Evolutionary Economic Geography in China [1st ed.] 978-981-13-3446-7;978-981-13-3447-4

Table of contents :
Front Matter ....Pages i-xxiii
Introduction (Canfei He, Shengjun Zhu)....Pages 1-24
How Has Production Space Evolved in China? (Canfei He, Shengjun Zhu)....Pages 25-46
How Does Regional Industrial Structure Evolve in China? (Canfei He, Shengjun Zhu)....Pages 47-76
What Matters for Regional Industrial Dynamics in China? (Canfei He, Shengjun Zhu)....Pages 77-95
What Facilitates New Firm Formation in China? (Canfei He, Shengjun Zhu)....Pages 97-122
Does Creative Destruction Work for Chinese Regions? (Canfei He, Shengjun Zhu)....Pages 123-146
What Causes Firm Failure in China? (Canfei He, Shengjun Zhu)....Pages 147-168
What Sustains Large Firms in China? (Canfei He, Shengjun Zhu)....Pages 169-194
How Do Agglomeration Economies Contribute to Firm Survival in China? (Canfei He, Shengjun Zhu)....Pages 195-214
How Does Geese Fly Domestically? Firm Demography and Spatial Restructuring in China’s Apparel Industry (Canfei He, Shengjun Zhu)....Pages 215-230
How Do Environmental Regulations Affect Industrial Dynamics in China? (Canfei He, Shengjun Zhu)....Pages 231-249
How to Jump Further? Path Dependence and Path-Breaking in an Uneven Industry Space (Canfei He, Shengjun Zhu)....Pages 251-280
What Drives the Evolution of Export Product Space in China? (Canfei He, Shengjun Zhu)....Pages 281-302
How Do Firm Dynamics Affect Regional Inequality of Productivity in China? (Canfei He, Shengjun Zhu)....Pages 303-323
Summary and Conclusion (Canfei He, Shengjun Zhu)....Pages 325-331

Citation preview

Economic Geography

Canfei He Shengjun Zhu

Evolutionary Economic Geography in China

Economic Geography

More information about this series at http://www.springer.com/series/15653

Canfei He • Shengjun Zhu

Evolutionary Economic Geography in China

Canfei He College of Urban and Environmental Sciences Peking University Beijing, China

Shengjun Zhu College of Urban and Environmental Sciences Peking University Beijing, China

ISSN 2520-1417 ISSN 2520-1425 (electronic) Economic Geography ISBN 978-981-13-3446-7 ISBN 978-981-13-3447-4 (eBook) https://doi.org/10.1007/978-981-13-3447-4 Library of Congress Control Number: 2018964119 © Springer Nature Singapore Pte Ltd. 2019 This work is subject to copyright. All rights are reserved by the Publisher, whether the whole or part of the material is concerned, specifically the rights of translation, reprinting, reuse of illustrations, recitation, broadcasting, reproduction on microfilms or in any other physical way, and transmission or information storage and retrieval, electronic adaptation, computer software, or by similar or dissimilar methodology now known or hereafter developed. The use of general descriptive names, registered names, trademarks, service marks, etc. in this publication does not imply, even in the absence of a specific statement, that such names are exempt from the relevant protective laws and regulations and therefore free for general use. The publisher, the authors, and the editors are safe to assume that the advice and information in this book are believed to be true and accurate at the date of publication. Neither the publisher nor the authors or the editors give a warranty, express or implied, with respect to the material contained herein or for any errors or omissions that may have been made. The publisher remains neutral with regard to jurisdictional claims in published maps and institutional affiliations. This Springer imprint is published by the registered company Springer Nature Singapore Pte Ltd. The registered company address is: 152 Beach Road, #21-01/04 Gateway East, Singapore 189721, Singapore

Foreword

Since the 1990s, there is an expanding scientific community that is working on the theoretical, conceptual, and empirical foundations of evolutionary economic geography (EEG). A lot of progress has been made since then: impacts on academic research and policy are clearly visible, but there is also awareness that EEG is evolving and still very much work in progress. During its formative years, EEG developed mainly in Western Europe and the USA. More recently, it has also diffused to other parts of the world, like Eastern Europe, Asia, and Latin America where the evolutionary turn in economic geography has been embraced and applied in a non-Western context. More importantly, scholars have made significant contributions taking into account the economic, cultural, and institutional peculiarities of their own worlds, leading to the further advancement of EEG. No doubt China is a prime example here. Many of us know China as a country that has a remarkable capacity to catch up in a relatively short period of time. Scientific work in EEG is no exception to that rule. Canfei He and Shengjun Zhu from Peking University are absolute pioneers in this respect. This new book of these two leading Chinese scholars bears witness to the tremendous work in EEG that Chinese scholars have produced in recent years. This book on EEG can be considered an absolute milestone in that respect. To me, they also confirm a well-known general wisdom about China: if the Chinese go for it, they do it well, and fast! This book by these two top Chinese scholars is an absolute must-read for everybody who is interested in EEG more in general and how EEG can be fruitfully applied to the Chinese context in particular. It contains the incredible number of 14 empirical chapters. All chapters provide important insights and valuable contributions in that part of the EEG literature that focuses on the dynamic interplay between firm dynamics, industrial change, and regional dynamics. In that sense, it comes close to the type of EEG that has developed in Europe and the USA. Building on evolutionary concepts like path dependency, proximity, related variety, product space, and related/unrelated diversification, the book presents very rich empirical

v

vi

Foreword

material on the changing fortunes of firms, industries, and regions in China during the last decades, showing, among other things, that the evolution of many regions in China has been subject to strong forces of path dependency, as in the USA and Europe. Another quality of the book is that it enters almost unexplored territories in EEG so far, like studies on environmental and inequality issues. Environmental sustainability and socioeconomic inequality are not only major challenges in the current context of China but also worldwide. There is no doubt that these topics will be high on the scientific and policy agendas for the years to come. In recent years, EEG is starting to explore these key topics and ready to contribute to a better understanding of how to tackle these pressing societal challenges from a specific evolutionary angle. A final key achievement of the book is that it gives testimony of the Chinese way of doing EEG, incorporating dimensions like the role of institutional arrangements and the (changing) role of the Chinese state at various spatial scales that are very specific to the Chinese context. We learn a lot about the role of the Chinese state, not only conditioning industrial and regional dynamics in China but also playing an active role itself in many forms and at various spatial scales. Doing so, this seminal book on EEG in China provides new key insights, opens up new challenges for academics and policy makers alike, and raises a number of fundamental theoretical, conceptual, and empirical questions that need to be taken up in future research in EEG. A must-read that comes at the right time! Utrecht University, Utrecht, Netherlands 12 October 2018

Ron Boschma

Parts of this monograph have been published in the following journal articles

1. Guo, Q. and C. He (2017) Production Space and Regional Industrial Evolution in China, GeoJournal, 82 (2), pp. 379–396. 2. He, C., Y. Yan and D. Rigby (2018) Regional Industrial Evolution in China, Papers in Regional Science, 97 (2), pp. 173–198. 3. He, C., S. Zhu and X. Yang (2017) What Matters for Regional Industrial Dynamics in a Transitional Economy?, Area Development and Policy, 2 (1), pp. 71–90. 4. Guo, Q., C. He and D. Li (2016) Entrepreneurship in China: The Role of Localisation and Urbanisation Economies, Urban Studies, 53 (12), pp. 2584–2606. 5. Zhou, Y., He, C. and Zhu, S. (2017), Does Creative Destruction Work for Chinese Regions? Growth and Change, 48: 274–296. 6. He, C. and Yang, R. (2016), Determinants of Firm Failure: Empirical Evidence from China. Growth and Change, 47: 72–92. 7. He, C., Q. Guo and D. Rigby (2017) What Sustains Larger Firms? Evidence from Chinese Manufacturing Industries, The Annals of Regional Science, 58 (2), pp. 275–300. 8. Howell, A., He, C., Yang, R., and Fan, C. C. (2018) Agglomeration, (un)‐related variety and new firm survival in China: Do local subsidies matter?. Papers in Regional Science, 97: 485–500. 9. Shi, J., He, C. and Guo, Q. (2016), How did geese fly domestically? Firm demography and spatial restructuring in China's apparel industry. Area, 48: 346–356. 10. Zhou, Y., S. Zhu and C. He (2017) How Do Environmental Regulations Affect Industrial Dynamics? Evidence from China's Pollution-Intensive Industries, Habitat International, 60: 10–18. 11. Zhu, S., C. He and Y. Zhou (2017) How to Jump Further and Catch Up? PathBreaking in an Uneven Industry Space, Journal of Economic Geography, 17 (3): 521–545.

vii

viii

Parts of this monograph have been published in the following journal articles

12. He, C. and Zhu, S. (2018), Evolution of Export Product Space in China: Technological Relatedness, National/Local Governance and Regional Industrial Diversification. Tijdschrift voor Economische en Sociale Geografie, 109(4): 575–593. 13. He, C., Zhou, Y. and Zhu, S. (2017), Firm Dynamics, Institutional Context, And Regional Inequality Of Productivity In China. Geographical Review, 107(2): 296–316.

Acknowledgments

We would like to thank the inputs of Qi Guo, Yi Zhou, Anthony Howell, Yan Yan, Xin Yang, Jin Shi, Rudai Yang, Yao Dong, and Yan Ma. We also acknowledge the financial support of the National Natural Science Foundation of China for Distinguished Young Scholars (No. 41425001) and the National Natural Science Foundation of China for Key Project (No. 41731278). The authors are responsible for all errors and interpretations.

ix

Contents

1

2

Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.1 An “Evolutionary” Turn in Economic Geography . . . . . . . . . . 1.2 The Case of China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3 Methodology . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.1 Data Sources . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.3.2 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.4 Synopsis of the Book . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.5 Part I: Industrial Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . 1.6 Part II: Firm Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.7 Part III Product Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . 1.8 Part IV Impact of Industrial Dynamics on Regional Inequality . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . . .

1 1 5 10 10 11 14 14 15 18

. .

