (2003) critique, we now perform similar analysis at the sectoral level, namely we explore firm size distribution, sectoral concentration, sectoral growth process, and sectoral growth rate distributions by two-digit industries. Here we focus on two aspects: First, whether the properties of industrial dynamics found at the aggregate level continue to hold at the disaggregated level; second, what is the role of the shift in the industrial policy regime from 2005 to 2006 in industrial dynamics. To start our exercise, we first normalize the logarithm firm size, as: Nj 1
sij (t) = log(Sij (t)) − log(Sij (t)), Nj i=1

(8.7)

where the subscript j denotes the sector (two-digit industries). Accordingly, the growth rate is defined as: gij,t = sij,t − sij,t−1 .

(8.8)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

173

0

.1

Density .2

.3

.4

Kernel density estimate

-4

-2

0 2 normalized log firm size 2002

2004

4

6

2006

Source: Enterprise Data, NBS, Beijing, 2001-2007

Figure 8.4. The Distribution of Firm Size in the Textile Sector. Source: Author’s own estimation from Enterprise Data, NBS, Beijing, 2001–2007.

8.5.1. Size distribution We build kernel density estimates of firm size distributions in some exemplary sectors, using the observations normalized according to Equation (8.7). Figures 8.4, 8.5, and 8.6 present the density estimations in the textiles industry (17), the chemical materials and chemical products manufacturing industry (26), and the non-metallic mineral products industry (31) in 2002, 2004, and 2006, respectively. Little variation is observed across the three years in the three sectors, suggesting a stationarity of the distributions. The shape of these distributions also does not differ significantly from that of the aggregate picture (see Figure 8.2). Therefore it may be safe to conclude that the shape of firm size distribution survives the disaggregation process, in contrast to the findings of Bottazzi et al. (2009).

8.5.2. Sectoral concentration Now let us turn to examine the upper tail behavior of the size distributions, using the concentration statistic introduced by Bottazzi et al. (2007),

March 5, 2013

174

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

0

.1

Density .2

.3

.4

Kernel density estimate

-4

-2

0 2 normalized log firm size 2002

4

2004

6

2006

Source: Enterprise Data, NBS, Beijing, 2001-2007

Figure 8.5. The Distribution of Firm Size in the Chemical Materials and Chemical Products Sector. Source: Author’s own estimation from Enterprise Data, NBS, Beijing, 2001–2007.

0

.1

Density .3 .2

.4

.5

Kernel density estimate

-4

-2

0 normalized log firm size 2002

2004

2

4 2006

Source: Enterprise Data, NBS, Beijing, 2001-2007

Figure 8.6. The Distribution of Firm Size in the Non-metallic Mineral Products Sector. Source: Author’s own estimation from Enterprise Data, NBS, Beijing, 2001–2007.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

175

as follows: 4 (t) = d20

C4 C20

t = 2001, . . . , 2007,

(8.9)

where C4 and C20 are the sum of market share of the top 4 and 20 firms in a sector respectively, and in our analysis the market share is calculated as the percentage of number of employees in the sector. One advantage of the 4 concentration statistic d20 (t) is that it overcomes the drawback arising from lack of data covering the whole sector (Bottazzi et al., 2007). The statistic is ranged between 0.2 and 1, and a bigger value indicates a higher degree of market concentration. As the concentration statistic of Equation (8.9) is calculated in each year for each sector, the average value of the concentration statistic is then calculated to obtain a more robust characterization of the upper tail, as follows: 4 d20

2007 1
4 = d (t). 7 t=2001 20

(8.10)

Table 8.1 also reports the calculated concentration statistic. Examining Table 8.1, we can find only little variation across sectors, with an average of 0.27, standard deviation of 0.03, minimum value of 0.23, and maximum value of 0.39. This further confirms our previous conclusion that the firm size distribution survives the disaggregation process.

8.5.3. Sectoral growth process The AR(1) model in the above (Equations (8.2) and (8.3)) is re-estimated at the disaggregated level over the seven years to investigate the sectoral growth process. The results are reported in Table 8.5. Three distinct features emerge from Table 8.5. First, the estimated coefficients of the lagged firm size (β1 ) are all highly significant, and their magnitude is close to one. The t tests for the null hypothesis of β1 = 1 reject the null in 9 out of the total 25 sectors at the 5% significance level. Therefore, the Gibrat’s Law holds in the majority of sectors, even though there exists substantial variation across sectors. The second feature is that, same as that of the aggregate regression, the coefficients of d115(β2 ), the dummy variable that captures the shift in the industrial policy regime, are all highly insignificant. Therefore, we conclude that the shift in the policy regime does not significantly affect the firm

Coefficient

0.0126 76.5954 −2.5377 −3.01 × 10−7 0.0190 50.8068 −1.9436 −1.10 × 10−7 0.0165 59.2782 −1.4324 5.07 × 10−8 0.0312 31.8528 −0.1728 1.00 × 10−8 0.0092 105.1979 −3.0009 −3.19 × 10−8 0.0123 78.3452 −3.2299 2.64 × 10−7 0.0172 56.8621 −1.3781 −1.74 × 10−7 0.0249 38.1682 −1.9824 7.39 × 10−8 0.0285 34.3381 −0.7862 1.91 × 10−7 0.0159 61.0410 −1.6971 −2.52 × 10−9 0.0190 50.9778 −1.6643 1.29 × 10−7 0.0205 47.8664 −0.9622 −4.66 × 10−8 0.0323 30.5463 −0.3768 −3.04 × 10−7 0.0092 106.3359 −2.4273 −1.06 × 10−7 0.0159 61.0544 −1.7502 −2.63 × 10−7

Robust Std. Err.

t

0.0101 0.0127 0.0141 0.0383 0.0059 0.0089 0.0126 0.0168 0.0221 0.0092 0.0084 0.0157 0.0197 0.0057 0.0094

−3 × 10−5 −8.64 × 10−6 3.60 × 10−6 2.62 × 10−7 −5.39 × 10−6 2.98 × 10−5 −1.4 × 10−5 4.41 × 10−6 8.61 × 10−6 −2.74 × 10−7 1.53 × 10−5 −2.97 × 10−6 −1.5 × 10−5 −1.9 × 10−5 −2.8 × 10−5

Robust Coefficient Std. Err. −0.2129 −0.1770 −0.1856 0.0022 −0.1224 −0.1950 −0.1761 −0.1375 −0.1489 −0.1608 −0.1425 −0.0910 −0.2283 −0.1429 −0.1502

0.0328 0.0456 0.0481 0.0771 0.0250 0.0300 0.0473 0.0610 0.0671 0.0432 0.0560 0.0520 0.0935 0.0257 0.0424

t −6.4932 −3.8830 −3.8537 0.0283 −4.9000 −6.5010 −3.7255 −2.2544 −2.2188 −3.7243 −2.5453 −1.7490 −2.4426 −5.5671 −3.5447

(Continued)

Regional Development and Economic Growth. . . b1491-ch08

0.9679 0.9632 0.9764 0.9946 0.9723 0.9604 0.9763 0.9506 0.9776 0.9729 0.9684 0.9803 0.9878 0.9777 0.9721

t(β1 = 1)

9in x 6in

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

t

ρ

11:55

Robust Sectors Coefficient Std. Err.

β2

Regional Development and Economic Growth in China

β1

March 5, 2013

176

Table 8.5. Sectoral Growth Process Regression Results.

March 5, 2013 11:55

Table 8.5. (Continued)

0.9710 0.9877 0.9667 0.9617 0.9850 0.9772 0.9764 0.9645 0.9644 0.9722

t(β1 = 1)

Coefficient

0.0642 15.1316 −0.4524 1.34 × 10−8 0.0274 35.9889 −0.4496 −3.31 × 10−7 0.0158 61.3125 −2.1104 −7.71 × 10−8 0.0090 107.4286 −4.2739 2.55 × 10−7 0.0222 44.4067 −0.6761 3.35 × 10−7 0.0295 33.1316 −0.7721 3.07 × 10−7 0.0134 72.9626 −1.7642 −2.56 × 10−7 0.0135 71.6180 −2.6389 2.19 × 10−8 0.0138 69.8740 −2.5814 −1.39 × 10−7 0.0166 58.7268 −1.6781 1.01 × 10−7

Robust Std. Err.

t

0.0228 5.85 × 10−7 0.0179 −1.8 × 10−5 0.0082 −9.45 × 10−6 0.0055 4.59 × 10−5 0.0144 2.33 × 10−5 0.0177 1.73 × 10−5 0.0084 −3.1 × 10−5 0.0057 3.85 × 10−6 0.0093 −1.5 × 10−5 0.0069 1.45 × 10−5

Robust Coefficient Std. Err. −0.0562 −0.1210 −0.1193 −0.1524 −0.1395 −0.1943 −0.1889 −0.1774 −0.1447 −0.0725

Note: The sector names can be found in Table 8.1; t(β1 = 1) column is the t statistic to test whether β1 = 1. Source: Author’s own estimation.

0.1790 0.0733 0.0411 0.0225 0.0653 0.0845 0.0357 0.0396 0.0382 0.0465

t −0.3137 −1.6499 −2.9036 −6.7837 −2.1366 −2.3004 −5.2930 −4.4740 −3.7928 −1.5572

177

Regional Development and Economic Growth. . . b1491-ch08

28 29 30 31 32 33 34 35 36 37

t

ρ Patterns of Industrial Dynamics in the Manufacturing Sector

Robust Sectors Coefficient Std. Err.

β2

9in x 6in

β1

March 5, 2013

178

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

growth process at both the aggregate and disaggregate levels. Third, firms experience a negative growth rate autocorrelation in most of the sectors. Only four sectors have an insignificant estimate of the ρ value. Comparing the results of aggregated and disaggregated regressions, it appears that the disaggregation process does not significantly alter the results.

8.5.4. Sectoral growth rate distribution Similar to before, we also fit the Subbotin distribution (Equation (8.6)) to the firm growth rates in the 25 sectors across the seven years. Table 8.6 reports the parameter estimates of 2002, 2004 and 2006, where the rest of the years are not reported to save space, and instead we report the descriptive statistics of the estimated parameters (a and b) in Table 8.7. Three characteristics can be observed from both Tables 8.6 and 8.7. First, most of the estimated shape parameters are consistently less than one, suggesting a fatter tail than that of the Laplace distribution. Even though some sectors have an estimated shape parameter (b) close to one, for example the cultural, educational and sporting goods manufacturing industry (24) in 2002, where the shape parameter is estimated to be 0.8335, almost all of the t tests reject the null hypothesis of being equal to one. There are only five cases where the null hypothesis of the shape parameter being equal to one cannot be rejected at the 5% significance level, which are the chemical fiber manufacturing industry (28) in 2003, the cultural, educational and sporting goods manufacturing industry (24) in 2007 and the tobacco manufacturing industry (16) in 2004, 2005, and 2006. Therefore, it is reasonable to conclude that the distribution of the firm growth rate in China has a fatter tail than that of the Laplace distribution, and this finding holds both at the aggregate and disaggregated levels. Second, there exists significant sectoral-level heterogeneity. For example, in 2001, the maximum value of the shape parameter b is estimated to be 0.7959, nearly three times that of the minimum value, and the standard deviation is also as big as 21% of the mean. Similar patterns exist for the dispersion parameter a as well. Third, the estimated parameters exhibit little variation in the time dimension, suggesting the stationarity of growth rate distributions. Up to now, the reader may want to ask: What is the relation between the growth rate distribution parameters and the average firm size? What is the role of the shift of industrial policy regime discussed previously? To explore

March 5, 2013

2004

2006

Std. Err

a

Std. Err

b

Std. Err

a

Std. Err

b

Std. Err

a

Std. Err

13 14 15 16 17 18 19 20 21 22 23 24 25 26 27

0.4386 0.5786 0.4670 0.5991 0.5930 0.5741 0.6701 0.5921 0.7691 0.5432 0.6049 0.8335 0.5150 0.4567 0.5705

0.0203 0.0410 0.0337 0.1390 0.0217 0.0262 0.0458 0.0590 0.0842 0.0291 0.0399 0.0704 0.0675 0.0154 0.0325

0.1375 0.1709 0.1117 0.0929 0.1290 0.1489 0.1772 0.1458 0.2481 0.1235 0.1151 0.2039 0.1199 0.0968 0.1290

0.0066 0.0110 0.0081 0.0192 0.0042 0.0062 0.0102 0.0130 0.0212 0.0062 0.0067 0.0129 0.0151 0.0033 0.0067

0.4569 0.4997 0.3970 0.7134 0.5788 0.5490 0.5691 0.4277 0.7273 0.4962 0.5051 0.7456 0.4104 0.5271 0.4263

0.0212 0.0347 0.0281 0.1703 0.0211 0.0249 0.0379 0.0408 0.0788 0.0263 0.0325 0.0616 0.0522 0.0181 0.0233

0.1598 0.1689 0.1169 0.1630 0.1639 0.1625 0.1722 0.1349 0.2119 0.1223 0.1063 0.2110 0.1288 0.1266 0.1099

0.0076 0.0114 0.0091 0.0316 0.0054 0.0068 0.0105 0.0136 0.0185 0.0063 0.0066 0.0139 0.0177 0.0041 0.0064

0.4564 0.5294 0.3561 0.7425 0.5854 0.4619 0.5133 0.5862 0.5433 0.4580 0.5406 0.7210 0.6138 0.4245 0.5723

0.0212 0.0370 0.0249 0.1785 0.0214 0.0204 0.0337 0.0583 0.0562 0.0240 0.0351 0.0593 0.0825 0.0142 0.0326

0.1221 0.1231 0.0791 0.1275 0.1104 0.1159 0.1296 0.1283 0.1443 0.0921 0.0828 0.1696 0.1053 0.0828 0.1109

0.0058 0.0081 0.0064 0.0244 0.0036 0.0052 0.0082 0.0115 0.0139 0.0049 0.0050 0.0113 0.0124 0.0029 0.0057

(Continued)

179

Regional Development and Economic Growth. . . b1491-ch08

b

Patterns of Industrial Dynamics in the Manufacturing Sector

Sectors

9in x 6in

2002

11:55

Table 8.6. Subbotin Estimation at the Disaggregate Level.

b

Std. Err

a

Std. Err

b

Std. Err

a

Std. Err

b

Std. Err

a

Std. Err

28 29 30 31 32 33 34 35 36 37

0.5662 0.5898 0.7222 0.2907 0.3918 0.4312 0.6186 0.5685 0.5088 0.7707

0.0988 0.0560 0.0363 0.0087 0.0307 0.0392 0.0303 0.0219 0.0287 0.0363

0.1108 0.1293 0.1618 0.0676 0.1027 0.0989 0.1673 0.1066 0.0984 0.1615

0.0177 0.0110 0.0066 0.0026 0.0089 0.0094 0.0072 0.0037 0.0054 0.0059

0.6474 0.4376 0.5667 0.5017 0.4276 0.4418 0.5857 0.4865 0.5580 0.5330

0.1154 0.0399 0.0274 0.0160 0.0338 0.0402 0.0285 0.0183 0.0318 0.0236

0.1790 0.1098 0.1534 0.1304 0.1274 0.1135 0.1667 0.1329 0.1537 0.1323

0.0273 0.0104 0.0068 0.0041 0.0107 0.0107 0.0073 0.0049 0.0081 0.0055

0.6884 0.6092 0.5215 0.5341 0.4745 0.5554 0.3460 0.5133 0.4360 0.5736

0.1240 0.0581 0.0249 0.0172 0.0381 0.0522 0.0158 0.0195 0.0241 0.0257

0.0995 0.1235 0.1108 0.1094 0.0968 0.1155 0.0826 0.0895 0.0881 0.1120

0.0148 0.0104 0.0050 0.0033 0.0078 0.0100 0.0044 0.0033 0.0051 0.0046

Note: a is the dispersion parameter, and b is the shape parameter; all estimates are significant at the 5% significance level. Source: Author’s own estimation.

Regional Development and Economic Growth. . . b1491-ch08

Sectors

9in x 6in

2006

11:55

2004

Regional Development and Economic Growth in China

2002

March 5, 2013

180

Table 8.6. (Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

181

Table 8.7. Summary of the Estimate of the Subbotin Parameters. Year

Parameters

Mean

Std. Dev.

Min

Max

2001

a b

0.1426 0.5747

0.0388 0.1193

0.0832 0.2894

0.2424 0.7959

2002

a b

0.1342 0.5706

0.0396 0.1234

0.0676 0.2907

0.2481 0.8335

2003

a b

0.1409 0.6315

0.0300 0.1048

0.0895 0.4116

0.2016 0.7751

2004

a b

0.1463 0.5286

0.0295 0.0985

0.1063 0.3970

0.2119 0.7456

2005

a b

0.1309 0.5660

0.0245 0.1098

0.0766 0.2784

0.1793 0.8903

2006

a b

0.1101 0.5343

0.0215 0.0986

0.0791 0.3460

0.1696 0.7425

2007

a b

0.1163 0.5250

0.0389 0.1479

0.0397 0.1481

0.2029 0.8792

Note: a is the dispersion parameter, and b is the shape parameter. Source: Author’s own estimation.

these two questions, we set up the following panel data model: yit = λ0 + λ1 afsit + λ2 d115it + φi + uit ,

(8.11)

where the subscript i and t denote sector and year respectively; y is the dependent variable and y ∈ {a, σa , b, σb }; afs denotes the average logarithm of the firm size; d115 is the dummy variable that captures the shift of industrial policy regime, and takes a value of one if a sector is in the 11th five-year plan period (2006 and 2007) and zero otherwise; φ is the industry fixed effect; and u is the i.i.d. error term. We use the fixed effect estimator to estimate Equation (8.11), with the standard error clustered by sectors to account for possible heteroskedasticity and serial correlation. Table 8.8 presents the regression results, where the top panel reports the regression results for the shape parameter and its standard error and the bottom panel is those of the dispersion parameter and its standard error.

March 5, 2013

182

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Table 8.8. Determinants of Firm Growth Rate Distributions. Coef.

Robust Std. Err.

t

Coef.

b afs −0.4178 d115 −0.0389 constant 2.8449 No. of obs. 175 F 8.82 R2 0.1

0.2057 0.0167 1.1163

0.0456 0.0032 0.2474

t

σb −2.03 −0.0097∗ −2.33 −0.0033 2.55 0.0983∗ 175 4.93 0.3

a afs −0.1631 d115 −0.0236 constant 1.0256 No. of obs. 175 F 58.85 R2 0.004

Robust Std. Err.

0.0287 0.0016 0.1559

−0.34 −2.1 0.63

σa −3.58 −0.0077 −7.41 −0.0014 4.15 0.0517 175 25.45 0.23

0.0021 0.0003 0.0112

−3.74 −4.75 4.6

Note: afs is the average firm size; and d115 denotes the policy regime dummy; All estimates, except those marked with ∗ , are significant at the 5% significance level. Source: Author’s own estimation.

In contrast to the finding of no relationship between the two Subbotin parameters and average firm size by Bottazzi et al. (2007), it is clear from Table 8.8 that a bigger sector (namely with a bigger average firm size) tends to have a fatter tail (smaller shape parameter b) and smaller variance (smaller dispersion parameter a) in its growth rate distribution. The average firm size appears not to significantly affect the standard error of estimated shape parameter b, but it negatively affects the standard error of the dispersion parameter a. Regarding the role of the shift in industrial policy regime, it exerts a similar impact as that of the average firm size. After entering the 11th five-year plan period (2006 and 2007), the shape parameter b is 0.0389 smaller than that of the 10th five-year plan period (from 2001 to 2005), indicating that the 11th five-year plan causes the growth rate distribution to have a fatter tail, however the dispersion of the growth rate distribution appears to be reduced as well. The 11th five-year plan also appears

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

183

to significantly reduce the estimated standard error of the two Subbotin parameters. Nevertheless it may be possible that the policy regime shift is endogenous. Even though it is unlikely that the Chinese government will make the policy by considering the growth rate distributions, it is still possible that the policy making procedure may be correlated with some factors that at the same time affect the distributions of the growth rate. In this case the endogeneity will exist and the estimates of the coefficients will be biased and inconsistent. To address this concern, we apply an instrumental variable fixed effect estimator, using the lagged d115 as the instrument, and test the endogeneity of d115, which fails to reject the null hypothesis that the policy regime shift (d115) is exogenous. Therefore, our previous finding is valid and robust. It is interesting to observe that a bigger sector tends to have a growth rate distribution with a fatter tail and smaller dispersion (the big sector effect) and the policy regime shift also exerts a similar impact (the policy regime shift effect). The big sector effect may be due to the fiercer competition that big sectors tend to have. The market competition on the one hand drives firms to grow at an average rate (smaller dispersion of the distribution), and on the other hand gives top firms opportunities to grow exceptionally fast and causes bottom firms to shrink quickly as well, creating a fatter tail in the distribution. The policy regime shift from the 10th five-year plan to the 11th five-year plan induces a similar effect, which causes more firms to grow at an average rate and at the same time allows good firms to expand quickly and bad firms to shrink quickly as well.

8.6. Conclusion Using a comprehensive longitudinal dataset from China, we investigate three dimensions of industrial dynamics in this chapter, namely firm size distribution, firm growth process (the Gibrat’s Law and the growth rate autocorrelation), and the distribution of growth rates. Firm size distributions appear to be stationary across time and robust to the aggregation and disaggregation procedures, even though they deviate significantly from the Gaussian distribution. Since a number of previous studies, for example Bottazzi and Secchi (2003) and Bottazzi et al. (2009),

March 5, 2013

184

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

have revealed an inconsistency of firm size distributions between the aggregate and disaggregated levels, the robustness of the firm size distribution to the aggregation and disaggregation procedures found here is worthy of noting as it highlights the peculiarity of the Chinese manufacturing sector. Compared with these previous studies, the Chinese manufacturing sector appears to be relatively more homogenous. The second dimension we investigate is the firm growth process. In tune with many previous studies, Gibrat’s Law, at least in its weakest form, appears to hold across both the aggregate and disaggregated levels. Besides, the shift of the industrial policy regime from the 10th five-year plan period to the 11th five-year plan period does not significantly affect the growth process, suggesting that firm growth is more governed by its internal mechanism. Most Chinese firms experience a negative autocorrelation in the growth process, indicating a mean-reversion anti-persistent tendency, namely a sort of “success brings failure” dynamic. As argued by Bottazzi et al. (2009), the mean reversion dynamic suggests two scenarios: First the relative advantages of firms lie in such ephemeral factors such as marketing strategies, oligopolistic market power, short-term commercial agreements, and scarcely appropriable innovation; second, the “success brings failure” dynamic may be due to the rapid adaptation of competitors through the efficient imitation and subsequent innovation. In the Chinese context, both scenarios may occur. On the one hand, firms may have a certain degree of market power and are engaged in short-term behaviors; and on the other hand, their competitors at the same time adapt themselves rapidly, which adversely affects firm growth. The imitation behavior, particularly domestic firms imitating the behavior of firms invested by the foreign direct investment, is widely observed. A number of previous studies have revealed that domestic firms benefit from the presence of foreign direct investment in China, for example Li et al. (2001), Liu (2002, 2008), and Sun (2009). The distribution of growth rates exhibit a typical tent shape, again confirming many previous studies. We discover much fatter tails in the growth rate distributions, compared with previous studies in the US (Bottazzi and Secchi, 2003), Italy (Bottazzi et al., 2007), and France (Bottazzi et al., 2009), suggesting another peculiarity of the Chinese manufacturing sector. In addition, we link the parameters of the growth rate distribution to the average firm size in the sector and the shift of industrial policy regime, and

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

185

find a “big sector effect” and “policy regime shift effect”. The “big sector effect” states that bigger sectors tend to have growth rate distributions with a fatter tail but smaller dispersion, which arises possibly due to the market competition. The “policy regime shift effect” is similar to the “big sector effect”. The shift of the industrial policy regime from the 10th five-year plan period to the 11th five-year plan period is pro-competitive in the sense that it induces most firms to grow at an average rate and while allowing good firms to expand quickly and bad firms to shrink rapidly.

References Amaral, L. A. N., Buldyrev, S. V., Havlin, S., Salinger, M. A., Stanley, H. E. and Stanley, M. H. R., “Scaling Behavior in Economics: The Problem of Quantifying Company Growth,” Physica A, 244: 1–24 (1997). Axtell, R. L., “Zipf Distribution of U.S. Firm Sizes,” Science, 293: 1818–1820 (2001). Bottazzi, G., Cefis, E., Dosi, G. and Secchi, A., “Invariances and Diversities in the Evolution of Italian Manufacturing Industry,” Small Business Economics, 29: 137–159 (2007). Bottazzi, G., Coad,A., Jacoby, N. and Secchi,A., “Corporate Growth and Industrial Dynamics: Evidence from French Manufacturing,” Applied Economics, 43(1): 103 (2009). Bottazzi, G., Dosi, G., Lippi, M., Pammolli, F. and Riccaboni, M., “Innovation and Corporate Growth in the Evolution of the Drug Industry,” International Journal of Industrial Organization, 19: 1161–1187 (2001). Bottazzi, G. and Secchi, A. “Common Properties and Sectoral Specificities in the Dynamics of US Manufacturing Companies,” Review of Industrial Organization, 23: 217–232 (2003). Bottazzi, G. and Secchi, A. “Growth and Diversification Patterns of the Worldwide Pharmaceutical Industry,” Review of Industrial Organization, 26: 195–216 (2005). Bottazzi, G. and Secchi, A. “Explaining the Distribution of Firms Growth Rates,” RAND Journal of Economics, 37: 234–263 (2006). Chen, Y. and Wu, Y., “Regional Economic Growth and Spillover Effects: An Analysis of China’s Pan Pearl River Delta Area,” China & World Economy, 20: 80–97 (2012).

March 5, 2013

186

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Regional Development and Economic Growth in China

Chesher, A., “Testing the Law of Proportionate Effect,” Journal of Industrial Economics, 27: 403–411 (1979). Clarke, R., “On the Lognormality of Firm and Plant Size Distributions: Some UK Evidence,” Applied Economics, 11: 415–433 (1979). Coad, A., “Firm Growth: A Survey,” Papers on Economics and Evolution 2007-03, Max Planck Institute of Economics, Evolutionary Economics Group (2007). De Wit, G., “Firm Size Distributions: An Overview of Steady-State Distributions Resulting from Firm Dynamics Models,” International Journal of Industrial Organization, 23: 423–450 (2005). Dunne, T., Roberts, M. J. and Samuelson, L., “The Growth and Failure of US Manufacturing Plants,” Quarterly Journal of Economics, 104: 671–698 (1988). Evans, D. S., “The Relationship Between Firm Growth, Size andAge: Estimates for 100 Manufacturing Industries,” Journal of Industrial Economics, 35: 567–581 (1987). Fan, C., “China’s Eleventh Five-Year Plan (2006–2010): From Getting Rich First to Common Prosperity,” Eurasian Geography and Economics, 47: 708–723 (2006). Gibrat, R., Les Inégalités Économiques. Paris: Librairie du Recueil Sirey (1931). Goddard, J., McMillan, D. and Wilson, J. O. S., “Do Firm Sizes and Profit Rates Converge? Evidence on Gibrat’s Law and the Persistence of Profits in the Long Run,” Applied Economics, 38: 267–278 (2006). Hall, B., “The Relationship Between Firm Size and Firm Growth in the US Manufacturing Sector,” Journal of Industrial Economics, 35: 583–600 (1987). Hart, P. E., “The Size and Growth of Firms,” Economica, 29: 29–39 (1962). Hart, P. E. and Prais, S. J., “The Analysis of Business Concentration,” Journal of the Royal Statistical Society, 119: 150–191 (1956). Hu, A. G. Z., Jefferson, G. H. and Qian, J., “R&D and Technology Transfer: FirmLevel Evidence from Chinese Industry,” Review of Economics and Statistics, 87: 780–786 (2005). Ijiri, Y. and Simon, H. A., Skew Distributions and the Sizes of Business Firms. Amsterdam: North-Holland (1977). Jefferson, G. H., Thomas, G. R. and Zhang, Y., “Productivity Growth and convergence across China’s Industrial Economy,” Journal of Chinese Economic and Business Studies, 6: 121–140 (2008). Li, X., Liu, X. and Parker, D., “Foreign Direct Investment and Productivity Spillovers in the Chinese Manufacturing Sector,” Economic System, 25: 305–321 (2001).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

Patterns of Industrial Dynamics in the Manufacturing Sector

187

Liu, Z., “Foreign Direct Investment and Technology Spillover: Evidence from China,” Journal of Comparative Economics, 30: 579–602 (2002). Liu, Z., “Foreign Direct Investment and Technology Spillovers: Theory and Evidence,” Journal of Development Economics, 85: 176–193 (2008). Lotti, F., Santarelli, E. and Vivarelli, M., “The Relationship between Size and Growth: The Case of Italian Newborn Firms,” Applied Economics Letters, 8: 451–454 (2001). Mansfield, E., “Entry, Gibrat’s Law, Innovation and the Growth of Firms,” American Economic Review, 52: 1023–1051 (1962). Quandt, R. E., “On the Size Distribution of Firms,” American Economic Review, 56: 416–432 (1966). Silberman, I. H., “On Lognormality as a Summary Measure of Concentration,” American Economic Review, 57: 807–831 (1967). Simon, H. A. and Bonini, C. P., “The Size Distribution of Business Firms,” American Economic Review, 48: 607–617 (1958). Stanley, M., Buldyrev, S., Havlin, S., Mantegna, R., Salinger, M. and Stanley, H. E., “Zipf Plots and the Size Distribution of Firms,” Economics Letters, 49: 453–457 (1995). Stanley, M. H. R., Amaral, L. A. N., Buldyrev, S., Havlin, S., Leschhorn, H., Maass, P., Salinger, M. A. and Stanley, H. E., “Scaling Behavior in the Growth of Companies,” Nature, 379: 804–806 (1996). Steindl, J., Random Processes and the Growth of Firms. London: Griffin (1965). Subbotin, M. T., “On the Law of Frequency of Errors,” Matematicheskii Sbornik, 31: 296–301 (1923). Sun, S., “How Does FDI Affect Domestic Firms Exports? Industrial Evidence,” World Economy, 32: 1203–1222 (2009). Sutton, J., “Gibrat’s Legacy,” Journal of Economic Literature, 35: 40–59 (1997). Wilson, J. O. S. and Williams, J. M., “The Size and Growth of Banks: Evidence from Four European Countries,” Applied Economics, 32: 1101–1109 (2000). Wu, Y., “The Role of Productivity in China’s Growth: New Estimates,” Journal of Chinese Economic and Business Studies, 6: 141–156 (2008a). Wu, Y., Productivity, Efficiency and Economic Growth in China. London UK: Palgrave Macmillan (2008b). Wu, Y., “Innovation and Economic Growth in China: Evidence at the Provincial Level,” Journal of the Asia Pacific Economy, 16: 129–142 (2011).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch08

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Chapter 9

Agglomeration and Export Performance of Manufacturing Firms Dahai Fu

9.1. Introduction The determinants of export participation at the firm level have been extensively investigated since the mid-1990s when Bernard and Jensen (1995) published the first of a series of papers to look at the differences between exporters and non-exporters in the case of the US manufacturing industry. During the years following their pioneering works, both empirical and theoretical studies on the export activities of firms, and its causes and consequences proliferated all over the world.1 One of the key findings is that exporting firms outperform those firms that supply the domestic market only with respect to productivity performance. Most studies point to the self-selection of more productive firms into foreign markets due to the sunk costs related to exporting activities as the most likely explanation for such differentials. While the existing economic literature mainly focuses on the role of firm heterogeneity such as productivity (Roberts and Tybout, 1997), firm size (Wagner, 2001), and innovation activities (Caldera, 2010) in determining the export decisions of firms, scholars from the international business field argue that it is insufficient to understand exporting behavior of a firm from a purely internal perspective. Exporting should be taken as a strategic response by management to the interplay of internal and external forces on firms (Cavusgil and Zou, 1994; Sousa et al., 2008). This chapter contributes to this argument by examining the impact of an agglomeration of exporting 1 See surveys by Greenaway and Kneller (2005) and Wagner (2007).

189

March 5, 2013

190

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

firms as a key external factor on export performance while controlling for firm heterogeneity using a very rich panel data of Chinese manufacturing firms over the period of 1998–2007. Following recent studies on export spillovers like Koenig (2009) and Koenig et al. (2010), this study adds to the literature from the perspective of a large and open economy in the Chinese context. In addition to examining the existence of export spillovers as is done in other works, this study goes further by exploring the curvilinear relationship between exporter clustering and export performance of manufacturing firms, possibly resulting from congestion costs. Moreover, whether or not the export spillovers are restricted to a specific industry and region is also examined. The analyses here are conducted in terms of both export participation and export intensity specifications. To preview some of the main findings of this chapter, it is found that in general agglomeration of exporting firms has had a positive effect on both the likelihood of export participation and the level of export intensity of Chinese manufacturing firms. It is found that this relationship resembles an inverted-U shape. That is, the diseconomies effect of export-agglomeration may appear if the degree of agglomeration of similar firms in the same region is too high. Moreover, export spillovers are likely to be industry and region-specific. The pool of local exporters only benefits firms in the same sector within the same region. The rest of the chapter is organized as follows. Section 9.2 discusses the theoretical background and hypotheses. Section 9.3 documents the econometric model and estimation strategy, while Section 9.4 presents the data and summary statistics. Section 9.5 presents the estimation results and sensitivity analysis and finally, Section 9.6 concludes, summarizing the main findings.

9.2. Theoretical Background and Hypotheses Since the seminal empirical and theoretical contributions of Bernard and Jensen (1995) and Melitz (2003), research focus in international trade has shifted from country and industry levels to the firm level. These studies emphasize the heterogeneity of firms and their critical role in shaping international trade flows. One of the fundamental insights in this strand of research is that more productive firms are more likely to export than

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

191

less productive firms, due to the fixed and variable costs of exporting (Roberts and Tybout, 1997).2 Based on these theories, it can be expected that any aspect of firm heterogeneity and business environment, which could affect the efficiency and trade costs of firms, would influence their exporting behavior. However, for a long time, numerous economic studies only tested the impact of the heterogeneous characteristics of firms, and in particular, productivity on the export decision, without considering the influence of external forces such as export-agglomeration economies on the export performance of firms. In fact, researchers of the international business field have long been aware that in the analysis of export performance, it is necessary to examine external factors that affect a firm’s efficiency, trade cost, as well as exportrelated strategic management, in addition to internal forces such as firm characteristics and product characteristics (Cavusgil and Zou, 1994). Sousa et al. (2008) claimed in a comprehensive review of 52 empirical articles on the determinants of export performance published between 1998 and 2005 that “export performance should be assessed at the two broad levels — the external environment level and the internal level” (Sousa et al., 2008, p. 363). In response to their call, a growing number of recent papers have investigated the effects of internal and external factors simultaneously. For example, Zhao and Zou (2002) and Clougherty and Zhang (2009) investigate the impact of domestic competition on a firm’s export behavior while controlling the firm heterogeneity. Aitken et al. (1997) and Greenaway et al. (2004) examine the effect of FDI inflows and find that the probability of a domestic plant’s exports is positively correlated with proximity to multinational firms. Although the export spillovers from the FDI firms have received much attention, few studies have specifically addressed the importance of having exporters in a firm’s geographic vicinity on the decision to export, and the existing evidence is mixed. Clerides et al. (1998) analyze the existence of spillovers by regressing export participation on regional and industry export intensity. They find that, for Colombia, the presence of many exporters increases a firm’s chances of being an exporter itself.Aitken et al. (1997) find 2 Economists usually relate sunk costs of exporting to the adaptation of products to foreign

customers’ demand and foreign market standards or the establishments of a distribution network abroad.