18 19

How Has Production Space Evolved in China? . . . . . . . . . . . . . . . 2.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.2 Literature Review and Analytical Framework . . . . . . . . . . . . . 2.3 Data and Model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.3.1 The Measurement of Industrial Relatedness . . . . . . . . 2.3.2 Model Specification and Variables . . . . . . . . . . . . . . . 2.4 The Evolution of Production Network and Regional Path Dependence . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.4.1 The Evolution of China’s Production Space . . . . . . . . 2.4.2 The Evolution of Regional Productive Structure . . . . . 2.5 The Regional Path Dependence and the Effect of Institutions . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2.6 Conclusion and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

25 25 26 29 29 30

. . .

30 30 34

. . .

37 42 44

xi

xii

3

4

5

Contents

How Does Regional Industrial Structure Evolve in China? . . . . . . 3.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.2 Understanding Regional Industrial Evolution . . . . . . . . . . . . . 3.2.1 Relatedness and Industrial Evolution of Regions . . . . 3.2.2 Global Linkages, Regional Institutions, and Industrial Evolution of Regions in China . . . . . . . . . . . . . . . . . 3.3 Data and the Relatedness Indicator . . . . . . . . . . . . . . . . . . . . . 3.4 Descriptive Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.4.1 Industrial Entries and Exits in China . . . . . . . . . . . . . 3.4.2 Relatedness and Regional Industrial Evolution in China . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5 Model Specifications and Empirical Findings . . . . . . . . . . . . . 3.5.1 Model Specifications . . . . . . . . . . . . . . . . . . . . . . . . 3.5.2 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . 3.5.3 Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . 3.6 Summary and Discussion . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . What Matters for Regional Industrial Dynamics in China? . . . . . . 4.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.2 Regional Industrial Dynamics: Region-Specific and Industry-Specific Factors . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.1 Technological Relatedness and Regional Institutions . . . . . . . . . . . . . . . . . . . . . . . . . 4.2.2 Industry-Specific Factors . . . . . . . . . . . . . . . . . . . . . 4.3 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.4 Regional Industrial Dynamics in China . . . . . . . . . . . . . . . . . . 4.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 4.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . What Facilitates New Firm Formation in China? . . . . . . . . . . . . . 5.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2 Literature Review and Hypothesis Development . . . . . . . . . . . 5.2.1 The Impact of Localization and Urbanization Economies on New Firm Formation . . . . . . . . . . . . . 5.2.2 The Impact of Related Variety and Unrelated Variety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.2.3 The Vernon-Chinitz Effect . . . . . . . . . . . . . . . . . . . . 5.3 New Firm Formation in China . . . . . . . . . . . . . . . . . . . . . . . . 5.4 Model Specification and Variables . . . . . . . . . . . . . . . . . . . . . 5.4.1 Dependent Variables and Model Specification . . . . . . 5.4.2 Localization Externalities and Supplier/Customer Linkages . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . .

47 47 49 49

. . . .

50 54 55 55

. . . . . . .

58 59 59 61 67 68 73

. .

77 77

.

79

. . . . . . .

79 80 81 83 86 92 93

. . .

97 97 99

.

99

. . . . .

100 101 102 106 106

. 106

Contents

xiii

5.4.3

Jacobs Externalities, Related Variety, and Unrelated Variety . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.4.4 The Vernon-Chinitz Effect . . . . . . . . . . . . . . . . . . . . 5.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 5.5.1 Impact of Agglomeration Economies . . . . . . . . . . . . . 5.5.2 The Vernon-Chinitz Effect . . . . . . . . . . . . . . . . . . . . 5.5.3 The Size Effect of Start-Ups . . . . . . . . . . . . . . . . . . . 5.5.4 Robustness Check . . . . . . . . . . . . . . . . . . . . . . . . . . 5.6 Conclusion and Implications . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . . . .

107 108 110 110 113 116 118 118 120

Does Creative Destruction Work for Chinese Regions? . . . . . . . . . 6.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.2 Articulation Between Firm Entry and Exit . . . . . . . . . . . . . . . 6.2.1 Firm-Specific Factors . . . . . . . . . . . . . . . . . . . . . . . . 6.2.2 Industry-Specific Factors . . . . . . . . . . . . . . . . . . . . . 6.2.3 Region-Specific Factors . . . . . . . . . . . . . . . . . . . . . . 6.3 Temporal and Spatial Changes of China’s Industrial Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.4 Model Specification and Variables . . . . . . . . . . . . . . . . . . . . . 6.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 6.5.1 Empirical Results on Firm-Specific Factors . . . . . . . . 6.5.2 Empirical Results on Industry- and Region-Specific Factors . . . . . . . . . . . . . . . . . . . . . . 6.5.3 Geographical Proximity in Creative Destruction . . . . . 6.6 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

123 123 125 126 126 127

. . . .

128 132 135 135

. . . .

138 140 141 143

7

What Causes Firm Failure in China? . . . . . . . . . . . . . . . . . . . . . . 7.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.2 Literature Review and Research Hypothesis . . . . . . . . . . . . . . 7.3 Data Source and Firm TFP Estimation . . . . . . . . . . . . . . . . . . 7.4 Pattern of Firm Failure in China . . . . . . . . . . . . . . . . . . . . . . . 7.4.1 Firm TFP and Firm Failure . . . . . . . . . . . . . . . . . . . . 7.4.2 Firm Age and Firm Failure . . . . . . . . . . . . . . . . . . . . 7.4.3 Local Support and Firm Failure . . . . . . . . . . . . . . . . . 7.5 Variables and Model Specification . . . . . . . . . . . . . . . . . . . . . 7.6 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7.7 Summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

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

147 147 148 151 152 152 154 155 157 159 163 165

8

What Sustains Large Firms in China? . . . . . . . . . . . . . . . . . . . . . 8.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2 Literature Review and Analytical Framework . . . . . . . . . . . . . 8.2.1 Firm Sustaining in China: An Analytical Framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 169 . 169 . 171

6

. 172

xiv

Contents

8.2.2

Globalization, Global-Local Interaction, and Firm Sustaining . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.2.3 Regional Decentralization, Regional-Local Interactions, and Firm Sustaining . . . . . . . . . . . . . . . . . 8.3 New Firm Sustaining in China During 1998–2005 . . . . . . . . . . 8.4 Model Specification and Variables . . . . . . . . . . . . . . . . . . . . . . 8.4.1 Cox Proportional Hazard Model . . . . . . . . . . . . . . . . . 8.4.2 Explanatory Variables . . . . . . . . . . . . . . . . . . . . . . . . 8.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 8.6 Conclusion and Implications . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9

10

How Do Agglomeration Economies Contribute to Firm Survival in China? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2 Literature Review . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.2.1 Agglomeration and the Role of State Support in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3 Data and Variable Development . . . . . . . . . . . . . . . . . . . . . . . . 9.3.1 Dependent Variable: Measuring and Interpreting Firm Survival . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.3.2 Agglomeration Measures . . . . . . . . . . . . . . . . . . . . . . 9.3.3 Firm-Level Covariates . . . . . . . . . . . . . . . . . . . . . . . . 9.4 Model Specification . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.4.1 Identification Issues . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5 Effects of Agglomeration on New Firm Survival . . . . . . . . . . . . 9.5.1 Alternative Model Specifications and Robustness Checks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 9.5.2 Effects of Local Subsidies on Firm Survival . . . . . . . . . 9.6 Concluding Remarks . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How Does Geese Fly Domestically? Firm Demography and Spatial Restructuring in China’s Apparel Industry . . . . . . . . 10.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2 Data and Method . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.2.1 The Firm-Level Dataset: The Annual Survey of Industrial Firms . . . . . . . . . . . . . . . . . . . . . . . . . 10.2.2 Firm Demography and Gross Job Flow Analysis . . . 10.2.3 Theil Index of Spatial Disparity and Its Decomposition . . . . . . . . . . . . . . . . . . . . . . . . . . . . 10.3 The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry . . . . . . . . . . . . . . . . . . . . . . . . . 10.3.1 Province Scale: Domestically Flying Geese Driven by Differences in Job Creation Rate . . . . . . .

173 174 176 179 179 179 182 189 190 195 195 197 198 199 199 201 202 202 204 208 208 211 212 213

. 215 . 215 . 216 . 216 . 218 . 220 . 221 . 222

Contents

xv

10.3.2

City Level: The Dominance of Firm Entry Through Start-Up Firms . . . . . . . . . . . . . . . . . . . . . 10.3.3 The Spatial Disparity of Start-Ups Between and Within Provinces . . . . . . . . . . . . . . . . . . . . . . . 10.4 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11

12

13

How Do Environmental Regulations Affect Industrial Dynamics in China? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.2 Environmental Regulation and Industrial Dynamics: The Role of Firm Heterogeneity and Government Intervention . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3 Data Source and the Dynamics of China’s Pollution-Intensive Industries . . . . . . . . . . . . . . . . . . . . . . . . 11.3.1 Data Source . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.3.2 Dynamics of China’s Pollution-Intensive Industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11.4 Model Specification and Empirical Results . . . . . . . . . . . . . . 11.4.1 Model Specification . . . . . . . . . . . . . . . . . . . . . . . . 11.4.2 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . 11.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . How to Jump Further? Path Dependence and Path-Breaking in an Uneven Industry Space . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.2 Path Dependence and Path-Breaking . . . . . . . . . . . . . . . . . . 12.2.1 Extra-regional Linkages . . . . . . . . . . . . . . . . . . . . . 12.2.2 Internal Innovation . . . . . . . . . . . . . . . . . . . . . . . . . 12.3 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.3.1 Model Specification . . . . . . . . . . . . . . . . . . . . . . . . 12.3.2 Variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.4 The Relationship Between Density and the Emergence of New Industries in Chinese Cities . . . . . . . . . . . . . . . . . . . 12.5 Empirical Results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 12.6 Discussion and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . What Drives the Evolution of Export Product Space in China? . . 13.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2 Theoretical Framework and Research Hypotheses . . . . . . . . . 13.2.1 Path-Dependent Evolution of Productive Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.2.2 State Power and Evolution of Productive Structure . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. 223 . 227 . 229 . 230 . 231 . 231

. 233 . 236 . 236 . . . . . .