March 5, 2013

192

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

that general export activity for both domestic exporters and multinational exporters has no effect on the probability of exporting for domestic firms, such as in the case of Mexico. However, the exports from multinationals increase the probability of exporting for other firms. Bernard and Jensen (2004) use a panel of US manufacturing plants to test the role of export spillovers in export decisions and find that there is no evidence for the importance of export activity by other firms in the same industry. Koenig (2009) investigates whether export spillovers are destination specific and finds evidence for destination and industry specific export spillovers. Koenig et al. (2010) go one step further in the disaggregation and find that positive export spillovers are destination-product specific. Positive export spillovers, which in the literature are referred to as the effect of exporters on neighboring firms’ export performance, are always expected because the variable or fixed costs of exports could be reduced via sharing, matching and learning mechanisms (Duranton and Puga, 2003). Exporter clustering favors the creation of pools of specialized workers. This implies that it will be easy for other firms to find the employees of their needs and thus minimize the costs associated with search and training. It thus could increase firm productivity and consequently foreign entry. Additionally, agglomeration effects could result in better infrastructure and specialized service, allowing local firms to share specialized services, save transportation costs, or manage efficiently their purchases of inputs. Moreover, concentration of exporting firms is thought to facilitate the exchange of information and knowledge about foreign markets through face-to-face communication of tacit knowledge, or the mobility of human capital among firms. Yet, the agglomeration of exporting firms could also hinder the international expansions of firms via congestion costs. Such diseconomies of scale can be realized not only through increased traffic and pollution, but also via intense competition in the labor and input markets. One of the stylized facts in international trade is that exporting firms pay higher wages than nonexporting firms, especially to high-skilled workers (Bernard and Jensen, 1999). Therefore, when non-exporting firms decide to start exporting, the export wage premium forces them to poach skilled workers from other nonexporting firms, not from exporting firms. In this situation, non-exporting firms are always exposed to the threat of labor poaching by exporting firms.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

193

As a result, the pool where these non-exporting firms can select skilled workers is dependent on what fraction of local firms are non-exporters. Consequently, agglomeration of exporting firms could have either positive or negative effects on the entry of exporters depending on which ones dominate. This may be the reason that Barrios et al. (2003), Bernard and Jensen (2004) and Lawless (2010) do not find evidence for export spillovers. According to these discussions, it can be hypothesized that the effect of agglomeration on the probability of being an exporter may be non-linear. In addition to the non-linear relationship, the export spillovers may be restricted to the same region and same industry. First, geographic proximity and industrial similarity could be important for transmitting and utilizing knowledge. Second, if the workers cannot move freely from one region or industry to another within a country, the inter-regional or industrial spillovers might be weak. Adams (2002) shows that knowledge spillovers are stronger within a given distance. Xu and Sheng (2011) find that domestic firms benefit more from the presence of foreign firms in the same sector within the same region. In this chapter, the interest lies in testing whether the spillovers from the export neighbors are region and/or industry-specific.

9.3. Econometric Approach In this section, empirical approaches are discussed to investigate the relationship between agglomeration of exporters and export performance, which is measured by their export status and export intensity, respectively. The different measurements of export performance guide us to use different estimation methods.

9.3.1. Agglomeration and export participation Following a similar approach as Roberts and Tybout (1997) and Bernard and Jensen (2004), a firm’s decision to export as a binary choice model is specified using a latent variable framework: yit∗ = xit β + µi + εit ,
 

(9.1)

eit

yit =

1[yit∗

> 0],

(9.2)

March 5, 2013

194

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

where yit denotes the export status of firm i at time t, taking value one when the firm reports positive export sales yit∗ and zero otherwise; xit represents the covariates which vary across the firms, regions and industries. The composite error term eit is made up of two components; µi is a firmspecific effect which captures time-invariant unobserved firm heterogeneity (such as managerial ability) that may affect the export decision, and ε−it is an unobserved shock which can be rationalized as the demand or profit shock in the export market and may affect the firm’s decision to export in a given year. As we are interested in the effect of agglomeration on the overall propensity to export, we empirically include the agglomeration variables in a dynamic panel probit model of the form: Pr(yit = 1|xit−1 , yit−1 , µi ) = (xit−1 β + ρyit−1 + µi + εit ),
 

(9.3)

eit

where yit is a dummy variable which is equal to one if firm i has positive export sales in year t and zero otherwise;  is the cumulative normal distribution function. yit−1 is the lagged export status of firm i, which is included to capture the persistence in the firm exporting decision due to the presence of sunk costs (Roberts and Tybout, 1997). All other components of the model are defined as before. The most restrictive version of the model assumes that the error ε−it is independent across time and firms, and imposes the restriction ρ = 0. This produces the standard pooled probit estimators that ignore possible serial dependence and unobserved heterogeneity, which cannot be attributed to the variable xit−1 . However, firms are heterogeneous in some unobserved attributes and this might affect their individual propensity to export. In order to take these differences into account, fixed and random effect panel models could be used. However, unlike the linear models, it is impossible to eliminate µi by means of first differencing or using within transformation in the probit model. Moreover, if we attempt to estimate µi directly by adding N − 1 individual dummy variables to the probit specification, this will result in severely biased estimates of β due to the incidental parameter problem. This identification problem restricts the analysis to a random effect approach.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

195

A random effect model is proposed by Butler and Moffitt (1982), in which the error term is specified as: eit = µi + εit , εit ∼ i.i.d.N(0, 1), µi ∼ i.i.d.N(0, σµ2 ).

(9.4)

The firm-specific term µi captures possible permanent latent differences in the propensity to export. Furthermore, it is assumed that µi and εit are independent of xit−1 . Nevertheless, it has been suggested that the use of the random-effect probit may come at a cost if the firm-specific effects µi are correlated with the regressors (Bernard and Jensen, 2004). In fact, this strong assumption can be relaxed following the approach proposed by Mundlak (1978) in which he assumes µi = x¯ i ξ + γi , where x¯ i is the time average of xit and γi ∼ i.i.d.N(0, σγ2 ). More generally, Chamberlain (1982) uses the vector of all explanatory variables across all time periods instead of x¯ i . The intuition behind this transformation is that differences in time averaged explanatory variables contain information about underlying individual-specific characteristics, which implies that the individual differences that are left (γi ) may be more plausibly independent of the explanatory variables. Since we estimate a nonlinear dynamic random-effect panel data model, we must deal with another crucial estimation issue, that is, the initial condition problem. This problem occurs when the history of a stochastic process is not observed from the very beginning. The exogenous initial value assumption is very naïve and may lead to severe biases if the initial observations have been created with the evolution of observed and unobserved characteristics in the past. Wooldridge (2005) recently provided a very simple solution to the initial condition problem. This method leads to a very simple and tractable likelihood that is not different from the standard static random-effect model. The Wooldridge method suggests specifying a model for unobserved individual-effects which is conditioned on the initial values, yi1 , and the within-means of time-variant explanatory variables, x¯ i : µi = ξ0 + ξ1 yi1 + ξ2 x¯ i + ηi ,

(9.5)

where ηi is a new unobserved individual-effect which is simply assumed as ηi ∼ i.i.d.N(0, ση2 ); µi |¯xi , yi1 ∼ N(ξ0 + ξ1 yi1 + ξ2 x¯ , ση2 ) and

March 5, 2013

196

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

 x¯ i = T1 Tt=1 xit . Thus, we obtain a conditional likelihood which is based on the joint distribution of observations conditional on initial values. The resulting likelihood function will be like those in a standard static randomeffect probit model. Following this approach, we estimate the following specification: Pr(yit = 1|xit−1 , x¯ i , yi1 , yit−1 , ηi ) = (xit−1 β + ρyit−1 + ξ0 + ξ1 yi1 + ξ2 x¯ i + ηi ).

(9.6)

9.3.2. Agglomeration and export intensity Besides the export participation, we are also interested in the effect of agglomeration on the export intensity measured as the proportion of exports to total sales. We thus estimate the following equation: yit∗ = xit β + µi + ωit ,

(9.7)

where yit∗ is export intensity of firm i in year t and xit includes the agglomeration variables of interest and firm characteristics variables. Since the dependent variable is proportional and bounded between zero and one, the effects of the explanatory variables tend to be non-linear and their variance tends to decrease when the mean gets closer to one of the boundaries. Moreover, much more observations are clustering at zero. Such features make linear regressions unattractive. Instead, Santos Silva and Teneyro (2006) suggest using Poisson pseudo-maximum likelihood (PPML) estimators, which are better than the OLS estimators when the proportion of zero is very large and can produce consistent estimates in the presence of heteroskedasticity. However, this method does not take panel characteristics into account. Alternatively, Papke and Wooldridge (1996) point out that quasilikelihood estimation will yield consistent estimates of the parameters in the model as long as the conditional expectation is correctly specified in analyzing the bounded nature of the dependent variable. So they suggest the usage of fractional logit models in analyzing cross-sectional data. Papke and Wooldridge (2008) and Wooldridge (2009) further propose non-linear panel data methods for fractional response variables to recognize the bounded

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

197

nature of the dependent variable, that is, the fractional probit model.3 They argue that for the case of panel data, as in our application, the fractional probit is better, and this is the procedure used in this chapter. More importantly, it allows the correlation between unobserved firm-specific effects and covariates using Mundlak’s (1978) approach. This means that we can assume the conditional mean of unobservable firm specific effects to be linear in the mean value of some of the covariates. Another important statistical issue regarding the estimation is sample censoring. We can only observe the export intensity for the exporting firms. Given that the process of determining a firm’s export participation is a nonrandom process, estimating the export intensity equation without taking into account that the truncated sample, which suffers from the omitted variable problem, would produce biased estimates. To deal with the sample selection problem, we apply Wooldridge’s (2002) extension of Heckman’s two-step sample selection method to the panel dimension. Wooldridge (2002) shows that the Heckman’s approach can be applied to panel data by estimating a time-varying inverse Mills ratio. Following this approach, we first run a probit model for each period yit = 1[xit−1 β + εit > 0], where εit is error term and εit ∼ N(0, 1); xit−1 includes variables that explain a firm’s export decision. We calculate inverse Mill’s ratio, λˆ it , for every period and then we estimate the following equation in the second stage: yit∗ = xit−1 · β + ρ · λˆ it + µi + ωit , t ≥ 2.

(9.8)

Here xit−1 includes the spillover variables and other control variables. Estimation of this equation can provide consistent estimates of the impact of agglomeration economies on export sales, with control for sample selection problems.

9.4. Data, Variables, and Descriptive Analysis 9.4.1. The dataset The sample used in this study is derived from the Annual Enterprise Census conducted by the National Bureau of Statistics of China (NBSC). This 3 This method is easy to estimate using Stata’s “glm” command with “family (binomial) and

link (probit)” with robust standard errors.

March 5, 2013

198

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

census covers all “above scale” industrial firms including all state-owned and non-state-owned enterprises with annual sales above 5 million Chinese Yuan in the mining, manufacturing and public-utility sectors in China. This is the most comprehensive firm level data for China, spanning 40 two-digit manufacturing industries and 31 provinces or province-equivalent municipal cities. It provides detailed information on firm identification, industrial and geographic codes, output value, export value, added value, fixed assets, intermediate input costs, total employment, the year of establishment etc. According to Chen et al. (2011), this dataset is used as the basis to compile basic statistics for the aggregate manufacturing sectors that are summarized in the China Statistical Yearbook (NBSC, 1999–2008), and statistics on two-digit manufacturing industries that are summarized in the China Industry Economy Statistical Yearbook (NBSC, 1999–2008). The same dataset or sub-samples from the same source have been used to study productivity growth (Jefferson et al., 2008), corporate tax avoidance (Cai and Liu, 2009), and productivity spillovers from FDI (Du et al., 2011). We use a 10-year unbalanced panel dataset from 1998 to 2007, which ranges from 162,033 firms in 1999 to 336,768 firms in 2007. According to Jefferson et al. (2008), there are at least three causes leading to the variation in the number of enterprises across years: closure of the firm; an increase or decrease in sales that pushes the annual total sales above or below the 5 million Yuan threshold; or a change in ID related with some changes in the organization. However, the data shows that 5 million Yuan is not a strict rule. Less than 5% of the total firms with annual sales less than 5 million Yuan are also reported. The original dataset includes 2,223,359 observations and contains identifiers that can be used to track firms over time. Since the study focuses on manufacturing firms, we eliminate non-manufacturing observations. That means we only keep the observations whose two-digit industry codes are between 13 and 42. Following Jefferson et al. (2008) and Cai and Liu (2009) the sample size is further reduced by excluding observations if either of the following restrictions is violated: (1) a firm’s identification number cannot be missing and must be unique; (2) information on key variables such as the number of total employees, export value, total output, and net value of fixed assets cannot be missing; (3) net values of fixed assets must

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

199

be non-negative, while output, and value-added must be positive; (4) the number of employees of each firm must not be less than eight since these firms may not have a reliable accounting system; (5) the firm must be operating in that year; (6) the value of total assets minus the value of total fixed assets, and total assets minus the net value of fixed assets must be nonnegative; (7) the ratios of value-added to output value and export value to output must lie between zero and one. We end up with a sample of 1,850,332 observations representing 538,542 firms which account for 83.2% of the observations and 87.5% of the firms from the original dataset. All monetary variables, such as exports, output, value-added are then deflated to the 1998 price using the producer price index (PPI) at the SCI two-digit level. We finally restrict our analysis to the sample of the firms that appear in the sample period for at least five years continuously, so as to reduce the influence of entry and exit. However, the spillover variables are still calculated based on the whole sample.

9.4.2. The variables 9.4.2.1. Dependent variables Following popular practice, we consider two variables to measure a firm’s export performance. The first is an export dummy variable identifying a firm’s export status in general, which equals one if firm i reports positive export sales in year t, and zero otherwise. The second is a dependent variable measured by the ratio of the value of exports to the total sales value (that is, export intensity) of firm i in year t.

9.4.2.2. Spillover variables The agglomeration of exporting activities has been measured in different ways in the existing literature: the number of other exporters (Koenig, 2009) or the number of employees working in local exporting firms (Henderson, 2003), or a monetary measure of export activity (Aitken et al., 1997). In the following, the baseline results are based on the number of exporting neighbors following Koenig (2009), while the number of workers in export neighbors is used for robustness checks. Moreover, to examine

March 5, 2013

200

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

whether export spillovers are region- or industry-specific, we differentiate four types of agglomerations: the number of exporting firms in the same industry within the same region, the same industry outside the region, different industry within the same region, and different industry outside the region.

9.4.2.3. Control variables Besides the agglomeration variables of greatest interest, we also control for other factors that may affect the export behavior of firms according to the previous studies. Following Bernard and Jensen (1999) and Bernard and Wagner (2001), we choose the following variables to capture the heterogeneous characteristics of firms, such as labor productivity, firm size, firm age, capital intensity, average wage, and foreign equity share. • Labor productivity (LPit ) is defined as the value-added per capita of each firm, is expected to have a positive impact on both a firm’s decision on whether or not to export and how much to export, since labor productivity indicates the firm’s efficiency to produce goods. • Firm size (SIZE it ) is equal to a firm’s total assets, which is expected to be positively related to our dependent variable since larger firms are believed to have production and trade cost advantages over small ones. • Firm age (AGE it ) is included to account for the impact of both business experience and late-starter advantage. No prior expectation can be made since older firms have more experience in exporting, while younger firms are able to adopt new technology and learn from other exporting firms. • Capital intensity (KL it ), is defined as the ratio of the net fixed assets to the number of total employees. It is used as a proxy for the technological level of firms. However, no prior expectation can be made, as on the one hand, a capital-intensive firm tends to be larger and use more advanced technology, while on the other, higher capital intensity may become a disadvantage if the industry as a whole has a comparative advantage in labor-intensive goods in foreign markets. • We also control for the effect of a firm’s human resources using average wage (WAGE it ) which is equal to total salary divided by the total number

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

201

of employees for each firm. This can have either positive or negative impact on exports similar to the impact of capital intensity. Finally, we control for the impact of the firm’s foreign equity share (FS it ) on the firm’s export behavior. This is expected to have a positive impact on a firm’s export decision and export value, since firms with a higher foreign equity share seem to have better knowledge about foreign markets as well as relationships with foreign customers. In addition to the firms’ characteristics, industrial and regional characteristics as well as macroeconomic shocks can also affect a firms’ export behavior. Although the agglomeration variables have captured some part of the effect of industrial heterogeneity, we further introduce two-digit dummies to control the impact of other unobserved industry characteristics. Regional dummies are added to capture the geographical effect. We divide China’s 31 provinces and province-equivalent municipal cities into six groups. Five dummy variables are added to the model to capture variations among firms located in the areas of Pearl River Delta (PRD), Yangtze River Delta (YRD), Bohai Economic Rim (BER), Northeast China (NEC) and Central China (CEC), with Western China (WEC) being chosen as the reference region.4 Finally, year dummies are included to proxy for timespecific macroeconomic conditions that firms are facing when they make export decisions.

9.4.3. Descriptive analysis A snapshot is first produced into the panel dimension of the dataset. Table 9.1 documents the distribution of firms across different years according to their status of exporting. A firm is classified as an exporter if it reports positive export sales. It shows that on average over one-fourth (27%) of firms exported during 1998–2007. The share of exporters to total 4 Specifically, China’s 31 administrative regions in the mainland are partitioned into six

groups and represented by six dummy variables, namely PRD (Guangdong, Fujian, Guangxi, and Hainan), YRD (Jiangsu, Shanghai, and Zhejiang), BER (Beijing, Tianjin, Hebei, and Shandong), NEC (Liaoning, Jilin, and Heilongjiang), CEC (Shanxi, Anhui, Jiangxi, Hubei, Hunan, and Henan), WEC (Inner Mongolia, Ningxia, Tibet, Xinjiang, Gansu, Guizhou, Qinghai, Shaanxi, Sichuan, Yunnan, and Chongqing).

March 5, 2013

202

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.1. Export Patterns in the Data. Exporter Year 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 Total/ Average

All Firmsa 117,151 123,493 125,716 139,606 146,080 169,530 240,315 229,790 256,733 301,918 1,850,332

Exporterb (%) Export Intensityc Pure Exporterd (%) 24.58 24.75 26.14 26.16 27.47 28.92 30.59 28.83 27.56 25.49 27.31

0.590 0.596 0.609 0.615 0.616 0.637 0.639 0.650 0.642 0.633 0.629

25.23 25.32 26.72 27.05 26.25 28.89 30.36 30.03 29.58 28.66 28.41

aAll firms refers to all Chinese manufacturing firms. b Exporters are those firms that report positive export sales. c Export intensity is the average of the ratio of export sales to total sales calculated across

all sectors and exporting firms for each year. d Pure exporters refer to those firms that exported all their output. Sources: Authors’ own calculation.

manufacturing firms gradually increased from 24.6% in 1998 to the peak of 30.6% in 2004, and then decreased slowly to 25.5% at the end of the sample period. The average export intensity for all exporting firms is presented in the fourth column of the table, which reports the average ratio of export sales to total sales by year. The average export intensity was very high during 1998–2007 and it shows that exporting firms on average sold 63% of their output abroad, which peaked at 65% in 2005. Moreover, among all exporters during the sample period, it is found that 28% of them exported all their output, and are termed “pure exporters” in the last column of the table. We take a closer look at the industrial distribution of exporters in Table 9.2, which shows that about 40% of the exporting firms are from the top five industries. Noticeably, the top two concentrated industries both in 1998 (Panel A) and 2007 (Panel B) are textile-related industries with an export intensity higher than the average. They account for one-fourth of all

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

203

Table 9.2. Industrial Distribution of Exporters in 1998 and 2007.

Two-digit Industry

Average No. of % of National Export Exporters Exporters Intensity

Panel A. 1998 Top five concentrated industries 17 Textile 3,834 18 Clothes, shoes, and hats 3,722 34 Metal products 1,824 35 General purpose machinery 1,822 26 Raw chemicals and chemical products 1,778

13.32 12.93 6.34 6.33 6.18

0.606 0.849 0.618 0.391 0.398

8,104 7,490 5,634

10.53 9.73 7.32

0.650 0.816 0.621

5,332

6.93

0.660

5,268

6.85

0.499

Panel B. 2007 Top five concentrated industries 17 Textile 18 Clothes, shoes and hats 39 Electric machines and apparatuses manufacturing 40 Communications equipment, computer and other electronic equipment 35 General purpose machinery

Sources: The numbers are calculated by the author using the sample data.

exporting firms. Moreover, we find that two of the top five industries in 2007 were high-tech sectors. This may imply changes in China’s export structure during this period. However, recent evidence shows that the success of Chinese high-tech exports is due to processing trade rather than to their R&D investment and technological progress (Huang et al., 2008). Table 9.3 shows that among the 25,185 exporters in 1998, 87% are located in the coastal areas. Among them, more than one-half of the exporters cluster in Guangdong, Zhejiang and Jiangsu provinces, followed by Shandong, Fujian, and Shanghai. The pattern is similar for 2007. Export intensity in the top nine provinces is always greater than 50%. Exporting firms in Fujian and Guangdong provinces have the highest export intensity (more than 70% each year). This is hardly surprising, as firms in the coastal region can have relatively easy access to international markets in comparison with those inland.

March 5, 2013

204

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.3. Regional Distribution of Exporters in 1998 and 2007. 1998

Provinces Guangdong Zhejiang Jiangsu Shanghai Shandong Liaoning Hebei Fujian Tianjin Beijing Total/ average

2007

% of Average % of Average No. of National Export No. of National Export Exporters Exporters Intensity Exporters Exporters Intensity 6,752 4,664 4,901 2,593 2,578 1,124 778 702 559 534 25,185

23.45 16.20 17.02 9.01 8.95 3.90 2.70 2.44 1.94 1.85 87.46

0.776 0.595 0.548 0.559 0.510 0.551 0.512 0.798 0.530 0.456 0.590

17,928 20,230 9,693 4,323 5,980 2,517 1,275 5,087 1,644 1,225 69,902

23.30 26.29 12.60 5.62 7.77 3.27 1.66 6.61 2.14 1.59 90.85

0.733 0.657 0.570 0.572 0.580 0.597 0.513 0.780 0.566 0.414 0.633

Sources: The numbers are calculated by the author using the sample data.

9.5. Empirical Results 9.5.1. Agglomeration and export participation We start by estimating a pooled probit model of the export participation as the baseline model under the assumption that the errors are independent across time. The estimates are presented in Table 9.4. The static models are reported in the left two columns and the dynamic specifications, including the lagged dependent variable, are presented in the right two columns. In columns (2) and (4), the squared term of the spillover variable is included to test the non-linearity between agglomeration and export propensity of a firm in the same industry and same region. According to column (1), the coefficient on the spillover variable is positive and statistically significant. The positive sign indicates that export spillovers are present and indeed influence the export decisions of firms in the same industry within the same region. In column (2), we include the quadratic form of the spillover variable to identify the non-linear

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

205

Table 9.4. Baseline Results for Export Participation: Pooled Probit Models. Dependent Variable: Export Participation SISR [Same industry same region] SISR2 [Squared term] LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] Pseudo R2 Log-likelihood N

Static Model

Dynamic Model

(1)

(2)

(3)

(4)

0.259∗∗∗ (0.002)

0.381∗∗∗ (0.006)

0.123∗∗∗ (0.003)

0.188∗∗∗ (0.008)

−0.128∗∗∗ (0.002)

−0.013∗∗∗ (0.001) −0.130∗∗∗ (0.002)

−0.060∗∗∗ (0.003)

−0.007∗∗∗ (0.001) −0.061∗∗∗ (0.003)

0.283∗∗∗ (0.002) −0.024∗∗∗ (0.003) −0.191∗∗∗ (0.002)

0.284∗∗∗ (0.002) −0.024∗∗∗ (0.003) −0.190∗∗∗ (0.002)

0.148∗∗∗ (0.002) −0.095∗∗∗ (0.005) −0.088∗∗∗ (0.002)

0.148∗∗∗ (0.002) −0.095∗∗∗ (0.005) −0.087∗∗∗ (0.002)

0.183∗∗∗ (0.004) 1.744∗∗∗ (0.008)

0.184∗∗∗ (0.004) 1.732∗∗∗ (0.008)

0.100∗∗∗ (0.005) 0.765∗∗∗ (0.011)

0.100∗∗∗ (0.005) 0.759∗∗∗ (0.011)

2.586∗∗∗ (0.005)

2.584∗∗∗ (0.005)

0.253 0.253 0.632 0.632 −346,779.66 −346,555.14 −170,913.95 −170,877.08 719,830 719,830 719,830 719,830

Note: All the independent variables except for dummies are in logarithmic values and are lagged by one-year.All regressions contain year, industry and region dummies. *** indicates statistical significance at 1%, 5% and 10% level, respectively. Standard errors are reported in parentheses.

relationship between agglomeration of exporters and export propensity. As shown in the table, the coefficient of the linear term of the spillover variable is still positive and significant, and the coefficient of the quadratic term is significantly negative, which demonstrates that there is an inverted-U

March 5, 2013

206

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

relationship between agglomeration and the probability of export. When the degree of agglomeration is low, the export spillover effects are more likely to dominate the congestion cost in local markets. But as more and more similar exporters cluster in a specific region, the congestion cost is more likely to dominate the export spillover effect. The inclusion of the lagged dependent variable in columns (3) and (4) reduces the effect of spillovers on the export propensity by one half, but it improves the fitness of the model as indicated by the highly significant increase of the maximized log-likelihood value and pseudo R2 . The results for the spillover variables are the same in the static model. The estimated coefficient on the lagged export status (EXPD) is positive and significant at the 1% significance level. This implies that current export participation significantly increases the probability of export participation in the following year. This may suggest that most firms in China face high costs associated with entering foreign markets, which appear to be sunk in nature, and it finally leads to the phenomenon of high persistence in export status. A series of robustness checks are considered. Given the estimation of a pooled probit model in the baseline specification may be biased, the panel characteristics were taken into account. Turning to the maximum likelihood estimates, the random effects probit model allows for the time-invariant unobserved heterogeneity and serially correlated idiosyncratic errors. For each regression, the assumption of independence between unobserved firmspecific effects and covariates is relaxed in the standard random effects probit model following the Mundlak’s approach. The estimates in Table 9.5 show that the findings from the baseline regressions still hold. The clusters of exporters operating in the same industry and that are located in the same region positively affect the exporting likelihood of other firms but have a negative effect after a threshold. Comparing the values of log-likelihood, we choose the dynamic random effect probit model as our preferred specification to estimate the export participation equation, which will be used in the following analyses. Table 9.6 shows the results from an investigation into whether the export spillover is only restricted to the same industry and same region (SISR). Hence four types of export spillovers are differentiated in terms of the number of exporting firms according to their industries and regions. We first differentiate those from different regions but in the same industry (SIDR) in

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

207

Table 9.5. Robustness Check (I) for Export Participation. Dependent Variable: Export Participation SISR [Same industry same region] SISR2 [Squared term] LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] Log-likelihood N

Static Model

Dynamic Model

(1)

(2)

(3)

(4)

0.435∗∗∗ (0.007)

0.745∗∗∗ (0.021)

0.180∗∗∗ (0.005)

0.282∗∗∗ (0.015)

−0.082∗∗∗ (0.005)

−0.033∗∗∗ (0.002) −0.085∗∗∗ (0.005)

−0.047∗∗∗ (0.004)

−0.011∗∗∗ (0.001) −0.048∗∗∗ (0.004)

0.561∗∗∗ (0.006) −0.065∗∗∗ (0.011) −0.205∗∗∗ (0.005)

0.565∗∗∗ (0.006) −0.063∗∗∗ (0.011) −0.205∗∗∗ (0.005)

0.089∗∗∗ (0.008) −0.190∗∗∗ (0.008) −0.026∗∗∗ (0.006)

0.091∗∗∗ (0.008) −0.189∗∗∗ (0.008) −0.026∗∗∗ (0.006)

0.179∗∗∗ (0.008) 2.833∗∗∗ (0.030)

0.178∗∗∗ (0.008) 2.814∗∗∗ (0.030)

0.097∗∗∗ (0.007) 1.019∗∗∗ (0.020)

0.097∗∗∗ (0.007) 1.011∗∗∗ (0.020)

1.374∗∗∗ (0.009)

1.373∗∗∗ (0.009)

−201,092.15 719,830

−200.946 719,830

−158,247.21 −158,219.24 719,830 719,830

Notes: All the independent variables except for dummies are in logarithmic values and are lagged by one-year. All regressions contain year, industry and region dummies. ∗∗∗ indicates staistical significance at 1% level. Standard errors are reported in parenthesis.

column (1). The coefficient of SISR is positive and statistically significant but that of SIDR is negative and significant, suggesting that firms are more likely to be positively affected by the exporting firms nearby than those located outside the province. In other words, export spillovers are more likely to take place within the same region. In column (2), spillovers are classified by sectors. The coefficients of SISR and DISR are positive and statistically

March 5, 2013

208

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.6. Robustness Check (II) for Export Participation. Dependent Variable: Export Participation Spillover variables SISR [Same industry same region] SIDR [Same industry different region] DISR [Different industry same region] DIDR [Different industry different region] Control variables LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] Log-likelihood N

(1)

(2)

0.192∗∗∗ (0.005)

0.153∗∗∗ (0.006)

−0.054∗∗∗ (0.009)

(3) 0.153∗∗∗ (0.007) −0.047∗∗∗ (0.010)

0.069∗∗∗ (0.009)

−0.013 (0.011) −1.069∗∗∗ (0.105)

−0.047∗∗∗ (0.004) 0.089∗∗∗ (0.008) −0.190∗∗∗ (0.008) −0.026∗∗∗ (0.006) 0.097∗∗∗ (0.007) 1.024∗∗∗ (0.020) 1.373∗∗∗ (0.009) −158,229.37 719,830

−0.047∗∗∗ (0.004) 0.088∗∗∗ (0.008) −0.190∗∗∗ (0.008) −0.025∗∗∗ (0.006) 0.098∗∗∗ (0.007) 1.024∗∗∗ (0.020) 1.373∗∗∗ (0.009) −158,214.44 719,830

−0.045∗∗∗ (0.004) 0.085∗∗∗ (0.008) −0.188∗∗∗ (0.008) −0.027∗∗∗ (0.006) 0.097∗∗∗ (0.007) 1.048∗∗∗ (0.020) 1.371∗∗∗ (0.009) −158,158.83 719,830

Note: The estimates are obtained from Wooldridge’s estimator by estimating a dynamic random effect probit model. All the independent variables except for dummies are in logarithmic values and are lagged by one-year. All regressions contain year, industry and region dummies. *** indicates statistical significance at 1% level. Standard errors are reported in parentheses.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

209

significant, suggesting there are positive inter-industry and intra-industry export spillovers within the region. In column (3), four spillover variables are included in the regression and it is found that only the coefficient of SISR is positive and significant, which confirms again that the export spillovers are more likely to occur in the same industry and same region. In Table 9.7, we check the different channels of spillovers measured by the number of workers in other exporting firms. Henderson (2003) suggests that knowledge spillovers occur mainly from the proximity of firms in the same industry. He further argues that, if the measure of employment performs better than the number of establishments, it means that the labor-pool externality is more important than knowledge spillovers. The results show that the pool of exporting workers is more important for export spillovers, as we do not find that the square term is positive and insignificant, and inter-industry spillovers are positive and statistically significant. We also investigate whether export spillovers vary according to firm size. In the literature, it has been argued that smaller firms could be more affected by external factors, such as export spillovers (Bernard and Jensen, 2004), possibly because of their limited amount of internal resources. We categorize firms according to their employment levels, that is, firms below 300 employees are grouped together as small firms; firms between 301 and 2,000 employees are labeled as medium-sized firms and the rest of the firms with more than 2,000 employees are considered as large firms. We indeed find that export spillovers are stronger for smaller firms than larger firms.5 As for other control variables, the coefficients of labor productivity are consistently negative and significant, which apparently contradicts the theoretical prediction (Melitz, 2003). Lu (2010) incorporated factor endowments into a Melitz model to explain why China’s exporters are typically less productive. She argues that the sectors that are intensive in the locally abundant factor face more competition in the domestic market than in foreign markets. Hence domestic markets rather than export markets select the most efficient firms. Moreover, the effect of firm size is positive, implying large firms are more likely to be exporters. The coefficient of age is significantly negative, implying the late-starter advantage of younger firms. The relationship between capital intensity and export status is also negative, suggesting that 5 The results are available upon request.

March 5, 2013

210

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.7. Robustness Check (III) for Export Participation. Dependent Variable: Export Participation Spillover variables SISR [Same industry same region] SISR2 [Squared term] SIDR [Same industry different region] DISR [Different industry same region] DIDR [Different industry different region] Control variables LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] Log likelihood N

(1)

(2)

0.090∗∗∗ (0.004)

0.079∗∗∗ (0.012)

(3) 0.084∗∗∗ (0.005)

0.001 (0.001) −0.015 (0.011) 0.072∗∗∗ (0.012) 0.475∗∗∗ (0.125)

−0.045∗∗∗ (0.004) 0.090∗∗∗ (0.008) −0.203∗∗∗ (0.008) −0.025∗∗∗ (0.006) 0.096∗∗∗ (0.007) 1.009∗∗∗ (0.020)

−0.045∗∗∗ (0.004) 0.090∗∗∗ (0.008) −0.203∗∗∗ (0.008) −0.025∗∗∗ (0.006) 0.096∗∗∗ (0.007) 1.010∗∗∗ (0.020)

−0.047∗∗∗ (0.004) 0.089∗∗∗ (0.008) −0.202∗∗∗ (0.008) −0.024∗∗∗ (0.006) 0.100∗∗∗ (0.007) 1.006∗∗∗ (0.020)

1.374∗∗∗ (0.009)

1.374∗∗∗ (0.009)

1.375∗∗∗ (0.009)

−158,636.04 719,830

−158,635.62 719,830

−158,606.33 719,830

Notes: Spillover variables are measured by the number of employees working in other exporting firms. Estimates are from dynamic random effect probit model. ∗∗∗ denotes 5% significance level.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

211

those firms that use labor-intensive technology have a higher probability of exporting. The effect of average wage, which is used as a proxy for the labor skill, is positive and indicates that firms with more skilled labor are more likely to export. Finally, we find that the foreign equity share of the firms’ matters in the likelihood of exporting. Firms with a higher share of foreign equity are more likely to be exporters. Summarizing the results from Tables 9.4–9.7, we conclude that the agglomeration of exporters has a positive impact on the exporting propensity of firms in general but its effect becomes negative when the level of clustering becomes too high. Exporter spillovers are also found to benefit those firms from the same sector and the same region. Younger, laborintensive and foreign-invested firms are more likely to enter an export market, while labor productivity and age are negatively correlated with export propensity.

9.5.2. Agglomeration and export intensity Export spillovers may also be relevant for export intensity. In Table 9.8, regressions are run with the same explanatory variables using the PPML method first. The overall picture from this table is comparable to that for export participation. Results in columns (1) and (2) indicate that having other exporting firms from the same industry nearby positively affects export sales. However, the marginal effect on the export intensity deteriorates as the agglomeration level of exporters becomes more clustered, since the coefficient of the squared term of the spillover variable is negative and statistically significant. In Table 9.9, the same estimations are presented as in Table 9.8 using the fractional probit model for panel data to take panel characteristics into account. The results confirm the findings from the PPML estimations. The non-linearity between agglomeration and export intensity is also present. In addition, a positive spillover effect is also found only when exporting firms from the same industry are present. This finding is in line with the interpretation of Rauch and Watson’s (2003) model by Koenig et al. (2010) who argues that the clustering of similar firms within a region may send positive signals to foreign buyers. Finally, we present the estimations using the Heckit model to address the sample selection bias. It is noted that the inverse Mill’s ratio is significant

March 5, 2013

212

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.8. Baseline Results for Export Intensity: PPML Estimation. Dependent Variable: Export Intensity Spillover variables SISR [Same industry same region] SISR2 [Squared term] SIDR [Same industry different region] DISR [Different industry same region] DIDR [Different industry different region] Control variables LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] R-squared Pseudo log-likelihood N

(1)

(2)

0.108∗∗∗ (0.003)

0.375∗∗∗ (0.015)

(3) 0.119∗∗∗ (0.004)

−0.025∗∗∗ (0.001) 0.007 (0.007) −0.018∗∗∗ (0.009) 0.072 (0.062)

−0.065∗∗∗ (0.002) −0.059∗∗∗ (0.002) −0.172∗∗∗ (0.006) −0.067∗∗∗ (0.002) 0.011∗∗∗ (0.004) 0.579∗∗∗ (0.008) 2.898∗∗∗ (0.009) 0.600 −246,008.2 719,830

−0.067∗∗∗ (0.002) −0.058∗∗∗ (0.002) −0.171∗∗∗ (0.006) −0.065∗∗∗ (0.002) 0.012∗∗∗ (0.004) 0.564∗∗∗ (0.008) 2.878∗∗∗ (0.011) 0.601 −245,813.21 719,830

−0.065∗∗∗ (0.002) −0.059∗∗∗ (0.002) −0.171∗∗∗ (0.006) −0.067∗∗∗ (0.002) 0.011∗∗∗ (0.004) 0.576∗∗∗ (0.009) 2.898∗∗∗ (0.011) 0.600 −245,995.84 719,830

Notes: The estimates are obtained from Wooldridge’s estimator by estimating a dynamic random effect probit model. All the independent variables except for dummies are in logarithmic values and are lagged by one-year. All regressions contain year, industry and region dummies. *** indicates statistical significance at 1% level. Standard errors are reported in parentheses.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

213

Table 9.9. Robustness Check (I) for Export Intensity: Fractional Probit Models. Dependent Variable: Export Intensity Spillover variables SISR [Same industry same region] SISR2 [Squared term] SIDR [Same industry different region] DISR [Different industry same region] DIDR [Different industry different region] Control variables LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] EXPD [Lagged export status] Pseudo log-likelihood N

(1)

(2)

0.141∗∗∗ (0.002)

0.225∗∗∗ (0.008) −0.008∗∗∗ (0.001)

(3) 0.153∗∗∗ (0.003)

−0.030∗∗∗ (0.005) −0.062∗∗∗ (0.005) −0.393∗∗∗ (0.043)

−0.100∗∗∗ (0.002) −0.098∗∗∗ (0.005) −0.188∗∗∗ (0.004) −0.195∗∗∗ (0.004) 0.006 (0.004) 0.968∗∗∗ (0.007) 1.981∗∗∗ (0.005) −166,076.72 719,830

−0.101∗∗∗ (0.002) −0.097∗∗∗ (0.005) −0.188∗∗∗ (0.004) −0.193∗∗∗ (0.004) 0.006 (0.004) 0.962∗∗∗ (0.007) 1.979∗∗∗ (0.005) −166,038.12 719,830

−0.098∗∗∗ (0.002) −0.099∗∗∗ (0.005) −0.186∗∗∗ (0.004) −0.195∗∗∗ (0.004) 0.004 (0.004) 0.977∗∗∗ (0.008) 1.981∗∗∗ (0.005) −188,033.24 719,830

Notes: The estimates are obtained from Wooldridge’s estimator by estimating a dynamic random effect probit model. All the independent variables except for dummies are in logarithmic values and are lagged by one-year. All regressions contain year, industry and region dummies. *** indicates statistical significance at 1% level. Standard errors are reported in parentheses.

March 5, 2013

214

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Table 9.10. Robustness Check (I) for Export Intensity: Heckit Model. Dependent Variable: Export Intensity Spillover variables SISR [Same industry same region] SISR2 [Squared term] SIDR [Same industry different region] DISR [Different industry same region] DIDR [Different industry different region] Control variables LP [Labor productivity] SIZE [Firm size] AGE [Firm age] KL [Capital–labor ratio] WAGE [Average wage] FS [Foreign equity share] IMR [Inverse Mill ratio] N

(1)

(2)

0.048∗∗∗ (0.001)

0.073∗∗∗ (0.003) −0.002∗∗∗ (0.000)

−0.035∗∗∗ (0.001) −0.041∗∗∗ (0.001) −0.055∗∗∗ (0.001) −0.041∗∗∗ (0.001) −0.002 (0.002) 0.317∗∗∗ (0.007) 0.037∗∗∗ (0.007) 248,832

−0.036∗∗∗ (0.001) 0.039∗∗∗ (0.001) −0.056∗∗∗ (0.001) −0.041∗∗∗ (0.001) −0.001 (0.002) 0.327∗∗∗ (0.007) 0.048∗∗∗ (0.007) 248,832

(3) 0.058∗∗∗ (0.001) −0.013∗∗∗ (0.002) −0.029∗∗∗ (0.002) −0.087∗∗∗ (0.015) −0.035∗∗∗ (0.001) −0.042∗∗∗ (0.001) −0.042∗∗∗ (0.001) −0.040∗∗∗ (0.001) −0.003 (0.002) 0.319∗∗∗ (0.007) 0.036∗∗∗ (0.007) 248,832

Notes: Same with Table 9.6.

in three columns in Table 9.10 suggesting that a selection bias would be present. However, we find no big change in the coefficients of the explanatory variables compared to previous regression results. The squared term of the spillover variable, in particular, still remains negative and

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

215

statistically significant. The hypothesis that the export spillovers are regionand industry-specific is consistent with those in other models. The coefficients of the variable representing labor productivity, age, capital intensity, wage and foreign equity share have the same sign as those of the variables in determining export propensity. However, firm size now has an opposite sign in the export intensity regressions, which suggests that, conditional on exporting, large firms are more likely to export at a lower level, although they have the greater probability of exporting.

9.6. Conclusion This chapter has investigated the spillover effects of exporter clustering and their impact on the export performance of individual firms using a dataset of Chinese firms during the period 1998–2007. The agglomeration of exporters is found to exert a significant positive impact on the export participations of individual firms. However, this study suggests that exporter clustering is not always desirable since we find that there is an inverted-U shape relationship between agglomeration and export participation possibly due to the congestion costs. A similar pattern is also found when export intensity is used as a measurement of export performance. Apart from a possible negative effect of exporter clustering, the agglomeration of exporters is found to benefit firms in the same industry within the same region only. It thus demonstrates that geographical proximity plays an important role in export spillovers. In addition, the inter-industry export spillovers are less significant since it is found that the presence of many exporters only increases a firm’s chance of being an exporter if they come from the same industry. In summary, exporter clustering is one of the important determinants of export participation in China, but its negative effects should be avoided. The inter-region and inter-industry export spillovers should be promoted. It should, however, be noted that the decision to export is not a once-and-forever action, though export entry is an important first step. If new exporters exit foreign markets in a short period, the learning-by-exporting effects of firms will be limited and will further weaken aggregate export growth. Therefore, the export duration after entering the export market of a firm should be of more interest.

March 5, 2013

216

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

References Adams, J. D., “Comparative Localization of Academic and Industrial Spillovers,” Journal of Economic Geography, 2(3): 253–278 (2002). Aitken, B., Hanson, G. H. and Harrison, A. E., “Spillovers, Foreign Investment, and Export Behavior,” Journal of International Economics, 43(1/2): 103–132 (1997). Barrios, S., Goerg, H. and Strobl, E., “Explaining Firms Export Behaviour: R&D, Spillovers and the Destination Market,” Oxford Bulletin of Economics and Statistics, 65(4): 475–496 (2003). Bernard, A. B. and Jensen, J. B., “Exporters, Jobs, and Wages in U.S. Manufacturing: 1976–1987,” Brookings Papers on Economic Activity: Microeconomics, 1995: 67–119 (1995). Bernard, A. B. and Jensen, J. B., “Exceptional Exporter Performance: Cause, Effect, or Both?” Journal of International Economics, 47(1): 1–25 (1999). Bernard, A. B. and Jensen, J. B., “Why Some Firms Export?” The Review of Economics and Statistics, 86(2): 561–569 (2004). Bernard, A. B. and Wagner, J., “Export Entry and Exit by German Firms,” Review of World Economics (Weltwirtschaftliches Archiv), 137(1): 105–123 (2001). Butler, J. S. and Moffitt, R., “A Computationally Efficient Quadrature Procedure for the One-Factor Multinomial Probit Model,” Econometrica, 50(3): 761–764 (1982). Cai, H. and Liu, Q., “Competition and Corporate Tax Avoidance: Evidence from Chinese Industrial Firms,” The Economic Journal, 119(537): 764–795 (2009). Caldera, A., “Innovation and Exporting: Evidence from Spanish Manufacturing Firms,” Review of World Economics (Weltwirtschaftliches Archiv), 146(4): 657–689 (2010). Cavusgil, S. T. and Zou, S., “Marketing Strategy-Performance Relationship: An Investigation of the Empirical Link in Export Market Ventures,” The Journal of Marketing, 58(1): 1–21 (1994). Chamberlain, G., “Multivariate Regression Models for Panel Data,” Journal of Econometrics, 18(1): 5–46 (1982). Chen, Z., Ge, Y. and Lai, H., “Foreign Direct Investment and Wage Inequality: Evidence from China,” World Development, 39(8): 1322–1332 (2011). Clerides, S. K., Lach, S. and Tybout, J. R., “Is Learning By Exporting Important? Micro-Dynamic Evidence from Colombia, Mexico, and Morocco,” The Quarterly Journal of Economics, 113(3): 903–947 (1998).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Agglomeration and Export Performance of Manufacturing Firms

217

Clougherty, J. A. and Zhang, A., “Domestic Rivalry and Export Performance: Theory and Evidence from International Airline Markets,” Canadian Journal of Economics, 42(2): 440–468 (2009). Du, L., Harrison, A. and Jefferson, G. H., “Testing for Horizontal and Vertical Foreign Investment Spillovers in China, 1998–2007,” Journal of Asian Economics 23(3): 234–243 (2012). Duranton, G. and Puga, D., “Micro-Foundations of Urban Agglomeration Economies,” NBER Working Papers 9931, National Bureau of Economic Research, Inc (2003). Greenaway, D. and Kneller, R., “Exporting and Productivity: Theory, Evidence and Future Research,” The Singapore Economic Review, 50: 303–312 (2005). Greenaway, D., Sousa, N. and Wakelin, K., “Do Domestic Firms Learn to Export from Multinationals?” European Journal of Political Economy, 20(4): 1027–1043 (2004). Henderson, J. V., “Marshall’s Scale Economies,” Journal of Urban Economics, 53(1): 1–28 (2003). Jefferson, G. H., Rawski, T. and Zhang, Y., “Productivity Growth and Convergence across China’s Industrial Economy,” Journal of Chinese Economic and Business Studies, 6(2): 121–140 (2008). Koenig, P., “Agglomeration and the Export Decisions of French Firms,” Journal of Urban Economics, 66(3): 186–195 (2009). Koenig, P., Florian, M. and Sandra, P., “Local Export Spillovers in France,” European Economic Review, 54(4): 622–641 (2010). Lawless, M., “Geography and Firm Exports: New Evidence on the Nature of Sunk Costs,” Review of World Economics (Weltwirtschaftliches Archiv), 146(4): 691–707 (2010). Lu, D., Exceptional Exporter Performance? Evidence from Chinese Manufacturing Firms. Chicago, USA: Job Market Paper, University of Chicago, USA (2010). Melitz, M. J., “The Impact of Trade on Intra-Industry Reallocations and Aggregate Industry Productivity,” Econometrica, 71(6): 1695–1725 (2003). Mundlak, Y., “On the Pooling of Time Series and Cross Section Data,” Econometrica, 46(1): 69–85 (1978). Papke, L. E. and Wooldridge, J., “Panel Data Methods for Fractional Response Variables with an Application to Test Pass Rates,” Journal of Econometrics, 145(1/2): 121–133 (2008).

March 5, 2013

218

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch09

Regional Development and Economic Growth in China

Papke, L. E. and Wooldridge, J. M., “Econometric Methods for Fractional Response Variables with an Application to 401(K) Plan Participation Rates,” Journal of Applied Econometrics, 11(6): 619–632 (1996). Rauch, J. E. and Watson, J., “Starting Small in An Unfamiliar Environment,” International Journal of Industrial Organization, 21(7): 1021–1042 (2003). Roberts, M. and Tybout, J. R., “The Decisions to Export in Colombia:An Empirical Model of Entry with Sunk Costs,” American Economic Review, 87(4): 545–564 (1997). Santos Silva, J. M. C. and Tenreyro, S., “The Log of Gravity,” The Review of Economics and Statistics, 88(4): 641–658 (2006). Sousa, C. M. P., Martínez-López, F. J. and Coelho, F., “The Determinants of Export Performance: A Review of the Research in the Literature between 1998–2005,” International Journal of Management Reviews, 10(4): 343–374 (2008). Wagner, J., “A Note on the Firm Size-Export Relationship,” Small Business Economics, 17(4): 229–237 (2001). Wagner, J., “Exports and Productivity: A Survey of the Evidence from Firm-Level Data,” The World Economy, 30(1): 60–82 (2007). Wooldridge, J., Economic Analysis of Cross-Section and Panel Data. Cambridge, Mass: MIT Press (2002). Wooldridge, J., “Simple Solutions to the Initial Conditions Problem in Dynamic, Nonlinear Panel Data Models with Unobserved Heterogeneity,” Journal of Applied Econometrics, 20(1): 39–54 (2005). Wooldridge, J., Correlated Random Effects Models with Unbalanced Panels. Michigan, USA: Manuscript (version July 2009), Department of Economics, Michigan State University, USA (2009). Xu, X. and Sheng, Y., “Are FDI Spillovers Regional? Firm-level Evidence from China,” Journal of Asian Economics 23(3): 244–258 (2012). Zhao, X. and Zou, S., “The Impact of Industry Concentration and Firm Location on Export Propensity and Intensity: An Empirical Analysis of Chinese Manufacturing Firms,” Journal of International Marketing, 10(1): 52–71 (2002).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Part III

China and Neighboring Economies

219

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

This page intentionally left blank

220

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Chapter 10

Individual Country Approaches to Agriculture in the ASEAN–China FTA Ray Trewin, David Vanzetti, Nur Rakhman Setyoko, and Nguyen Ngoc Que1

10.1. Introduction China has negotiated a free trade agreement (FTA) withASEAN (ACFTA) in which its Members can independently negotiate their tariff reductions. FTAs generally do not go beyond this, such as with non-tariff barriers (NTBs). Being significant traders with China, ASEAN Members are aware of the opportunities that the Chinese market presents in large for accessing it’s markets, but individual Members are concerned to differing degrees about being flooded with Chinese imports including in agriculture. For example, as the time for implementation approaches, Indonesia has expressed a desire to renegotiate its tariff reduction schedules to protect sensitive sectors, including agriculture (Patunru and von Nuebuke, 2010). By contrast, Vietnam, just over the border from China and with a long history of informal trade and a more recent history of the benefits of trade liberalization, seems to accept more of the prospects. Despite this concern with China, it has its own sensitive agricultural sub-sectors, as evident from the differential access offered. Why this is the case? Is Vietnam more accepting

1 The authors thank ACIAR for funding this project and Paul Bartlett, Malcolm Bosworth

and an anonymous referee for comments on drafts of the chapter.

221

March 5, 2013

222

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

of Chinese competition because of its location or other factors such as more recent awareness of the benefits of trade liberalization, mainly unilateral, and of allowing the imports of products from the world’s cheapest supplier? Has Indonesia negotiated a worse deal with China under the ACFTA than Vietnam and wanting to redress this, or has a changed political economy made them less committed to trade liberalization? Does China, despite benefiting massively from trade liberalization to become the ominous economic power it is seen as today, still have its own non-economic drivers of sensitive sectors? The purpose of this chapter is to analyze such questions through comparing past trade flows, tariffs, other trade-related constraints and the agricultural political economy, as well as potential impacts of the ACFTA on the agricultural sectors of China and representative ASEAN Member States of Indonesia and Vietnam, using a global general equilibrium model, Global Trade Analysis Project (GTAP) (2008). Aggregated tariff line data with some modification enables the differential impact of separate sensitive sectors for China, Indonesia and Vietnam to be identified and analyzed. The simulated results following full implementation indicate all countries, which include China, one of the world’s cheapest suppliers, would improve their trade and welfare if the agreement is implemented as negotiated and tariff cuts are effective, although the extent of exemptions for sensitive products represent differing degrees of missed opportunities for each country. At the sectoral level, all countries can expect some reductions, compared with the baseline, in output of some agricultural sectors. However, generally these changes are relatively small apart from when significant NTBs are taken into account. Therefore, structural adjustment improves economic welfare. The chapter is structured as follows. Section 10.2 presents trade flows, tariffs and NTBs, plus aspects of ACFTA such as exemptions of sensitive sectors. Section 10.3 describes the GTAP Computable General Equilibrium (CGE) model, the data, sectors and regions, and scenarios that analyze the ACFTA. Section 10.4 presents the results, setting out trade, welfare and sector impacts, and finally conclusions, limitations and implications are drawn in the Section 10.5.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

223

10.2. Existing Trade Flows, Tariffs and Institutional Arrangements 10.2.1. Trade flows China is a major world trader, mainly with other major traders such as the EU, Japan, Korea, and the United States. Although both Indonesia and Vietnam do not sit amongst China’s major traders, China is a significant trader in their trades. In 2010, Indonesia had China as its third largest destination of exports and second largest source of imports whilst Vietnam had China as its fourth largest destination of exports and largest source of imports. Indonesia and Vietnam gain more from Chinese imports than the other way around. Chinese, Indonesian and Vietnamese trade in food show some similarities as well as differences (see Table 10.1 depicting trade flows and shares). Food export values had been growing at about the same rate until 2009 when those for both Indonesia and Vietnam fell despite high prices, again by about the same rate. This could be the result of the Global Financial Crisis lowering overall trade and countries increasing their self-sufficiency in 2009 following the shortages and price hikes in 2008. China’s food export values followed the same pattern but at lower rates, the 2009 fall only being around 1–2% prior to a large increase in 2010. Indonesian food export values increased back to around 2008 levels in 2010 but Vietnam’s continued to fall though only slightly. In general, food import values have also been increasing, apart from falls in 2007, Vietnam’s relatively more so than Indonesia’s. China followed Indonesia’s pattern but at a higher level. Net food trade followed the pattern of food export values for Indonesia and Vietnam, increasing up to 2008 and falling in 2009. In 2010, Indonesia rose back to 2008 levels whilst Vietnam fell further. China’s fell earlier and actually became negative in 2008, less so in 2009 but more so in 2010. A notable difference which offers one explanation of the counter movements in net food trade values in 2010 of Indonesia, and those of China and Vietnam, is in the share of food trade in total trade. Indonesia’s food export and import shares increased, exports quite substantially to nearly double in 2010 of what they were in 2000 (see Table 10.1). In contrast, Vietnam’s export share fell significantly by about a third whilst its import

March 5, 2013

224

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

Table 10.1. Chinese, Indonesian and Vietnamese Food Trade Flows and Shares of Total Merchandise Trade. Share Share 1990 China — food exports — food imports — net food trade Indonesia — food exports — food imports — net food trade Vietnam — food exports — food imports — net food trade

2000

2010

US$m US$m US$m US$m US$m 7,868 13,559 35,887 35,319 44,168 4,619 9,043 49,522 45,248 59,540 3,249 4,516 −13,635 −9,929 −15,372

% 5.4 4.0

% 2.8 4.3

5,526 24,090† 19,998 3,336 9,383† 8,639 2,190 14,707 11,359

25,630 11,470 14,160

8.4 7.7

16.2 8.5

12,487 11,823 11,352‡ 5,444 5,501 5,719‡ 7,043 6,322 5,633

25.3 5.2

15.7 6.7

2,853 1,104 1,749 na na na

2000

3,666 814 2,852

2008

2009

2010

Source: WTO International Trade Statistics 2012. † denotes break in series. ‡ denotes estimate.

share increased by around a third. Vietnam has obviously diversified its exports away from agriculture/food, as is normally the case with development, whilst Indonesian exports have concentrated relatively into agriculture/food. Vietnam’s pattern reflects that of China’s development path where exports declined more and imports rose less, both from lower share values. This relative diversification of Vietnam and China away from, and the relative concentration of Indonesia towards, agricultural/food production and trade, are also evident from looking at the relative Gross Domestic Product (GDP) per agricultural worker in all countries. In Indonesia, post the Asian Financial Crisis, GDP per agricultural worker halved, contrary to the usual pattern in growing economies, and this was in conjunction with no shortage of government policies and increases in related government expenditures. In Vietnam, as in China, labor has been successfully pulled out of agriculture, facilitated by the introduction of labor-saving techniques of production. As a result, agricultural GDP per capita grew due to the decrease in workers as well as an increase in value added. In Indonesia, agriculture’s

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

225

employment share exceeds its GDP share, which is indicative of the “labor shift” factor of a much lower productivity per agricultural worker relative to other workers, but this is common relative to other developing countries such as China and Vietnam to varying degrees. There are other notable differences in trade between China, Indonesia and Vietnam, more evident when adjustments to total trade are made for the different sizes of the countries in terms of populations and GDP. In terms of trade per capita, Vietnam’s is around 45% higher than Indonesia’s, US$1,764 over 2008–2010 compared to US$1,221. China’s is US$2,135, about three-quarters higher than Indonesia but only around 20% higher than Vietnam. The same relationship holds with the trade to GDP ratio which is a measure of a country’s openness to trade. The ratios were 154% for Vietnam versus 49% for Indonesia over 2008–2010, illustrating that Vietnam’s openness to trade is much greater than Indonesia’s. China’s was in between the two at 55%. More generally, the countries’ macroeconomic situations differ with indicators like Indonesia’s GDP per capita growth, as well as its degree and growth of trade openness, lagging that of China and Vietnam. China started its agricultural policy reform much earlier than Vietnam’s in 1978, transforming from a centrally planned economy to a socialist market economy (OECD, 2005). Since 1978 it has moved from a focus on increasing production (e.g., from lowering its self-sufficiency target below 100%) to income support and most recently to environmental objectives. Somewhat uniquely, its shift in employment out of agriculture, improving agricultural productivity, has been within the rural economy. Government still has an involvement, most significantly through state trading in grains which drives a wedge between domestic and world prices, input subsidies and price supports, which are the most inefficient and trade distorting forms of assistance, as well as more positive infrastructure investment. Vietnam has undertaken many macro-level reforms in its transition towards a market economy, for example closing or selling off many unprofitable State Owned Enterprises (SOEs), removing production and consumption subsidies from the state budget, as well as interest rate subsidies to SOEs (though some still appear to have preferential access to credit). The exchange rate was stabilized and devalued, raising incentives for exports which were encouraged

March 5, 2013

226

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

by progressively lifting barriers to trade, including inputs for agricultural production (there are now few restrictions on exports, tariffs are down to around 11–12%, and Quantitative Restrictions are on only 1.2% of imports) (OECD, 2010). In contrast, Indonesia still has a logistics agency in Bulog that controls trade, storage, distribution and so on in some key commodities like rice, and also funds production and consumption subsidies (including on interest rates through credit inputs) from the state budget.

10.2.2. Tariffs, non-tariff barriers and other trade-related policies As renegotiating tariffs are being looked at by Indonesia in its commitments under the ACFTA, it is useful to look at what has happened in the past with tariff reductions and trade in general, and more specifically in relation to China and agriculture. As mentioned in the last section, Vietnam’s final bound and applied simple average tariffs in 2010 were both around 10% and tariff binding coverage was 100%, as might be expected for a country that had recently undergone World Trade Organization (WTO) accession in 2007 (see Table 10.2 for details on such selected tariffs). For agricultural goods, both these tariffs were respectively around 18%. Indonesia’s final bound tariffs in 2010 were 37%, applied tariffs were much lower at around 7%, and tariff binding coverage was around 96%. For agricultural goods, they were respectively 47% and 8%. Vietnam’s bilateral applied tariffs on China’s exports are 5%, having recently come down from 23%, another reflection Table 10.2. Bound and Applied Simple Average Tariffs 2010.

Bound tariff Bound tariff agriculture Applied tariff Applied tariff agriculture Tariff binding coverage Applied tariffs on imports from China

China %

Indonesia %

Vietnam %

10.0 15.7 9.6 15.6 100 —

37.1 47.1 6.8 8.4 95.8 6

11.4 18.5 9.8 17.0 100 5

Source: WTO/ITC/UNCTAD (2010) and WITS.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

227

of its openness to trade with China, and Indonesia’s are 6%. Indonesian applied tariffs are lower and bound tariffs are higher than those for Vietnam. China’s tariff structure is more similar to Vietnam’s than Indonesia’s. If applied tariffs were reflecting the true relative levels of protection of Indonesia and Vietnam then this could be an explanation of why Indonesia was more concerned about trade with China than Vietnam is, regardless of the implementation of the ACFTA — it appears much more open to increased imports of Chinese products than Vietnam which seem to have higher domestic protection. But the relative tariffs go against the fact that China and Vietnam have had greater openness to trade than Indonesia as measured by the ratio of its trade to GDP. Moreover, the relative bound positions taken by the countries suggest Indonesia is much more cautious in its trade liberalization than China and Vietnam. The impression is that China and Vietnam have reformed more in the recent past (but from high levels of protection) and that this has been responsible for the large growth in trade. One possible explanation of this conundrum is that tariffs are only part of the trade constraints or barriers story. Countries may have low tariffs than do constraining trade between more than countries with much larger tariffs through the use of a maze of NTBs such as monopoly traders, licensing, anti-dumping actions, and restrictive Sanitary and Phyto-Sanitary (SPS) settings. As can been seen from the Table 10.3 of WTO notifications, measures in force and dispute numbers, after taking into account the relative sizes of the economies, Indonesia is much more active in anti-dumping, safeguards and disputes which are often in Table 10.3. Number of WTO Notifications and Measures in Force, and Number of Disputes. Indonesia Vietnam China Anti-dumping Safeguards Request for consultation (complainant-defendant) Original panel/Appellate body reports (“) Compliance panel/Appellate body reports (“) Arbitration awards (“) Source: WTO Country Profiles 2010.

16 9 5-4 2-4 1-0 0-0

— 0 2-0 1-0 0-0 0-0

115 0 8-26 6-8 0-0 0-0

March 5, 2013

228

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

areas where NTBs are prevalent. China has a lot of anti-dumping actions but it is the defendant in requests for consultation nearly three times more often than it is the complainant. There are more quantitative measures that incorporate tariffs and some NTBs, such as Nominal Rates of Assistance to producers (NRAs) which have been measured via comparisons of domestic and border prices across a range of agricultural commodities for many countries, including China, Indonesia and Vietnam, in a major World Bank project (Anderson and Valenzuela, 2009). NRAs for China, Indonesia, and Vietnam are provided in Table 10.4. Table 10.4. NRAs to all Agricultural Products, Indonesia, Vietnam and China, 1996 to 2005. 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 Indonesia −10 Vietnam −3 China 4

−7 −5 7

−24 −8 10

6 21 5

16 15 8

17 24 4

15 11 4

18 32 7

12 23 7

— 11 7

Note: Indonesian figures at farm level covered agricultural products including the fertilizer subsidy. Source: World Bank Agricultural Distortions Research Project.

The final available years of NRAs are around the same order for Indonesia and Vietnam, with China’s about half these, in contrast to the situation with the tariffs, as a result of taking account of NTBs. Moreover, Vietnam NRAs do not take into account intermediate goods produced by SOEs with high tariffs which would lower these NRAs. Given that Vietnam tariffs were nearly double than those of Indonesia, if Vietnam had no NTBs then Indonesian NTBs would have to be of the same order as its tariffs for the NRAs to be of the same order for both countries. China’s NRAs have been positive but low relative to Indonesia and Vietnam. However, not all NTBs will necessarily be taken into account in these NRA measures. For example, using the restricted issuing of licensing to constrain imports, as in the case of Indonesian beef, would most likely not to be picked up as a consistent NTB. Monopoly importers like Bulog often have similar non-transparent behavior. SPS and Technical Barriers to Trade

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

229

(TBT) issues are a grey area where it is difficult to differentiate between genuine health-related constraints and those that are basically aimed at protection (for example, see Bosworth and Cutbush (2010) on Australian SPS arrangements in relation to New Zealand apples). Anti-dumping is sometimes treated similarly despite it having little economic justification in terms of the predatory pricing argument which is rarely, if ever, proven in practice. Identifying all NTBs requires detailed analysis of the countries policies and their implementation. An upper bound approach to assessing NTBs is to use the difference between domestic and international prices, assuming none of the difference is due to aspects like differences in quality; that is, all of the difference is due to NTBs. Indonesia’s agricultural policies are focused on self-sufficiency and price stability, and mainly in respect of rice. Bulog, the Indonesian logistics agency, has acted as a monopoly trader (thus making tariffs irrelevant), undertaking domestic market purchases, stockholding, sales, and implementing floor and ceiling prices. High tariffs or import bans have also been imposed. These trade policies and Bulog’s operations have led to large nominal protection or assistance rates (Warr, 2005). Input subsidies feature prominently (some on fertilizer have been removed but then reimposed). These generally require complementary interventionist trade or border policy (e.g., the above-mentioned rice price support and sugar tariffs). Sugar policies also involve forced plantings, regulated distribution chain and import licenses. There are also constraints on major exports such as tree crops (export bans and taxes, coffee export quotas, and bio-fuel mixing regulations). One aspect evident from the NRAs table is that these have jumped around quite a lot, turning from negative to positive in recent years and varying yearon-year within such groupings. Changes in international prices offer some explanation, even when Indonesian policies do not change, and this aspect needs to be taken into account when trying to estimate representative costs of policies. But domestic factors are also at work. Fane and Warr (2007) offer a political economy explanation of this changing protection. In general, they observed that Indonesia has followed a pattern of “good economic times, bad policies and bad economic times, good policies” where good/bad policies refer to their “good” economic efficiency or their “bad” protection. During bad economic times, technocrats introducing economically efficient

March 5, 2013

230

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

policies, supported by institutions like the World Bank that needed to be on-side to encourage loans, held sway with the President. In good economic times, nationalists with popular support held sway with protectionist policies to support nationalist industries that were very expensive and only fundable during good economic times. This situation changed after 1998 when NRAs started another increasing stint following the Asian Financial Crisis with the move to a much more democratic and populist form of government that has reduced the influence of technocrats and promoted populist economic nationalism. Despite considerable liberalization due to policy reforms, Chinese agriculture still has policy distortions, mainly in the form of import tariffs but also through export subsidies, often disguised in the form of domestic marketing, transport and storage subsidies which are allowed under WTO rules for developing countries (Huang et al., 2007). Given this relatively high level of liberalization in the Chinese economy, including in agriculture, why does China have some sensitivities in agriculture? Again there is a political economy story. There were incidences, still in the living memory of some Chinese citizens, of food shortages due to famines and blockades that have given support to a grain self-sufficiency policy that is driven more by political than economic factors. Food selfsufficiency is ineffective in addressing such concerns (for example, being more vulnerable to local farming conditions, facing a growing threat of water shortages and other environmental constraints), as well as being economically inefficient, inequitable and more environmentally damaging. Political factors also influence the level of assistance provided to agriculture. The plight of Chinese peasants has been thought to pose a political threat to social stability, and grain self-sufficiency policies were seen as one means of assisting peasant farmers. The self-sufficiency policy is narrowly based on grains as it is not financially feasible to subsidize all of agriculture. Consumers tend to tolerate such assistance as on a per capita basis, the individual costs are small and consumption of grain staples is declining as a share of consumption. There are no electoral reasons as in the US and Japan for supporting agricultural assistance policies such as grain self-sufficiency. There are also no farmer associations to lobby for such support. Thus the grain self-sufficiency policy has weak and weakening underpinnings, and is susceptible to change, especially with China having had more recent

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

231

experiences of the economic benefits of opening up markets. Change is already happening with assistance to products such as soy and corn being removed following pressure from more efficient downstream industries seeking cheaper imported inputs, which offer better employment opportunities to peasants. Assistance is moving away from being production-related (self-sufficiency targets have been lowered from 100%) to income-related and more recently, also environmentally-related. Vietnam has become a major exporter of rice and coffee with its reforms. Unilateral liberalization under Doi Moi, which abandoned central planning for effective property rights over land and production decisions based on market signals, increased production incentives, production and in some cases exports. Vietnam subsequently entered into multilateral, regional and bilateral trade agreements following these more significant unilateral reforms. Vietnam’s recent levels of NRAs are dominated by one importable commodity, sugar, which receives high tariff and NTB protection. There is a political economy element behind Vietnam’s changing NRAs as well. There is not a “more democratic government” story here but a selfsufficiency or sectoral assistance one concerning a single commodity in sugar. Sugar has its own political determinants of a strict licensing regime for governing sugar imports and being the focus of government rural development and agricultural diversification programs. These political factors were strong enough to have survived the opportunity for reform during the WTO accession (Athukorala et al., 2007). Vietnam does not have, like Indonesia, a wider number of commodities that it assists or protects through input subsidies and border protection through tariffs and NTBs, as well as export sectors that it taxes — sugar, rice, dairy, livestock and so on in the first instance and tree crops like palm oil and cocoa in the second instance. Vietnam has diversified away from agriculture in terms of contribution to GDP, employment, SOEs, budget dependence and exports in what has become a very open economy with few export constraints and is highly dependent on trade. Agriculture, though still contributing significantly to the Vietnamese economy, is becoming less important politically than other sectors. There are other agricultural trade-related policies that would not be picked up in measures such as NRAs that can have a positive effect on agriculture, for example agricultural-related Research and Development

March 5, 2013

232

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

(R&D). There is evidence in Indonesia of a slowdown in agricultural production and potential trade as a consequence of a long-run downward trend in related public investment — the growth rate in spending on agricultural research is negative (Cervantes-Godoy and Dewbre, 2010). In contrast, in China and especiallyVietnam there has been a rapid growth in government investment in R&D which is felt to have contributed to the growth in food production and trade, for example in aquaculture.

10.2.3. The ACFTA agreement — exemptions for sensitive and highly sensitive products The ACFTA was signed in 2002 and renegotiated in 2006 when the more recent ASEAN members, Vietnam, Cambodia, Laos and Myanmar, specified their exemptions for sensitive and highly sensitive products. Implementation was to commence in 2010. As far as trade in goods is concerned, tariff reductions (based on in-quota Most Favored Nation (MFN) rates) are phased in over a number of years. Tariffs on products in the sensitive list were to be reduced to 20% by 2012 and to between 0% and 5% within the implementation period, and highly sensitive track products were to be reduced to a maximum of 50%. Each ASEAN member has a different list of exemptions. Countries tend to exempt products with high tariffs although not exclusively (see Scollay and Trewin (2006) for analysis of this issue in ASEAN which showed member states exempt products that they did not need to protect for survival, as well as products that were always going to require protection to survive). Indonesia has 47 exemptions, most notably in Chapters‘10 (rice), 17 (sugar), 22 (alcohol), 64 (footwear), and 87 (motor vehicles) (ASEAN Secretariat, 2006). Indonesia is currently renegotiating its highly sensitive list. This involves removing some items and replacing them with others. It must get an agreement with China before the list can be revised and this is looking more unlikely to be granted as time goes by. Less developed Vietnam was allowed 150 items in its highly sensitive list plus a longer implementation period. The main chapters include 17 (sugar), 24 (tobacco), 40 (rubber), 69 (ceramics), 70 (glass), 72 (steel), 84 (motor bikes), 85 (audio devices), and 87 (motor vehicles). China with its much broader and larger economy has 101 items in its highly sensitive list. The main items are Chapters 10 (rice), 11 (maize),

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

233

15 (oils), 17 (sugar), 24 (tobacco), 40 (rubber), 44 (wood products), 48 (paper products), 52 (cotton), and 87 (motor vehicles). These exemptions are specified at the six-digit level from a possible list of 5,113 tariffs (so for example, Vietnam’s sensitive list is about 3% in number of tariff lines but is generally much larger in terms of the domestic production they are attempting to protect). Bilateral tariffs reductions are calculated at the six-digit level, using the Gempack utility TASTE, and aggregated to the 23 user-specified GTAP sectors shown in Table 10.5a and Table 10.5b. The bilateral tariffs before and after the simulations are Table 10.5(a). Base and Final Indonesian and Chinese Bilateral Tariffs. Indonesian Tariffs on Imports from China

Chinese Tariffs on Imports from Indonesia

Sector

Base %

Final %

Base %

Final %

Rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables, fruit and nuts Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages and tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufactures

20.0 1.2 4.9 0.7 35.1 5.0 4.7 4.7 5.1 4.9 2.3 5.2 4.9 5.8 28.3 10.2 5.6 6.6 5.8 6.3

20.0 1.2 4.9 0.6 35.1 5.0 4.7 4.7 4.5 4.9 2.3 5.0 4.9 5.8 20.1 5.0 3.8 3.6 4.5 3.9

0.0 0.0 5.2 2.6 7.0 7.4 7.2 2.9 5.8 2.8 0.8 6.2 3.8 6.8 11.6 7.1 8.3 3.8 3.1 6.1

0.0 0.0 3.4 1.0 4.3 4.2 3.8 1.3 2.9 1.0 0.7 3.0 1.5 3.0 4.2 4.0 6.6 3.3 2.7 1.8

Source: GTAP version 7 database and author’s calculations.