236 239 239 240 244 247

. . . . . . . .

251 251 253 254 256 258 258 259

. . . .

261 264 274 278

. 281 . 281 . 283 . 283 . 285

xvi

Contents

13.3 13.4

Evolution of Export Product Space in China . . . . . . . . . . . . . Econometric Analysis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 13.4.1 Variables and Model Specification . . . . . . . . . . . . . . 13.4.2 Econometric Results . . . . . . . . . . . . . . . . . . . . . . . . 13.5 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14

15

. . . . . .

How Do Firm Dynamics Affect Regional Inequality of Productivity in China? . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.1 Introduction . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.2 Regional Inequality of Productivity, Firm Dynamics, and Institutional Context . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.3 Research Design . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.4 Regional Inequality of Productivity in China . . . . . . . . . . . . . 14.5 Decomposition of Regional TFP Growth . . . . . . . . . . . . . . . 14.5.1 Within Share . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.5.2 Between Share . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.5.3 Entry and Exit Share . . . . . . . . . . . . . . . . . . . . . . . . 14.6 Institutional Context and Regional Inequality of Productivity . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 14.7 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . . .

Summary and Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.1 Industrial Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.2 Firm Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.3 Product Dynamics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15.4 Impact of Industrial Dynamics on Regional Inequality . . . . . . References . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

. . . . . .

287 291 291 295 298 300

. 303 . 303 304 306 308 310 311 313 314

. 317 . 319 . 321 325 326 327 329 330 331

List of Figures

Fig. 2.1 Fig. 2.2 Fig. 2.3 Fig. 2.4

Fig. 2.5 Fig. 2.6 Fig. 2.7 Fig. 2.8 Fig. 2.9 Fig. 3.1 Fig. 3.2 Fig. 3.3 Fig. 3.4 Fig. 3.5

Distribution and statistic index of industry relatedness, 1999 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The production network of China, 1999 . . . . . . . . . . . . . . . . . . . . . . . . . . . . The production network of China, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . The mean value of industry relatedness for all links (Total), links within the same classification (Within), and between different ones (Between), 1999–2007 . . . . . . . . . . . . . . . . Localization of the productive structure for different regions in the production space of China, 1999 . . . . . . . . . . . . . . . . . . . . The productive structure of different regions in the production space in China, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Distribution of density for transition sectors and undeveloped sectors during 1999–2007 . . . . . . . . . . . . . . . . . . . . . . . . The relationship between density and sectoral productivity for the transition sectors . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The transition sectors with low density and high productivity in North West and South West . . . . . . . . . . . . . . . . . . . . . . . . Industrial diversification of Chinese prefectures (Left 1998, Right 2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Number of industry entries (Left 1998–2003, Right 2003–2008) . . . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . Number of industry exits (Left 1998–2003, Right 2003–2008) . . . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . . . . .. . . Relationship between average relatedness of entry industries and nonentry industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Relationship between average relatedness of exit and non-exit industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

31 32 33

33 35 36 37 42 43 56 57 57 59 59

xvii

xviii

Fig. 4.1

Fig. 4.2

Fig. 4.3

Fig. 4.4

Fig. 5.1 Fig. 5.2 Fig. 5.3 Fig. 6.1 Fig. 6.2 Fig. 6.3 Fig. 6.4 Fig. 6.5

Fig. 7.1 Fig. 7.2 Fig. 7.3 Fig. 7.4

List of Figures

(a) (Left) and (b) (right) the relationship between the average density of the industries that were absent in a prefecture in 1998 (or 2003) and the probability of this prefecture creating or attracting a new industry in 2003 (or 2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (a) (Left) and (b) (right) the relationship between the average density of the industries that a prefecture had in 1998 (or 2003) and the probability of these industries exiting from this prefecture in 2003 (or 2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (a) (Left) and (b) (right) the relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of this prefecture creating or attracting a new industry in 2003 (or 2008) . . . . . . . . . . . . . . . . . . . . . (a) (Left) and (b) (right) the relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of industries exiting from this prefecture in 2003 (or 2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

84

84

85

85

The number of private owned start-ups by 2-digit manufacturing industries, 2001 and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 Spatial pattern of private-owned manufacturing start-ups, 2001, 2003, 2005, and 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Spatial pattern of agglomeration and new firm formation in China, 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 Two perspectives on the articulation of firm entry and exit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Firm exit and entry in China at provincial level in 1998–2003 and 2003–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Four types of growth regimes based on firm entry and exit rate . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Growth regimes in China at prefecture level in 1998–2003 and 2003–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . (a) TFP of exiting and entering firms in Central, East, and West China during 1999–2007 (left); (b) TFP of exiting and entering firms during 1999–2007 at provincial level (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Failing firms, surviving firms, and firm TFP . . . . . . . . . . . . . . . . . . . . . . . Average age of Chinese firms during 1998–2006 . . . . . . . . . . . . . . . . . . Firm age and firm failure in China . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Local supports and firm exit . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

125 131 132 132

133 154 155 155 156

List of Figures

xix

Fig. 7.5 Fig. 7.6

Firm age, local supports, and firm failure . . . . . . . . . . . . . . . . . . . . . . . . . . . 156 Local supports and firm TFP . . . . . . .. . . . . . .. . . . . . . .. . . . . . . .. . . . . . . .. . . 157

Fig. 8.1

New firms’ survival rates for different number of years (1998–2005) . . .. . .. . .. . .. . .. . .. . .. . .. .. . .. . .. . .. . .. . 177 Survival rates of new firms born in 1998 across regions (left, survival rates; right, region classification of China) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 Provincial distribution of survival rate of firms born in 1998 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178

Fig. 8.2

Fig. 8.3 Fig. 9.1

Average 3-year survival rates, 1998–2007 .. . . . . . . . . . . . . . . . . .. . . . . . . 200

Fig. 10.1

Net employment change of China’s apparel industry by province and city for 1999–2003 and 2004–2008 . . . . . . . . . . . . . . Job creation and destruction of China’s apparel industry by province for 1999–2003 and 2004–2008 . . . . . . . . . . . . . . Decomposition of the Theil index for spatial disparity of start-ups in China’s apparel industry (1998–2008) . . . . . . . . . . . . . Theil index for spatial disparity of start-ups within provinces in China’s apparel industry during 1999–2003 and 2004–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Fig. 10.2 Fig. 10.3 Fig. 10.4

Fig. 11.1 Fig. 11.2

Fig. 11.3

Fig. 11.4

Fig. 12.1 Fig. 12.2

Fig. 12.3 Fig. 12.4

The percentage of industrial SO2 meeting standard for emission at the city level in 2003 (left) and 2008 (right) . . . . . . . . . . Firm entry and exit rate of pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Employment change of incumbent firms in pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . TFP change of incumbent firms in pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Two types of new path creation and regional development . . . . . . . Extra-regional linkages and internal innovation. (HQs, headquarters; TNC, transnational corporation; FDI, foreign direct investment) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The share of transition industries in China’s five big regions during 2002–2006 and 2007–2011 . . . . . . . . . . . . . . . . . . . . (a) (top) and (b) (bottom) Relationship between the average density of the industries without a comparative advantage in a city in 2002 (or 2007) and the probability of this city developing a comparative advantage in a new industry in 2006 (or 2011) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

222 224 227

228 237

237

238

238 254

256 262

263

xx

Fig. 13.1 Fig. 13.2 Fig. 13.3

Fig. 13.4 Fig. 13.5 Fig. 14.1 Fig. 14.2 Fig. 14.3 Fig. 14.4 Fig. 14.5 Fig. 14.6 Fig. 14.7

List of Figures

Distribution of China’s export product relatedness in 2000 and 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . . . . . . . . . . . . . . . . . .. . . . . . . . . . . . (a) The export product space in China, 2000. (b) The export product space in China, 2011 . . . . . . .. . . . . . .. . . . . .. . . Relationship between the probability of developing a comparative advantage in a new product (5 years later) and average export/import density of products without a comparative advantage at the beginning . . . . . . . . . . . . . . . . . . . . . . . . . . Export tax rebate rates in different types of products (2002–2011) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Share of 4-digit products for which FDIs are encouraged or limited/prohibited in 2-digit products (2002) . . . . . . . . . . . . . . . . . . . . Spatial variation of regional TFP in China in 1999 (left) and 2007 (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spatial distribution of regional TFP change from 1999 to 2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Spatial distribution of the within share of regional TFP growth from 1999 to 2007 .. . . .. . . .. . . .. . . .. . . .. . . .. . .. . . .. . . .. . Spatial distribution of the within share generated by SOEs (left) and private firms (right) . . . . . . . . . . . . .. . . . . . . . . . . . . .. . . . . . Spatial distribution of the between share of regional TFP change from 1999 to 2007 .. . . .. . . .. . . .. . . .. . . .. . . .. . .. . . .. . . .. . Spatial distribution of the entry (left) and the exit (right) share of regional TFP change from 1999 to 2007 . . . . . . . . . . . . . . . . . . The most important component of TFP growth in China’s regions during 1999–2007 (left), 1999–2003 (middle) and 2003–2007 (right) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

288 290

292 293 294 309 310 313 313 314 315

316

List of Tables

Table 2.1 Table 2.2 Table 2.3 Table 2.4 Table 3.1 Table 3.2 Table 3.3 Table 3.4 Table 3.5 Table 3.6 Table 3.7 Table 3.8 Table 3.9 Table 4.1 Table 4.2 Table 4.3 Table 5.1 Table 5.2 Table 5.3

Correlation coefficients of production relatedness during 1999–2007 .. . . .. . . .. . .. . . .. . . .. . .. . . .. . . .. . . .. . .. . . .. . . .. . .. . The correlation coefficients of regional advantage industries between 1999 and 2007 .. . .. . .. . .. .. . .. . .. . .. .. . .. . .. . .. The effect of production space (density) on industrial evolution . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . The effect of supportive policies on path creation . . . . . . . . . . . . . . . Average number of new entries and exits in Chinese prefectures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Definition of explanatory variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Correlation coefficients among independent variables . . . . . . . . . . . Logit regression results for entry equations . . . . . . . . . . . . . . . . . . . . . . . Logit regression results for exit equations . . . . . . . . . . . . . . . . . . . . . . . . Logit regression results with annual data for entry equations (LQ ¼ 1.0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Logit regression results with annual data for entry equations (LQ ¼ 0.5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Logit regression results with annual data for exit equations (LQ ¼ 1.0) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Logit regression results with annual data for exit equations (LQ ¼ 0.5) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Pearson correlation matrix . .. . .. . .. . . .. . .. . .. . .. . .. . . .. . .. . .. . .. . . .. Estimation results on industry characteristics and big region variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Estimation results on region- and industry-specific factors . . . . . .