March 5, 2013

234

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

Table 10.5(b). Base and Final Vietnamese and Chinese Bilateral Tariffs. Vietnamese Tariffs on Imports from China

Chinese Tariffs on Imports from Vietnam

Sector

Base %

Final %

Base %

Final %

Rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables, fruit and nuts Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages and tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufactures

20.3 3.1 5.2 2.1 20.6 15.1 13.9 5.8 4.2 10.7 17.9 10.0 15.1 19.4 78.4 12.8 2.4 6.5 15.3 14.2

5.0 3.1 3.4 0.4 18.9 5.0 12.9 4.9 4.0 5.0 4.9 5.0 5.1 5.2 70.4 5.0 1.6 4.5 6.1 10.2

62.4 16.2 8.0 21.2 6.9 13.5 9.1 4.2 6.2 4.1 0.4 10.7 1.6 7.6 4.3 10.0 12.2 5.7 1.6 6.2

45.8 11.9 5.0 5.9 4.4 4.6 4.5 1.7 3.1 1.7 0.4 3.0 0.5 3.1 2.0 4.1 10.3 3.6 1.5 2.6

Source: GTAP version 7 database and author’s calculations.

shown in this table. From an Indonesian perspective, the most significant changes are for “Beverages and tobacco” and “Textiles & apparel”. Notably, there are no changes to rice and sugar, both of which have relatively high tariffs. From the perspective of Indonesia’s exports to China, most tariffs are reduced to less than 5%. From aVietnamese perspective, the most significant changes are for agricultural products of “Rice” and “Vegetables, fruit and nuts” as well as “Textiles and apparel”. There are relatively small changes to highly protected “Sugar” and “Other crops”, as well as “Beverages and Tobacco”. From the perspective of Vietnam’s exports to China, most tariffs are reduced to less than 5% with the exception of “Rice”, which maintains a

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

235

very high tariff, “Other cereals”, reflecting China’s history of a strong grain self-sufficiency policy and its protection against competitive suppliers like Vietnam, and “Chemicals”.

10.2.4. Some other relevant aspects of FTAs It is unusual for FTAs like ACFTA to address NTBs, though under the Australia New Zealand Closer Economic Relations Trade Agreement (ANZCERTA), a strongly economically integrated FTA between two longterm trading neighbors, anti-dumping is handled in a more economic way as part of competition policy. Generally, current WTO arrangements such as in relation to anti-dumping, SPS, and so on are accepted under FTAs. Other agricultural trade-related policies, such as trade facilitating R&D support, are generally not part of FTAs, though under the ASEAN Australia New Zealand FTA (AANZFTA) there has been some R&D funding through the ASEAN Secretariat to assist ASEAN Member States in strengthening the trade agreement. An example of this is funding a diagnostic study of constraints in trade in services and prioritizing capacity building that will assist trade liberalization. FTAs are more about the political economy than trade liberalization — “many tend to be ‘trade light’ tools of foreign policy and diplomacy” (Sally, 2008a, 2008b). Shifts in trade policy cause redistribution of gains and losses between sectors, regions, and socio-economic groups. Given these aspects, it is not surprising that China, Indonesia and Vietnam with their different political economies display different attitudes to the ACFTA. With strong political economy drivers, politically sensitive sectors and associated protection policies such as anti-dumping, SPS, and TBT are carved out of FTAs.

10.3. The Model The Global Trade Analysis Project (GTAP) model is used to measure the impact of changes in trade policy on the traded-goods sector. GTAP is suited for modeling preferential trade agreements because it contains bilateral trade and tariff data. It can also handle non-tariff measures if these can be converted into ad valorem equivalents. However, it has difficulty incorporating Rules of Origin (ROOs) in its analysis as the increase in the costs of production in specific sectors would need to be known. ROOs in

March 5, 2013

236

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

the ASEAN FTA tend to be flexible but this may not be the situation in the ACFTA. Ignoring the impact of ROOs requires the results to be qualified that they are overstated to the extent that ROOs limit imports of cheaper goods. A similar issue applies to utilization rates of the preferences which are low in the ASEAN FTA and would suggest cuts in preferential tariffs may not have the impact that similar cuts in MFN tariffs would have and may need to be qualified as “outer envelope” impacts. GTAP is a multicountry and multi-sectoral CGE model and it is fully documented in Hertel and Tsigas (1997). For each country or region, there are multistage production processes which combine primary factors of land, labor, capital and natural resources with intermediate inputs assuming a constant elasticity of substitution technology. Returns to factors, i.e., income, are taxed by the government, saved or spent by the single representative household. While there is no substitution between intermediate inputs and primary factors or among the intermediate inputs, there is substitution between different sources of intermediate inputs, namely domestic, and imports from each region. The regions are linked together by imports and exports of commodities. Similar commodities, which are produced by different countries, are assumed to be imperfect substitutes for one another. The degree of substitution is determined by Armington elasticities. In this application, the standard closure, or choice of exogenous variables set outside the model, is modified to allow capital to flow between countries in response to changes in demand for capital intensive goods. In addition, a semi-flexible labor market for unskilled labor is assumed, implying a change in the demand for labor leads to some increase in both wages and employment. Skilled labor is assumed to be mobile in each country but in a fixed supply, with no surplus labor. This is the standard GTAP closure. GTAP is used here to compare the trade and welfare effects of changes in bilateral tariffs once the impacts have worked through. There is no attempt to phase in the tariff changes nor trace the time profile of the impacts. Thus, we ignore changes such as growth in trade that may have occurred over the implementation period, but we incorporate differential changes in productivity suggested to be the result of differential expenditures on R&D as separate shocks to capture the effect of such changes over the implementation period. The main focus here is on changes in tariffs as outlined in the schedules. We also attempt to capture the impact of NTBs

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

237

such as mentioned earlier and other quantitative restrictions such as import bans or quarantine restrictions that result in differences between domestic and border prices in some separate scenarios. The regions used in the model are the European Union, the United States, Japan, Australia, Other Developed, China, Indonesia, Malaysia, the Philippines, Thailand, Vietnam, Rest of ASEAN, South Asia, Central America, Africa, and Rest of the World. The sectoral aggregation is shown in Table 10.8. This is similar to Table 10.5a and 10.5b with the addition of services. Four scenarios are modeled here: (i) FTA as negotiated. This involves reducing to no more than 5%, all the tariffs between China and Vietnam and Indonesia as of 2007 (when AFTA was in place, which is incorporated in the analysis unlike more recent FTAs such as the AANZFTA), with the exception of those in the highly sensitive list.2 These are reduced to a maximum of 50% if they are above this or left unchanged otherwise. (ii) FTA without exemptions. All tariffs reduced to no more than 5%. (iii) Productivity. Scenario 1 plus annual productivity increases of 3.7% for China, 2.9% for Vietnam, and 1.5% for Indonesia.3 (iv) NTB. Scenario 2 with nominal rates of assistance (NRAs) used to determine tariff equivalents for Indonesian sugar. In the absence of definite data, we use a baseline rate of 400% and reduce this to 5% for all potential exporters.4 2 The transaction costs associated with obtaining preferential access have been estimated at

around 5% on average so that a more realistic scenario to capture the utilization rates would be to reduce the (effective) tariffs over 5% to only 5% rather than zero, and leave those under 5% unchanged. 3 These were annual estimates for agriculture obtained from Fuglie (2008) applied over the whole period of simulation and to all sectors which will isolate the individual country impacts but which could be refined in further simulations. 4 The NTB values are of the order estimated in some earlier research (Warr, 2005) and supported by recent research (Marks and Rahaardja, 2012) but again are mainly used here to illustrate the relative impacts of NTBs. Rice estimates of 200% were also available from this source but the 2004 data in GTAP had Indonesian rice imports unrepresentatively low and this low base led to an overstatement of absolute increases in imports and other distorted estimates following the shock of the “removal of NTBs” so rice was dropped from the scenario. This left the sugar shock to illustrate the significance of removing NTBs. Non-agricultural NTBs have not been incorporated into the analysis. Data on NTBs is hard

March 5, 2013

238

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

10.4. Results The estimated annual changes in welfare under the scenarios are shown in Table 10.6. The changes are positive, suggesting each country benefits from the tariff reductions. This need not always be the case. FTA agreements can make members worse off because of trade diversion and losses in tariff revenue. Non-members can be worse off for similar reasons, and these losses are a common criticism of such agreements. China gains the most in absolute terms, by virtue of having the largest economy. Relative to the size of its economy, Vietnam benefits most. Overall, members of the FTA benefit by $7,154 million, but global gains are little more than this, indicating that non-members do not share the gains. Indeed, Japan and “Other Developed Countries” are negatively affected. Table 10.6. Welfare Impacts.

China Indonesia Vietnam

FTA as Negotiated $m

FTA Without Exemptions $m

Productivity $m

NTB $m

3,421 759 403

4,446 908 527

436,016 22,778 5,709

4,432 1,910 526

Source: GTAP simulation.

to obtain (e.g., the tariff equivalent of recent Indonesian horticultural NTBs of restricting ports of entry for imports). Some estimates have been derived from differences between actual trade and estimates based on gravity model predictions that do not incorporate NTB effects on trade (Marks and Rahardja (2012) apply a combination of price differences and some individual NTB’s impacts). Feridhanusetyawan (2005) using World Bank Trade Restrictiveness Indices found NTBs at an aggregate level were correlated with tariffs. However, this was not the case in agriculture, where NTBs tend to be concentrated (Anderson and Valenzuela, 2009), Petri et al. (2010) assumed all NTBs were 120% and went down to 60% with liberalization whereas this simulation assumes a less uniform shock of sugar at 400% going down to a universal level of 5%. There is a non-linear relationship between tariffs (or tariff equivalents), imports and welfare so different assumptions like sugar NTBs reducing to the baseline of 120% instead of 5% could result in quite different relationships between the figures provided in the results. A chart shows the impacts of sugar imports and welfare if the NTBs were reduced by the full amount.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

239

All non-members experience a fall in exports, although the magnitudes are small, less than 0.1% in each case. In welfare terms at least, all countries would have done better by removing tariffs on their highly sensitive products. This can be seen by comparing the first two scenarios in Table 10.6. As negotiated, all three countries capture around three quarters of the available gains. The remainder represents the cost of protecting a few particular sectors. The third scenario shows the benefits of productivity growth. In fact these benefits swamp the allocative efficiency gains from trade liberalization, although the technical change enhances the allocative efficiency effects and the value of additional endowments, labor, and capital. However, there are negative terms of trade effects. Reducing the Indonesian sugar NTBs, estimated to be equivalent of 400% tariffs for all potential exporters, down to 5% increases Indonesian annual welfare gains by over double to $1,910 million. China’s and Vietnam’s welfare gains hardly change as might have been expected with a purely Indonesian shock and little sugar trade between the countries. Under this scenario, domestic production of sugar is estimated to fall from $1,594 million to $310 million. Sugar imports rise from $102 to $470 million and imports account for more than half domestic production. The sugar self-sufficiency ratio falls from 76% to 43%. Thailand continues to supply most of the additional sugar as expected seeing as the shock was the same for all exporters. In Indonesia, producer prices for sugar fall 19%, taking into account that domestic and imported goods are regarded as differentiated products (Armington assumption). Indonesia currently has some small exports ($20 million) which could be re-exports or have some value-added which increase with the price fall as would other downstream sector-like beverages. The source of the welfare changes is shown in Table 10.7. The bulk of the welfare gains stem mainly from using resources better (allocative efficiency); using resources that were previously under-utilized (endowments); and more favorable prices for imports or exports (terms of trade). For Indonesia, the second scenario of no exemptions delivers few additional allocative efficiency gains (many tariffs are under 50% so removal of exemptions does not lower them), but there are slight improvements in its terms of trade and an increased demand for unskilled labor-intensive products.

March 5, 2013

240

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

Table 10.7. Source of Welfare Gains. Allocative Efficiency $m FTA as negotiated China Indonesia Vietnam FTA without exemptions China Indonesia Vietnam Productivity China Indonesia Vietnam

Endowments $m

Terms of Technical Trade Change $m $m

Total $m

560 151 273

2,172 561 267

773 49 −133

0 0 0

3,421 759 403

822 177 338

2,831 677 367

897 55 −155

0 0 0

4,446 908 527

55,703 2,517 1,443

171,819 11,896 1,667

−17,282 −450 −525

221,482 8,606 3,278

436,017 22,778 5,709

819 780 337

2,823 1,174 366

896 −55 −155

−105 11 −22

4,432 1,910 526

NTB China Indonesia Vietnam Source: GTAP simulation.

Vietnam makes some allocative efficiency gains (mainly from its resources, textiles and manufacturing sectors) but its terms of trade decline further. China gains from all three sources. This is mainly related to trade with matching up Vietnam, with Vietnam gaining more, relative to the size of its economy, than Indonesia from the tariff reductions. Removing NTBs of 400% on sugar significantly improves Indonesia’s welfare, again mainly in terms of allocative efficiency and endowments. These measures, assuming they could be removed, would increase Indonesia’s welfare by as much as all the tariff measures combined. To show the trade importance of the factors captured in the scenarios, the change in exports and imports by sector and for each economy in total are shown in Tables 10.8 and 10.9. China’s increase in aggregate exports of 0.7% is three-quarters of what could be achieved without exemptions,

March 14, 2013

Table 10.8. Change in Exports. China

11.2 0.0 0.6 1.5 −0.5 2.5 0.5 0.0 −0.4 0.3 5.0 4.5 −1.1 2.4 6.4 1.3 1.1 1.1 0.7 0.5 −0.1 −0.3 −0.2 0.74

19.4 0.1 0.4 1.5 48.8 4.0 12.9 −0.2 −0.6 0.3 5.0 4.2 5.3 2.5 27.4 1.2 1.4 1.2 0.7 0.9 −0.1 −0.4 −0.3 0.96

−1.59 −0.87 −1.05 0.6 −0.24 −1.12 −0.68 2.7 5.53 0.93 −0.28 −1.01 −0.91 −0.2 −6.38 0.69 1.25 0.08 −0.33 3.54 −0.28 −0.39 −0.3 0.93

FTA Without Exemptions (%) 4.11 −0.64 −0.55 0.57 0.11 −1.16 −3.15 3.06 5.52 0.98 −0.32 −0.95 −1 −0.2 −21.7 0.59 3.65 −0.01 −0.34 3.43 −0.32 −0.45 −0.34 1.03

FTA (%)

FTA Without Exemptions (%)

1.2 −0.33 −2.96 38.84 0.46 3.55 −0.54 1.85 6.6 −0.78 1.32 1.03 4.48 0.89 −1.54 5.57 3.09 1.64 0.21 3.09 4.09 −0.66 −0.42 2.94

7.27 2.64 −4.28 45.5 1.14 2.61 −1.52 0.83 6.33 −0.79 1.43 0.66 3.14 0.44 −5.27 6.21 12.39 3.26 0.72 5.69 4.29 −0.33 0.3 4.03

241

(Continued)

Regional Development and Economic Growth. . . b1491-ch10

FTA (%)

9in x 6in

FTA Without Exemptions (%)

11:1

FTA (%)

Vietnam

Individual Country Approaches to Agriculture in the ASEAN–China FTA

Paddy rice & proc rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables and fruit Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages & tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufacturing Transport & communications Business services Services etc NES Total

Indonesia

−15.8 −18 −12.4 −1.2 −2.8 −5 −14.6 19.3 32.8 13.3 3.8 −13.2 −4.4 2.4 −5.3 8.2 13.5 9.7 6.2 17.1 2.8 1.9 2.7 8.4

6.1 0.7 0.8 1.4 108.9 −0.1 −1.8 4.4 5.3 0.6 −0.5 1.6 0.7 2.0 −20.7 1.2 4.4 0.4 0.2 4.0 0.0 −0.1 0.1 1.6

Productivity (%) −3.2 −9.1 −9.5 64.8 5.7 8.6 −5.2 −1.6 36.3 −2.3 6.5 −5.9 28.4 2.6 −0.6 28.8 18.2 16.5 5.4 17.5 8.7 −4.3 0 14.4

NTB (%) 7.3 2.7 −4.3 45.3 2.2 2.6 −1.5 0.8 6.3 −0.8 1.5 0.7 3.2 0.4 −5.3 6.2 12.4 3.3 0.7 5.7 4.3 −0.3 0.3 4.0

Regional Development and Economic Growth. . . b1491-ch10

19.2 0.1 0.4 1.5 22.6 4.0 12.8 −0.2 −0.6 0.3 5.1 4.2 5.3 2.4 27.4 1.2 1.4 1.2 0.7 0.9 −0.1 −0.4 −0.3 1.0

NTB (%)

9in x 6in

Source: GTAP simulation.

−7.2 −27.4 −28 0.7 −21.8 −15.5 −25.2 −25.1 −31.2 −14.4 −0.2 2.3 −34.1 −8.4 13 8.8 17.9 20.7 8.8 22.1 4 3 4.3 14

Productivity (%)

11:1

Paddy rice & proc rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables and fruit Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages & tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufacturing Transport & communications Business services Services etc NES Total

NTB (%)

Vietnam

March 14, 2013

Productivity (%)

Indonesia

Regional Development and Economic Growth in China

China

242

Table 10.8. (Continued)

March 14, 2013

Table 10.9. Change in Imports. China

50.3 1.1 0.5 1.5 1.1 5.2 1.8 1.5 1.0 1.2 1.2 0.8 1.9 2.2 0.8 1.5 1.7 0.9 0.8 1.0 0.5 0.5 0.5 1.1

2.9 0.8 1.2 1.0 0.5 0.9 0.2 1.0 0.4 0.8 0.4 1.4 1.2 0.6 −0.2 3.6 1.1 1.7 0.3 1.2 0.6 0.7 0.6 1.0

FTA Without Exemptions (%) 0.6 0.7 1.1 1.0 0.8 0.8 −0.2 1.0 0.5 0.9 0.5 1.5 1.3 1.0 −1.4 3.7 1.3 1.7 0.3 1.3 0.7 0.8 0.7 1.2

FTA (%) 396.3 1.0 9.8 0.9 1.0 6.2 2.6 1.1 1.2 4.1 4.1 1.9 5.4 2.5 1.0 6.8 1.9 2.1 2.4 2.4 −1.1 1.6 1.9 3.1

FTA Without Exemptions (%) 449.2 0.8 11.8 1.2 1.1 6.6 5.1 2.0 1.9 4.1 4.7 2.1 6.5 2.6 1.2 7.3 2.9 2.6 2.8 4.8 −0.8 1.9 2.1 4.2

243

(Continued)

Regional Development and Economic Growth. . . b1491-ch10

3.6 0.9 0.5 1.4 0.8 5.1 1.2 1.2 0.9 1.1 1.1 0.7 1.5 2.1 0.5 1.4 0.9 0.7 0.7 0.9 0.4 0.4 0.4 0.9

FTA (%)

9in x 6in

FTA Without Exemptions (%)

11:1

FTA (%)

Vietnam

Individual Country Approaches to Agriculture in the ASEAN–China FTA

Paddy rice & proc rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables and fruit Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages & tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufacturing Transport & communications Business services Services etc NES Total

Indonesia

54.3 15.8 14.6 7.6 9.3 20.8 11.1 16 11.2 18.2 7.2 17.5 13.6 7.5 4.8 12.3 7.1 8.9 5.5 9.2 6.2 7.1 9 8.6

−4.3 1.0 1.0 1.1 358.2 0.4 0.1 0.5 1.2 1.9 0.8 0.6 0.8 0.1 −1.4 4.0 1.4 2.0 0.6 1.6 0.9 1.0 0.8 1.8

Productivity (%) 558.5 6.3 26.9 11.1 9.2 25.1 18.8 22.3 13.9 26.2 18.7 13.9 30.4 10.6 10.5 26.1 11.5 13.6 11.6 13.6 7.5 17.5 17.8 15.9

NTB (%) 448.7 0.8 11.7 1.2 4.6 6.6 5.1 1.9 4.1 1.1 4.7 2.1 6.5 2.6 1.2 7.3 2.9 2.6 2.8 4.8 −0.8 1.9 2.1 4.2

Regional Development and Economic Growth. . . b1491-ch10

49.2 1.1 0.5 1.5 0.9 5.1 1.8 1.5 1.0 1.2 1.2 0.8 1.9 2.2 0.8 1.5 1.7 0.9 0.8 1.0 0.5 0.5 0.5 1.1

NTB (%)

9in x 6in

Source: GTAP simulation.

103.4 35.7 19 12.2 37.7 59.2 28.6 44.7 49.7 58.8 20.5 15.5 52.3 27.8 17.2 10.8 9.2 9 11.9 13.9 11 14.3 18.4 14.4

Productivity (%)

11:1

Paddy rice & proc rice Other cereals Oilseeds Vegetable oils and fats Sugar Vegetables and fruit Other crops Livestock Forestry Fishing Petroleum and coal products Ruminant meat Non-ruminant meat Other processed agriculture Beverages & tobacco Textiles & apparel Chemicals Metal manufactures Wood & paper products Manufacturing Transport & communications Business services Services etc NES Total

NTB (%)

Vietnam

March 14, 2013

Productivity (%)

Indonesia

Regional Development and Economic Growth in China

China

244

Table 10.9. (Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

245

whereas Indonesia, who is not a big trader with China in rice and sugar, has less scope to improve (in fact there is some “policy space” that would allow tariffs to increase up to the sensitive sector level of 50% under Scenario 1). However, using NTBs against Indonesian sugar imports has a significant effect at a sectoral level. Vietnam’s exports of 2.9% under Scenario 1 are also short of its potential, 4% without exemptions. For Indonesia, the largest relative changes under Scenario 1 are in the non-agricultural sectors of forestry (5%) and manufacturing (3%). Vietnam shows significant growth in a number of areas, most notably vegetable oils and fats, forestry, vegetables and fruit, non-ruminant meat, textiles and apparel, and manufactured goods. For China, rice, sugar and beverages and tobacco could increase markedly if ASEAN countries opened up their markets completely as modeled in Scenario 2. Also important is the utilization rate. Here it is assumed that importers would not take advantage of any tariffs below 5% because of the transaction costs involved. Hence, in the simulations the standard tariff reduction for non-exempt goods is to a maximum of 5%. This assumption makes quite a difference, and effectively cuts the exports and welfare gains almost in half. This is because a large share of tariffs is at 5% or less, and a large share of trade occurs at these rates. Avoiding such transaction costs, for example through unilateral liberalization, would lead to large welfare gains. On the import side, there are no significant increases in Indonesian imports following the various FTA scenarios, with the possible exception of textiles and apparel. Removal of NTBs on sugar would lead to a large increase in sugar imports. If Indonesia is not required to reduce support for sugar, and thus few jobs are thought to be at risk, there is a question as to why Indonesia is expressing concern with the negotiated arrangements under ACFTA, ignoring for the time being any change in the political economy towards greater opposition to trade liberalization. The modeling shows more significant import increases for Vietnam, particularly vegetable oils and fats, non-ruminant meats, textiles and chemicals. Comparing the first two scenarios shows where the protection is maintained by the exemptions — other crops, livestock, and manufactures (which includes motor vehicles). One line of thinking on the insignificant increase in Indonesian imports is that in spite of rather low tariffs, some Indonesian agricultural sectors are protected by high NTBs. Removing NTBs against Indonesian sugar

March 5, 2013

246

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China 2500 2000 1500 $m, %

Welfare $m Imports %

1000 500 0 0

-10 -20 -30 -40 -50 -60 -70 -80 -90 -99 Sugar tariff equivalent cut

Figure 10.1.

Indonesian Sugar Imports and Change in Welfare.

Source: GTAP simulation.

imports has a significant effect at a sectoral level. Figure 10.1 shows the estimated change in sugar imports (%) and national welfare ($m) assuming tariff equivalents of non-tariffs barriers of 400% on sugar imports from all sources were reduced in intervals of 10% of the base, i.e., from 400 to 360, 320, and so on. Imports would increase in an almost linear fashion. National welfare would increase in a similar linear fashion to a nearly $2 billion annual change.

10.5. Implications, Limitations and Concluding Comments In comparing the response of China, Indonesia and Vietnam to the ACFTA, it seems Vietnam has obtained greater protection from China with its exemptions than Indonesia. Without exemptions, Vietnam’s imports would rise from 3.1% to 4.2% whereas Indonesia’s would increase only marginally (Table 10.9). China’s imports also rise only marginally from 0.9% to 1.1%. Less developed and more diversified Vietnam has 150 products in its sensitive list whereas Indonesia has only 47, but China with its broader and larger economy has 101. A more important reason why Vietnam may have obtained a greater degree of protection than Indonesia is that it had higher tariffs to start with. Indonesia’s average applied agricultural tariff is 8%

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

247

compared with Vietnam’s 17% (Table 10.2). On trade with China, Indonesia has high tariffs on rice and sugar, but low tariffs on most other goods. By contrast, Vietnam has much higher tariffs across a range of imports from China and vice versa. But this conclusion of greater tariff protection changes once NTBs are brought into the assessment (Table 10.6). Removing NTBs from the Indonesian sugar sector has large positive welfare impacts although the additional imports come from other ASEAN countries rather than China (there is little trade with China in this commodity). The potential gains, the absence of an overall surge in imports plus the apparent difficulties in removing NTBs suggest extreme caution should be taken with introducing any new NTBs as these would lead to the gains from tariff liberalization not being optimized. NTBs have a significant effect on aspects such as allocative efficiency at the sectoral level at which they are applied. Benefits of productivity growth dominate those from allocative efficiency but allocative efficiency enhances this domination along with additional resource endowments (Table 10.7). In the Introduction, the question was asked about the reason behind Vietnam’s preference to ACFTA than Indonesia. It was proffered that this could be a result of Vietnam’s location next to China and the threat of informal trade without any trade agreements, plus its recent experience of the benefits of trade liberalization, for example in the form of cheap imports. Evidence of Vietnam’s greater acceptance of ACFTA was obtained from an industry survey in a recent (Vanzetti et al., 2010) study of Vietnam FTAs where threatened industries, such as pulp and paper, remained optimistic that they could develop a niche off cheap Chinese inputs. Indonesia is not a neighbor of China and the benefits of trade liberalization are not as recently evident. And as just outlined, it has committed to bigger tariff cuts under ACFTA than Vietnam which would concern some protected industries even though the Indonesian economy would be the biggest beneficiary of these cuts. There has also been a large change in the Indonesian political economy with a move to a decentralized political system that has given greater power to minority interests including those that represent agriculture and see a threat in opening up Indonesian agriculture to greater international competition. The political economy matters. Perhaps because of its lack of tariff protection (as distinct from less transparent and higher protection from NTBs), Indonesia is currently attempting to renegotiate its highly sensitive

March 5, 2013

248

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

list with China. However, if items are to be included in the list, others must be removed. This creates an inevitable trade-off among Indonesian domestic producers, and China must be persuaded to agree. To date, getting this agreement has proved increasingly difficult. As with all modeling, the analysis has limitations. Producers and consumers may not respond to tariff changes as readily as the modeling suggests. Furthermore, the tariff changes modeled here may not occur causing the estimates to be “outer envelope” ones (PC, 2010). Already we have seen further negotiations trying to slow down the reform process which can also occur through the greater use of NTBs outside the ACFTA. This “outer envelope” approach was defended in the PC inquiry (PC, 2010) by the CIE on the basis that modeling what was expected from the negotiations in terms of less ambitious liberalization would be pre-empting them and lead to less ambitious outcomes. At any rate, the government accepted the PC recommendations that transparent and credible modeling be undertaken at the feasibility stage, and there need not be any independent body overseeing the modeling and any further assessment following the final text at the end of the negotiations and before signing (DFAT, 2011).5 Modeling large shocks such as a reduction in an NTB of 400%, is somewhat speculative. A constant elasticity of substitution is assumed although it is not clear that this functional form would hold for such large changes. Nor is it clear that all the NTBs would apply to all importers as assumed here, though something like the closure of Bulog would result in a more level playing field for all importers. Finally, some “water” in the tariff might exist, for example the barriers might be just as prohibitive at 100% as at 400%. If so, any change in imports would be an overestimate. There are two important groups of implications from the above analysis, both for GTAP modeling and for Australian and other countries’ negotiations with countries like China, Indonesia, and Vietnam. The GTAP modeling implications include that if only tariff trade constraints are available in the database for analysis then the results of the modeling can be misleading in terms of the benefits of trade liberalization. Other trade constraints such 5 The Policy Statement is not clear on these last aspects, stating in the main text that an

independent body and further assessment should take place but stating the opposite in the table of responses to the PC recommendations on these aspects.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

249

as NTBs need to be incorporated, especially in situations like in Indonesia where these dominate the trade constraints, if their costs, or the benefits of trade liberalization, are to realistically estimated. Even where just modeling of reductions of tariffs is appropriate, care needs to be taken to avoid the “outer envelope” criticisms of CGE modeling of FTAs, for example assuming liberalization is fully implemented when this is most likely not to be the case (PC, 2010). There are also important implications for trade negotiations from the analysis. The negotiated agreements need to be comprehensive, not only in terms of covering agriculture and other goods, and services that enable trade-offs in the political economy, such as in agricultural liberalization in the longer term with better services access under TAFTA (Bosworth and Trewin, 2008), but also in terms of trade constraints. Tariffs are not the whole trade liberalization story and as their importance has diminished, NTBs’ has tended to grow. However, NTBs have proved difficult to address in FTAs as, like with many service trade constraints, they are entwined with domestic policies which are best addressed via unilateral liberalization. What needs to be shown is that such measures are not in the best interests of the country imposing them.6 Some NTBs were addressed in the AANZCERTA where it was agreed anti-dumping should be outlawed and related concerns handled through competition policy. Indonesia has liberalized tariffs but has not opened its agriculture to international competition and realized the benefits of reallocating resources to better uses and developing fully its domestic agriculture to compete internationally. There are better ways for countries like Indonesia, as well as China and Vietnam, to achieve its legitimate objectives in agriculture and it needs to be encouraged to unilaterally reform in its own interest, not on the basis of any trade agreement. Trade facilitation is important in this so if trade agreements offer help in this regard, such as with R&D that increases productivity as in the AANZFTA, then this should be encouraged. At least this would lower the costs of closed policies if they were maintained but at the same time encourage more openness to take 6 In the Australian Productivity Commission’s latest annual Trade and Assistance Review (PC, 2012), it warns that assistance is growing in the form of less transparent budgetary measures like regulatory restrictions on competition, mainly in agriculture, and not all assistance, such as SPS, is captured in these estimates.

March 5, 2013

250

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

advantage of growing trade opportunities. Such approaches can address the important political economy constraints. Why are Indonesian tariff cuts under ACFTA of concern when in reality they would have little impact because of the prevalence of NTBs? Do stakeholders working against liberalizing not appreciate what little impact tariff cuts have with a prevalence of NTBs due to a lack of transparency? Or, as is often the case in trade reform, do a few that are under threat of being disadvantaged complain more loudly than a silent, often hard to organize majority of consumers and others who would gain? “Others” disadvantaged by high domestic commodity prices would include value-added sectors such as drink manufacturers in respect of sugar. These manufacturers do not have the same political economy strength as the commodity groups. Or is it just a reflection of misplaced Indonesian skepticism of the benefits of trade liberalization which they have enjoyed in the past, mainly unilateral liberalization, but are not fully aware of now, unlike Vietnam where the benefits are more recently obvious? Fane and Warr (2007) put forward a credible answer to the question of why there are different Indonesian and Vietnamese responses to the ACFTA, which is that a changing political economy towards populist economic nationalism, or even provincialism, which is anti-trade followed the decentralization of Indonesia’s political system. Finally in relation to Chinese agricultural sensitivities, these seem a relic of a past where poor economic signals and a closed economy (internally and externally) led to grain shortages and strong support for (provincial and national) grain self-sufficiency. The strength of this grain self-sufficiency policy with its political economy origins is starting to wane as the Chinese economy prospers, imports of many foods grow and its agricultural sector shrinks in relative terms. These sensitivities could be removed unilaterally along with the high tariffs on Vietnamese rice imports, with China gaining in economic welfare terms. Similarly, Vietnam would gain from removing its sensitivities especially in products where it has a comparative advantage as with rice production.

References Anderson, K. and Valenzuela, E., “Estimates of Global Distortions to Agricultural Incentives 1955–2007,” World Bank, Washington DC (2009).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Individual Country Approaches to Agriculture in the ASEAN–China FTA

251

ASEAN Secretariat, “Protocol to Amend the Agreement on Trade in Goods of the Framework Agreement on Comprehensive Economic Co-Operation Between ASEAN and the People’s Republic of China,” Cebu, The Philippines, 8 December (2006). Athukorala, P., Huong, P. L. and Thanh, V. T., “Distortions to Agricultural Incentives in Vietnam,” Agricultural Distortions Research Project Working Paper 26, December (2007). Bosworth, M. and Cutbush, G., “Australia’s Quarantine Mess: The Case of New Zealand Apples,” Unpublished Policy Discussion Paper, October (2010). Bosworth, M. and Trewin, R., “Domestic Dynamics of Preferential Services Liberalisation–Experience of Australia and Thailand,” in Marchetti and Roy (eds.), Liberalising Trade in Services: Bilateral, Regional and Multilateral Perspectives in the 21st Century (2008). Cambridge University Press, Cambridge, UK and WTO, Geneva, Switzerland. Cervantes-Godoy, D. and Dewbre, J., “Economic Importance of Agriculture for Poverty Reduction,” OECD Food, Agriculture and Fisheries Working Paper No. 23 (2010). DFAT, “Trading Our Way to More Jobs and Prosperity,” Gillard Government Trade Policy Statement, April (2011). Fane, G. and Warr, P., “Distortions to Agricultural Incentives in Indonesia,” Agricultural Distortions Research Project Working Paper 24, December (2007). Feridhanusetyawan, T., “Preferential Trade Agreements in the Asia-Pacific Region,” IMF Working Paper WP/05/149, International Monetary Fund, Washington (2005). Fuglie, K., “Is a Slowdown in Agricultural Productivity Growth Contributing to the Rise in Commodity Prices,” Agricultural Economics 39, Supplement (2008). GTAP (Global Trade Analysis Project) (2008). http://www.gtap.org. Hertel, T. and Tsigas, M., “Structure of GTAP,” in Hertel, T. W. (ed.), Global Trade Analysis: Modeling and Applications, Chapter 2, pp. 38–46. New York: Cambridge University Press (1997). Huang, J., Rozelle, S., Martin, W. and Liu, Y., “Distortions to Agricultural Incentives in China,” Agricultural Distortions Working Paper 29, World Bank (2007). Marks, S. V. and Rahardja, S., “Effective Rates of Protection Revisited for Indonesia,” Bulletin of Indonesian Economic Studies, 48(1): 57–84 (2012). OECD, “Agricultural Policy Reform in China,” Policy Brief, Paris, October (2005).

March 5, 2013

252

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch10

Regional Development and Economic Growth in China

OECD, “Policies for Agricultural Development, Poverty Reduction and Food Security,” OECD Global Forum on Agriculture, Paris 29–30 November (2010). Patunru, A. A. and von Luebke, C., “Survey of Recent Developments,” Bulletin of Indonesian Economic Studies (BIES), 46(1): 7–31 (2010). Petri, P. A., Plummer, M. G. and Zhai, F., “The Economics of the ASEAN Economic Community,” Working Paper 13, Brandeis University, Department of Economics and International Business School (2010). Productivity Commission (PC), “Bilateral and Regional Trade Agreements,” Research Report, December (2010). Productivity Commission (PC), “Trade and Assistance Review 2010–2011,” Annual Report Series, Productivity Commission, Canberra, May (2012). Sally, R., “Globalisation and the Political Economy of Trade Liberalisation in the BRICS” (2008a). www.ecipe.org/people/razeen-sally. Sally, R., “Trade Policy, New Century: The WTO, FTAs and Asia Rising,” IEA (2008b). Scollay, R. and Trewin, R., “Australia and New Zealand Bilateral CEPs/FTAs with the ASEAN countries and their implications on the AANZFTA,” Final Report, REPSF Project No. 05/003, June (2006). Vanzetti, D., Trewin, R. and Cassing, J., “Impact Assessment of Trade Agreements on Vietnam’s Economy”, Unpublished monograph, Multilateral Trade Assistance Project (MUTRAP III), Hanoi (2010). www.mutrap.org.vn. Warr, P., “Food Policy and Poverty in Indonesia: A General Equilibrium Analysis,” Australian Journal of Agricultural and Resource Economics, 49: 429–451 (2005). WTO/ITC/UNCTAD, “World Tariff Profiles 2009,” Geneva (2010).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Chapter 11

Environmental Regulation and Productivity Growth in APEC Economies Bing Wang and Yanrui Wu

11.1. Introduction There is an abundant literature on economic growth, in particular the sources of growth across countries as well as among individual economies. However, investigations of the relationship between economic growth and the environment only appeared in the early 1990s.1 Since then, the effect of environmental regulation on economic growth has attracted a lot of attention among both policy-makers and academia. Underlining the increased interest in this topic is the growing awareness of the environmental consequences of economic growth in the world. The latter has led to the United Nations Framework Convention on Climate Change (UNFCCC) with the aim to stabilize greenhouse gas (GHG) concentration in the atmosphere at a desirable level while maintaining economic growth. The UNFCCC was negotiated at the Earth Summit in Rio de Janeiro in 1992. Subsequently, according to the 1997 Kyoto Protocol, industrialized countries agreed to reduce carbon dioxide (CO2 ) emissions by about 5% of the 1990 level during the period of 2008–2012. A total of 168 countries and one regional economic bloc have ratified this Protocol to date. With the recent policy changes towards climate change in the US, the campaigns for environmental protection among the nations are likely to gain new momentum and bring about tough regulations. Environmental regulations may result in resources being diverted away from the production of goods to pollution abatement activities (Färe et al.,

1 Examples include Selden and Song (1994) and Grossman and Krueger (1995).

253

March 5, 2013

254

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

2001a). How would these changes affect economic growth and in particular productivity growth? The latter is the main driving force of economic growth. A number of studies have focused on the effect of environmental regulations on traditional measures of total factor productivity (TFP) (e.g., Jaffe et al., 1995). However, traditional measures of TFP, e.g., Törnquist and Fischer indices, concentrate only on the production of desirable or good outputs and fail to consider environmentally hazardous (undesirable or bad) by-products of production processes because no prices are available for the undesirable outputs. In the meantime, the cost of abatement activities is included in the inputs. Hence, traditional approaches may yield biased measures of productivity growth. This problem may be overcome by considering the Malmquist productivity index which does not require information on prices. For example, Färe et al. (1994) developed an approach which can decompose the Malmquist productivity index into technological progress and efficiency change components. This decomposition has been used to compare the differences and similarities in growth patterns across regions. In an application to Swedish paper and pulp mills, Chung et al. (1997) further introduced a directional distance function approach, i.e., the Malmquist–Luenberger (ML) productivity index, to analyze models of joint production of good outputs and bad outputs. This index considers the reduction of bad outputs as well as the increase in good outputs. It also possesses all the desirable properties of the Malmquist productivity index. The objective of this chapter is to apply the ML index method to a sample of 17 APEC economies over the period of 1980–2004. This chapter contributes to the existing literature in at least two directions. First, three types of productivity indices are estimated and compared according to different policy scenarios, i.e., no regulatory constraints, no change in current emissions levels and a partial reduction of emissions. Second, the determinants of productivity changes are also examined. The remainder of the chapter is organized as follows. In Section 11.2, a brief review of the related literature is conducted. In Section 11.3, the analytical framework is presented. In Section 11.4, the data issues and empirical results are discussed. In Section 11.5, the sources of productivity variation are investigated. Finally, Section 11.6 concludes the chapter.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

255

11.2. Literature Review When resources are employed for pollution abatement activities, measured inputs in an economy increase. As a result, traditional measures of TFP as the ratio of outputs over combined inputs are likely to be lower. This bias has led some observers to suggest that current methods of productivity measures almost always lead to the conclusion that environmental protection efforts and productivity performance are inversely related (Repetto et al., 1997). This may distort our assessment of economic performance and resultant changes in social well-being and hence lead to potentially misguided policy recommendations (Hailu et al., 2000). Economists have long recognized that failure to account for non-market activities may lead to biases in the measurement of productivity change. Pittman (1983) provided the earliest attempt by introducing shadow prices and thus incorporating undesirable outputs in efficiency measurement.2 Chung et al. (1997) extended the literature and developed the ML productivity index, which allows producers to increase the production of desirable outputs and reduce the production of undesirables simultaneously. This approach also accommodates the decomposition of changes in TFP into changes in efficiency and technological progress. Although the Malmquist productivity index has been used widely, only a limited number of empirical studies have employed the ML index to measure productivity growth.A brief review of these studies is presented here. Using micro-level panel data, Färe et al. (2001a) estimated the ML indices for the US state manufacturing sectors during the period of 1974– 1986. They found that average annual productivity growth was 3.6%, whereas it was 1.7% when emissions are ignored. Similar conclusions were drawn by Domazlicky and Weber (2004), who applied the same technique to six US chemical industries, at the three-digit SIC level for the period of 1988–1993. Domazlicky and Weber (2004) further argued that while there are costs associated with environmental regulations, those costs are overwhelmed by subsequent productivity growth. Lindmark et al. (2003) adopted a similar approach to analyze global convergence in productivity 2 Other studies following the same concept include Färe et al. (1993), Coggins and Swinton

(1996), Swinton (1998), and Reig-Martínez et al. (2001).

March 5, 2013

256

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

by using a sample of 59 countries for the period of 1965–1990. They found that, when bad outputs are included, TFP growth is lower and so are the growth rates of technological progress and efficienc change. Another study by Jeon and Sickles (2004) who applied both Malmquist and ML indices to examine the impact on productivity growth due to the consideration of CO2 as a bad output in OECD andAsian economies over the period 1980–1990 and 1980–1995, respectively. They found little change in average growth rates of TFP for OECD countries and significant negative productivity growth in Asian economies except Japan. However, they could not decide whether changes are due to catch-up (efficiency change) or innovation (technological progress). Yörük and Zaim (2005) also applied both Malmquist and ML indices to measure productivity growth for all but two OECD countries over the period 1985–1998. The Malmquist index showed an average productivity growth of at least 10% for the OECD countries from 1985 to 1998, while the index that includes nitrogen oxide and organic water pollutant emissions implied a productivity growth of 20%. In comparison with the conventional Malmquist indices, the ML indices record at least 7% higher productivity growth for OECD countries. In addition, they also investigated the determinants of the variation in productivity growth across countries. They found that the dummy variable reflecting the ratification of the UNFCCC protocol on CO2 emissions has a significant, positive effect on the ML index. More recently, Kumar (2006) employed the ML index to examine conventional and environmentally sensitive TFP in 41 developed and developing countries over the period 1971–1992. It is found that TFP indices are not different when CO2 emissions are assumed to be freely disposable. As for the productivity growth components, i.e., technological progress and efficiency changes, the null hypothesis of no changes under two different scenarios could not be accepted. Kumar also examined global catch-up and convergence or divergence in productivity growth, which is environmentally sensitive. Finally, several studies exclusively focused on productivity growth in APEC economies (Table 11.1). Chambers et al. (1996) calculated productivity growth and its components for 17 APEC economies over the period 1975–1990 using a Luenberger productivity indicator, which is based on the concept of a directional distance function. They computed three versions

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

257

Table 11.1. Summary of Main Studies on Productivity Growth in APEC Economies. Author

Sample Method

Period

TFP (%) EC (%) TP (%)

Chambers et al. (1996)

17

DEA

1975–1990

−0.26 −2.46 −0.52

−0.84 −4.39 −0.37

0.58 1.93 −0.15

Chang and Luh (1999)

19

DEA

1970–1980 1980–1990

−1.38 0.03

−1.50 0.35

0.10 −0.32

Färe et al. (2001b)

17

DEA

1975–1990 1975–1996

0.07 0.28

−1.13 0.48

1.21 0.76

Wu (2004)

16

SFA

1980s 1990s

3.98 2.71

1.32 −0.66

2.66 3.38

Note: TFP, EC and TP represent the rate of TFP growth, efficiency changes and technological progress respectively. Chambers et al. (1996) computed three versions of the indices by specifying three different “directions” for the component distance functions. Efficiency changes in Wu (2004) include scale efficiency. DEA and SFA are abbreviations for data envelopment analysis and stochastic frontier analysis.

of the productivity index by specifying three different “directions” for the distance function. Generally speaking, average annual productivity growth declined due to falling efficiency while technological progress was generally positive. Chang and Luh (1999) calculated productivity growth and its components using the Malmquist productivity indices for 19 APEC member economies over the periods 1970–1980 and 1980–1990, respectively. Regression analyses are also conducted to investigate the role of FDI and education in catch-up (moving along the production frontier) and innovation (shifting the production frontier). Their results indicate that the United States was not the sole innovator among the 19 APEC member economies. Instead, Hong Kong and Singapore have shown their capability to shift the grand frontier of the APEC economies during the 1980s. This result is quite inspiring because it implies that the NIEs are not only good at moving towards the frontier, but are also potential innovators. Chang and Luh (1999)

March 5, 2013

258

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

showed that FDI contributed to TFP growth either through catch-up or technological progress. Färe et al. (2001b) also employed the Malmquist index to measure TFP growth and its two components, i.e., efficiency change and technological progress, in a sample of 17 APEC economies over the period of 1975–1996. In all economies, the main cause of low TFP growth was a poor (negative) efficiency record. The average TFP growth rate for Japan and Malaysia was positive during 1975–1996, but the efficiency change component remained negative. They found that among APEC economies the main contributor to labor productivity growth was capital accumulation. Unlike previous studies, they found no evidence of a poor TFP growth performance for Singapore. Among the studies reviewed so far, Wu (2004) is an exception. Wu applied a stochastic frontier (parametric) approach to analyze the relationship between openness, productivity and growth among the APEC economies for the period of 1980–1997. He found that openness affects not only efficiency changes but also the structure of production technology. In general, the empirical analyses have shown that, in terms of productivity growth, APEC developed members have performed better than their developing counterparts. In particular, APEC developed economies, led by the US, are found to be more innovative than APEC developing members. However, Japan appears to lag behind other developed economies in terms of technological progress. According to Wu (2004), APEC developing members have shown rapid catch-up with their rich neighbors. Korea and Taiwan were the lead performers in the 1980s. Mainland China took over to become the leader in the 1990s. The brief review above suggests several gaps in the existing literature. While Chambers et al. (1996), Chang and Luh (1999), Färe et al. (2001b) and Wu (2004) examined APEC economies, their productivity estimates ignored undesirable outputs. Färe et al. (2001a) and Domazlicky and Weber (2004) focused on micro-level productivity growth. Other studies applied macro-level data, but only considered two types of productivity indices. Jeon and Sickles (2004) is an exception. However, they did not examine the causes of productivity changes. This chapter attempts to fill these gaps in the literature. The analytical framework is introduced next.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

259

11.3. Analytical Framework To introduce the analytical framework, it is assumed that a vector of inputs x = (x1 , . . . , xN ) ∈ RN + are employed to produce a vector of good outputs y = (y1 , . . . , yM ) ∈ RM + , and undesirable or bad outputs b = (b1 , . . . , bI ) ∈ I R+ . Let P(x) be the feasible output set for the given input vector x. The technology is modeled by its output sets: P(x) = {(y, b) : x can produce (y, b)},

x ∈ RN +.

(11.1)

It is assumed that the output sets are closed and bounded and that inputs are freely disposable.3 In addition, three axioms are proposed: If (y, b) ∈ P(x)

and

b=0

If (y, b) ∈ P(x)

and

0≤θ≤1

If (y, b) ∈ P(x)

and



y ≤y

then y = 0. then (θy, θb) ∈ P(x). 

imply (y , b) ∈ P(x).

(11.2) (11.3) (11.4)

The first axiom in Equation (11.2) known as the null-jointness implies that the country cannot produce good outputs in the absence of bad outputs. The second axiom in Equation (11.3) means that good and bad outputs are weakly disposable, implying that there is a cost for pollution control and that abatement activities would typically divert resources away from the production of desirable outputs and thus affect the good output negatively. The third axiom in Equation (11.4) indicates that good outputs are strongly disposable. That is, the good output is freely disposable, but this is not a maintained condition for the bad output. To formulate a DEA model that satisfies the above conditions, it is assumed that for each period t = 1, . . . , T , there are k = 1, . . . , K observations of inputs and outputs, that is, (xkt , ykt , bkt ). This database is then employed to construct the following output set that satisfies the above three axioms
K  t t t t P (x ) = (yt , bt ) : ztk ykm ≥ ym , m = 1, . . . , M k=1 K 

t ztk bki = bit , i = 1, . . . , I

k=1 3 For more details, refer to Färe and Primont (1995).

March 5, 2013

260

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China K 

t ztk xkn ≤ xnt , n = 1, . . . , N

k=1

ztk



≥ 0, k = 1, . . . , K K 

(11.5)

t bki > 0, i = 1, . . . , I

(11.6)

t bki > 0, k = 1, . . . , K,

(11.7)

k=1 I  i=1

ztk are the nonnegative weights assigned to each observation when constructing the production set, and imply that the production technology exhibits constant returns to scale. The latter ensures that TFP indices are computed (Färe and Grosskopf, 1996). The inequality constraints in Equation (11.5) on the good outputs and input variables imply that these outputs and inputs are freely disposable. Furthermore, Equations (11.6) and (11.7) ensure the property of the null-jointness of outputs. Although the representation of the technology in Equations (11.5), (11.6) and (11.7) is conceptually useful, it is not very helpful from a computational viewpoint. For functional representation of the technology, the following directional output distance function is employed.4 This technique accommodates the production of byproducts, and is conceptually consistent with the above axiomatic approach.

11.3.1. Directional output distance functions The objective of environmental protection is to reduce pollution (bad output) while economic growth (good output growth) is still maintained. To model such a production process, the directional output distance function is used. It is a generalization of the Shephard output distance function, and can accommodate non-proportional changes in output. Formally, it is defined as  o (x, y, b; g) = sup{β : (y, b) + βg ∈ P(x)}, D

(11.8)

4 The directional output distance function is a variation of Luenberger’s shortage function, see Luenberger (1992, 1995).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

261

where g = (gy , gb ) is the vector of directions in which outputs can be scaled. The directional distance function allows for a variety of direction vectors which depend on whether the technology exhibits free or weak disposal of bad outputs.5 This study mainly considers three scenarios, that is, • Scenario 1 (S1): The direction vector is g = (y, 0) and the bad outputs are ignored in constructing the reference technology. • Scenario 2 (S2): The direction vector is g = (y, 0) and the technology exhibits weak disposability in bad output • Scenario 3 (S3): The direction vector is g = (y, −b) and the technology exhibits weak disposability in bad outputs. The first scenario implies that no environmental regulations exist. Under the second scenario, environmental regulations allow good outputs to increase while bad outputs are held constant. This is a direction that seems most in agreement with the goals of the Kyoto Protocols in terms of CO2 emissions (Jeon and Sickles, 2004). The third scenario deals with reduction of bad outputs at the same proportion that good outputs are allowed to increase. This direction can be viewed as a compromise between the goals of the pro-growth and anti-growth environmental movements (Jeon and Sickles, 2004). It is also consistent with current practices and the objectives of UNFCCC as far as the control of CO2 emissions is concerned. To simulate the three proposed scenarios, the following linear programming (LP) problem is to be solved  ot (xkt  , ykt  , bkt  ; ykt  , −bkt  ) = Max β D s.t. K 

t ztk ykm ≥ (1 + β)ykt  m ,

m = 1, . . . , M

k=1 K 

t ztk bki = (1 − β)bkt  i ,

i = 1, . . . , I

k=1

5 For more discussions, see Chambers et al. (1996) and Färe et al. (2005).

March 5, 2013

262

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China K 

t ztk xkn ≤ xkt  n ,

n = 1, . . . , N

k=1

ztk ≥ 0,

k = 1, . . . , K.

(11.9)

This LP problem corresponds to Scenario 3 of which Scenarios 1 and 2 are special cases. A description of the latter is presented in the appendices. The directional output distance function takes a minimum value of zero for countries that are technically efficient, that is, they operate on the frontier of P(x). A value of the directional output distance function greater than zero indicates technical inefficiency. The derived directional distance functions are then used to construct TFP indices.

11.3.2. Productivity indices Following Chung et al. (1997), the output oriented ML productivity index between period t and t + 1 is expressed as
 ot (xt , yt , bt ; gt )] [1 + D ML t+1 = t  ot (xt+1 , yt+1 , bt+1 ; gt+1 )] [1 + D  ot+1 (xt , yt , bt ; gt )] [1 + D ×  ot+1 (xt+1 , yt+1 , bt+1 ; gt+1 )] [1 + D

 21 .

(11.10)

The ML index can be decomposed into an index of efficiency change (EFFCH) and an index of technological progress (TECH): ML = EFFCH × TECH,

(11.11)

where  ot (xt , yt , bt ; gt )] [1 + D ,  ot+1 (xt+1 , yt+1 , bt+1 ; gt+1 )] [1 + D
 ot+1 (xt , yt , bt ; gt )] [1 + D =  ot (xt , yt , bt ; gt )] [1 + D

= EFFCH t+1 t TECH t+1 t

 ot+1 (xt+1 , yt+1 , bt+1 ; gt+1 )] [1 + D ×  ot (xt+1 , yt+1 , bt+1 ; gt+1 )] [1 + D

(11.12)

 21 . (11.13)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

263

A value of greater (less) than one for ML, EFFCH and TECH indicates productivity growth (decline), efficiency improvement (deterioration) and technical progress (regress), respectively. Under the three scenarios considered, there are three directional distance functions and hence three productivity indices. In order to get each productivity index, four programs need to be solved. Two programs involve observations and technology from the same time period t or t + 1, and the other two use observations and technology of a different time period, for example, period t technology with observations from period t + 1. The latter problems can cause difficulties in calculation if the observed data in period t + 1 is not feasible in period t. To reduce the number of infeasible solutions in computing the ML index, each year’s reference technology is determined by observations of the current and the past two periods.6 Hence the reference technology for 2000, for example, would be constructed from data in 2000, 1999, and 1998. Following this approach, the productivity index and its two components are estimated for 17 APEC economies over the period of 1980 to 2004.

11.4. Data and Empirical Results 11.4.1. Data issues CO2 emissions account for over 80% of total GHG emissions. In this study, GDP and CO2 are considered as proxies of good and bad outputs respectively, and labor force and capital stock as inputs. The real GDP measured in 2000 US dollars is obtained by using population and real GDP per capita (RGDPCH) data from the Penn World Tables PWT6.2 (Heston et al., 2006). Labor force is obtained by dividing the real GDP by the real GDP per worker (RGDPW) in the PWT6.2. Capital stock values are estimated using capital formation statistics drawn from the PWT6.2 (see the appendices for details about capital stock derivation). World Development Indicators (World Bank, 2007) is the source for CO2 emissions measured in thousand metric tons.7 6 This technique was also adopted by Färe et al. (2001b) who provided the technical details

about infeasible solutions in constructing index numbers. 7 Taiwan’s CO emission figures were drawn from the Oak Ridge Dataset (Marland et al., 2

2003) and multiplied by 3.664 in order to be consistent with the world development indicators (WDI) statistics (because CO2 emission is expressed in thousand metric tons of carbon in the Oak Ridge Dataset). CO2 emissions data for 2003 and 2004 are estimated using data from Euromonitor (2007) and WDI CO2 data for 2000–2002.

March 5, 2013

264

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

Summary statistics of the sample are presented in Table 11.2. China, Malaysia, Thailand and the four East Asian NIEs have indeed achieved high growth since 1978. This high growth was matched by the rapid expansion of capital stock and CO2 emissions in those economies. To take into account the possible impact of UNFCCC and the Kyoto Protocol on growth in CO2 emissions, the discussions here focus on two sub-periods, i.e., without UNFCCC (1978–1991) and with UNFCCC (1992–2004), and two economic groups: Annex-I countries (Canada, USA, Japan, Australia, and New Zealand) and Non-Annex-I countries.8 Non-Annex-I countries can be further grouped into developing countries (Mexico, Chile, China, Indonesia, Malaysia, the Philippines, Peru, and Thailand) and East Asian Newly Industrialized Economies (NIEs) (Hong Kong, South Korea, Singapore, and Taiwan). By 1994, all APEC members but Singapore had ratified UNFCCC. Australia and USA have not yet ratified the Kyoto Protocol. Annex-I countries and China are the major contributors of CO2 emissions, accounting for 87.5% of the total emissions from the APEC group. During the entire sample period, the highest growth rate with respect to CO2 emissions was observed in Thailand (8.05%). It is also shown that the average annual growth in CO2 emissions has slowed down since 1992 indicating the potentially positive impact of UNFCCC as supported by Yörük and Zaim (2005).

11.4.2. Estimation results A summary of the empirical findings about productivity growth and its components under three scenarios is presented in Table 11.3. Under Scenario 1 (the presence of CO2 emissions is ignored), the average productivity index (PI) value of 1.0025 indicates that the annual productivity growth for the sample countries was 0.25% over the entire period, 1980–2004. On average, this growth was due to a technical efficiency change (EC) of 0.20% and technological progress (TP) growth of 0.05%. A comparison across country groups indicates that, over the entire period, productivity growth 8 The impact of the Kyoto Protocol is not considered since the earliest date of enforcement was February 16, 2005 which is out of our sample period. The Annex-I parties to the UNFCCC are listed in the Annex-I of the Climate Convention. They mainly include developed countries and regional organizations (EU).

March 5, 2013

L

K

3.3 2.89 4.49 9.57 5.52 4.54 2.46 6.65 2.84 6.48 2.52 1.89 3.1 6.66

1.78 1.57 2.39 1.56 2.36 2.8 0.71 2.06 3.09 2.85 1.67 3.18 2.73 3.23

3.7 3.68 4.32 9.61 6.68 6.53 3.62 9.42 3.58 8.25 2.51 1.61 3.4 5.84

Date of Ratification

1978–1991 1992–2004 1978–2004 Shares (%) UNFCCC 2.57 0.24 2.54 4.17 4.51 4.37 1.36 7.24 3.83 8.58 2.36 −0.38 1.59 2.63

2.64 2.15 4.05 3.01 2.7 4.0 0.89 4.02 0.99 6.66 2.59 2.12 3.58 3.00

2.65 1.18 3.31 3.66 4.05 4.54 1.13 5.82 2.62 7.73 2.59 0.81 2.82 2.76

2.58 4.08 0.36 23.46 0.27 1.64 9.85 2.61 3.22 0.75 0.24 0.22 0.49 0.42

30/12/92(21/3/94) 4/12/92(21/3/94) 22/12/94(22/3/95) 5/1/93(21/3/94) —(5/5/03) 23/8/94(21/11/94) 28/5/1992(21/3/94) 14/12/92(21/3/94) 11/3/1993(21/3/94) 13/7/94(11/10/94) 16/9/93(21/3/94) 7/6/93(21/3/94) 2/8/94(31/10/94) 29/5/97(27/8/97)

Kyoto Protocol — 7/12/02(16/2/05) 26/8/02(16/2/05) 30/8/02(16/2/05) — 3/12/04(3/3/05) 2/6/04(16/2/05) 8/11/02(16/2/05) 7/9/00(16/2/05) 4/9/02(16/2/05) 29/12/02(16/2/05) 12/9/02(16/2/05) 20/11/03(16/2/05) 12/4/06(11/7/06) (Continued)

265

Regional Development and Economic Growth. . . b1491-ch11

AUS CAN CHL CHN HKG IDN JPN KOR MEX MYS NZL PER PHL SGP

Y

Emission

9in x 6in

Country code

Average Growth in CO2 Emissions (%)

Environmental Regulation and Productivity Growth in APEC Economies

Average Growth (%, 1978–2004)

11:55

Table 11.2. Summary Statistics of the Sample.

K

5.59 6.52 3.04 2.84 5.32 4.81 6.34 4.59

1.93 1.39 1.36 1.42 2.46 2.57 2.26 2.16

6.5 8.04 3.9 3.48 6.15 5.48 7.5 5.36

1978–1991 1992–2004 1978–2004 Shares (%) UNFCCC 9.65 5.44 0.11 1.33 4.51 4.29 4.96 3.58

6.24 4.53 1.64 1.6 3.74 3.1 4.01 2.25

8.05 5.09 0.82 1.67 4.27 4.19 4.43 3.51

1.17 1.34 47.29 64.04 35.95 31.31 4.64 99.99

Kyoto Protocol

28/12/94(28/3/95) 28/8/02(16/2/05) — — 15/0/92(21/3/94) — — — — — —

— — —

Note: Means 1–5 correspond to the group means of Annex-I, Non-Annex-I, Developing Countries, East Asian NIEs (Hong Kong, South Korea, Singapore, and Taiwan) and APEC. The country codes represent in turn Australia (AUS), Canada (CAN), Chile (CHL), China (CHN), Hong Kong (HKG), Indonesia (IDN), Japan (JPN), Korea (KOR), Mexico (MEX), Malaysia (MYS), New Zealand (NZL), Peru (PER), the Philippines (PHL), Singapore (SGP), Thailand (THA), Taiwan (TWN) and the United States (USA). The column “emission shares” reports a country’s total percentage contribution to APEC CO2 emissions for the period from 1980 to 2005. The date in the parentheses indicates the date of ratification for each economy.

Regional Development and Economic Growth. . . b1491-ch11

L

Date of Ratification

9in x 6in

THA TWN USA Mean 1 Mean 2 Mean 3 Mean 4 Mean 5

Y

Emission

11:55

Country code

Average Growth in CO2 Emissions (%)

Regional Development and Economic Growth in China

Average Growth (%, 1978–2004)

March 5, 2013

266

Table 11.2. (Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

267

Table 11.3. Average Productivity Growth, 1980–2004. Scenario 1

AUS CAN CHL CHN HKG IDN JPN KOR MEX MYS NZL PER PHL SGP THA TWN USA Mean 1 Mean 2 Mean 3 Mean 4 Mean 5

Scenario 2

Scenario 3

PI

EC

TP

PI

EC

TP

PI

EC

TP

1.0010 0.9992 0.9909 1.0042 1.0001 0.9853 1.0160 1.0003 0.9980 0.9946 1.0020 0.9972 1.0057 1.0306 1.0001 1.0066 1.0112 1.0059 1.0011 0.9970 1.0093 1.0025

1.0003 1.0017 0.9939 1.0190 0.9989 0.9936 0.9985 0.9988 1.0007 0.9976 1.0032 0.9972 1.0052 1.0127 1.0088 1.0023 1.0013 1.0010 1.0024 1.0020 1.0032 1.0020

1.0007 0.9975 0.9970 0.9854 1.0012 0.9918 1.0175 1.0015 0.9972 0.9970 0.9988 0.9999 1.0006 1.0177 0.9914 1.0043 1.0099 1.0049 0.9987 0.9950 1.0062 1.0005

1.0021 1.0076 1.0029 0.9653 1.0199 0.9921 1.0151 1.0038 0.9929 0.9919 1.0031 1.0066 1.0113 1.0305 1.0009 1.0184 1.0144 1.0084 1.0029 0.9954 1.0181 1.0045

0.9957 0.9979 1.0058 0.9981 1.0000 1.0016 0.9973 0.9984 0.9927 0.9953 0.9945 1.0046 1.0000 1.0117 0.9971 1.0007 1.0006 0.9972 1.0005 0.9994 1.0027 0.9995

1.0064 1.0097 0.9971 0.9672 1.0200 0.9905 1.0179 1.0054 1.0002 0.9966 1.0087 1.0020 1.0114 1.0186 1.0038 1.0176 1.0138 1.0113 1.0024 0.9960 1.0154 1.005

1.0071 1.0075 1.0041 1.0002 1.0140 0.9935 1.0125 0.9961 0.9936 0.9930 1.0016 1.0038 1.0071 1.0250 0.9954 1.0084 1.0035 1.0064 1.0028 0.9988 1.0108 1.0039

0.9963 0.9978 1.0055 1.0114 1.0000 1.0010 0.9980 0.9998 0.9944 0.9965 0.9952 1.0023 1.0000 1.0094 0.9928 1.0007 1.0005 0.9976 1.0011 1.0005 1.0025 1.0001

1.0108 1.0097 0.9986 0.9889 1.0140 0.9925 1.0145 0.9963 0.9992 0.9966 1.0065 1.0015 1.0071 1.0154 1.0026 1.0077 1.0030 1.0089 1.0017 0.9984 1.0083 1.0038

Note: The country codes are the same as in Table 11.2. These annual average growth indices are geometric means. The LP problems required for these exercises are solved using the software package GAMS. The authors are grateful to Carl Pasurka for providing us the GAMS codes used in Färe et al. (2001a). These codes have been the starting point for preparing the codes for this paper.

and technological progress were higher in the Annex-I countries (0.59% and 0.49%) than in the Non-Annex-I countries (0.11% and −0.13%), but technical efficiency change was higher in the Non-Annex-I countries. For the sample of the four East Asian NIEs the average productivity index value is 1.0093, which is due to a technological progress of 0.62% and an improvement in efficiency of 0.32%. The average productivity growth for

March 5, 2013

268

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

the developing economy group is −0.3%. This negative growth is largely due to technical regress. Among individual economies, 65% (11/17) of APEC members showed a positive productivity growth rate during 1980– 2004. The economies that showed the highest productivity growth were Singapore (3.06%), Japan (1.60%), and USA (1.12%). In these countries, technological progress accounted for a greater portion of productivity growth than efficiency changes, in particular, Japan has recorded a negative rate of technical efficiency change. This finding is consistent with that of Färe et al. (2001b) who found Singapore was ranked first among APEC in terms of efficiency change during the period of 1975–1996. Under Scenario 2 (CO2 emissions are held constant), the average productivity index value of 1.0045 is slightly higher than the value under Scenario 1. This result is supported by the findings of Jeon and Sickles (2004). In their research, the productivity indices under Scenario 2 are on an average higher than those under Scenario 1 for both OECD and Asian economies. Table 11.3 also shows that productivity growth was due to a technical efficiency change of −0.05% and a technological progress of 0.50%. A comparison across country groups indicates that, over the entire period, productivity growth and technological progress were higher in the Annex-I countries (0.84% and 1.13%) than in the Non-Annex-I countries (0.29% and 0.24%), but technical efficiency change were relatively high in the Non-Annex-I countries. At the economy level, 76% (13/17) of the economies showed a positive growth rate of productivity during 1980–2004. The countries that showed the highest productivity growth within the APEC group were Singapore (3.05%), Hong Kong (1.99%), and Taiwan (1.84%). In these economies, technological progress accounted for a greater portion of productivity growth than efficiency change. Overall, the productivity index under Scenario 2 has a higher value than that under Scenario 1 for Canada, Chile, Hong Kong, Indonesia, Mexico, Peru, the Philippines, and Thailand. However, the opposite is true for the developing countries and the East Asian NIEs. Thus, generalization of the results is difficult. Higher CO2 growth does not necessarily imply lower productivity growth. In order to incorporate negative externalities into the measures of productivity, weights have to be assigned to the bad outputs. The ML productivity index under Scenario 3 imposes a restriction on CO2 emissions

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

269

and is consistent with concerns of global warming. The idea is to recognize producers for simultaneously increasing outputs and reducing CO2 emissions. This technique thus offers an alternative way of assigning weights to the bad outputs. The average ML productivity index value of 1.0039 indicates that the annual productivity growth for the sample countries was 0.39% over the entire period of 1980–2004. This is higher than the rate under Scenario 1 but is lower than the value under Scenario 2, a finding supported by Jeon and Sickles (2004). On average, this growth was due to a technical efficiency change of 0.01% and a technological progress of 0.38%. A comparison across sub-groups indicates that, over the entire period, productivity growth and technological progress were higher in the Annex-I countries (0.28% and 0.17%) than in the Non-Annex-I countries (0.29% and 0.24%). Among the APEC members, 71% (12/17) of the economies showed a positive productivity growth rate over the entire period. The economies that showed the highest productivity growth were Singapore (2.50%), Hong Kong (1.40%), and Japan (1.25%). The productivity indices under Scenario 3 have relatively high values in comparison with the values under Scenario 1 for Australia, Canada, Chile, China, Hong Kong, Indonesia, Mexico, Peru, the Philippines, and Thailand. On average, productivity indices under Scenario 1 are higher than those under Scenario 3 for Annex-I countries, but the reverse is true for the Non-Annex-I countries. This finding is different from the conclusion by Kumar (2006). However, technological progress is higher in the Annex-I countries than in the Non-Annex-I countries if the goal is to reduce CO2 emissions. Kopp (1998) argued that developed countries experienced technological progress in a way that economizes on CO2 emissions but the same did not happen in the developing economies during 1970–1990. Finally, under the three scenarios, productivity growth, efficiency change and technological progress for the two sub-periods, i.e., 1980–1991 and 1992–2004, are also calculated and reported in the appendices. Under Scenario 1, between 1980–1991 and 1992–2004 periods, 41% (7/17) of APEC members showed improvement in productivity with the greatest gains being obtained in Indonesia, Peru, and Canada. Under Scenario 2, between 1980– 1991 and 1992–2004 periods, 47% (8/17) of the APEC economies showed productivity improvement with the largest gains being recorded in Peru, New Zealand, and Malaysia. Under Scenario 3, between 1980–1991 and

March 5, 2013

270

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

1992–2004 periods, 47% (8/17) of the economies showed an increase in productivity with the largest gains being shown in New Zealand, Malaysia, and Mexico.