31 36 39 41 56 61 62 63 64 69 70 71 72 87 88 90

The employment and number of all start-ups and start-ups of different ownership, 2001–2007 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 Definition of independent variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 Correlation coefficients of independent variables . . . . . . . . . . . . . . . . 111 xxi

xxii

Table 5.4 Table 5.5 Table 5.6

Table 5.7 Table 5.8 Table 5.9 Table 6.1 Table 6.2 Table 6.3

List of Tables

Correlation coefficients of independent variables . . . . . . . . . . . . . . . . Correlation coefficients of independent variables . . . . . . . . . . . . . . . . The effect of agglomeration economies on new firm formation (hereafter, city-sector fixed effect negative binomial models) . . .. . . . . .. . . . . . .. . . . . . .. . . . . .. . . . . . .. . . . . . .. . . . . .. . . Decomposition of localization economies for the Vernon-Chinitz effect . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . .. . . . . . . . . . . .. . . . Decomposition of Jacobs externalities for the Vernon-Chinitz effect . . . . . . . .. . . . . . . . . . . .. . . . . . . . . . .. . . . . . . . . . . .. . . . The Vernon-Chinitz effect for start-ups with different sizes in 2007 . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . .

112 113

114 115 117 118

Firm entry and exit in China (1998–2008) . . . .. . . .. . . . .. . . . .. . . .. . Regression results on firm-specific factors (1999–2008) . . . . . . . . Regression results on industry-specific and region-specific factors (1999–2008) . . . . . . . . . . . . . . . . . . . . . . . . . . Regression results on geography proximity (1999–2008) . . . . . . .

129 136

Firm entry and failure in China during 1998–2007 . . . . . . . . . . . . . . Definitions of dependent and explanatory variables . . . . . . . . . . . . . Correlation coefficients among key explanatory variables . . . . . . Regression results from LPM for panel data .. . .. . .. . .. .. . .. . .. . .. Regression results from LPM for panel data (excluding the sample of year 2004) . . . . . . . . . . . . . . . . . . . . . . . . .

153 158 160 161

181 182

Table 8.4 Table 8.5 Table 8.6

Definitions of explanatory variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regression results of PH Cox models for all new firms . . . . . . . . . Regression results of PH Cox models for decomposition of ELI . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Regression results of PH Cox models for SOEs . . . . . . . . . . . . . . . . . Regression results of PH Cox models for foreign firms . . . . . . . . . Regression results of PH Cox models for private firms . . . . . . . . .

184 186 187 188

Table 9.1 Table 9.2 Table 9.3 Table 9.4 Table 9.5

Descriptive statistics . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Sample selection model . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects of agglomeration and subsidies on firm survival . . . . . . . . Alternative model estimations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Effects of local subsidies on firm survival . . . . . . . . . . . . . . . . . . . . . . . .

205 207 209 210 211

Table 10.1

Evolution of the panel of apparel manufacturing enterprises in China (1999–2008) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 217 Components of employment change . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 219 Employment dynamics of China’s apparel industry at the city level during 1999–2003 and 2004–2008 . . . . . . . . . . . . . . . . . . . . . 225

Table 6.4 Table 7.1 Table 7.2 Table 7.3 Table 7.4 Table 7.5 Table 8.1 Table 8.2 Table 8.3

Table 10.2 Table 10.3

139 141

164

List of Tables

xxiii

Table 10.4

Start-ups vs. non-start-up firms within China’s apparel industry at the city level during 1999–2003 and 2004–2008 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 226

Table 11.1 Table 11.2 Table 11.3

Definition of variables and descriptive statistics . . . . . . . . . . . . . . . . . 241 Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 242 Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 245

Table 12.1 Table 12.2

Correlation matrix . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . . .. . . Determinants of having developed industries in China (LPM) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Determinants of having developed industries in East and Northeast China . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . .. . Determinants of having developed industries in different industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Determinants of having developed industries in China (probit) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Table 12.3 Table 12.4 Table 12.5 Table 13.1 Table 13.2 Table 13.3 Table 13.4 Table 13.5 Table 14.1 Table 14.2 Table 14.3 Table 14.4 Table 14.5

265 266 270 272 275

Correlation analysis of export product relatedness during 2000–2011 .. . . .. . . .. . .. . . .. . . .. . .. . . .. . . .. . . .. . .. . . .. . . .. . .. . Pearson correlation matrix . .. . .. . .. . . .. . .. . .. . .. . .. . . .. . .. . .. . .. . . .. Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Empirical results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

289 296 296 297 299

Temporal change of regional TFP in China (1998–2007) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . Decomposition of regional TFP growth (1999–2007) . . . . . . . . . . . Within share generated by various types of firms . . . . . . . . . . . . . . . . Entry and exit share generated by various types of firms . . . . . . . . Regression results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

309 311 312 315 318

Chapter 1

Introduction

1.1

An “Evolutionary” Turn in Economic Geography

Regional industrial dynamics can be seen as a complex process, consisting of both qualitative and quantitative changes. Even though quantitative changes like regional growth in terms of employment, value added, and output may reflect the rise and fall of regional economies and the restructuring of industrial areas, these changes often take place as a result of qualitative changes in regional industrial structures (Neffke et al. 2011). Regions are shaped by a never-ending process of creative destruction (Schumpeter 1939, 1942), referring not only to the capability of local entrepreneurs to develop new products or processes that can replace traditional ones and render the latter obsolete in the short term but also to the capability of a certain region to generate and attract new industries to offset the destruction caused by industrial exit and decline in the long term. Such a dual process is closely related to the resilience of regional economies in face of disturbances and shocks (e.g., institutional transformation, policy changes, fluctuation of currency exchange rates, and technological shifts) that may result in decline, atrophy, or even shutdown of an entire industry in regions (Martin and Sunley 2015a, b). This view inspires plenty of economic geographers to examine the geographical implications of creative destruction (Hassink and Shin 2005; Stam and Martin 2012), and we now have several examples of where such process has forced the transformation of regional industrial structures in Europe, North America, and Japan (Grabher 1993; Hassink 2007; Schamp 2005), the Asian newly industrialized economies since the mid-1990s (Cho and Hassink 2009; Van Grunsven and Smakman 2005), and emerging regional economies such as mainland China since the 2000s (He and Wang 2012; Zhu and He 2016; Zhu et al. 2014). The idea of how regions develop new industries and how new regional growth paths emerge has long been a concern of economic geography in the past (Scott 1988b; Storper and Walker 1989) and present (Martin 2010; Martin and Sunley 2006). Traditional economic geography and regional industrial development © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_1

1

2

1 Introduction

literatures have stressed the importance of a variety of region-specific assets, such as local institutional contexts, policy-making, local economic agents, and vertical and horizontal linkages that generate collective efficiency and knowledge spillovers (Marshall 1920 [1890]), institutional thickness (Amin and Thrift 1994), embeddedness (Granovetter 1985), traded and untraded interdependencies (Scott 1988a; Storper 1997), and localized capabilities (Maskell and Malmberg 1999). Regional economic development can thus be seen as the historical product of the combination of layers of activities, consisting of economic, social, cultural, institutional, and political strata (Massey 1984). In the past few years, an “evolutionary turn” has begun to emerge in economic geography (Boschma and Frenken 2006), and this turn, along with one of its core concepts—path dependence—has emerged to reframe how we understand regional economic change and the evolution of clusters, industrial districts, and other localized forms of industrial specialization (Hassink 2007; Martin and Sunley 2006; Maskell and Malmberg 2007). Some scholars have re-examined their previous studies of regional development and its association with local knowledge spillovers, the cultivation of a local pool of specialized labor, local interfirm synergies and the division of labor, and institutional contexts, by employing “evolutionary” metaphors, concepts, and terminology (Bathelt and Boggs 2003; Gertler 2005; Hassink 2005; Hassink and Shin 2005). In the wake of the evolutionary turn, economic geography has witnessed lively debates on path dependence and the determinants of path creation (Boschma and Frenken 2006). This strand of literature on evolutionary economic geography can be further divided into two main camps. One camp can be traced back to some works in traditional economic geography on path dependence, lock-in, learning, and their roles in the development of industrial clusters (Grabher 1993; Hassink and Shin 2005; Rigby and Essletzbichler 1997; Storper 1997; Storper and Walker 1989). According to the key findings of the first camp’s research, industrial clustering has long been recognized as important in the ways in which firms are able to develop interfirm linkages, expand their forms of interaction, and deepen their networks of trust (Barnes and Gertler 1999; Storper 1997). In this view, geographical clustering of industrial activities positively affects competitiveness, and studies of such clustering and networking effects have been vital in reshaping institutional approaches to economic geography. However, recent studies focusing on the role of path dependence and lock-in have begun to also focus on the negative effects of clustering, particularly in explaining the decline of old industrial districts or clusters and the inability of firms in them to respond adequately (Cho and Hassink 2009; Grabher 1993; Schamp 2005; Stam and Martin 2012; Yang 2012). As Grabher (1993, p. 256) argued much earlier, “the initial strengths of the industrial districts of the past—their industrial atmosphere, highly developed and specialized infrastructure, the close inter-firm linkages, and strong political support by regional institutions—have, in some cases, turned into stubborn obstacles to innovation.” Such obstacles have given rise to what Hassink and Shin (2005) have termed the “rigid specialization” trap, where geographically concentrated clusters become insular and inward-looking systems. Here, the notion of “lock-in” describes those situations where the processes