11.4.3. Identifying innovators The estimation results reported so far have shown technological progress indices for the economies between two adjacent years, but they do not allow us to identity which countries are shifting the frontier over time, that is, the innovators. In order to identify the innovators who actually cause the bestpractice frontier to shift, Färe et al. (2001a) and Kumar (2006) used the following criteria: TECH t+1 >1 t t  o (xt+1 , yt+1 , bt+1 ; yt+1 , −bt+1 ) < 0 D

(11.14)

 ot+1 (xt+1 , yt+1 , bt+1 ; yt+1 , −bt+1 ) = 0. D Economies satisfying the above criteria are regarded as innovators. The first condition, TECH t+1 > 1, ensures that the production possibility t frontier shifts in the direction with more good and fewer bad outputs. It implies that, given the input vector in period t + 1, it is possible to increase the good output and reduce the bad output (CO2 emissions) relative to period t. The second condition guarantees that the production in period t +1 occurs outside the production frontier of period t (i.e., technological progress has occurred). Thus, the technology of period t cannot produce the output vector of period t + 1 given the input vector of period t + 1. Hence, the value of the directional distance function relative to the reference technology of period t is less than zero. The third condition implies that the country must be on the production frontier in period t + 1. According to these criteria, the innovating countries are identified and listed in Table 11.4. Out of 24 two-year periods, under Scenario 1 where CO2 emissions are ignored, USA shifted the frontier 19 times and Taiwan shifted the frontier 11 times. Under Scenario 2 where CO2 emissions are held constant, USA shifted the frontier 20 times and Hong Kong and Taiwan shifted the frontier 17 times, respectively. Under Scenario 3 where CO2 emissions are reduced, Hong Kong, Taiwan and USA shifted the frontier 17, 16 and 13 times respectively. Overall, eight

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

271

Table 11.4. Countries Shifting the Frontiers. Scenario 1

Scenario 2

Scenario 3

1980–1981

—∗

Hong Kong, Hong Kong, New Zealand, the New Zealand, the Philippines, the USA Philippines, USA

1981–1982

—

Philippines

Philippines

1982–1983 USA

Hong Kong, New Zealand, the Philippines, Taiwan, USA

Hong Kong, New Zealand, the Philippines, Taiwan, USA

1983–1984 Taiwan, USA

Hong Kong, New Zealand, the Philippines, Taiwan, USA

Hong Kong, New Zealand, the Philippines, Taiwan, USA

1984–1985 Taiwan, USA

Taiwan, USA

Taiwan, USA

1985–1986 Taiwan, USA

Hong Kong, the Philippines, Taiwan, USA

Hong Kong, the Philippines, Taiwan, USA

1986–1987 Taiwan, USA

China, Hong Kong, Taiwan, USA

Hong Kong, Taiwan

1987–1988 Taiwan, USA

Hong Kong, Hong Kong, the the Philippines, Taiwan Philippines, Taiwan

1988–1989 Taiwan, USA

Hong Kong, the Philippines, Taiwan, USA

Hong Kong, Japan, the Philippines, Taiwan, USA

1989–1990 USA

Hong Kong, Taiwan, USA

Hong Kong, Taiwan, USA

1990–1991 Taiwan

Hong Kong, Taiwan

Hong Kong, Taiwan

1991–1992 Taiwan, USA

Chile, Hong Kong, Taiwan, USA

Chile, Hong Kong, Taiwan, USA

1992–1993 The Philippines, Taiwan, USA

Chile, Hong Kong, Taiwan, USA

Chile, Hong Kong, Taiwan (Continued)

March 5, 2013

272

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

Table 11.4. (Continued) Scenario 1

Scenario 2

Scenario 3

1993–1994

The Philippines, Taiwan, USA

Chile, Hong Kong, Taiwan, USA

Hong Kong, Taiwan

1994–1995

USA

Taiwan, USA

Chile, Hong Kong, Taiwan, USA

1995–1996

USA

Hong Kong, Taiwan, USA

Hong Kong, Taiwan

1996–1997

The Philippines, USA

Hong Kong, Taiwan, USA

Hong Kong, Taiwan

1997–1998

USA

USA

USA

1998–1999

USA

Taiwan, USA

Taiwan

1999–2000

The Philippines, USA

The Philippines, Taiwan, USA

The Philippines, Taiwan

— Hong Kong, the Philippines, Taiwan, USA

— Hong Kong, the Philippines, USA

2000–2001 2001–2002

— —

2002–2003

Taiwan, USA

Hong Kong, Taiwan, USA

Hong Kong, USA

2003–2004

The Philippines, USA

Chile, Hong Kong, Indonesia, the Philippines, Taiwan, USA

Chile, Hong Kong, Indonesia, the Philippines, USA

∗ This implies that the frontier shifted backward slightly.

different countries shifted the frontier at least once. In addition, according to Table 11.4, only one country shifted the frontier during 1997–1998 (i.e., immediately after the 1997 Asian Economic Crisis) and no country shifted the frontier during 2000–2001 due to the world economic recession. Färe et al. (2001a) argued that there might exist a relationship between the business cycle and the number of countries that shift the frontier in a given year.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

273

11.5. Explaining Productivity Growth The estimation results in the preceding section have shown considerable variation in productivity performance across APEC economies. An examination of the sources of the cross-economy variation would contribute to the understanding of productivity growth with the presence of environmental regulations. There is no formal theory identifying the factors that affect productivity growth. Researchers often resort to previous studies and their own beliefs. In some cases, the choice of the factors is also dictated by the availability of cross-economy statistics. This study is subjected to those constraints too. To examine the relationship between productivity growth and its determinants, the following simple regression involving panel data is employed here PI = α + αi zi + u,

(11.15)

where PI and zi represent productivity indices (the dependent variable) and their determinants (the explanatory variables), α s are parameters to be estimated and u is the standard white noise. To take environmental regulations into consideration, the PI’s from both Scenarios 2 and 3 are employed in Equation (11.15). The explanatory variables are GDP per capita (GDPPC) in constant prices, the share of industrial value-add over GDP (IND), technical inefficiency in the previous year (TI t−1 ), capital–labor ratios (KL), energy use per capita (EPC), openness index (OPEN) and a dummy variable (UNFCCC) that takes the value of one for the year in which the sample country ratified the UNFCCC and all subsequent years, and zero otherwise.9 The squares of both GDP per capita and the share of industrial value-added over GDP are included to capture any quadratic relationships between the productivity index and these variables. Data for the GDP per capita and openness index are taken from PWT6.2. Both the share of industrial valueadded over GDP and energy use per capita are drawn from the World Development Indicators database (World Bank, 2007).10 9As Färe et al. (2001a) argued that a change in the composition of the industry sector of a

country can also affect the level of CO2 emissions. For example, presumably a shift away from a pollution-intensive sector would yield a decline in CO2 emissions. 10 Some missing cells are filled by mean values of the observations in the past five years.

March 5, 2013

274

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

The estimation results are presented in Table 11.5. The Hausman statistics indicate that the fixed-effect specification is preferred for both regressions. All coefficients are statistically significant. Table 11.5 shows a positive relationship between the GDP per capita and productivity index. In the meantime, the negative coefficient of the squared income variable implies that the relationship between the productivity index and income per capita follows an inverted-U shape with a turning point at approximately $39003 (Scenario 2) or $38235 (Scenario 3). Hence, once an “average” APEC economy reaches this threshold income level, a downward trend in productivity growth is observed. This may reflect the catch-up movement of less developed APEC economies. In contrast, Yörük and Zaim (2005) showed an U-shaped relationship for OECD economies probably because these economies are more homogeneous in terms of the level of development. Table 11.5 also shows a negative relationship between the share of industrial value-added over GDP and the productivity index. However, the coefficient of the squared term demonstrates that the quadratic relationship is U-shaped with a turning point at approximately 22% (Scenario 2) or 18% (Scenario 3). Hence, once the share of industrial value-added over GDP exceeds this threshold for an economy, productivity growth trends upwards. Yörük and Zaim (2005) made the similar observation for the OECD group. This phenomenon may be due to the fact that productivity grows relatively fast as an economy becomes more industrialized. Furthermore, it is shown that the productivity index and the lagged technical inefficiency are positively related while the coefficient of the capital– labor ratio variable is negative. One argument is that these relationships indicate convergence between APEC economies. Economies producing closer to the production frontier would have a lower level of productivity growth than those being farther away so that the latter can catch up with the former group (Lall et al., 2002). Kumar (2006) also supports this convergence hypothesis. Finally, the openness and energy use per capita variables are both negatively related to the productivity indices. The openness variable could be a proxy for institutional and policy framework of an economy and capture the impact of international trade on productivity growth in particular (Etkins et al., 1994; Taskin and Zaim, 2001; Kumar, 2006). Thus the results imply that the environmentally undesirable effects may stem from the increased

March 5, 2013

Table 11.5. Factors Associated with Changes in Productivity. Scenario 2

2.0468∗ 24.4003 2.2700∗ 11.3577 −2.9100∗ −9.8123 −0.3235∗ −2.8261 0.7432∗ 4.589 0.1385∗ 9.6554 −0.1142∗ −13.5134 −1.3400∗ −4.349 −0.0171∗ −4.082 −0.0058∗ −3.1022 39,003 0.22 0.5881 408

Coefficient

t-Statistic

Coefficient

0.9936∗ 0.2520‡ −0.3140 0.2119 −0.3635‡ 0.0162‡ −0.0037 −0.2070 0.0046∗∗ −0.007∗∗ 40,127

15.5336 1.3311 −0.795 1.2438 −1.4247 1.3912 −0.5009 −1.1264 1.9239 −1.9253

1.7229∗ 26.1308 9.9009 1.6900∗ −2.2100∗ −8.5357 −0.1291† −1.8244 0.3608∗ 3.5179 0.1606∗ 11.2917 −0.082∗ −11.3659 −1.0100∗ −4.8698 −0.0047‡ −1.4138 −0.0073∗ −3.8877 38,235

0.29 0.0717 81.8104 408

0.18 0.4807 408

t-Statistic

Coefficient

t-Statistic

1.1259∗ 0.5090∗ −0.6410† 0.2298‡ −0.3676‡ 0.0582∗ −0.0193∗ −0.3260† 0.0033‡ −0.0069∗∗ 39,704

18.8822 2.8462 −1.6971 1.4494 −1.5445 4.5809 −2.9312 −1.7134 1.3335 −2.1134

0.31 0.076 51.3601 408

275

Note: The Hausman test indicates that the fixed-effects specification is preferred in both cases. ∗ Significance at the 1% level. ∗∗ Significance at the 5% level. † Significance at the 10% level. ‡ Significance at the 20% level.

Regional Development and Economic Growth. . . b1491-ch11

Constant GDPPC GDPPC2 IND IND2 TIt−1 LN(KL) EPC OPEN UNFCCC Turning point (GDPPC) Turning point (INDS) R2 Hausman test Number of observations

t-Statistic

Random Effect

9in x 6in

Coefficient

Fixed Effect

11:55

Variable

Random Effect

Environmental Regulation and Productivity Growth in APEC Economies

Fixed Effect

Scenario 3

March 5, 2013

276

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

volume of trade and use of energy. In addition, the coefficient of the dummy variable is negative and statistically significant in both cases. This is contradictory to Yörük and Zaim (2005) who provided empirical evidence of a positive impact of UNFCCC on productivity growth in OECD countries that have ratified the convention.

11.6. Conclusions To sum up, traditional measures of productivity ignore the undesirable outputs and abatement activities and hence are likely to be biased. This study applied a well-developed approach to examine productivity growth under three policy options for environmental regulation. Using a sample of 17 APEC economies during the period 1980–2004, it is found that in the absence of environmental regulations the average productivity growth was 0.25% which was largely due to technical efficiency change. However, if the policy objective is to maintain or reduce the current level of CO2 emissions, average productivity growth is estimated to be 0.45% or 0.39% which was largely due to technological progress. Thus, with environmental regulations, on average the TFP growth for 17 APEC economies is slightly higher than that without regulations. This finding is supported by other studies.11 This study also shows that out of 17 countries eight shifted the frontier at least once. The determinants of the variation in productivity growth among APEC members are also investigated under two regulatory options. In general, more industrialized and advanced economies have shown better productivity performance. However, the productivity index is found to be negatively associated with technical efficiency and the capital–labor ratio, indicating the possibility of catch-up movement among the economies. In addition, energy intensity and openness of an economy are shown to be negatively related to productivity growth. Thus there are potentially undesirable effects from energy inefficiency and expanded international trade. 11 Examples include Boyd et al. (1999), Ball et al. (2001), and Jeon and Sickles (2004).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

277

Appendices to Chapter 11 A11.1 Linear programming problems for Scenarios 1 and 2: Scenario 1 (no environmental regulations)  ot (xkt  , ykt  , 0; ykt  , 0) = Maxβ D s.t. K 

t ztk ykm ≥ (1 + β)ykt  m ,

m = 1, . . . , M (A11.1)

k=1 K 

t ztk xkn ≤ xkt  n ,

n = 1, . . . , N

k=1

ztk ≥ 0,

k = 1, . . . , K

Scenario 2 (bad output, CO2 emissions, is held constant)  ot (xkt  , ykt  , bkt  ; ykt  , 0) = Maxβ D s.t. K 

t ztk ykm ≥ (1 + β)ykt  m ,

m = 1, . . . , M

k=1 K 

t ztk bki = bkt  i ,

i = 1, . . . , I

(A11.2)

k=1 K 

t ztk xkn ≤ xkt  n ,

n = 1, . . . , N

k=1

ztk ≥ 0,

k = 1, . . . , K

A11.2 Capital stock estimates Capital stock data are derived following Wu (2004) who applied the conventional perpetual inventory approach, that is, Kt = Kt + (1 − δ)Kt−1 ,

(A11.3)

March 5, 2013

278

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

where Kt is the capital stock at time t for each economy, δ a given rate of depreciation and Kt the incremental capital at time t. Kt is computed from the real investment share of GDP presented in the PWT6.2 for the period 1950–2004 for most economies (data for Hong Kong, Indonesia and Singapore cover the period 1960–2004). The data series for Kt are backcasted to the year 1900. Accordingly, Equation (A11.3) is expanded into: Kt =

t−1901 

(1 − δ)k Kt−k + (1 − δ)t−1900 K1900 .

(A11.4)

k=0

Equation (A11.4) implies that, given the value of capital stock in 1900 and an appropriate rate of depreciation, a capital stock series for each economy can be derived. In this study, K1900 is assumed to be zero. This assumption is made due to the fact that the value of capital stock existed in 1900 would be zero by the 1980s and 1990s due to capital decay. While the potential impact of the choice of the rate of depreciation is noted, due to data constraints this chapter applies a unified rate of depreciation of 7% for all economies in the sample. A sensitivity analysis is applied to shed some light on the possible impact of effective depreciation rates. We choose a rate of depreciation of 4% for developing countries and Taiwan, and 7% for other economies. The estimation results hardly change.

A11.3 Estimation results for sub-periods Table A11.1. Average Indices and Changes under Scenario 1. 1980–1991

AUS CAN CHL CHN HKG IDN JPN KOR

1992–2004

Change

PI

EC

TP

PI

EC

TP

PI

EC

TP

1.0022 0.9919 1.0031 1.0083 1.0010 0.9788 1.0278 1.0053

0.9939 0.9910 0.9965 1.0484 0.9997 1.0006 1.0133 0.9971

1.0084 1.0009 1.0067 0.9617 1.0014 0.9782 1.0142 1.0082

1.0000 1.0059 0.9791 1.0007 0.9992 0.9902 1.0066 0.9957

1.0063 1.0117 0.9909 0.9944 0.9982 0.9866 0.9850 1.0003

0.9938 0.9942 0.9880 1.0064 1.0011 1.0036 1.0219 0.9955

−0.0022 0.0140 −0.0240 −0.0076 −0.0018 0.0114 −0.0212 −0.0096

0.0124 0.0207 −0.0056 −0.0540 −0.0015 −0.0140 −0.0283 0.0032

−0.0146 −0.0067 −0.0187 0.0447 −0.0003 0.0254 0.0077 −0.0127

(Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

279

Table A11.1. (Continued) 1980–1991

MEX MYS NZL PER PHL SGP THA TWN USA Mean 1 Mean 2 Mean 3 Mean 4 Mean5

1992–2004

Change

PI

EC

TP

PI

EC

TP

PI

EC

TP

0.9954 0.9842 1.0002 0.9876 1.0028 1.0314 1.0016 1.0153 1.0094 1.0062 1.0011 0.9952 1.0132 1.0026

0.9867 0.9795 1.0001 0.9826 1.0103 1.0169 1.0074 1.0050 1.0014 0.9999 1.0024 1.0013 1.0046 1.0017

1.0088 1.0047 1.0002 1.0050 0.9926 1.0142 0.9943 1.0103 1.0080 1.0063 0.9987 0.9939 1.0085 1.0010

1.0002 1.0038 1.0037 1.0058 1.0088 1.0324 0.9987 0.9993 1.0138 1.0060 1.0011 0.9984 1.0065 1.0025

1.0139 1.0143 1.0064 1.0105 1.0009 1.0099 1.0108 1.0000 1.0014 1.0021 1.0025 1.0027 1.0021 1.0024

0.9865 0.9897 0.9974 0.9953 1.0079 1.0223 0.9880 0.9993 1.0124 1.0039 0.9986 0.9956 1.0045 1.0001

0.0048 0.0196 0.0035 0.0182 0.0060 0.0010 −0.0029 −0.0160 0.0044 −0.0002 0.0000 0.0032 −0.0067 –1E–04

0.0272 0.0348 0.0063 0.0279 −0.0094 −0.0070 0.0034 −0.0050 0.0000 0.0022 0.0000 0.0014 −0.0025 0.0007

−0.0223 −0.0150 −0.0028 −0.0097 0.0153 0.0081 −0.0063 −0.0110 0.0044 −0.0024 −0.0000 0.0017 −0.0040 −0.0009

Table A11.2. Average Indices and Changes under Scenario 2. 1980–1991

AUS CAN CHL CHN HKG IDN JPN KOR MEX MYS NZL PER PHL SGP THA TWN USA Mean 1 Mean 2 Mean 3 Mean 4 Mean 5

1992–2004

Change

PI

EC

TP

PI

EC

TP

PI

EC

TP

1.0053 1.0062 1.0063 0.9785 1.0256 0.9844 1.0271 1.0071 0.9872 0.9841 0.9945 0.9971 1.0143 1.0306 1.0116 1.0321 1.0128 1.0091 1.0047 0.9954 1.0238 1.006

0.9868 0.9885 1.0091 1.0181 1.0000 1.0034 1.0059 0.9963 0.9777 0.9844 0.9757 0.9899 0.9987 1.0108 0.9968 1.0016 1.0000 0.9913 0.9988 0.9972 1.0022 0.9966

1.0188 1.0180 0.9972 0.9611 1.0256 0.9811 1.0211 1.0109 1.0097 0.9996 1.0192 1.0074 1.0156 1.0196 1.0149 1.0304 1.0128 1.0180 1.0059 0.9982 1.0216 1.0094

0.9993 1.0095 1.0000 0.9506 1.0165 0.9985 1.0056 1.0010 0.9975 0.9985 1.0113 1.0160 1.0096 1.0330 0.9912 1.0075 1.0171 1.0085 1.0015 0.9951 1.0144 1.0036

1.0037 1.0065 1.0033 0.9800 1.0000 1.0001 0.9892 1.0002 1.0061 1.0050 1.0115 1.0187 1.0012 1.0134 0.9970 1.0000 1.0012 1.0024 1.002 1.0014 1.0034 1.0021

0.9956 1.0030 0.9967 0.9701 1.0165 0.9983 1.0165 1.0008 0.9915 0.9935 0.9998 0.9974 1.0084 1.0193 0.9941 1.0075 1.0158 1.0061 0.9994 0.9937 1.0110 1.0014

−0.0060 0.0033 −0.0063 −0.0279 −0.0091 0.0141 −0.0215 −0.0061 0.0103 0.0144 0.0168 0.0189 −0.0047 0.0024 −0.0204 −0.0246 0.0043 −0.0006 −0.0032 −0.0003 −0.0094 −0.0024

0.0169 0.0180 −0.0058 −0.0381 0.0000 −0.0033 −0.0167 0.0039 0.0284 0.0206 0.0358 0.0288 0.0025 0.0026 0.0002 −0.0016 0.0012 0.0111 0.0032 0.0042 0.0012 0.0055

−0.0232 −0.0150 −0.0005 0.009 −0.0091 0.0172 −0.0046 −0.0101 −0.0182 −0.0061 −0.0194 −0.0100 −0.0072 −0.0003 −0.0208 −0.0229 0.003 −0.0119 −0.0065 −0.0045 −0.0106 −0.0080

(Continued)

March 5, 2013

280

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

Table A11.3. Average Indices and Changes under Scenario 3. 1980–1991

AUS CAN CHL CHN HKG IDN JPN KOR MEX MYS NZL PER PHL SGP THA TWN USA Mean 1 Mean 2 Mean 3 Mean 4 Mean 5

1992–2004

Change

PI

EC

TP

PI

EC

TP

PI

EC

TP

1.0041 1.0043 1.0092 1.0248 1.0191 0.9895 1.0220 1.0001 0.9884 0.9869 0.9946 0.9994 1.0079 1.0237 0.9966 1.0162 1.0063 1.0062 1.0051 1.0003 1.0147 1.0054

0.9888 0.9896 1.0090 1.0196 1.0000 1.0021 1.0049 1.0008 0.9839 0.9884 0.9763 0.9958 0.9988 1.0078 0.9903 1.0015 1.0000 0.9919 0.9998 0.9984 1.0025 0.9975

1.0154 1.0148 1.0002 1.0051 1.0191 0.9874 1.0170 0.9993 1.0046 0.9985 1.0188 1.0036 1.0091 1.0157 1.0064 1.0147 1.0063 1.0145 1.0053 1.0018 1.0122 1.0080

1.0105 1.0111 0.9998 0.9782 1.0105 0.9967 1.0049 0.9921 0.9978 0.9981 1.0082 1.0082 1.0070 1.0282 0.9939 1.0021 1.0012 1.0072 1.0010 0.9974 1.0081 1.0028

1.0028 1.0052 1.0027 1.0050 1.0000 1.0001 0.9915 0.9988 1.0036 1.0036 1.0125 1.0084 1.0011 1.0116 0.9946 1.0000 1.0010 1.0026 1.0024 1.0024 1.0026 1.0025

1.0076 1.0058 0.9970 0.9734 1.0105 0.9966 1.0135 0.9933 0.9942 0.9945 0.9959 0.9997 1.0059 1.0165 0.9993 1.0021 1.0002 1.0046 0.9985 0.9950 1.0056 1.0003

0.0064 0.0068 −0.0094 −0.0466 −0.0086 0.0072 −0.0171 −0.0080 0.0094 0.0112 0.0136 0.0088 −0.0009 0.0045 −0.0027 −0.0141 −0.0051 0.001 −0.0041 −0.0029 −0.0066 −0.0026

0.014 0.0156 −0.0063 −0.0146 0.000 −0.0020 −0.0134 −0.0020 0.0197 0.0152 0.0362 0.0126 0.0023 0.0038 0.0043 −0.0015 0.001 0.0107 0.0026 0.004 0.000 0.005

−0.0078 −0.009 −0.0032 −0.0317 −0.0086 0.0092 −0.0035 −0.006 −0.0104 −0.004 −0.0229 −0.0039 −0.0032 0.0008 −0.0071 −0.0126 −0.0061 −0.0099 −0.0068 −0.0068 −0.0066 −0.0077

References Chambers, R. G., Färe, R. and Grosskopf, S., “Productivity Growth in APEC Countries,” Pacific Economic Review, 1(3): 181–190 (1996). Chang, C. and Luh, Y., “Efficiency Change and Growth in Productivity: The Asian Growth Experience,” Journal of Asian Economics, 10: 551–570 (1999). Chung, Y. H., Färe, R. and Grosskopf, S., “Productivity and Undesirable Outputs: A Directional Distance Function Approach,” Journal of Environmental Management, 51: 229–240 (1997). Coggins, J. S. and Swinton, J. R., “The Price of Pollution: A Dual Approach to Valuing SO2 Allowances,” Journal of Environmental Economics and Management, 30: 58–72 (1996).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

281

Domazlicky, B. and Weber, W., “Does Environmental Protection Lead to Slower Productivity Growth in the Chemical Industry?” Environmental and Resource Economics, 28: 301–324 (2004). Etkins, P., Folke, C. and Costanza, R., “Trade, Environment and Development: The Issues in Perspective,” Ecological Economics, 9: 1–12 (1994). Färe, R. and Grosskopf, S., Intertemporal Production Frontiers: With Dynamic DEA. Boston: Kluwer Academic Publishers (1996). Färe, R. and Grosskopf, S., New Directions: Efficiency and Productivity. Boston/London/Dordrecht: Kluwer Academic Publishers (2004). Färe, R. and Primont, D., Multi-output Production and Duality: Theory and Applications. Boston/London/Dordrecht: Kluwer Academic Publishers (1995). Färe, R., Grosskopf, S. and Pasurka, C., “Accounting for Air Pollution Emissions in Measuring State Manufacturing Productivity Growth,” Journal of Regional Science, 41: 381–409 (2001a). Färe, R., Grosskopf, S. and Margaritis, D., “APEC and the Asian Economic Crisis: Early Signals from Productivity Trends,” Asian Economic Journal, 15: 325–342 (2001b). Färe, R., Grosskopf, S., Noh, D. W. and Weber, W., “Characteristics of a Polluting Technology: Theory and Practice,” Journal of Economics, 126: 469–492 (2005). Färe, R., Grosskopf, S., Lovell, K. C.A. and Yaisawarng, S., “Derivation of Shadow Prices for Undesirable Outputs: A Distance Function Approach,” Review of Economics and Statistics, 75: 374–380 (1993). Färe, R., Grosskopf, S., Norris M. and Zhang, Z., “Productivity Growth, Technological Progress, and Efficiency Change in Industrialized Countries,” American Economic Review, 84(1): 66–82 (1994). Grossman, G. and Krueger, A., “Economic Growth and the Environment,” Quarterly Journal of Economics, 110(2): 352–377 (1995). Hailu, A. and Veeman, T. S., “Environmentally Sensitive Productivity Analysis of the Canadian Pulp and Paper Industry, 1959–1994: An Input Distance Function Approach,” Journal of Environmental Economics and Management, 40: 251–274 (2000). Heston, A., Summers, R. and Aten, B., “Penn World Table Version 6.2,” Center for International Comparisons of Production, Income and Prices at the University of Pennsylvania, September (2006).

March 5, 2013

282

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Regional Development and Economic Growth in China

Jaffe, A. B., Peterson, S., Portney, P. and Stavins, R., “Environmental Regulation and the Competitiveness of U.S. Manufacturing: What Does the Evidence Tell Us?” Journal of Economic Literature, 33: 132–163 (1995). Jeon, B. M. and Sickles, R. C., “The Role of Environmental Factors in Growth Accounting,” Journal of Applied Econometrics, 19: 567–591 (2004). Kopp, G., “Carbon Dioxide Emissions and Economic Growth: A Structural Approach,” Journal of Applied Statistics, 25(4): 489–515 (1998). Kumar, S., “Environmentally Sensitive Productivity Growth: A Global Analysis Using Malmquist–Luenberger Index,” Ecological Economics, 56: 280–293 (2006). Lall, P., Featherstone, A. M. and Norman, D. W., “Productivity Growth in the Western Hemisphere (1978–94): The Caribbean in Perspective,” Journal of Productivity Analysis, 17: 213–231 (2002). Lindmark, M. and Vikström, P., “Global Convergence in Productivity — A Distance Function Approach to Technological Progress and Efficiency Improvements,” Paper for the Conference Catching-up growth and technology transfers in Asia and Western Europe, Groningen 17–18 October (2003). http://www.ggdc.net/conf/paper-vikstromlindmark.pdf. Luenberger, D. G., “Benefit Functions and Quality,” Journal of Mathematical Economics, 21: 461–481 (1992). Luenberger, D. G., Microeconomic Theory. Boston: McGraw-Hill (1995). Pittman, R. W., “Multilateral Productivity Comparisons with Undesirable Outputs,” Economic Journal, 93: 883–891 (1983). Reig-Martínez, E., Picazo-Tadeo, A. J. and Hernández-Sancho, F., “Shadow Prices and Distance Functions: An Analysis for Firms of the Spanish Ceramic Pavements Industry,” International Journal of Production Economics, 69: 277–285 (2001). Repetto, R., Rothman, D., Faeth, P. and Austin, D., “Has Environmental Protection Really Reduced Productivity Growth?” Challenge, January–February: 46–57 (1997). Selden, T. M. and Song, D., “Environmental Quality and Development: Is there a Kuznets Curve for Air Pollution Emissions?” Journal of Environmental Economics and Management, 27: 147–162 (1994). Swinton, J. R., “At What Cost do We Reduce Pollution? Shadow Prices of SO2 Emissions,” Energy Journal, 19: 63–83 (1998).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

Environmental Regulation and Productivity Growth in APEC Economies

283

Taskin, F. and Zaim, O., “The Role of International Trade on Environmental Efficiency: A DEA Approach,” Economic Modelling, 18: 1–17 (2001). World Bank, World Development Indicators. Washington, DC: World Bank (2007). Wu, Y., “Openness, Productivity and Growth in the APEC Economies,” Empirical Economics, 29: 593–604 (2004). Yörük, B. K. and Zaim, O., “Productivity Growth in OECD Countries: A Comparison with Malmquist Indices,” Journal of Comparative Economics, 33: 401–442 (2005).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch11

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Chapter 12

Inflation Transmission in China’s Goods and Asset Markets Huawei Liu and Juan Yang

12.1. Introduction In the last decade, China’s economy has experienced a golden age, with an average GDP growth of 10% and relatively stable inflation (around 2%). This achievement is at least partly attributable to a successful monetary policy. In China, money supply has significantly increased in the past decade (almost doubling during the 2008–2009 financial crisis), while GDP, CPI and interest rate have remained relatively stable over the same period (see Figure 12.1). Recall that the classical money demand function denoting the equilibrium condition for money market is MPs = L(i, Y ). Considering China’s case, there is a puzzling effect of monetary policy; we have not observed the proportional rise in inflation in line with increases in the money supply. Does that mean the famous Friedman assertion “inflation is everywhere and anytime monetary phenomenon” was untrue in China? At the same time, a great volatility in various asset markets was observed in China, compared to a relatively stable price level. The stock market in China has experienced dramatic fluctuations in the past decade; the Shanghai Composite Stock Index peaked at point 6,124 in the year 2007 and then dropped to 1,728.79 in the year 2008.1 It is also worth noting that housing prices have soared in the past decade, to the extent that the Chinese government has started an administrative intervention to stabilize it. This trend in asset price changes can be found in Figure 12.2. We doubt

1 Data is from China Economic Network Statistic Database.

285

March 5, 2013

286

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China 30.00% 25.00% 20.00% 15.00% 10.00% 5.00% 0.00% -5.00%

2006

2007 M2

2008 CPI

2009

R

2010

GDP

Figure 12.1. The Growth Rate of Money Supply, CPI, GDP and the Interest Rate from 2006 to 2010 in China.

6000 5000 4000 3000 2000 1000 0

2005

2006

2007

2008 SP

2009

2010

HP

Figure 12.2. Time Trend for the Housing Price and Shanghai Composite Stock Index from 2005 to 2010.

whether these movements can help explain such a disproportional response in inflation, represented by the change in consumer price index (CPI), to the change of money supply existing in China. Thus, we introduce the major asset markets (housing, stock market and so on) in China, which are supposed to act as the major supplement of price level to deflate the real balance circulating in the system. Therefore the new equilibrium condition Ms for the money market would be f(Pc,Pa) = L(i, Y ), where the denominator, to derive the real balance, is defined as the function of the goods price

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

287

and asset prices. Under this assumption, the price of the housing market, stock market, and commodity future prices will all be influenced by monetary policy, and a proportionally positive increase will be expected in each market, ceteris paribus, following an expansionary monetary policy. Our study seeks to answer two questions. First, can adding a different perspective of the asset market explain the disproportional change in inflation to the change in money supply in China? Second, what are the dynamic interactions that occur among good and asset markets for the inflation transmission?

12.2. Literature Review The issue of how asset price and consumption price interact with each other has generated a lot of discussion in the literature. In one aspect, a common view is that an increase in asset price will generate a wealth effect, which means that higher asset prices will encourage people to consume more, and thus increase the CPI (Goodhart and Hofmann, 2002; Kontonikas, 2005). On the other hand, some scholars believe that when inflation is severe, people will turn to the asset market to hedge against inflation and that a high CPI will lead to high asset prices (Bernanke and Gerter, 2000). Li et al. (2010) investigated whether stocks provide a hedge against inflation using UK data, and found that the UK stock market fails to hedge against inflation in the short term, whilst in the mid-term the results are mixed and is dependent on different inflation regimes. Gao (2009) conducted empirical research on the relationship between asset prices and the inflation rate in China based on the VAR model using monthly data from 1998 to 2008. The results show that the price of real estate and equities are closely linked to inflation. The policy implication is that asset prices should be brought into the consideration of the monetary authority to manage inflation. However, the relationship between asset prices and inflation in China has been ambiguous in the literature. Our study will contribute to the identification of the inflation transmission between the asset market and good market in China. Currently, the primary method of controlling inflation is through monetary policy. Most central banks aim to keep a stable inflation rate, normally with a target rate of around 2% to 3% per annum, and within a targeted low

March 5, 2013

288

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

inflation range, somewhere from about 2% to 6% per annum.2 However, without a comprehensive understanding of inflation dynamics between the asset market and good market, the efficiency of monetary policy is in doubt. In order to determine the inflation dynamics of the asset price and good price in China, we establish a vector error correction model (VECM) to investigate the dynamic system of money supply, interest rate, CPI, housing price, stock price, commodity price, and the foreign exchange rate. This provides an informative policy implication to the monetary authority for their future operation and practice.

12.3. Model and Variable Selection The underlying theoretical model of this study is the traditional Cagan money demand function augmented with asset price. In China, the major asset prices refer to stock and housing prices. Moreover, the expectation of inflation will affect the money demand greatly, and the commodity future price is also widely used to capture the inflation expectation in the literature (Sims, 1992; Bernanke and Mihov, 1998). Since China launched their foreign exchange rate reform in 2005, the RMB has appreciated by 20%. Following on from this, China now has a large current and capital account surplus, and has accumulated massive foreign exchange reserves. With strong expectations of the RMB appreciating, people will choose to exchange the Dollar for the RMB, which will greatly influence the money demand. Seymur (2011) investigated the causes of inflation, pointing out that in the long-run exchange rate appreciation has a strong influence on inflation. Therefore, we add the exchange rate to the model, with the basic empirical model constructed as follows: Ms = L(i, GDP, er), f(CPI, FP, HP, SP)

(12.1)

where CPI represents the consumer price index, FP is the commodity future price, HP is the housing price, SP is the stock price, GDP is the real output, 2 This viewpoint is quoted from Wikipedia.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

289

and ER represents the nominal foreign exchange rate. An expansionary monetary shock would be expected to lead to an increase in CPI, FP, HP, SP and GDP, and a decrease in the interest rate and exchange rate (appreciation), with all other conditions unchanged in this model.3 In our preliminary study, the commodity future price is not included, and in that case the price puzzle appears, meaning a decline in inflation will follow an expansionary monetary shock. A number of studies have tried to address this problem in different ways. Some include potential output to remove it (Giordani, 2004), while others estimate the model by adding different measurements of inflation expectation such as the commodity price (Sims, 1992; Chari et al., 1995; Bernanke and Mihov, 1998). In this chapter, we add the commodity future price to remove the price puzzle. In the literature, interest rate is often used as an indicator of monetary policy. In such models, an output puzzle often emerges which shows that loose monetary policy will slow down economic growth (Eichenbaum, 1992; Gordon and Leeper, 1994; Angeloni, 2003; Cyrille, 2010). This contradicts real economy theory and may be caused by the exclusion of important variables. This chapter uses money supply (M2) to represent monetary policy and also incorporates the interest rate of the three-month central bank note as an exogenous variable in the system. This is because the interest rate in China has not been liberalized yet and is still under the strict control of the central bank. The preliminary Granger Causality/Block Exogeneity test shows that we should include such a variable in the system, while paired Granger causality reveals no significant causality from the variables such as CPI, money supply and GDP to the interest rate. The test results are robust to the altered measurement of interest rate, i.e., the threemonth deposit rate, one year central bank note rate etc, and simply reveals the fact that there is no obvious feedback from other variables to the interest

3 Traditional purchasing power theory suggests that the increase in money supply will

increase the domestic price level and thus depreciate the exchange rate. In our model, we focus on the effect of the exchange rate on the demand side which posits that a decline in exchange rate (appreciation) will induce demand for domestic money, thus given all other conditions unchanged, the increase in money supply causes insufficient money demand and to restock money market equilibrium the exchange rate has to appreciate to induce more money demand.

March 5, 2013

290

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

rate. Therefore, we treat the interest rate as an exogenous variable in this study. By using this setting, our empirical model removes the price puzzle but still records the output puzzle. In the subsequent analysis, we will estimate the VECM model defined by Equation (12.2).
Xt = αβ
Xt−1 +

k


i
Xt−i + µ + et ,

(12.2)

i=1

where
X = (M2, CPI, HP, FP, SP, ER, GDP, Interest Rate),  = αβ
is a coefficient matrix, α can be viewed as the matrix of the speed of the adjustment parameters, β is the matrix of co-integrating parameters, i is a matrix of short-run dynamics coefficients, et is a vector of innovations and µ represents the time trend (constant). The one-step Schwarz Loss test discussed in the next section will show if the time trend exists in the model. The parameters in the VECM above can be partitioned to provide information on the long-run, short-run and contemporaneous structure. The longrun structure can be examined through testing the hypothesis on β and the short-run structure can be identified through testing hypotheses on the variables α and i . For this we will calculate the impulse response function and variance decomposition to find out the dynamic interactions among the variables we are interested in.

12.4. Data China has been implementing its foreign exchange rate reform since 2005, transitioning from a fixed exchange rate system to a relatively floated exchange rate. To avoid the obvious structure break associated with this, we only use the data collected from 2005 to 2010. We have picked up seven endogenous variables (M2, GDP, CPI, SP, FP, HP, and ER) and one exogenous variable (R) in our study, with all series used being monthly data. R represents the interest rate of three-month central bank notes. M2 is the broad money supply in China coming from the WIND database, which is created by a leading financial data company in China. CPI is sourced from the China Macro-Economic Database and has

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

291

been processed to use the 2000 January level as the base level. HP represents the housing price, which can be derived by dividing the total sales value over total areas of sales. GDP used in this study is the real-term (nominal GDP deflated by CPI). Since China never publishes monthly GDP data, we use electricity usage as a weight to generate the monthly GDP. ER is the nominal foreign exchange rate which is reported as the direct exchange rate (the RMB value of US dollar). FP represents the Chinese commodity futures index, which is developed by the Peking University HSBC Business School. This commodity price index is a weighted average price index for the most actively traded commodity contracts. The weights of the composite commodity contracts are calculated based on the average value for the past two years. All the series except for the interest rate have been seasonally adjusted and processed with a logarithm transformation. We conduct the stationarity test for each series respectively, and the results from the test are stated in Tables 12.1 and 12.2. Based on the specification of the model without intercept and trend, the test results have been reported. We also test using the other two common specifications with intercept and/or trend, finding that the conclusions are robust and we accept the null hypothesis with the unit root in each series (except for M2, which rejects the null hypothesis with trend in the model, suggesting that M2 might be trend stationary and still non-stationary at this level). In each test, the optimal lag length is chosen based on the Schwartz Information Criterion.

Table 12.1. Augmented Dickey–Fuller Test for the Level Series. Variable CPI GDP SP HP FP ER M2

t-statistics

p-value

−0.159 −0.712 −1.850 −1.136 −1.648 −1.246 1.219

0.938 0.836 0.350 0.697 0.453 0.650 0.942

March 5, 2013

292

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

Table 12.2. Augmented Dickey–Fuller Test for the First Difference of Each Series. Variable

t-statistics

p-value


CPI
GDP
SP
HP
FP
ER
M2

−6.498 −10.325 −3.779 −10.778 −6.516 −3.308 −8.200

0.000 0.000 0.004 0.000 0.000 0.018 0.000

Results in Tables 12.1 and 12.2 show that all the variables are following the I(1) process and thus there might exist co-integration among the variables in the long run.

12.5. Empirical Results The commonly used procedure in recent studies for a co-integration process is to use either a trace test or information criterion to determine the lag order of the unrestricted VAR in the first step. The second step is to then use the same criterion to determine the co-integration rank and appropriate specification for ECM. However, this chapter will use the one step Schwart Loss Criterion (SLC) to determine the lag length and the co-integration vectors in the VECM simultaneously. This has been proven to work at least as well as or even better than the traditional trace test or two steps approach, in terms of both efficiency and consistency (Wang and Bessler, 2005). Step by step, we check the Schwart Loss for each rank = 1, 2 . . . and each model specification by lags to find the one yielding the lowest SL. We start our search from lag 1 and continue to lag 12 (detailed statistics are listed in Table 12.3). According to the Schwarz Loss Criterion, the optimal model is the one with one lag and no deterministic trend in data, and no intercept or trend in co-integrating equation (CE).4 4 In our preliminary study, the Johansen test is conducted; the results support cointegration

with three cointegrating vectors and linear deterministic trend in the co-integrating vector.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

293

Table 12.3. One Step Schwarz Loss Criteria by Lags on the Number of Co-Integrating Vectors (R) and Model Specifications Fit over Period January 2005–December 2010. Model Lag 1 1 1 1 1 1 2 2 2 2 2 2 3 3 3 3 3 3

Rank

No intercept No trenda

Intercept No trend Ib

Intercept No trend IIc

Intercept Trend Id

Intercept Trend IIe

1 2 3 4 5 6 1 2 3 4 5 6 1 2 3 4 5 6

− 20.33 −20.13 −19.67 −19.13 −18.38 −17.57 −18.61 −18.27 −17.82 −17.28 −16.57 −15.78 −16.62 −16.39 −15.96 −15.42 −14.77 −14.01

−20.27 −20.02 −19.50 −18.90 −18.20 −17.39 −18.54 −18.19 −17.70 −17.18 −16.44 −15.65 −16.56 −16.26 −15.89 −15.32 −14.70 −13.98

−20.16 −19.96 −19.50 −18.86 −18.13 −17.36 −18.49 −18.19 −17.74 −17.07 −16.36 −15.61 −16.63 −16.33 −15.90 −15.34 −14.70 −13.92

−20.30 −20.04 −19.71 −19.13 −18.43 −17.64 −18.54 −18.24 −17.77 −17.22 −16.49 −15.72 −16.65 −16.36 −16.00 −15.38 −14.71 −14.00

−20.06 −19.83 −19.56 −19.05 −18.41 −17.66 −18.31 −18.07 −17.67 −17.17 −16.50 −15.76 −16.51 −16.28 −15.96 −15.41 −14.80 −14.02

Notes: The optimal lag and rank combination is marked in bold in the table. a This test assumes no deterministic trend in data, and no intercept or trend in CE or test VAR; b This test assumes no deterministic trend in data and that it has intercept (no trend) in CE, but no intercept in VAR; c This test allows for a linear deterministic trend in data, and assumes intercept (no trend) in CE and tests VAR; d This test allows for linear deterministic trend in data, and assumes intercept and trend in CE, but no trend in VAR; e This test allows for quadratic deterministic trend in the data, and assumes intercept and trend in CE and linear trend in VAR.

We use the software EVIEWS6.0 to estimate the model as specified in Equation (12.3). The diagnostic checks for this modeling are conducted. Two types of serial correlation tests for residuals (LM test and Portmanteau test) both support the null hypothesis for no serial correlation in all short-run

March 14, 2013

294

10:58

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

and long-run lag levels. The co-integration relationship is expressed in Equation (12.4).    T −0.754
Mt 1.000 
ER  −0.007  1.379      
FPt  −0.015 −0.296      
SP  = −0.050  0.165      
CPIt  −0.003  14.941      
HPt  −0.017 −1.874 −0.017 −6.927 t
Yt 



−0.213  0.004  −0.003  + −0.032 −0.001   0.005 0.005



 Mt−1  ERt−1     FPt−1     SPt−1    CPI t−1     HPt−1  Yt−1

   M −2.028 −0.474 1.876 −10.919 2.861 4.270 −0.015  t−1  ERt−1  0.381 0.008 0.025 −0.036 −0.017 0.029 −0.002   FPt−1    −0.011 0.171 0.018 1.554 0.273 −0.071 −0.0003  SPt−1    −0.079 0.070 0.064 −2.032 0.355 0.017 0.0004  CPI t−1    −0.055 0.0156 −0.001 0.076 0.023 0.012 0.0001  HPt−1   −0.479 0.001 0.236 −1.396 −0.309 0.144 −0.001   Yt−1  −0.102 0.209 0.138 −0.509 −0.183 −0.531 0.001 R

(12.3) M2 = − 1.19 ER + 0.30 FP − 0.17 SP + 11.9 5CPI (5.03)

(0.33)

(0.79)

+ 2.19 HP − 5.67 GDP, (−1.75)

(3.01)

(−3.86)

(12.4)

where t-statistics are included in parenthesis. It shows that in the long-run equilibrium, when M2 increases, CPI, housing prices and commodity prices will increase, while exchange rate, stock prices and GDP will decrease. As we expect, it explains the flow of liquidity to different markets. However, for the VECM, over-parameterization is a common problem and the coefficients do not mean much. More reliable analysis can come from innovation accounting. To obtain accurate innovation accounting, we need to identify the contemporaneous causality among the variables. Table 12.4 shows that the correlation among different variable residues is insignificant, with most of them being smaller than 40%. In this case, contemporaneous causality among different variables does not matter much to the final result of the impulse response analysis and variance decomposition. The estimated impulse responses are reported in Figures 12.3 and 12.4 to address two basic research questions. From Figure 12.3, the impulse responses confirm our theoretical hypothesis that good and asset prices

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

295

Table 12.4. The Correlation Matrix among the Variables. M2

ER

FP

SP

CPI

HP

M2 1.000 ER 0.108 1.000 FP 0.506 −0.084 1.000 SP 0.323 0.021 0.401 1.000 CPI −0.085 0.027 0.046 −0.041 1.000 HP 0.181 0.095 0.071 0.112 −0.046 1.000 GDP −0.118 −0.087 −0.015 0.052 −0.109 −0.194

GDP

1.000

rise (housing price, stock price, and commodity price) following an expansionary monetary shock. Moreover, all asset prices respond more significantly to the change of money supply, indicating that the excessive liquidity could be absorbed in the asset sectors. We also observe a decline in the GDP and exchange rate after a positive shock, which implies that the monetary policy in China is not as effective as we expect it to be, and that it also coincides with previous studies in the output puzzle. A possible explanation for such an output puzzle in China may be the unmatched timing for monetary policy implementation. The decrease in the exchange rate following a positive monetary shock is consistent with our theory that the rising money supply must be followed by an appreciation of the exchange rate to induce more foreign demand for money (given all other conditions are unchanged). In Figure 12.4, we find out that the CPI responds positively to all other variables except for the exchange rate. This finding supports the wealth effect hypothesis; higher asset prices will accelerate inflation in the long run. If the monetary authority would like to counter inflation, then asset prices should also be monitored. If we use the commodity price to approximately measure the inflation expectation from the impulse response, it is clear that a high expectation about inflation will influence the current inflation, and that a positive inflation expectation will actually be realized over time. The negative response of CPI to the exchange rate is puzzling; the lower exchange rate (the larger appreciation) should dampen the domestic price level, but instead we have the opposite results, that the appreciation will foster inflation. This is related to the role the exchange rate plays in the

March 5, 2013

296

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China Response to Cholesky One S.D. Innovations Response of M2 to M2

.8

Response of ER to M2

.000

.7

-.002

.6 .5

-.004

.4

-.006

.3 -.008

.2 .1

-.010 2

4

6

8 10 12 14 16 18 20 22 24

Response of FP to M2

.05

2

4

6

8 10 12 14 16 18 20 22 24

Response of SP to M2

.09 .08

.04

.07

.03

.06 .02

.05

.01

.04

.00

.03 2

4

6

8 10 12 14 16 18 20 22 24

Response of CPI to M2

.005

2

4

6

8 10 12 14 16 18 20 22 24

Response of HP to M2

.012 .011

.004

.010

.003

.009 .002

.008

.001

.007

.000

.006 2

4

6

8 10 12 14 16 18 20 22 24

2

4

6

8 10 12 14 16 18 20 22 24

Response of GDP to M2

-.002 -.003 -.004 -.005 -.006 -.007 2

4

6

8 10 12 14 16 18 20 22 24

Figure 12.3. The Response of CPI, FP, HP, SP, GDP, ER to Impulse of M2.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets Response to Cholesky One S.D. Innovations Response of CPI to M2

Response of CPI to ER

.006

.006

.004

.004

.002

.002

.000

.000

-.002

-.002

-.004

-.004 2

4

6

8 10 12 14 16 18 20 22 24

2

4

6

Response of CPI to FP

8 10 12 14 16 18 20 22 24

Response of CPI to SP

.006

.006

.004

.004

.002

.002

.000

.000

-.002

-.002

-.004 2

4

6

8 10 12 14 16 18 20 22 24

Response of CPI to CPI

.006

-.004 2

4

6

.006

.004

.004

.002

.002

.000

.000

-.002

-.002

-.004

8 10 12 14 16 18 20 22 24

Response of CPI to HP

-.004 2

4

6

8 10 12 14 16 18 20 22 24

2

4

6

8 10 12 14 16 18 20 22 24

Response of CPI to GDP

.006 .004 .002 .000 -.002 -.004 2

4

6

8 10 12 14 16 18 20 22 24

Figure 12.4. The Response of CPI to Impulses of Other Variables.

297

March 5, 2013

298

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

Table 12.5. Variance Decomposition of CPI (Unit %). Period 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24

S.E.

M2

ER

FP

SP

CPI

HP

GDP

0.773 0.835 0.884 0.916 0.945 0.973 0.999 1.024 1.049 1.074 1.0975 1.121 1.144 1.166 1.1880 1.2096 1.231 1.251 1.272 1.292 1.312 1.332 1.351 1.370

2.069 16.278 24.188 29.301 33.118 35.662 37.590 38.976 40.052 40.877 41.537 42.069 42.508 42.875 43.186 43.453 43.684 43.887 44.066 44.225 44.367 44.496 44.611 44.717

0.680 9.921 16.350 18.950 20.152 20.619 20.794 20.843 20.840 20.820 20.794 20.770 20.747 20.727 20.709 20.694 20.680 20.669 20.658 20.649 20.640 20.633 20.626 20.620

3.256 5.457 7.835 8.214 8.459 8.527 8.575 8.599 8.619 8.632 8.643 8.652 8.660 8.666 8.671 8.676 8.680 8.683 8.686 8.689 8.691 8.693 8.695 8.697

0.027 0.932 0.728 0.697 0.993 1.210 1.441 1.650 1.763 1.880 1.975 2.052 2.118 2.172 2.218 2.258 2.292 2.322 2.349 2.372 2.394 2.413 2.430 2.446

92.412 64.484 48.553 39.986 34.397 30.904 28.440 26.706 25.403 24.412 23.628 22.998 22.480 22.048 21.681 21.368 21.095 20.857 20.646 20.460 20.292 20.142 20.005 19.881

0.000 2.195 1.788 2.129 1.984 2.019 1.968 1.964 1.943 1.935 1.925 1.919 1.912 1.908 1.903 1.900 1.897 1.894 1.892 1.890 1.888 1.886 1.884 1.883

1.556 0.732 0.557 0.724 0.898 1.059 1.192 1.296 1.379 1.444 1.497 1.540 1.576 1.606 1.631 1.653 1.672 1.688 1.703 1.716 1.727 1.738 1.747 1.756

model. Since exchange rate acts as a demand factor, not a supply factor, the implication is that the exchange rate has more power to influence the capital account than the current account. This appreciation will induce more money flow in China and will push up the domestic price level. We can also obtain variance decomposition for the price level (CPI) as shown in Table 12.5. From the variance decomposition of CPI, it can be seen that besides its own persistent impact on itself, the significant part of the innovations of inflation over time could be attributed to money supply (M2), and then the exchange rate and the commodity price, representing the expected inflation.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

299

The stock price, housing price and GDP have barely any significant impact on inflation over time. Asset prices (especially the housing price and stock price) account for a small fraction of changes in inflation, which is reasonable in the sense that the wealth effect of the asset in China is weak compared to the US or other developed countries. The reason for the weak wealth effect in China is possibly that the growth rate of consumption price greatly exceeds the growth rate of salary and that there is no reliable medical and pension system. Combining the impulse response and variance decomposition above, we can draw conclusions on the two major research questions in this study. First, the disproportional relationship between the money supply (M2) and inflation (CPI) does not mean that Friedman’s famous rule is wrong. Instead, the money supply can still explain most changes in inflation in the long run. The excessive liquidity might be absorbed by the asset markets to restore the money market equilibrium in the short run. Second, the correlation between inflation and asset prices is not static but dynamic. In the short run, the asset prices might divert inflation pressure for a while, while in the long run, the rising asset prices will accelerate and magnify the progress of inflation. For the monetary authority to better handle inflation, asset prices should be an important reference taken under greater consideration.

12.6. Conclusion In this chapter, we modify the classical Cagan money demand function and build a VECM model to investigate the interactive relationship amongst a series of variables, including the money supply, CPI, housing price, stock price, commodity price, GDP and exchange rate in China. Put altogether, this chapter answers two questions. First, how to explain the disproportional change in inflation to the change in money supply in China by reshaping the role of asset markets and second, what are the dynamic interactions amongst good markets and asset markets for the inflation transmission. We find a significant long-run co-integration relationship amongst the variables, indicating that the fundamental factors, asset prices and inflation will follow the long-run equilibrium and are closely linked to each other in the long run. Based on the co-integration relationship and the analysis of the impulse response and variance decomposition, we conclude that the money

March 5, 2013

300

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Regional Development and Economic Growth in China

supply is still the major factor affecting inflation in the long run, which is in line with the monetarist viewpoint that inflation is a monetary phenomenon in the long term. At the same time, asset markets play an important role in absorbing the excessive liquidity, and help to maintain the money market equilibrium. This role of asset markets explains why inflation can remain relatively stable whilst the money supply displays high volatility. The relationship between asset prices and inflation is further investigated through empirical analysis. From the impulse response function and variance decomposition, we find that first, CPI responds positively to all asset prices (the housing price, stock price, and commodity price). Second, the money supply, exchange rate and commodity price have a larger influence on inflation over time. To better predict future inflation and to design an appropriate monetary policy, the inflation dynamics between the asset market and goods market should be taken into consideration. At present, the focus of China’s inflation fight is still on the CPI. Our study shows that asset prices do have important implication for inflation management. To reduce inflation, the monetary authority should not put asset price aside, and both asset prices and good prices should be consistently targeted and monitored.

References Angeloni, I., “The Output Composition Puzzle: A Difference in the Monetary Transmission Mechanism in the Euro Area and United States,” Journal of Money, Credit and Banking, 35(6): 1265–1306 (2003). Bernanke, B. S. and Gertler, M., “Monetary Policy and Asset Price Volatility,” NBER Working Paper, No. 7559 (2000). Bernanke, B. S. and Mihov, I., “Measuring Monetary Policy,” Quarterly Journal of Economics, 113(3): 869–902 (1998). Chari, V. V., Christiano, L. J. and Eichenbaum, M., “Inside Money, Outside Money and Short-Term Interest Rates,” Journal of Money, Credit and Banking, 27(4): 1354–1386 (1995). Cyrille, S., “On the Liquidity Effect of Monetary Policy in the CEMAC Countries: An Empirical Investigation,” International Journal of Economics and Finance, 2(3): 208–221 (2010).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

Inflation Transmission in China’s Goods and Asset Markets

301

Eichenbaum, M., “Comments:Interpreting the Macroeconomic Time Series Facts: The Effects of Monetary Policy by Christopher Sims,” European Economic Review, 36(5): 1001–1011 (1992). Gao, Q. H., “Asset Price and Inflation Rate in China: Empirical Research Based on VAR Model,” Wireless Communications, Networking and Mobile Computing, 5th International Conference (2009). Goodhart, C. and Hofmann, B., “Do Asset Prices Help to Predict Consumer Price Inflation,” The Manchester School, 68: 122–140 (2000). Gordon, D. B. and Leeper, E. M., “The Dynamic Impacts of Monetary Policy: An Exercise in Tentative Identification,” Journal of Political Economy, 102(6): 1228–1247 (1994). Kontonikas, A. and Ioannidis, C., “Should Monetary Policy Respond to Asset Price Misalignments,” Economic Modeling, 22: 1105–1121 (2005). Li, L., Narayan, P. K. and Zheng, X., “An Analysis of Inflation and Stock Returns for UK,” Journal of International Financial Market, Institutions & Money, 20(5): 519–532 (2010). Giordani, P., “An Alternative Explanation of the Price Puzzle,” Journal of Monetary Economics, 51: 1271–1296 (2004). Seymur, A., “Exchange Rate, Wage, Money: What Explains Inflation in CIS Countries: Pannel Causality and Panel Fixed Effects Analysis,” Middle Eastern Finance And Economics, 9: 6–13 (2011). Sims, C. A., “Interpreting the Macroeconomic Time Series Facts: The Effects of Monetary Policy,” European Economic Review, 36(5): 975–1000 (1992). Wang, Z. and Bessler, D. A., “A Monte Carlo Study on the Selection of Cointegrating Rank Using Information Criteria,” Econometric Theory, 21: 593–620 (2005).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch12

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Chapter 13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents Fei Yu

13.1. Introduction Stronger intellectual property rights (IPR) protection is usually acknowledged beneficial for a country’s technology progress (Briggs, 2008; Dinopoulos and Kottaridi, 2008; Grossman and Lai, 2004; Li, 2008; Sequeira, 1998; Tvedt, 2010). However, what if the IPR protection is carried out unequally towards domestic and foreign patents? More specifically, do some national patent offices’ impose strong IPR protection to domestic patents but weak ones to foreign patents? If it is true, then why and how does a patent office treat native and foreign patent applicants differently? Many scholars have recognized the malfunction of patent offices. It can be caused by the congestion of mass applications, inconsistent standards among different patent examiners, incapability of examiners in updating their science and background knowledge over time, or budgetary constraints of patent offices (Abbott, 2004; Burke and Reitzig, 2007; Caillaud and Duchêne, 2011; Mack, 2006; Thomas, 2002). However, these failures, due to imperfections of examiners and institutions, should be distinguished from deliberate discrimination against non-citizen patent applications. If foreign patent applications receive such deliberate non-national treatment, which is used as a strategy to protect domestic innovations, we identify it as a kind of strategic patent policy. This chapter exploits the possible strategic patent policy or inconsistency in the assessment of domestic and foreign patents in the Chinese patent office by the analysis of patent data. The content is organized as follows. We first review the literature in Section 13.2, then specify the econometric models in Section 13.3

303

March 5, 2013

304

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

and analyze the characteristics of the patent data in Section 13.4. The regression results are discussed in Section 13.5. Conclusions are presented in Section 13.6.

13.2. Literature Review Strategic patent policy is isomorphic to the well-documented strategic trade policy (Spencer and Bredner, 2008). Both of the two policies aim to raise the level of domestic welfare in a given country by shifting profits from foreign to domestic firms. Strategic trade policies are aimed to foster domestic firms by export subsidies, import tariffs and subsidies to R&D or investment. Similarly, strategic patent policies can be achieved by prolonging foreign patent application examinations or even rejecting those applications to encourage domestic inventors to catch-up or leap-frog. Two examples that are discussed below, the Japaneses and Brazilian patent offices, can partially reveal how the strategic patent policies are practiced in the real world. Japan leads the world in the volume of patent applications filed each year, but companies and inventors have gotten into the habit of seeking patents merely to assert their ownership rather than to exploit it. Watts (2000) argues that Japanese patents are viewed defensively as a way to prevent ideas from being used by competitors, whereas in other countries, they are increasingly seen in more aggressive terms as commodities that can be sold or from which royalties can be earned. Wolfson (1993) labels the US patent laws as fostering technological development via protection of inventors whereas hallmarks the Japanese ones as fostering technological development via dissemination of technology. He compares the Japanese patent laws with the US ones and criticizes the Japanese patent system for its fostering of patent flooding and unique provision of pre-grant oppositions. Patent flooding occurs when a Japan-based company files a flood of patents with minor changes around the core technology of a patent held by a foreign company. After patent flooding, Japanese companies typically try to force the original patent applicant or patent owner to cross-license the technology.1 Wolfson argues 1 Cross-licensing occurs when two or more competitors exchange use of each others’patents, often at minimal or no cost. The exchange results in the mutual non-enforcement of these patents against those competitors.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

305

that compulsory, or forced, cross-licensing may be unfair to the foreign corporation because, as a result of patent flooding, the Japanese company can gain valuable technology while giving up patents that consist of only minor changes to the foreign patent’s technology. Thus, while the foreign company allows use of its invention and technology, the Japanese company only allows use of minor changes to the same foreign invention and technology. These settlements are often shrouded in secrecy, a practice that has restricted patent flooding publicity. Wolfson (1993) also points out an alternative strategy that Japanese companies employ is to oppose foreign companies’ applications in order to delay issuance. Different from other patent systems in the world, Japanese patent laws allow re-grant opposition, which means that a patent application can be opposed before it is granted rather than after. Consequently, the Japanese patent office must resolve the pre-grant oppositions before deciding whether to issue a patent. Until the oppositions are resolved, the Japanese company may use the newly revealed technology without fear of serious reprisal and continue to file floods of patents. Worse yet, the Japanese patent office protects an invention for 15 years from the date the application is published for the purposes of pre-grant opposition, but for no more than 20 years from the filing date of the application.2 Therefore, even if patent issuance is delayed for more than five years due to amendments, pre-grant oppositions, or backlog at the Japanese patent office, the patent will still expire with the fixed 20-year period from the date of filing. Consequently, any delay in patent issuance reduces the effective life and value of the patent. Judicial remedies for patent infringement are also difficult to enforce. Dinwiddie (1995) shows that the patent enforcement system in Japan provides limited judicial remedies. The full value of monetary damages is extremely difficult to prove, and the possibility for equitable recovery of damages in excess of those proved does not exist. From January 1, 1996 the pre-grant opposition was replaced by a sixmonth post-grant opposition. However, another provision, kind-code B1 2 The European and Chinese patent offices’ grant patents for a term which begins with the

date of filing too. However, the period of patent protection in US patent system starts with the date of the grant.

March 5, 2013

306

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

patent, has been introduced since then so that Japanese companies can still seek for favorable treatment. Normal patent applications are granted after early publication. In contrast, kind-code B2 patents are granted without early publication. O’Keeffe (2000) argues that two effects are generated from kind-code B1 patents. First, these patents are hard for surveillance in English because they are only partly covered in patent databases such as Derwent World Patents Index, Patent Abstracts of Japan and International Patent Documentation Center. They can be found in Japanese databases such as Industrial Property Digital Library, CyberPatent Desk and GreenNet, but only in Japanese. Second, B1 patents are granted almost exclusively to Japanese applicants and are usually granted very quickly. Given that those patents are concentrated in electronics and telecommunications, the kindcode B1 patent stipulation is highly suspicious as a deliberate design for manipulating unfair technology protection. The Brazilian patent office is also under suspicion for adopting strategic patent policies. Brazil shares some characteristics with other developing countries, such as important activities in applying patents by foreign-owned firms and low firm involvement in R&D activities (Albuquerque, 2000). Hence, large amounts of backlog for patent applications can be applied as a strategy to counter foreign patent protection requirements. In fact, the top official in Brazil’s patent office confirmed the complaints of multinational pharmaceutical firms that his office is extremely slow in processing patent requests on drugs (Josephber et al., 2003). Many US and other multinational firms believe that this situation means Brazil does not, in fact, protect patents on pharmaceutical products. The Pharmaceutical Research and Manufacturers of America, a US pharmaceutical trade association, charges that Brazil’s patent law is simply not working, and as a result international patents are not respected. Besides the backlog strategy in the cases of Japan and Brazil, patent offices’ can also manipulate the probability of granting to discriminate against foreign patent applications. Kotabe (1992) compared the practices of national patent offices in Germany, Japan, the UK, and the US. He observed that the Japanese patent office takes more time to examine foreign applicants than domestic applicants, while the probability of a foreign patent being granted by American, British and German offices is lower than that of a domestic application.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

307

Guellec and van Pottelsberghe (2002) explore the determinants of the probability for a patent application to result in a grant and find that technological diversity, cross-border ownership of inventions, domestic and international research co-operation, the number of applicants, the combination of designated states for protection, the patenting procedure, the technological category and the geographical origin of an invention are all characteristics that significantly affect the probability of a patent application to be granted. They also find that in Japan, inventions with a Japanese inventor and a nonJapanese patentee have a lower probability to be granted. Webster et al. (2007) found that foreign patent applications in Japanese patent office are less likely to be granted than Japanese ones, meanwhile US patent applications in European patent office also have a relatively low grant rate. In a related study, Palangkaraya et al. (2006) argued that both Japanese and European offices give preferential treatments to local inventors and that such treatments are more severe in industrial sectors where the native technologies are relatively advanced, especially in Japan. The literature clearly shows that foreign patent applicants tend to receive non-national treatments in certain countries. What remains unclear is the mechanism underlying such treatments. There are institutional reasons. For example, foreign applicants face many disadvantages such as cultural and language barriers and lags of adjustment to foreign patent system changes. There are also barriers due to the existing differences in patent systems, which may also increase foreign patentees’ information cost. However, foreign applications may be deliberately delayed or rejected if patent examination is used as a kind of protection for domestic inventors. This is similar to strategic trade behavior and is thus distinguished from institutional barriers (Palangkaraya et al., 2006; Linck and McGarry, 1993). Several findings support the view that certain discriminative strategies can actually foster the catch-up of domestic companies, especially in the industries where MNCs dominate the market or format industrial standards (Li, 2007; Wang et al., 2007). Up to date, the existing literature only detects the non-national treatment that foreign applicants face in national patent offices. However, little has been done to detect whether such non-national treatment is caused by institutional factors or strategic behaviors. To make a contribution, this chapter offers an insight of this issue by checking the patent data in China’s State Intellectual Property Office.

March 5, 2013

308

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

13.3. Econometric Modeling In this section, two econometric models are proposed. The first model checks if foreign patent applications face a lower grant probability than domestic ones, other conditions being equal. Denote gi as a binary variable where gi = 1 stands for the success of grant for patent i and gi = 0 for the failure of grant. The probability of grant Pi (gi = 1|X) is conditioned on X, the vector of explanatory variables. To ensure that Pi (gi = 1|X) only takes a value between zero and one, while allowing the explanatory variables X to take any values, we assume the logistic function for Pi (gi = 1|X): Pi (gi = 1|X) =

1


1 + e−Xiβ

,

(13.1)

where vector β stands for the coefficients to be estimated. The second model focuses on whether foreign patent applications face a longer grant lag than domestic ones. The grant lag (Li ) for patent i can be featured using a count model. By definition, the lag can be calculated as the time period between a patent’s grant date and filing date. Since Li is a count number and always positive, it can be modeled as follows:


Li = eXi θ +εi ,

(13.2)

where θ represents the coefficient vector and ε is the error term. The vector X includes four explanatory variables. Some studies show that important innovations would be approved quickly (Harhoff and Wagner, 2009; Regibeau and Rockett, 2010). We measure the importance of a patent with the number of patents (N) in a patent family (Lanjouw et al., 1998; Burke and Reitzig, 2007). A patent family refers to the same invention patented in more than one country.3 A patent applicant only lodges applications in countries where the invention can potentially earn a profit, because each patent application incurs a registration fee and maintenance costs. Thus a larger N may imply a larger market value for that invention. Patent applications from different sectors may also be treated with different speed (Regibeau and Rockett, 2010). Each patent application in a patent office is assigned with one or more International Patent Classification 3 The web site (http://www.uspto.gov/main/glossary/#patentfamily) of United States Patent

and Trademark Office.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

309

(IPC) codes. However, the IPC codes are technological categories rather than industrial categories (e.g., NACE) used in most economic indicators.4 Hence, we converted each IPC code to a NACE code by following Schmoch et al. (2003).5 The NACE codes cover four groups of industries, namely, high-tech, medium-high-tech, medium-low-tech and low-tech industries.6 Therefore, we construct the industry indicator (Ii ) by assigning integrals 1–4 to the four industrial categories (from high-tech to low-tech). Note that a patent may have more than one integral if it is assigned with multiple IPC codes. In such a case, we calculate Ii as the average value of the integrals. As a result, Ii also takes frictional values. If strategic patent policies are employed, then they are possibly connected with the country’s innovative capability in certain technological fields. The innovative capability can be measured by the native application ratio. We define the native application ratio Ri for patent i as the ratio of the total number of native patent applications over the total number of native and foreign applications in a certain patent field. If a patent is categorized into multiple codes, then only the main (first) code is considered. Patents in different technological fields have different Ri values. A higher R indicates that the country has accumulated a relatively larger stock of knowledge in that technological field than in other fields. Finally, we include dummy variables Cij which distinguish patent applications from seven regions, namely, China (CN), Japan (JP), South Korea (KR), the United States (US), member countries of the European patent office (EP), Canada–Australia–New Zealand (CA) and the rest of the world (RW ). Hence, we can elaborate vector X into Equation (13.3): Xi
β = β0 + βN Ni + βI Ii + βR Ri +

6


βC Cij .

(13.3)

j=1

4 NACE stands for Nomenclature Générale des Activités Économiques dans I‘Union

Européenne and is used in European Union countries. ISIC is the acronym for International Standard Industrial Classification which is developed by the United Nations. 5 Currently two versions of NACE are available, NACE Rev 1.1 and NACE Rev 2. NACE Rev 1.1 is used in this paper. 6 The four industrial levels are defined by the statistical office of the European Union (Eurostat). See Eurostat website (http://epp.eurostat.ec.europa.eu/cache/ITY_SDDS/ Annexes/htec_esms_an2.pdf).

March 5, 2013

310

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

A test of the estimated coefficients of Cij will reveal whether domestic and foreign patent applications receive the same treatments. However, they are not sufficient to distinguish whether non-national treatments are caused by institutional barriers or strategic patent policies. Therefore, we introduce the interaction terms in our extended models. The estimated coefficients of the interaction terms will give a clue of strategic patent policies. Hence, Equation (13.3) is extended into Equation (13.4). Xi
β = β0 + βN Ni + βI Ii + βR Ri +

6


(βj + βNj N + βIj I + βRj R) × Cj .

j=1

(13.4)

13.4. Data We explore the European Patent Organization (EPO) Worldwide Patent Statistical Database (also known as PATSTAT) which covers 70 million patent records from over 80 countries. Chinese patents are distinguished as patents for invention, patents for utility model and patents for design.7 In this study, we only cover applications of invention patent. Figure 13.1 shows that approximately one third of patent applications fall into the category of

Number of patent applications (Thousands)

300 250 200 150

Invention patents (recorded in the PATSTAT database) Utility models (PATSTAT) Designes (PATSTAT) Invention patents (reported by the Chinese patent office) Utility models (Chinese patent office) Designes (Chinese patent office)

100 50 0 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

Figure 13.1.

PATSTAT Coverage of the Chinese Patent Data.

7 See the web site (http://www.epo.org/patents/patent-information/east-asian/helpdesk/ china/faq.html#what) of the European Patent Office.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

311

140

Number of patent applications (Thousands)

120

China

Japan

South Korea

United States

Other OECDs

The rest countries

100 80 60 40 20 0 1985

1987

1989

1991

Figure 13.2.

1993

1995

1997

1999

2001

2003

2005

2007

Number of Patents by Source of Origin.

invention patents, and the coverage of the PATSTAT database is extensive in invention patents and utility models but very poor in designs. Figure 13.2 shows that most of the foreign patents (we use “patent” to mean “invention patent” hereafter) are from Japan, the United States, South Korea and other OECD countries.8 In the regression, the group of Chinese patents is taken as the reference group of the country dummy variable (Cij ). Figure 13.3 shows that the application times (N) of the patents lodged in the Chinese patent office vary across country origins. For example, only 2% of the applications from China are also lodged in other countries and most of them are only lodged in one foreign country. The majority of the applications from Japan and South Korea are also lodged in two to four other non-native countries. That implies that China is an important destination for the Japanese and Korean patent applications. However, it is not that important for patents from other origins, demonstrated by the peaks of the distributions being at five foreign destinations. Figure 13.4 illustrates the comparison between patent applications being granted and rejected. There are two turning points at 1994 and 2004. Before 1994, both the number of granted patents and rejected ones are 8 For simplicity, here we use OECD countries rather than two categories, the member coun-

tries in the European patent office and Canada, Australia, and New Zealand.

March 5, 2013

312

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

China Japan South Korea United States Other OECDs Rest of the world

60

Thousands

50 40 30 20 10 0 1

Figure 13.3.

2

3

4

5

6

7

8

9

10

Number of Patents in Patent Families (N), by Origins.

(Thousands)

Note: (1) 98% of patents lodged by Chinese citizens with N = 0 (not shown); (2) Patents with N ranges from 11 to 50 are not shown (decreasing trend). 200 180

Rejected

160

Granted

Number of patent applications

140 120 100 80 60 40 20 0 1985 1987 1989 1991 1993 1995 1997 1999 2001 2003 2005 2007

Figure 13.4.

Patents Granted and Rejected in the Chinese Patent Office.

small; however, the number of granted patents starts to grow after 1994 and exceeds that of the rejected patents. This phenomenon may reflect the effect of the first amendment to Chinese patent law in 1992.9 The 1992 amendment expanded patentable subject matter under Chinese patent law 9 The 1992 amendment was made in accordance with the “Memorandum of Understanding

between the Government of the United States and the Government of the People’s Republic of China on the Protection of Intellectual Property”.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

313

RW OECD US KR JP CN 1985

1990

Figure 13.5.

1995

2000

2005

Granted Patents, by Origins.

to include chemical inventions, including pharmaceuticals (Yang and Yen, 2009). Figure 13.5 decomposes the granted patents by the source of origins, and it reveals that the increase in patent numbers after 1994 is largely due to the growth of foreign patents. After 2004 another turning point appeared where the number of rejected applications surpassed the number of granted patents. This may have been caused by two factors. First there is a longer examination delay for granted patents than rejected ones. Hence the trend for granted patents is probably increasing after 2004 instead of decreasing, once all the applications are fully published. The other possible reason is the second amendment to the Chinese patent law in 2000. The 2000 amendment was made in anticipation of China’s accession to the World Trade Organization (WTO) in 2001. Various provisions of the 2000 amendment facilitated the patent application process for foreign entities in China (Yang and Yen, 2009). Despite experiencing a huge growth, the chance of foreign applications being approved has dropped (see Figure 13.6). Therefore, the sharp increase of rejected applications may possibly indicate the relentlessness of the Chinese patent office to grant patents to foreign applications which crowd into China after 2000. Policy and regime changes may initiate structural breaks of the data. For instance, China’s entry in the WTO may have had profound effects on its patent policy. The Agreement on Trade Related Aspects of Intellectual Property Rights (TRIPS) was established in 1994, which explicitly prohibits any discrimination against foreign patent applications in WTO member

March 5, 2013

314

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China 100% 90% 80% 70% 60% 50% 40% 30% 20% 10% 0%

RW OECD US KR JP CN 1985

1990

Figure 13.6.