1.1 An “Evolutionary” Turn in Economic Geography

3

that cause path dependence gradually lead to increasing fixity or rigidities in the patterns of industrial activity (Arthur 1989; Martin and Sunley 2006). The line between successful and vigorous cluster development and insular, inward-looking and inflexible clusters can be very thin (Hassink and Shin 2005; Saxenian 1994). Indeed, although the idea of lock-in has usually been assigned a negative interpretation, path dependence and lock-in, broadly defined, may have both negative and positive effects on regional economic performance (Henning et al. 2013; Martin and Sunley 2006). On the one hand, lock-in may contribute to economic performance as increasing returns and positive externalities reinforce local industrial dynamism. On the other hand, the same forms of lock-in may subsequently contribute to economic decline as established structures and configurations that once offered positive effects to firms lead to growing fixity or rigidity. The thinness of the line between negative and positive lock-in has resulted in the rise and fall of regional economies and the restructuring of industrial areas (Hassink and Shin 2005; Stam and Martin 2012), and we now have several examples of where such negative effects have forced the restructuring of regional economies in Europe, North America, and Japan (Grabher 1993; Hassink 2007; Schamp 2005), the Asian newly industrialized economies since the mid-1990s (Cho and Hassink 2009; Van Grunsven and Smakman 2005), and emerging regional economies such in mainland China since the 2000s (Li et al. 2012; Wei et al. 2007, 2009). Even many of the classical industrial clusters in Italy have experienced serious recent declines (Dunford 2006; Dunford and Greco 2006, 2007). In short, the studies of this camp tend to examine industrial restructuring and rigidity, particularly in industrial clusters, by using qualitative methods and lay the theoretical foundation of EEG studies. The other camp covers a group of works on technological relatedness, cognitive proximity, and regional branching (Boschma and Iammarino 2009; Boschma et al. 2012, 2013; Frenken et al. 2007; Neffke et al. 2011). Since the seminal work of Boschma (2005), there has been an increasing awareness that geographical proximity is neither a necessary nor a sufficient condition for spurring knowledge diffusion between firms. Empirical work in economic geography and regional science has confirmed that the right extent of cognitive distance among economic agents must be in place for knowledge spillovers to occur (Boschma 2005; Nooteboom 2000). The notion that an optimal level of cognitive distance may exist in knowledge spillovers and sharing predicates on Nooteboom’s (2000) work, who claims that “information is useless if it is not new, but it is also useless if it is so new that it cannot be understood” (p. 153). Hence, the role of geographical proximity in determining economic phenomena is complemented and enriched by the notion of cognitive proximity (Lo Turco and Maggioni 2015). This body of literature has emphasized the role of technological relatedness for creating or attracting new industrial sectors in a region (Boschma and Iammarino 2009; Boschma et al. 2012; Neffke et al. 2011) and argued that long-term regional industrial diversification demonstrates strong path dependence. Frenken et al. (2007) and Boschma and Iammarino (2009) have divided Jacobs’s externalities into related and unrelated variety and argued that Jacobs’s externalities do not necessarily result in knowledge spillover; instead, it only takes place effectively when

4

1 Introduction

complementarities and technological relatedness exist among industrial sectors in terms of shared competences. It is thus argued that related variety has the potential to provide more learning opportunities for local industries, generate more inter-sectoral knowledge spillovers, and finally facilitate regional economic development (Boschma and Iammarino 2009; Boschma et al. 2012, 2013; Frenken et al. 2007; Neffke et al. 2011). On the other hand, in addition to driving growth of existing industries in the short term, relatedness also plays a key role in new path creation in the long term. The gist of their argument is that new regional growth paths do not start from scratch, but are strongly rooted in the region’s historical industrial structure (Neffke et al. 2011). This idea is also supported by Maskell and Malmberg (2007) who have maintained that, from a microlevel perspective, a region can be seen as having a memory that directs the path of subsequent development. This new perspective based on technological relatedness, defined as interactive learning among economic actors stemming from the existence of a certain degree of cognitive proximity among them (Boschma and Iammarino 2009; Boschma et al. 2012, 2013; Frenken et al. 2007; Neffke et al. 2011), has given new momentum to debates in economic geography on the relative importance of specialization versus diversity economies in the mechanics of agglomeration externalities for firm performance and regional development (Lo Turco and Maggioni 2015). In addition to driving growth of existing industries through agglomeration externalities in the short term, relatedness also plays a key role in new path creation in the long term (Boschma and Capone 2015; Boschma et al. 2013; Delgado et al. 2016; Neffke et al. 2011). It is argued that new industries/paths do not originate in purely random events, but emerge out of preexisting industrial structures through technologically related localized knowledge spillovers (Boschma and Frenken 2011; Boschma and Martin 2007). Regions often diversify into new industries through a process of branching, where, driven by technological relatedness, a new industrial sector spawns either from related ones (Klepper and Simons 2000) or from the recombination of capabilities from multiple related sectors (Klepper 2002; Tanner 2016). Empirical studies also confirm that it is easier for regions to create new industries that are related to the existing set of industries (Binz et al. 2016; Colombelli et al. 2014; Simmie 2012; Tanner 2014). The idea that relatedness between economic activities, understood as cognitive proximity, has an impact on the direction of regional economic development was first testified at the national level (Hidalgo et al. 2007). Boschma et al. (2013) have further shown that relatedness to the regional industrial structure plays a much larger role in the emergence of new comparative advantage industries in regions than does relatedness to the national industrial structure. His work also implies that geographical and cognitive proximity complement with one another in promoting knowledge sharing. As a result, regions with a wide range of related industries are more likely to enjoy higher economic growth rates (Boschma et al. 2012). By analyzing the economic evolution of 70 Swedish regions during 1969–2002, Neffke et al. (2011) have demonstrated that technological relatedness is important in rising regions’ technological cohesion. Industries have a higher probability of entering regions where the regional industrial structure is related to that industry, whereas existing

1.2 The Case of China

5

industries unrelated to the region’s industry are more likely to exit. The concept of technological relatedness has become an important element in understanding local knowledge spillovers, the creation of new varieties, and the emergence of new regional growth paths (Boschma and Frenken 2011; Neffke et al. 2011). Relatedness between economic activities also matters at the firm level and is found to affect a firm’s economic performance. Firm growth, indeed, can be seen as a process of exploitation of productive opportunities. Firms are unique bundles of resources where firm-specific abilities are combined with product-specific competences related to the production of a particular good (Lo Turco and Maggioni 2015). With respect to product-specific competences, firms that manufacture products closely related to the regional industrial structure should grow faster, as they have access to more productive opportunities by capitalizing on localized capabilities (Maskell and Malmberg 1999). Analogous to the national- and regional-level effect, we expect high level of relatedness between a firm’s products and the regional industrial structure to give rise to a greater firm performance (Poncet and Starosta de Waldemar 2015). For instance, by examining the evolution of the car industry in Britain during 1895–1968, Boschma and Wenting (2007) found that the presence of related industries has a positive impact on the survival of automobile firms. Neffke et al. (2012) have shown that plant survival is positively and significantly affected by technologically related localization externalities; plants related to other local plants are more likely to maintain their competitive advantage and less likely to fail. The positive impact of relatedness between firms and the local pool of knowledge over firm performance is also confirmed by Boschma et al. (2009) who focused on the relationship between intra-regional-related skill mobility and firms’ productivity growth. These firm-level studies, thus, suggest that technologically related knowledge spillovers do not only play a role in national and regional economic development but are also a key driver of firm innovation and performance.

1.2

The Case of China

Working with this strand of literature that call for more attention directed toward key concepts such as path dependence, technological relatedness, and cognitive proximity as well as their relationship with national and regional economic development, new path creation, firm performance, and industrial and firm dynamics, this book extends this debate on how studies in the Chinese context can contribute to current debates in EEG. Here, we draw upon our recent work to argue that the rapidly changing economic geographies in China may serve as one of the key sites to reproduce the literature on EEG, or to “theorize back” (Yeung 2007) at mainstream, Western theories of economic-geographic dynamics in specific locales ranging from macro-regions and nation-states to subnational regions and cities. The sheer pace and rapidity of regional dynamics in China provide a fertile ground for new conceptual development, methodological innovation, and empirical insights.

6

1 Introduction

Studying and interpreting China’s economic-geographic dynamics will definitely allow us to advance new knowledge and move the subject matter forward. Yeung (2007) has pointed out that to “theorize back” at mainstream, Westerncentric economic geography, we can either propose original theory that emanates from research on sites outside Western countries or remake key economic geography concepts and ideas in light of new insights from China. In this book, we provide examples on the China case’s contribution to Western theories in economic geography. Recent studies on the economic geographies of manufacturing in China have pointed out the need to take into account some factors that are often neglected in traditional analyses based on EEG. First, one issue that has not received much attention in the public debate as well as in the academic literature is the role of institutional context in regional industrial diversification. Institutions are responsible for regulating learning processes, supporting the formation of mutual trust, and facilitating the transmission of knowledge (Rodríguez-Pose 2013). Regions with high-quality institutions are found to perform better in terms of knowledge spillovers and innovation (Rodríguez-Pose 2013; Rodríguez-Pose and Di Cataldo 2015). Hence, regional industrial diversification that relies on technological relatedness and knowledge spillovers should be also affected by institutions (Cortinovis et al. 2016). Boschma and Capone (2016) have pointed out that recent EEG studies pay little attention to the differences that the industrial diversification process can display across regions and, more importantly, to the possible effect of regional institutions on the intensity and nature of the industrial diversification process. Such an underestimation of regional variations of institutional frameworks is problematic (Rodríguez-Pose 2013), particularly in transitional economies like China where economic, fiscal, and political reform has resulted in enormous spatial variations of economic and institutional landscape (He et al. 2008). Since the reform, China has gradually transformed from a state-owned, collective economy dominated by SOEs to one with growing level of private ownership and market orientation (marketization) and from a centrally planned to a decentralized economy (decentralization) (He et al. 2008). Such transformations have fundamentally changed China’s overt and covert rules, further resulting in a process of institutional transformations since institutions develop upon the former. The process of economic marketization and privatization of selected state-owned sectors not only lifted the restrictions on factor mobility and commodity exchanges (i.e., formal institutions) but also changed public and private economic actors’ attitude toward the market economy (i.e., informal institutions). This generated a favorable, nurturing environment for industrial linkages, knowledge spillovers, and entrepreneurial activities, further stimulating the formation of new industries. As a result, technological relatedness started to play an increasingly crucial role in regional industrial evolution (Guo and He 2017). Since institutions develop in relation to certain rules, regulations, laws, norms, and conventions that have a territorial basis, the resulting formal and informal institutions are often context- and place-specific (Bathelt and Glückler 2014). In our case, the process of economic liberalization is by no means regionally balanced