1995

2000

2005

Rejected Patent Applications, by Origins.

countries.10 Thus if China employed strategic patent policies, those policies should be amended or dropped in accordance with the requirement by the WTO. If such changes are ignored, pooled regressions possibly introduce biased and inconsistent estimations. Thus we constrain the publication date of the patents to 1994–2007 and introduce a dummy variable T90 to control the effect of China’s joining the WTO. T90 is set to one if a patent is published during 1994–2000 and to zero otherwise. After introducing the dummy variable T90 , Equation (13.4) is further extended into Equation (13.5). Xi
β

= β0 + βN Ni + βI Ii + βR Ri + 

6


(βj + βNj N + βIj I + βRj R) × Cj

j=1

+ T90 × 1 + βNT Ni + βIT Ii + βRT Ri

+

6


 (βjT + βNjT N + βijT I + βRjT R) × Cj .

(13.5)

j=1

As for the examination delay or grant lag (L), China applies a twostage examination system for patent applications, namely, preliminary and 10Article 27.1 requires that “patents shall be available and patent rights enjoyable without

discrimination as to the place of invention, the field of technology and whether products are imported or locally produced”.

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

315

substantial examination. The reason of two-stage examinations is that a lot of patent applications are not targeted for the grant of patents, but are used as a competition strategy against business rivals. Patent applicants can disclose innovation information to create prior art that might stop rivals from patenting or to harden the patent race for rivals (Baker and Mezzetti, 2005). According to Article 34 of China’s Patent Law, after receiving an application for an invention patent, once the patent administration department finds that the application conforms to the requirements of the patent law upon preliminary examination, it shall publish the application promptly after the expiration of 18 months from the date of filing.11 A patent applicant can request for earlier-publication to avoid the 18-month waiting. If so, the patent office shall publish the application immediately after a preliminary examination of it, unless it is to be rejected.12 These publications and earlier publication arrangements removes the need for defensive patent applications after the preliminary examination stage, and filters out the granting-aimed applications to the substantial examination stage. The side effect of this regime is that inventors who want to patent their innovation may face the risk that their knowledge is made public even though their patent applications are eventually rejected in the substantial examination stage. Figures 13.7 and 13.8 illustrate the examination duration of patent application in the Chinese patent office. The examination duration for rejected applications is mostly within two years, as shown by Figure 13.7, where a large portion of rejected applications is for defensive patenting. However, for granted patents, most are examined for more than four years (Figure 13.8). Hence, if the patent examiners in the Chinese patent office deliberately prolong foreign patent applications, such behaviors are more likely observed in the group of granted patents. This kind of strategy blocks the foreign corporations’ technological pressure over native companies and meanwhile allows extra time for technology spillovers to native companies. An extreme example happened in Japan. Texas Instruments, which received 11 See the website (http://www.sipo.gov.cn/sipo_English/FAQ/200904/t20090408_449696.

html) of the State Intellectual Property Office of China. 12 See Rule 46, Implementing Regulations of The Patent Law of The People’s Republic of

China (Promulgated by Order No. 306 of the State Council of the People’s Republic of China on June 15, 2001, amended by the Decision of the State Council Revising the Implementing Regulations of the Patent Law of the People’s Republic of China of December 28, 2002).

March 5, 2013

316

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China 100% 90%

>4 years

80% 4 years

70% 60%

3 years

50% 40%

2 years

30% 20%

1 year

10% 0% 1988 1990 1992 1994 1996 1998 2000 2002 2004 2006

Figure 13.7.

Rejected Patent Applications, by Grant Lag (L).

100% 90%

>6 years

80%

6 years

70%

5 years

60% 50%

4 years

40%

3 years

30%

2 years

20% 1 year

10% 0% 1993

1995

1997

Figure 13.8.

1999

2001

2003

2005

2007

Granted Patents, by Grant Lag (L).

an American patent on semiconductors in 1964, first applied for the patent in Japan in 1960. However, that patent was not issued until 1989 (Zipser, 1989).

13.5. Regression Results and Discussions The regression results are reported in Table 13.1. We used both rejected and granted patents in the Logit regression, and there were 823,915 observations in total. However, our regression in the count model was restricted to granted patents only (made up of 334,136 observations). Two types of

March 5, 2013

Table 13.1. Regression Results.

0.073(0.004)∗∗∗

−0.008(0.001)∗∗∗

R US

−0.320(0.018)∗∗∗ −0.396(0.038)∗∗∗

−0.190(0.005)∗∗∗ 0.272(0.009)∗∗∗

CA

−0.605(0.073)∗∗∗

0.293(0.017)∗∗∗

EP

−0.255(0.039)∗∗∗

0.262(0.009)∗∗∗

KR JP

0.241(0.041)∗∗∗

0.066(0.010)∗∗∗

0.499(0.038)∗∗∗

0.185(0.009)∗∗∗

RW US ∗ N

−0.342(0.057)∗∗∗ 0.093(0.012)∗∗∗

0.310(0.014)∗∗∗ −0.023(0.003)∗∗∗

CA∗ N EP∗ N

0.180(0.016)∗∗∗ 0.126(0.012)∗∗∗

−0.024(0.003)∗∗∗ −0.031(0.003)∗∗∗

KR∗ N

0.154(0.013)∗∗∗

−0.018(0.003)∗∗∗

JP∗ N

0.157(0.012)∗∗∗

−0.010(0.003)∗∗∗

RW ∗ N

0.108(0.013)∗∗∗

−0.033(0.003)∗∗∗

I

0.969(0.176)∗∗∗

0.207(0.032)∗∗∗

T90∗ N T90∗ I

0.507(0.059)∗∗∗ −0.198(0.011)∗∗∗

−0.040(0.011)∗∗∗ −0.007(0.003)∗∗

T90∗ R T90∗ US

0.171(0.044)∗∗∗ −1.527(0.179)∗∗∗

0.183(0.013)∗∗∗ −0.140(0.033)∗∗∗

T90∗ CA T90∗ EP

−1.472(0.235)∗∗∗ −1.339(0.179)∗∗∗

−0.089(0.050)∗ −0.172(0.033)∗∗∗

T90∗ KR

−1.909(0.190)∗∗∗

−0.011(0.037)

T90∗ JP T90∗ RW T90∗ US ∗ N T90∗ CA∗ N T90∗ EP∗ N T90∗ KR∗ N T90∗ JP∗ N T90∗ RW ∗ N

−1.421(0.179)∗∗∗

−0.079(0.033)∗∗

−1.779(0.221)∗∗∗ −0.493(0.059)∗∗∗

−0.223(0.048)∗∗∗ 0.033(0.011)∗∗∗

−0.527(0.062)∗∗∗ −0.543(0.059)∗∗∗

0.033(0.011)∗∗∗ 0.039(0.011)∗∗∗

−0.529(0.062)∗∗∗

0.021(0.012)∗

−0.493(0.059)∗∗∗

0.023(0.011)∗∗ 0.044(0.011)∗∗∗

T90

−0.504(0.060)∗∗∗

317

(Continued)

Regional Development and Economic Growth. . . b1491-ch13

7.195(0.008)∗∗∗ 0.039(0.003)∗∗∗

Count Model

9in x 6in

−0.550(0.036)∗∗∗ 0.001(0.012)

Intercept N

Logit Model

11:55

Count Model

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

Logit Model

−0.024(0.003)∗∗∗

CA∗ I EP∗ I

0.214(0.042)∗∗∗

−0.061(0.010)∗∗∗

0.198(0.010)∗∗∗

−0.036(0.002)∗∗∗

KR∗ I JP∗ I

0.297(0.020)∗∗∗

−0.040(0.005)∗∗∗

RW ∗ I US ∗ R

0.113(0.010)∗∗∗ 0.281(0.035)∗∗∗

−0.014(0.002)∗∗∗ −0.042(0.008)∗∗∗

−1.064(0.046)∗∗∗

0.210(0.012)∗∗∗

CA∗ R EP∗ R

−0.540(0.192)∗∗∗ −0.653(0.046)∗∗∗

0.295(0.042)∗∗∗ 0.208(0.011)∗∗∗

KR∗ R JP∗ R

−0.629(0.078)∗∗∗ −0.422(0.042)∗∗∗

0.234(0.009)∗∗∗

RW ∗ R

−0.444(0.155)∗∗∗

0.213(0.037)∗∗∗

−0.018(0.019)

T90∗ US ∗ I

0.184(0.027)∗∗∗

0.035(0.007)∗∗∗

0.527(0.114)∗∗∗

0.010(0.027)

T90∗ EP∗ I T90∗ KR∗ I T90∗ JP∗ I T90∗ RW ∗ I T90∗ US ∗ R T90∗ CA∗ R T90∗ EP∗ R T90∗ KR∗ R T90∗ JP∗ R T90∗ RW ∗ R

0.345(0.027)∗∗∗ −0.315(0.077)∗∗∗

0.020(0.007)∗∗∗ 0.041(0.021)∗

0.413(0.03)∗∗∗ 0.250(0.101)∗∗

0.004(0.007) 0.050(0.027)∗∗∗

T90∗ CA∗ I

0.763(0.124)∗∗∗

−0.099(0.032)∗∗∗

−0.542(0.459) 0.489(0.115)∗∗∗

−0.116(0.109) −0.091(0.029)∗∗∗

−0.188(0.288)

0.001(0.079)

0.281(0.125)∗∗

−0.101(0.030)∗∗∗

0.452(0.436)

−0.048(0.114)

Note: Standard errors in parentheses. ∗ p < 0.1, ∗∗ p < 0.05, ∗∗∗ p < 0.01. N, I and R are centered to 3, 2 and 0.5 (close to their means), respectively.

Regional Development and Economic Growth. . . b1491-ch13

0.238(0.010)∗∗∗

Count Model

9in x 6in

US ∗ I

Logit Model

11:55

Count Model

Regional Development and Economic Growth in China

Logit Model

March 5, 2013

318

Table 13.1. (Continued)

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

319

distributions are usually applied to count models, namely, the Poisson distribution and the negative binomial distribution (Hausman et al., 1984). The Poisson distribution has an implicit restriction, that is, the variance of the sample is equal to its mean. Since real count data are commonly observed to have inconsistency between its variance and mean, therefore, researchers routinely employ a more commonly used general specification, such as the negative binomial distribution, which allows for over-dispersion, where the variance is larger than the mean. The degree of dispersion can be measured by the dispersion value. If the dispersion value equals zero, the model is reduced to a Poisson model. The dispersion value in our count model is 0.104, which is greater than zero. It confirms that the dependent variable is over-dispersed. This means that the negative binomial distribution fits the count model better than the Poisson distribution. The interpretation of the regression results is not straightforward because the models in Equations (13.1) and (13.2) are nonlinear. Therefore, we transform the results into Figures 13.9 and 13.10. The two figures show that the China patent office practices strategic patent policies all the time. However, they also capture a substantial structural change between the period 1994–2000 and 2001–2007. The strategic policies are more apparent in terms of grant probability during 1994–2000. The foreign patent applicants face a substantially lower grant probability than the Chinese ones. Such treatment is more apparent in the more valuable (high N) and hightech (low I) patent fields as well as where China’s innovative capability is stronger (high R) (see Figures 13.9(b), 13.9(d), and 13.9(f )). By contrast, this strategy is hardly observable in the next period from 2001 to 2007. However, Figure 13.10 suggests that after 2001 Chinese patent applications were examined much more quickly than foreign ones, while such a gap in terms of the grant lag is not as apparent during 1994–2000. The above observation suggests that strategic policy changed from a lower probability of grant issue to a longer grant lag to foreigners. This change might be explained by China joining the WTO in 2001. As mentioned, the WTO contains the TRIPS clause, which explicitly requires its members to abandon any form of discrimination toward foreign patent applications. Thus, the strategic patent policy in China might be changed from a stronger and easily detected form of manipulation on grant probability to a less detectable form of extending grant lag to foreigners.

March 5, 2013

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China (a) P(N|I=2, R=0.5), 2001–2007 CN KR

0.8

US JP

CA RW

EP

0.6 0.4 0.2 N

0 1

3

4

5

6

7

8

0.6 0.4 0.2 I

1

2

3

0.4 0.2 R

0 0.6

0.8

N 1

1

2

3

4

5

6

7

8

9

10

(d) P(I|N=3, R=0.5), 1994–2000

0.8 0.6 0.4 0.2 I

0 1 1

0.6

0.4

0.2

4

0.8

0.2

0.4

0

(e) P(R|N=3, I=2), 2001–2007

0

0.6

1

0.8

1

0.8

10

(c) P(I|N=3, R=0.5), 2001–2007

0

Grant Probability

9

Grant Probability

Grant Probability

1

2

(b) P(N|I=2, R=0.5), 1994–2000

1 Grant Probability

1 Grant Probability

9in x 6in

Grant Probability

320

11:55

2

3

4

(f) P(R|N=3, I=2), 1994–2000

0.8 0.6 0.4 0.2 R

0 0

0.2

0.4

0.6

0.8

1

Figure 13.9. Transformed Results from the Logit Regression.

Another characteristic in the two figures also supports the hypothesis that strategic patent policy exists in China. Even though the grant probability for Chinese patent applicants is not substantially higher than that for foreigners in the 2000s, there is a remarkable pattern only observed among the Chinese applications. Figures 13.9(a), 13.9(c) and 13.9(e) show that the grant probability for the Chinese applications does not vary much in different groups of N, I, and R. If the foreign applications are treated indifferently with the domestic ones, then the same constancy should be observed. However, the figures show that foreign patent applications in the high valuable, low tech and China’s weak innovative patent fields tend to receive higher grant probability than those in the other groups. Finally, patents from different countries are also possibly treated differently. The figures show Korean and Japanese patents tend to receive a

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

US JP

CA RW

EP

1.5 N

1 Grant Lag (thousand days)

1

2

3

4

5

6

7

8

9

(c) L(I|N=3, R=0.5), 2001–2007

2.5

2

1.5 I

1 Grant Lag (thousand days)

1

2.5

2

3

2

1.5 R

1 0

0.2

0.4

0.6

2

1.5

0.8

1

N

1 1 2.5

2

3

4

5

6

7

8

9

10

(d) L(I|N=3, R=0.5), 1994–2000

2

1.5 I

1

4

(e) L(R|N=3, I=2), 2001–2007

321

(b) L(N|I=2, R=0.5), 1994–2000

2.5

10 Grant Lag (thousand days)

CN KR

2

Grant Lag (thousand days)

(a) L(N|I=2, R=0.5), 2001–2007

2.5

1 Grant Lag (thousand days)

Grant Lag (thousand days)

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

2.5

2

3

4

(f) L(R|N=3, I=2), 1994–2000

2

1.5 R

1 0

0.2

0.4

0.6

0.8

1

Figure 13.10. Transformed Results from the Count Regression.

higher grant probability or shorter grant lag than patents from other foreign countries. The reason is not yet clear. It can be related to the characteristics of the Korean and Japanese patents in China. As Figure 13.3 suggests, China is an important destination for Japanese and Korean patent applicants. Thus, these applicants may exert extraordinary efforts to obtain the patents quickly and successfully. It could also be related to the characteristics of the Korean and Japanese FDIs. For example, Park and Lee (2003) find that the Korean FDI firms in China aimed to use China as an export-processing base, whereas the American FDI firms tended to target local Chinese markets. Similarly, Greaney and Li (2009) find a much higher degree of export-orientation for the Japanese affiliates than the American affiliates in China, with the latter tending to make the vast majority of their sales in the Chinese market. Therefore, if the Chinese patent office employs

March 5, 2013

322

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

strategic protection to Chinese corporations, such protectionism might be less severe against the Korean and Japanese opponents because they are not the direct competitors of Chinese native firms in the local markets.

13.6. Conclusions This chapter examines the strategic patent policy of the Chinese patent office in protection of domestic innovators. A country’s patent system may affect its technology progress in many ways. Previous scholars have contributed a lot on topics such as IPR protection or imperfection of patent offices. However, the deliberate discriminative strategy employed by national patent office’s is still rarely tested with empirical evidence. Based on the information of a world patent database, we construct four variables that may reflect the discriminative strategy, namely, the importance of innovation, industry indicator, the native application ratio and a dummy variable that distinguishes the source regions of the patent applications. These variables are then used in two regression models to test two complementary questions, that is, whether Chinese patent applications are more likely or quicker to get approved, ceteris paribus. In the regressions, we separate the data into two periods, 1994–2000 and 2001–2007. The regression results suggest that during 1994–2000 the Chinese patent applications are more likely to get approved than the foreign ones. However, during 2001–2007 the Chinese patent applications are more quickly to get approved; meanwhile the gap on the grant probability is largely diminished. China’s joining in WTO might cause this policy shift, since WTO prohibits any discriminative practices in patent fields and the strategy by extending grant lag is less detectable than by decreasing grant probability to foreign patent applications.

References Abbott, A., “Pressured Staff Lose Faith in Patent Quality,” Nature, 429(6991): 493 (2004). Albuquerque, E. M., “Domestic Patents and Developing Countries: Arguments for their Study and Data from Brazil (1980–1995),” Research Policy, 29(9): 1047–1060 (2000).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

323

Baker, S. and Mezzetti, C., “Disclosure as a Strategy in the Patent Race,” Journal of Law and Economics, 48(1): 173–194 (2005). Briggs, K., “Three Essays on Intellectual Property Rights in Developing Countries,” Ph.D. thesis, The University of North Carolina, Chapel Hill (2008). Burke, P. and Reitzig, M., “Measuring Patent Assessment Quality-analyzing the Degree and Kind of (in) Consistency in Patent Offices’ Decision Making,” Research Policy, 36(9): 1404–1430 (2007). Caillaud, B. and Duchêne, A., “Patent Office in Innovation Policy: Nobody’s Perfect,” International Journal of Industrial Organization, 29(2): 242–252 (2011). Dinopoulos, E. and Kottaridi, C., “The Growth Effects of National Patent Policies,” Review of International Economics, 16(3): 499–515 (2008). Dinwiddie, S. K., “A Shifting Barrier? Difficulties Obtaining Patent Infringement Damages in Japan,” Washington Law Review, 70(3): 833–858 (1995). Greaney, T. M. and Li, Y., “Assessing Foreign Direct Investment Relationships Between China, Japan, and the United States,” Journal of Asian Economics, 20(6): 611–625 (2009). Grossman, G. M. and Lai, E. L. C., “International Protection of Intellectual Property,” American Economic Review, 94(5): 1635–1653 (2004). Guellec, D. and van Pottelsberghe, B., “The Value of Patents and Patenting Strategies: Countries and Technology Areas Patterns,” Economics of Innovation and New Technology, 11(2): 133–148 (2002). Harhoff, D. and Wagner, S., “The Duration of Patent Examination at the European Patent Office,” Management Science, 55(12): 1969–1984 (2009). Hausman, J., Hall, B. H. and Griliches, Z., “Econometric Models for Count Data with an Application to the Patents-R&D Relationship,” Econometrica, 52: 909–938 (1984). Kotabe, M., “A Comparative Study of U.S. and Japanese Patent Systems,” Journal of International Business Studies, 23(1): 147–168 (1992). Lanjouw, J. O., Pakes, A. and Putnam, J., “How to Count Patents and Value Intellectual Property: Uses of Patent Renewal and Application Data,” The Journal of Industrial Economics, 46(4): 405–433 (1998). Li, X., “The Impact of Higher Standards in Patent Protection for Pharmaceutical Industries Under the TRIPS Agreement: A Comparative Study of China and India,” Working Papers, RP2008/36, World Institute for Development Economic Research (UNU-WIDER) (2008).

March 5, 2013

324

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Regional Development and Economic Growth in China

Li,Y., “From Imitation to Innovation: The Role of Patent in China’s Biotechnology and Pharmaceutical Industries,” Ph.D. thesis, Stanford University (2007). Linck, N. J. and McGarry, J. E., “Patent Procurement and Enforcement in Japan — A Trade Barrier,” George Washington Journal of International Law and Economics, 27: 411–431 (1993). Mack, K., “Reforming Inequitable Conduct to Improve Patent Quality: Cleansing Unclean Hands,” Berkeley Technology Law Journal, 21(1): 147–175 (2006). O’Keeffe, M., “Japanese Submarine Patents: Examined Patents Within a Year of Filing!” World Patent Information, 22(4): 283–286 (2000). Palangkaraya, A., Jensen, P. and Webster, E., “Patent Examination Decisions and Strategic Trade Behavior,” Working Papers, 2006-03, Centre for International Economic Studies, University of Adelaide (2006). Park, B. and Lee, K., “Comparative Analysis of Foreign Direct Investment in China,” Journal of the Asia Pacific Economy, 8(1): 57–84 (2003). Regibeau, P. and Rockett, K., “Innovation Cycles and Learning at the Patent Office: Does the Early Patent Get the Delay?” The Journal of Industrial Economics, 58(2): 222–246 (2010). Schmoch, U., Laville, F., Patel, P. and Frietsch, R., “Linking Technology Areas to Industrial Sectors,” Final Report to the European Commission, DG Research (2003). Sequeira, K. P., “The Patent System and Technological Development in Late Industrializing Countries: The Case of the Spanish Pharmaceutical Industry,” Ph.D. thesis, University of Sussex (1998). Spencer, B. and Bredner, J. A., “Strategic Trade Policy,” in Durlauf, S. N. and Blume, L. E. (eds.), The New Palgrave Dictionary of Economics. Hampshire: Palgrave Macmillan (2008). Thomas, J., “The Responsibility of the Rulemaker: Comparative Approaches to Patent Administration Reform,” Berkeley Technology Law Journal, 17(2): 727–761 (2002). Tvedt, M., “One Worldwide Patent System: What’s in it for Developing Countries?” Third World Quarterly, 31(2): 277–293 (2010). Wang, Q., Wang, H. and Li, M., “Industrial Standard Based Competition and Chinese Firm Strategic Choices,” International Journal of Technology and Globalisation, 3(4): 422–436 (2007). Watts, J., “Seeking More Creativity, Japan Overhauls IP Laws,” Research Technology Management, 43(5): 4–5 (2000).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

Inconsistency in the Assessment of China’s Domestic and Foreign Patents

325

Webster, E., Palangkaraya, A. and Jensen, P. H., “Characteristics of International Patent Application Outcomes,” Economics Letters, 95(3): 362–368 (2007). Wolfson, J., “Patent Flooding in the Japanese Patent Office: Methods for Reducing Patent Flooding and Obtaining Effective Patent Protection,” The George Washington Journal of International Law and Economics, 27(2/3): 531–563 (1993). Yang, W. and Yen, A., “The Dragon Gets New IP Claws: The Latest Amendments to the Chinese Patent Law,” Intellectual Property & Technology Law Journal, 21(5): 18–27 (2009). Zipser,A., “Texas Instruments Gets Japanese Patent,Analysts See SizableAddition to Revenue,” Wall Street Journal (Eastern Edition), 11: 1 (1989).

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-ch13

This page intentionally left blank

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-index

Index

gross fixed, 117, 120–1 intensity, 200–1, 209, 215 Central and Western regions, 78, 132–8 Central China (CEC), 63, 66, 201 central regions, 5, 79–80 CES function, 53, 56 CGE (computable general equilibrium), 7, 49–50, 222 Model, 50–2, 66, 72 static, 50–2 Chongqing, 3–5, 19–20, 23, 57, 78–80, 101, 107, 133, 201 city districts, 19–21 city-only dataset, 16, 25, 43, 44 CO2 emissions, 10, 51, 256, 261, 263–4, 268–70, 273, 276–7 coastal, 2–3, 5, 8, 22, 25–7, 30, 32–3, 35–8, 41, 43–4, 97, 105–8, 203 China, 8, 97 region, 2–3, 5, 22, 25–7, 30, 32–3, 35–8, 41, 43–4, 105–8, 203 rank, 292 relationship, 10, 294, 299 co-integration, 10, 292, 294, 299 comparative advantage, 200, 250 concentration statistic, 173, 175 constant elasticity of substitution (CES), 52, 236, 248 constant elasticity of transformation (CET), 55, 56 Consumer Price Index, see CPI contemporaneous structure, 290 COP15, 48 county-level cities, 7, 19, 20, 22, 23, 25 county-level data, 6, 7, 15

ACFTA, 9, 221–2, 226–7, 232, 235–6, 245–8, 250 administrative regions, 20, 100, 108, 201 agglomeration, 189, 191, 193, 196–7, 199–201, 204–6, 211, 215 and export propensity, 204 effect of, 193–4, 196 level, 211 of exporters, 9, 193, 205, 211, 215 of exporting firms, 192–3 variables, 194, 196, 200–1 agricultural trade-related policies, 231, 235 APEC (Asian Pacific Economic Cooperation), 10, 258, 268, 280–1 economies, 254, 256–8, 263, 269, 273–4, 276, 283 member economies, 257 members, 264, 268–9, 276 Armington composite, 56, 68 ASEAN, 9, 221, 232, 235, 237, 245, 247 Member States (AMS), 9, 235 Australia New Zealand FTA (AANZFTA), 235, 237, 249 Asian Economic Crisis, 272 asset prices, 10, 287–8, 294–5, 299–300 augmented growth model, 8, 119 autocorrelated error term, 168 bad outputs, 254, 256, 259–61, 263, 268–70, 277 capital formation, 117, 119–20, 124 domestic, 120–1 domestic gross fixed, 120

327

March 5, 2013

328

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-index

Industrial Development in East Asia

county-level GRP data, 20 county-level units, 7, 15, 22 county-only dataset, 16, 25, 43, 44 CPI (Consumer Price Index), 10, 128, 285–92, 294–5, 298–300 creative class, 7, 75–83, 85–96 and regional growth, 94 density, 78–9, 82 distribution, 7, 78, 85, 87–8, 90, 92, 94 in China, 77 regional, 79 cross-regional investments, 138 culture index, 92 curvilinear relationship, 190 defensive patent applications, 315 defensive patenting, 315 demonstration effects, 126 direction vector, 261 directional output distance functions, 260, 262 discrimination, 303, 313, 314, 319 dispersion parameters, 171–2, 178, 181–2 dispersion value, 319 distribution of talent, 76, 77, 83, 92, 93, 94 distribution of creative class, 75, 76, 79, 88, 90 regional, 75, 79 domestic banking performance, 8, 141, 149, 153, 158 domestic capital, 127, 136 dummy variable, 128, 168, 175, 181, 194, 201, 256, 273, 276, 322 dynamic panel probit model, 194 dynamic random-effect panel data model, 195 ecology index, 85 economic zones, 15, 26 education industry, 80 educational attainment, 75, 82 technical, 99–100, 264, 267–9, 276

efficiency change (EC), 10, 99–100, 102, 105–6, 111, 254, 256–8, 262, 264, 267–9, 276, 280–1 endogeneity, 129, 183 energy intensity, 48, 276 energy tax, 7, 47, 49, 50, 55, 59, 61, 66–7 ad valorem, 7, 49, 59, 61, 66 pilot program, 49 entrepreneurial activities, 80, 83 environmental regulation, 10, 253, 254, 255, 261, 273, 276, 277 environmentally sensitive TFP, 256 environmentally undesirable effects, 274 export intensity, 9, 190, 193, 196–7, 199, 202–3, 211–15 non-linearity between agglomeration and, 211 export spillovers, 9, 163, 190–3, 200, 204, 206–7, 209, 211, 215 export wage premium, 192 externality, 209 FDI (foreign direct investment), 8, 117–42, 145, 148–60, 163, 184, 186–7, 198, 216, 258, 323–4 inflows, 117, 119–21, 123, 129, 132, 138, 145, 152, 155, 158, 191 productivity spillovers, 150–1 spillovers regional, 160, 218 stock, 128–9, 132, 136 spillover effects of, 125, 127–9, 132, 138 feasible output set, 259 firm heterogeneity, 9, 189–91 five-year plan period, 168, 181–2, 184–5 fixed effect estimator, 154–5, 181, 183 fixed effect model, 102 fixed exchange rate system, 290 foreign equity share, 200–1, 208, 211–15 foreign exchange reserves, 288 foreign firms, 117, 149–51, 155, 193 foreign capital (FK), 126, 127, 128, 131, 137 input, 126, 131, 137 foreign-invested enterprises, 125

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-index

Index fossil fuels, 7, 50 four-digit industries, 145, 149, 153, 155, 158 fractional logit model, 196 fractional probit models, 197, 211, 213 free trade agreement, see FTAs Friedman, M., 285, 299 frontrunners of economic liberalization, 105 FTA (free trade agreement), 9, 221, 235, 237–8, 240–1, 243, 249, 252 scenarios, 245 Gaussian distribution, 167, 171, 183 general purpose technologies, 118 Gibrat’s Law, 9, 168–70, 175, 183–4, 186–7 coefficients, 28–9, 34–5, 39–40, 43, 46 relative, 28–9, 31 Gini, 26–7, 32–3, 37–8 coefficient, 16, 25–7, 35, 38, 40 Granger causality, 289 grant lag, 308, 314, 316, 319, 321 grant probability, 319–20, 322 growth rate distributions, 9, 162–3, 167, 170–1, 178, 182–5 GRP (gross regional product), 1, 4–6, 17 per capita, 19–21 GTAP (Global Trade Analysis Project), 9, 222, 233–7, 251 Hainan, 3–5, 22–3, 28, 30–1, 33, 36–7, 39, 41–2, 78–80, 101, 133, 201 harmonious society, 1 Hausman, 103–4, 319, 323 Heckit model, 211, 214 heteroskedasticity, 169, 181, 196 high-tech, 7, 58, 81, 83, 86, 92, 93, 94, 166, 203, 309 industries, 81, 83, 93, 166 sector development, 7 household utility, 50, 57, 59, 63, 64, 65

329

Hukou, 24, 77, 84, 86, 90, 92 index, 84 policy, 77, 84 human capital, 2, 75–6, 81–2, 96, 125, 127–9, 132, 134, 136, 139, 152, 192 accumulation, 75 and entrepreneurship, 2 augmentation, 126 role of, 75 impulse response, 290, 294, 295, 299, 300 analysis, 294 index of efficiency change, 262 industrial policy regime, 162, 163, 168, 170, 172, 175, 178, 181, 184, 185 inequality measurement, 16, 19 inflation, 10, 20, 90, 165, 285, 286, 287, 288, 289, 295, 298, 299, 300 expectation, 288, 289, 295 regimes, 287 transmission, 10, 285, 287, 299 initial value of capital stock, 110 Inner Mongolia, 3, 5, 23, 30, 31, 33, 36, 37, 39, 41, 57, 79, 106, 133, 201 innovative economy, 94 innovators, 257, 270, 322 intellectual property rights (IPR), 303, 313 inter-city inequality, 16, 35, 36, 37, 39, 40, 42, 43 inter-county Gini coefficients, 33 intra-provincial, 34 inter-county inequality, 15, 32, 33, 34, 40, 43 inter-CU inequality, 25, 40, 41, 42, 43 International Labor Organization (ILO), 122 International Patent Classification (IPC), 308 inter-regional migration, 138 intra-provincial regional inequality, 7, 15, 22, 25, 28, 29, 44 inverse Mill’s ratio, 197, 211 inverted-U shape, 9, 190, 215, 274 IPR protection, 303, 322

March 5, 2013

330

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-index

Industrial Development in East Asia

knowledge economy, 76 diffusion of, 136, 137 knowledge spillovers, 119, 126, 136, 137, 193, 209 lack of, 137 knowledge-based industries, 77 kurtosis, 167 Kyoto Protocol, 47, 48, 253, 261, 264, 265, 266 labor productivity, 83, 200, 209–211, 213–215, 258 regional, 83 Lagrange and Hausman tests, 102 Laplace distribution, 161, 162, 170, 171, 178 learning-by-exporting effects, 215 least square regression, 90 likelihood function, 196 linear programming, 261, 277 local production frontier, 133–137 Lorenz curve, 16 Luenberger productivity indicator, 256 Malmquist productivity index, 254, 255 Malmquist–Luenberger (ML) productivity index (ML index), 254 Malmquist–Luenberger productivity, 10 maximum likelihood, 171, 196, 206 mean logarithmic deviation, 17 medium-sized firms, 209 MFN, 232, 236 tariffs, 236 ML productivity index, 254, 255, 262, 268, 269 multicollinearity, 77 multinational enterprises (MNEs), 118, 119, 126 multiple energy resources, 48 national Gini coefficient, 32, 37 net values of fixed assets, 163, 198 nominal foreign exchange rate, 289, 291 non-bank financial institutions (NBFIs), 142

non-national treatment, 303, 307, 310 non-state-owned enterprises, 198 non-tariff barriers (NTBs), 221, 222, 226–229, 231, 235–240, 245, 247–250 removal of, 237, 245 removing, 237, 240, 245, 247 Northeast regions, 5 null hypothesis, 170, 175, 178, 183, 256, 291, 293 null-jointness, 259, 260 number of patents, 7, 308, 311, 312 openness, 4, 5, 7, 76, 81, 84, 86, 88, 94, 225, 227, 249, 258, 273, 274, 276 index, 88, 94, 273 optimal lag length, 291 ordinary least square (OLS), 155, 169, 196 over-parameterization, 294 patent examinations, 10 patent family, 308 patent flooding, 304, 305 patent index, 93, 94 patent office behavior, 10 Pearl River Delta, 163, 185, 201 performance of domestic banks, 8, 141, 151, 152, 155 Pigou–Dalton condition, 16 Poisson pseudo-maximum likelihood (PPML), 196, 211, 212 pollution control, 259 pooled probit model, 204–206 population-weighted Gini coefficient, 17 producer price index (PPI), 199 production frontier, 119, 125–128, 132–135, 137, 257, 270, 274 productivity spillovers, 150, 151, 198 productivity-increasing, 126 pseudo R2, 205, 206 Pudong, 152 quadrant, 29–31, 34, 39 R&D, 83, 125, 134, 136, 138, 203, 232, 235, 236, 249, 304, 306

March 5, 2013

11:55

9in x 6in

Regional Development and Economic Growth. . . b1491-index

Index random effect model, 102, 195 static, 195 rates of depreciation, 100, 101, 108, 110 regional development and growth, 1, 2 regional diversity, 84 regression analysis, 85, 88, 90 resource tax, 6, 7, 49 resource-rich regions, 6 Return on Assets (ROA), 144, 146 Return on Equity (ROE), 144, 147 robustness checks, 199, 206 Rules of Origin, 235 scenario simulation, 61, 66 SCGE model, 7, 49, 51, 66 Schwartz Information Criterion, 291 science and technology, 47 sectoral capital stock, 100, 110 serial correlation, 181, 293 service amenity and the cultural index, 93 Shanghai, 1, 3, 4, 5, 20, 22, 57, 78–80, 101–103, 105, 133, 152, 201, 203, 204, 285, 286 skewness, 167 source of technology transfer, 126 spatial groupings, 7, 15, 20, 22, 25–27, 32, 33, 35, 37, 38, 43 special income levy, 49 standard deviation, 107, 109, 145, 148, 149, 153–158, 167, 175, 177 state-owned enterprises (SOEs), 98, 198, 225, 228, 231 stationarity, 166, 173, 178, 291 stochastic frontier analysis (SFA), 99, 257 strategic patent policies, 304, 306, 309, 310, 314, 319 strategic trade behavior, 307 sulfur dioxide, 49 sustainability, 8, 98, 108

technological progress, 8, 10, 98, 99, 100, 102, 105–108, 110, 119, 125–128, 132, 133, 135–138, 203, 254–258, 262, 264, 267–270, 27 index of, 262 indices, 270 rates of, 102, 105, 256 TFP growth, 99, 98, 102, 105–107, 125, 256–258, 276 Theil index, 17 the third line, 59 total factor productivity (TFP), 97, 125, 254 tourism industry, 80 TRIPS, 313, 319 undesirable outputs, 254, 255, 258, 276, 280 UNFCCC, 253, 256, 261, 264, 273, 276 unweighted Gini coefficient, 17 urban environmental infrastructure, 85, 87, 93 urbanization, 7, 83, 93 value-added, 83, 86, 100, 197, 200, 239, 250, 273, 274 variance decomposition, 290, 294, 298–300 vector error correction model (VECM), 10, 288, 290, 299 Vietnam, 9, 221–228, 231–235, 237–250 volatility, 285, 300 Western development, 5, 8, 106, 107, 108 program, 5, 8, 106, 107 World Bank, 4, 51, 228, 230, 238, 263, 273 WTO, 143, 224, 226, 227, 230, 231, 235, 299, 313, 314, 319, 322 Yangtze River Delta, 201

technical efficiency, 99, 100, 102, 111, 264, 267–269, 276

331

zonal-level analysis, 43