1.2 The Case of China

7

(Han and Pannell 1999), and national-level rules cannot override all else (see also (Gertler 2010) for a discussion on the inherent dangers of “methodological nationalism” in some studies). China is characterized by substantial regional disparities in institutions such as de facto property rights protections, government intervention in business operations, law enforcement capability, and contract enforcement (Du et al. 2008; Jiang and Murmann 2012). Generally speaking, coastal regions that have been one step ahead in reforms have become more economically liberalized with stronger market forces at work, while in inland China, where marketization and privatization have been implemented half-heartedly by various levels of governments, production is still heavily reliant on SOEs and affected by the state’s social, political, and military considerations. In more marketized regions, firms can have easy access to credits on financial markets and by venture capitalists that favor the establishment of new firms in new fields (Boschma and Capone 2015). Market-based interfirm relations allow firms to enter new industries by acquiring already established firms or by licensing new products. Weakly regulated product markets reduce probability of legal obstacles to the introduction of new products. It is thus expected that in regions with high levels of marketization, firms are more likely to create new paths. In addition, even regions inside coastal and inland China vary from one another in terms of their formal institutions based on laws and regulations regarding market orientation, as well as their informal institutions developing upon social attitudes toward the market economy and private sectors. Such place specificity has been reinforced by the other transformation introduced by the reform—decentralization— which enabled Chinese localities to pick different development routes. Specifically, it granted local governments more autonomy and allowed them to get involved in shaping the regional economy, as planners, developers, and policy-makers (He et al. 2008; Wei et al. 2007). Fiscal federalism theory believes fiscal decentralization would promote growth by transferring powers and resources to subnational administrations which would increase allocation efficiency by better matching policies to local preference and production efficiency through interjurisdictional competition (Peterson 1981). In the context of China, decentralization is a much more complex process, where local officials have become revenue maximizers with strong incentives to intervene regional economic development (Montinola et al. 1995; Oi 1995). Such a regionally decentralized authoritarian system is largely driven by the combination of political centralization—where high-level officials in the hierarchical political system promote low-level officials based on their economic performance— and economic regional decentralization, where local administrations have to selffinance their development (Hu and Hassink 2016; Xu 2011). This has further created a GDP-based interjurisdictional competition, also known as “tournament competition,” as local officials compete against one another in boosting regional economic development in order to maximize their chances of political promotion (Pan et al. 2016; Yu et al. 2016). Although all local authorities attempt to adopt pro-business policies in order to create new industries, some in China’s wealthy coastal regions are often more capable in providing a variety of subsidies, tax credits, and fiscal and administrative supports than their counterparts in less developed inland China. Local authorities do not produce immediate effects on local industrial dynamics, but rather

8

1 Introduction

through influencing and shaping institutions as “mediators” of economic practice and interaction (Bathelt and Glückler 2014; Hu and Hassink 2016). Such regional variations of institutional arrangements have been named by Hu and Hassink (2017) as “leadership of context” and must be taken into account to better understand China’s regional industrial dynamics. Financial, political, and technological supports provided by governments could reduce firms’ cost of entering new, and even unrelated, industries, triggering more dynamic regional industrial diversification (Cortinovis et al. 2016). Second, we need to acknowledge that although interfirm technologically related knowledge spillovers are important in the evolution of regional industrial structure, firm heterogeneity plays a role in firms’ choice of new industries (Neffke et al. 2014). In the canonical EEG model, a regional economy tends to be treated as a homogenous entity that form the unit of analysis, and there is a tacit assumption that all cluster firms are relatively homogenous and that they do not merit attention in their own right (Martin 2010). Martin (2010) has, however, pointed out that regional economies are composite systems, consisting of numerous heterogeneous firms with different ownership structures, market orientations, specific technologies, competences, resources, and business models, even though the firms may all belong to the same industry. Homogeneity is rare, and firms in a locality often vary dramatically in their innovative and productive performance and developmental trajectory (Antonelli and Scellato 2015). Since firm-specific competences, organizational routines, and knowledge bases affect the ways in which and the extents to which firms interact with each other, it would be problematic to examine the effect of technologically related knowledge spillovers on firm performance without paying attention to firm heterogeneity (Almeida and Kogut 1997; Lo Turco and Maggioni 2015; Martin 2010). For instance, (1) large firms may be particularly relevant to new industry creation and tend to be more path-breaking due to their scale and access to larger financial resources compared to small firms (Lecocq and Van Looy 2016). By creating local niches and/or intermediary markets, large firms may also encourage entrepreneurial activities in the region and attract high-quality suppliers. (2) High-productivity firms are more able to adopt the new technological opportunities offered by a disruptive, path-breaking technology and thus tend to play a much more important role in the development of new technologies/industries (Lecocq and Van Looy 2016). (3) Firms also differ in their access to regional resource base. Incumbent firms can access regional resource more easily and are more likely to build on regional existing capabilities and assets (Neffke et al. 2014). Hence, they are less likely to be pathbreaking and to create unrelated industries than new establishments. (4) In the context of China, after the reform, economic marketization has resulted in the co-existence of state-owned enterprises (SOEs) and privately owned enterprises (POEs), while foreign-owned firms (FOEs) have been also swarming into China driven by China’s Opening-up policies and its integration into the global economy (He et al. 2008; Yu et al. 2015). Such a firm-level variation in terms of ownership structure also affects firms’ abilities to explore new production fields. Specifically, SOEs may be more capable to diversify into new industries since they enjoy

1.2 The Case of China

9

institutional advantages such as preferential access to favorable policies, business information, and government subsidies (Bai et al. 2004; Sun and Tong 2003). FOEs may be not only more likely to introduce new industries to Chinese regions, but also more path-breaking, introducing less related industries, since they have access to more advanced know-how and ideas in the international market (Amighini and Sanfilippo 2014). The key role of FOEs is also important in the third point. By emphasizing endogenous regional branching and industrial diversification processes (Grillitsch and Trippl 2014; Tanner 2014), EEG risks embracing a regional fetishism (Binz et al. 2016; Martin and Sunley 2006), as it overlooks extra-regional linkages that may contribute to the renewal and restructuring of regional resource bases and the formation of specific regional growth paths (Bathelt and Cohendet 2014; Bathelt et al. 2004; Maskell 2014; Maskell et al. 2006). Extra-regional linkages which connect actors inside and outside the region may be important in enabling firms in the region to avert tendencies toward path dependence in the evolution of regional economy, thus enabling the region to remain innovative and competitive (Bathelt and Li 2014; Bathelt et al. 2004; Sydow et al. 2010). By deliberately investing in building extra-regional linkages to distant communities, regions may be able to increase the variety of knowledge, resources, and capabilities available to them. As China’s reform policies pushed forward the process of marketization in the economy, the opening-up policies globalized China’s economy and allowed Chinese enterprises to access financial capital, more advanced technologies, knowledge and management skills, and high-quality input factors on the international market (He et al. 2008; Wei et al. 2007). Enterprises benefitted from globalization as they learned from foreign direct investment (FDI) that started to swarm into China since the 1980s, on the one hand, and were increasingly encouraged to established extraregional linkages with firms in the North by participating in global production networks, on the other hand. Such knowledge spillover from global lead firms to local firms is thus likely to lower barrier for new path creation. FDI, as one type of extra-regional linkages, plays a critical role in promoting regional economic development in China through a variety of channels, such as the formation of forward and backward linkages, the existence of competitive and demonstration effects, the possibility for domestic firms to recruit more experienced and skilled workers that are released from foreign-owned firms, and finally the knowledge spillover effects between domestic- and foreign-owned firms (Görg and Greenaway 2004; Lall and Narula 2004; Poncet and Starosta de Waldemar 2013; Zhu and Fu 2013). FOEs are important for regional economy, as they not only contribute to productivity increase in existing industries but, more importantly for our present purpose, they also bring in new knowledge and ideas that may enable regions to break their old paths and diversify into new industries (Amighini and Sanfilippo 2014). Another example of extra-regional linkages—imports—expands the set of inputs available in the economy and thus increases regions’ productivity (Amighini and Sanfilippo 2014). The rising availability of inputs may encourage the creation of new domestic varieties (Goldberg et al. 2010). Imports can also provide more sophisticated inputs that enable regions to upgrade their production and export.

10

1 Introduction

More importantly, there is a certain degree of new knowledge embedded in imported products, which could translate into new learning opportunities involved in the use of new products (Dollar 1992; Schiff and Wang 2006). As a result, extra-regional linkages should be stressed in China’s regional industrial dynamics. In examining these issues, our book provides the first book-length, systematic treatment of the articulations between path dependence, path-breaking, technological relatedness, firm heterogeneity, and institutional contexts in China. It will be the first book examining the role played by technological relatedness and more importantly China’s peculiar institutional arrangements in regional industrial diversification. Specifically, this book contributes to extant literature in several ways: 1. The book provides the first detailed account of the complex geographical dynamics restructuring China’s manufacturing industries from the evolutionary economic geography perspective. 2. The geographical and industrial shifts analyzed in this book have enormous implications in and beyond China for what is possible in the post-crisis global economy. 3. This book is to demonstrate that the interface between evolutionary economic geography approaches and other approaches (e.g., institutional economic geography) could be a fertile area for further consideration. 4. The importance of this research lies mainly in its attempt to understand the different and interconnected roles of government policies and firm strategies and their effects on regional economic development. The two main audiences that this book appeals to are economic geographers and regional scientists. The topics covered in the book are also relevant to development studies, economics, economic sociology, and international studies, offering academics, international researchers, and post-graduate and advanced undergraduate students in these fields an accessible, grounded, yet theoretically sophisticated account of the evolutionary economic geography in China and its interaction with firm performance and regional economic development. The book is also attractive to national policy-makers, since it engages directly with economic and industrial policy issues, such as industrial competitiveness, regional and national development, industrial and employment restructuring, and trade regulation.

1.3 1.3.1

Methodology Data Sources

One database on firm-specific economic and financial variables is central to the firmlevel analysis in this research: China’s Annual Survey of Industrial Firms (ASIF) (1998–2009). The study time period from 1998 to 2009 is critical in terms of the development of China’s manufacturing industry. This time period is often described as a turning point laden with a variety of far-reaching events which have potentially

1.3 Methodology

11

transformed China’s manufacturing industry in fundamental ways, such as China’s entry into the WTO in 2001, the appreciation of China’s currency since 2005, and rising production costs along China’s coastal area since the early 2000s. The 2007–2008 global financial crisis which has stimulated Chinese industrial restructuring also falls into this time period, though its complete effects may take more than 2 or 3 years to be seen. The ASIF is administered by the National Bureau of Statistics of China and covers all Chinese industrial state-owned enterprises and non-state-owned enterprises with annual sales of 5 million RMB or more. The database provides firm-level data on firm structure and operation, including firm identification, location, capital structure, total profits, total shipments, exported shipments, intermediary inputs, asset value, inventory, employment, sales value, type of investment, output, value added, R&D expenses, education and training of staff, and wages, social insurance, and benefits paid. A comparison with the 2004 full census of industrial firms reveals that these firms employed roughly 70% of the industrial workforce and generated 90% of output and 98% of exports. Because this dataset suffers some problems, such as missing data on indicators, vague definition of variables, and measurement errors, we have adopted a systematic method, developed by Brandt et al. (2012), to clean this dataset and delete some incomplete items.1 This book also uses Chinese export data during 2002–2011 compiled by the Chinese Customs Trade Statistics (CCTS). The CCTS dataset records all merchandise transactions passing through Chinese customs and reports basic firm information (e.g., name, address, and ownership), export value and quantity, destination of exports, origin of imports, and processing or ordinary exports. The raw data contains a number of intermediary firms that mediate trade for other firms but do not directly engage in production. Their export behaviors are likely to be different from those of manufacturing firms. We exclude intermediary firms as our results may be distorted by these trading agents’ business networks. We follow Bernhofen et al. (2017) and use a list of keywords that are typically used by various types of intermediary firms in their names in China (e.g., “importer,” “exporter,” and “trading”). These intermediary firms represent around 4% of our observations.

1.3.2

Research Design

To investigate whether regional industrial, firm, and product dynamics have been affected by technological relatedness in China, we need an indicator to measure technological relatedness between industries. This indicator also reflects the distance between new industries and existing industrial structure, thus allowing us to observe whether regional industrial structure changes toward related or relatively unrelated industries. Boschma et al. (2012) have pointed out that the ex post relatedness

1

Please see Brandt et al. (2012) for further details.

12

1 Introduction

indicator developed by Hidalgo et al. (2007) based on proximity product index can better capture the essence of technological relatedness than can the conventional ex ante measure of related and unrelated variety (Boschma and Iammarino 2009; Frenken et al. 2007) and the cluster-based ex post indicator of industrial relatedness formulated by Porter (2003). Since a variety of factors may affect the relatedness between two industries, including similarities in the combination of productive factors, the use of technologies, the characteristics of customers, required institutions, and social norms, Hidalgo et al. (2007) have adopted an ex post approach to measure proximity between two industries. Two industries are considered to be related with each other if regions tend to have revealed comparative advantage (RCA) in both. A region has a comparative advantage in an industry when the share of this industry in the region’s total exports is larger than the share of this industry in the national total.2 The proximity (ϕ) between industry i and industry j in year t can be calculated as: 

 ϕi, j ¼ min P RCAc, i > 1jRCAc, j > 1 ; P RCAc, j > 1jRCAc, i > 1

ð1:1Þ

where: RCAc, i ¼

Export Valuec, i =

P i

Export Valuec, i

, 

P c

P Export Valuec, i = Export Valuec, i c, i



ð1:2Þ

RCAc,i is the revealed comparative advantage of industry i in city c. City c is considered as having a comparative advantage in industry i, if RCAc,i is above 1. The proximity between industry i and industry j is the minimum between the conditional probability of having a comparative advantage in industry i, given that the city c has a comparative advantage in industry j (i.e., P (RCAc,i > 1|RCAc, j > 1)), and the conditional probability of having a comparative advantage in in industry j, given a revealed comparative advantage in industry i (i.e, P (RCAc,j > 1| RCAc,i > 1)). The rationale behind this proximity indicator is that if two industries are related with each other, they probably demand similar institutions, infrastructure, factor inputs, capabilities, and technology and are likely to be produced together. The proximity indicator between industries is computed using CCTS dataset during 2002–2011. This dataset reports imports and exports by 6-digit product. In

2

We follow Neffke et al. (2011) and calculate proximity indicators based on data in one country. First, China is a large country with a high level of regional disparity, which means calculation based on data in such a big economy should be sufficient. Second, calculating proximity indicators based on Chinese export data rather than world trade data may allow us to better control China’s unique, nationwide political and economic environments.

1.3 Methodology

13

this book, we mainly focus on 4-digit level industries (1080 industries in the dataset).3 The geographical unit of analysis is China’s prefecture-level cities. A matrix of proximity indicators among all 4-digit industries can be estimated. This 1080*1080 matrix therefore defines the industry space. The RCA indicator measures export output specialization in international trade. In this book, we also use the location quotient (LQ) to measure domestic output specialization (Isard 1960). A region has a revealed locational advantage in an industry when the share of this industry in the region’s total employment is larger than the share of this industry in the national total. The proximity (ϕ) between industry i and industry j is calculated as: 

 ϕi, j ¼ min P LQc, i > 1jLQc, j > 1 ; P LQc, j > 1jLQc, i > 1

ð1:3Þ

where: LQc, i ¼

Employmentc, i =

P

Employmentc, i

,

i



P c

P Employmentc, i = Employmentc, i c, i



ð1:4Þ

and where Employmentc,i is the number of employees in industry i and city c. LQc,i is the location quotient of industry i in city c. City c is considered as having a revealed locational advantage in industry i, if LQc,i is above 1. The proximity between industry i and industry j is the minimum of the conditional probability of specializing in industry i, given that city c has an industrial specialization in industry j (i.e., P (LQc,i > 1|LQc,j > 1)), and the conditional probability of specializing in industry j, given a revealed locational advantage in industry i (i.e, P (LQc,j > 1|LQc, i > 1)). In this case, we use the ASIF dataset, based on Eqs. (1.3) and (1.4); a 424*424 matrix of proximity indicators for all 4-digit industries was estimated. To measure the distance between new industries and region’s existing industrial structure, the density indicator, developed by Hidalgo et al. (2007), was calculated. Recent empirical studies suggest that regional industrial diversification is pathdependent, because if a new industry is related to a number of industries in which the region is already specialized, the density of this new industry in the region is high, and the probability that this region will attract and create this new industry will also be high. The density indicator was therefore measured as follows: P

densityi, c

x j, c ϕi, j ¼ P ϕi , j j

j

3

We focus on the secondary industry and therefore exclude data on agriculture.

ð1:5Þ

14

1 Introduction

where xj,c,t takes the value of 1 if city c has a revealed locational advantage (or RCA) in industry j and zero otherwise. Density around a new industry will be high if a region is specialized in most of the industries related to the industry under consideration.

1.4

Synopsis of the Book

The rest of the book contains a conclusion chapter and 14 empirical chapters that have been organized in four parts: 1. Industrial dynamics: This part sets the stage by examining industrial dynamics, path dependence, and path creation in China from an evolutionary economic geography perspective. 2. Firm dynamics: this part then analyze the ways in which firm dynamics, such as firm survival, firm relocation, and firm entry have been shaped by technological relatedness as well as other factors. 3. Product dynamics: this part uses the export data to analyze the evolution of China’s product space. 4. Impact of industrial dynamics on regional inequality: this part seeks to understand the impact of regional industrial dynamics on regional inequality.

1.5

Part I: Industrial Dynamics

Chapter 2 How Has Production Space Evolved in China? A growing literature on EEG concludes that regional industrial evolution is pathdependent and is determined by the preexisting industries. This chapter more accurately calculates the industrial relatedness based on the co-occurrence approach to portray China’s production space and then examines the impact of industrial relatedness on regional industrial evolution. The findings report that industrial relatedness does underscore the regional structure change in China but shows significant regional differences in the evolution path. The coastal region has strong tendency of path dependence in its industrial evolution, while North West and South West break the path-dependent trajectory and transition into high productive sectors distant from their own production network. The results suggest that governmental policies can play its crucial role in creating new paths in the West. Institutions matter to allow the significant role of industry relatedness in driving regional industrial evolution. Chapter 3 How Does Regional Industrial Structure Evolve in China? EEG indicates that regional industrial development is path-dependent. The empirical studies in EEG however have not paid sufficient attention to the importance of global linkages nor the role of regional institutions in driving industrial dynamics. Based on

1.6 Part II: Firm Dynamics

15

firm-level data of 4-digit manufacturing industries during 1998–2008 in China, this chapter finds that Chinese regions branch into new industries technologically related to the existing industrial portfolio and related industries are less likely to exit. Further analysis reveals that global linkages, economic liberalization, and state involvement not only create favorable conditions to allow a larger role of technological relatedness but also generate opportunities for Chinese regions to create new paths of industrial development. Chapter 4 What Matters for Regional Industrial Dynamics in China? Recent EEG studies have argued that regional diversification emerges as a pathdependent process, as a region often branches into industries that are related to preexisting industrial structure, whereas industries that are not closely related have a high probability of exiting the region. This chapter contributes to the ongoing debates on path dependence and path creation with a new analytical framework that emphasizes the need to bring in a wider range of factors that contributes to regional industrial evolution. It suggests that regional industrial dynamics are not only conditioned by preexisting regional capabilities and technological relatedness but also by the ways in which technological relatedness is interconnected with industry attributes and more importantly region’s institutional context. Based on a firm-level dataset of China’s manufacturing industries during 1998–2008, this chapter has studied regional industrial evolution in China by examining the entry and exit of 4-digit industries at the prefectural level. It shows that regional industrial development is a path-dependent process in China, one that is also inflected by industry characteristics and institutional framework. Our results imply that EEG would profit from incorporating insights on institutional change and industry characteristics in its explanations of regional industrial evolution.

1.6

Part II: Firm Dynamics

Chapter 5 What Facilitates New Firm Formation in China? Why do some regions occur more entrepreneurial than others? This chapter explores the determinants of new firm formation at the prefectural city level in China by highlighting the influence of localization and urbanization economies and the significance of technological relatedness and small firm clusters. Descriptive analysis reports significant and increasing spatial variation of new firm formation in China during 2001–2007. The empirical results based on negative binomial model provide evidence to support the business network view of new firm formation. Localization economies can predict new firm formation well, while the effect of urbanization economies is mixed. In terms of localization economies, supplier-customer linkages play a very important and positive role in cultivating new firm formation. The mixed results of urbanization economies are mainly derived from the interweaving of related variety and unrelated variety. The former significantly promotes new firm formation, while the latter in most cases

16

1 Introduction

discourages it. The clustering of small firms has a larger effect on new firm formation, which is consistent with the view of Vernon and Chinitz effect. Chapter 6 Does Creative Destruction Work for Chinese Regions? Creative destruction is a key driving force behind industrial development. The continuing process of creative destruction provides an impetus to regional industrial renewal. Our analytical framework that emphasizes the ways in which firm exit creates a stimulus for firm entry is complementary to the process of technological change and industrial renewal articulated by Schumpeter who pays attention to how new entrants bring in radical innovation and new products, making incumbents’ products and technologies obsolete and forcing them to exit or catch up. Using firmlevel data of China’s industries during 1998–2008, this chapter seeks to argue that the relationship between firm exit and entry has been constantly shaped by an assemblage of various factors, including not only firm characteristics but also industrial linkages, and, most importantly, national and regional institutional contexts, particularly in the context of China where a triple process of decentralization, globalization, and privatization has resulted in enormous spatial and temporal variation of economic and institutional landscape. Chapter 7 What Causes Firm Failure in China? This chapter is to investigate the determinants of firm failure in China. Based on the annual survey of industrial firms during 1998–2007, it first describes the patterns of firm failure. On average, less productive and older firms are more likely to fail, while firms with governmental supports are more likely to survive. Statistical results based on the linear probability model for panel data indicate that market competition crowd out the less productive firms. Competition dominates learning effects and imposes challenges on the survival of older firms. There is an inverted U-shaped relationship between firm age and firm failure. Local protectionism and supportive policies can reduce the chance of firm failure. However, governmental intervention can generate negative externality, reducing the survival chances of firms without supports. The findings from this chapter enrich our understanding of industrial dynamics in the transitional China. Chapter 8 What Sustains Large Firms in China? This chapter investigates what sustains large firms in China and identifies the determinants of firm sustaining. With the understanding of the triple process of economic transition, it explores the influence of global-local interaction and regional factors on business sustaining. Based on the Annual Survey of Industrial Firms in China during 1998–2005, this chapter employs the Cox proportional hazard model to confirm that global, provincial, and local forces are critical for the sustaining of large firms. Particularly, the presence of foreign firms shows strong competition effect at the prefecture city level but no spillover effect at the provincial level. Provincial market-oriented institutions and market potential however are crucial to sustain businesses in China. Non-state-owned enterprises such as private and foreign firms are more dependent on market-oriented institutions than state own enterprises. Firms which are able to reap from agglomeration economies and local governmental

1.6 Part II: Firm Dynamics

17

supports are more likely to sustain. In addition, local factors show different impacts on business sustaining in different regions. The findings indicate that both market and state can play a substantial role in sustaining businesses in transitional economies. Chapter 9 How Do Agglomeration Economies Contribute to Firm Survival in China? This chapter studies empirically the effects of five different dimensions of agglomeration—specialization, diversity, related variety, unrelated variety, and city size— on the survival chances of new entrepreneurial firms in China. Consideration is further given to studying the mediating effects of local subsidies on new firm survival given different existing local industrial structures in those regions. In support of the “regional branching” hypothesis, this chapter finds that increasing local-related variety has a stronger positive effect on new firm survival than other types of agglomeration. Receiving comparatively fewer subsidies motivates firms to seek out and benefit from local existing economies, which, in turn, positively influence their chances of survival. By contrast, agglomerated firms that receive relatively more subsidies tend to be more likely to face financial distress leading to eventual market exit. The findings thus reveal that both the intensity and the location of state support matters in terms of optimizing positive agglomeration effects on firms’ post-entry performance and survival. Chapter 10 How Does Geese Fly Domestically? Firm Demography and Spatial Restructuring in China’s Apparel Industry Using a firm-level database from 1999 to 2008, this chapter sheds new light on the industrial dynamics of China’s clothing industry, highlighting the multi-scalar process of spatial restructuring primarily driven by differences in the rate of job creation, especially through start-up firms. At the interprovincial level, 2004 represents a turning point where the industry ceased to be concentrated in coastal areas and began to relocate inland, especially to central provinces. At the intra-provincial level, cities in coastal provinces remained the most attractive locations for start-up companies even after 2004. The results suggest that the regionally decentralized authoritarian regime, intertwined with market forces shaping the multi-scalar process of spatial restructuring, is pivotal to the understanding of the changing geography of China’s apparel industry. Chapter 11 How Do Environmental Regulations Affect Industrial Dynamics in China? Pollution haven hypothesis (PHH) and porter hypothesis (PH) offer two different perspectives to understand the relationship between industrial dynamics and environmental regulations. This chapter seeks to move beyond existing studies that are based on either the PHH or the PH while neglecting the other, toward an analytical framework that not only pays more attention to the ways in which the PHH and the PH co-exist but also acknowledges the role of firm heterogeneity and local government intervention. Based on a firm-level industrial dataset and a dataset on China’s polluting firms, this chapter studies the relationship between environmental

18

1 Introduction

regulations and industrial dynamics in China’s pollution-intensive industries at the firm level. Empirical results confirm the co-existence of the PH and the PHH. Furthermore, firm heterogeneity and government intervention both have the potential to inflect the relationship between environmental regulations and industrial dynamics.

1.7

Part III Product Dynamics

Chapter 12 How to Jump Further? Path Dependence and Path-Breaking in an Uneven Industry Space By using the proximity index, recent studies have argued regional diversification emerged as a path-dependent process. Developed countries that start from core, dense areas in the uneven industry space have more opportunities to jump to new related industries and to sustain economic growth than do developing countries that jump from peripheral, deserted areas. Can developing countries/regions jump further in the industry space to break path-dependent development trajectories? Based on China’s export data, this paper shows that regions can do so by investing in extraregional linkages and internal innovation. The effects of these two sets of variables vary across regions and industries. Chapter 13 What Drives the Evolution of Export Product Space in China? Regions tend to develop new products that are related to preexisting industrial structure. This stresses the critical role of relatedness in regional industrial diversification. Governments can also strongly affect the evolution of regional industrial structure by using a variety of industrial policies. Based on China’s custom productlevel data, this chapter explores the evolution of China’s export product space during 2000–2011. Econometric analyses suggest that the evolution of export product space in China is heavily driven by relatedness between export products, indicating that regional industrial diversification is path-dependent. Chinese national and local governments can exert significant influence on the evolution of export products by using export- and FDI-related policies and by establishing development zones. More importantly, national and local governments have played conflicting roles in China’s regional industrial dynamics.

1.8

Part IV Impact of Industrial Dynamics on Regional Inequality

Chapter 14 How Do Firm Dynamics Affect Regional Inequality of Productivity in China? This chapter focuses on the issue of regional inequality. It seeks to move away from focusing on the relationship between factor accumulation and regional inequality toward an analytical framework that not only pays more attention to the ways in

References

19

which regional inequality has articulated with factor productivity and firm dynamics but also acknowledges the role of institutional context. Based on one database on firm-specific economic and financial variables, this chapter measures factor productivity in China’s manufacturing industry at the regional level, decomposing it into various components. Empirical results confirm that regional inequality in terms of productivity has been declining in China. It also shows that regional inequality has been fundamentally shaped by regional institutional context, particularly in the context of China where a triple process of decentralization, globalization, and marketization has resulted in enormous spatial variation of economic and institutional landscape. Chapter 15 Summary and Conclusion The conclusion brings together the key arguments and returns to the core questions that inform the analysis. In this chapter, we argue that in addition to technological relatedness, a wide range of factors are also important and have co-shaped industrial restructuring and regional economic development in China alongside technological relatedness. Furthermore, we call for a more nuanced reading of the Western theories based on studies in developed countries. Regional industrial dynamics should be analyzed in ways that pay special attention to the articulation between governments and firms, firm heterogeneity, and the wider historical, political, institutional, economic, and social context.

References Almeida, P., & Kogut, B. (1997). The exploration of technological diversity and geographic localization in innovation: Start-up firms in the semiconductor industry. Small Business Economics, 9(1), 21–31. Amighini, A., & Sanfilippo, M. (2014). Impact of south–south Fdi and trade on the export upgrading of African economies. World Development, 64(0), 1–17. Amin, A., & Thrift, N. (Eds.). (1994). Globalization, institutions, and regional development in Europe. Oxford: Oxford University Press. Antonelli, C., & Scellato, G. (2015). Firms size and directed technological change. Small Business Economics, 44(1), 207–218. Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal, 99(394), 116–131. Bai, C.-E., Du, Y., Tao, Z., & Tong, S. Y. (2004). Local protectionism and regional specialization: Evidence from China’s industries. Journal of International Economics, 63(2), 397–417. Barnes, T., & Gertler, M. (1999). The new industrial geography: Regions, regulation and institutions. London: Routledge. Bathelt, H., & Boggs, J. S. (2003). Toward a reconceptualization of regional development paths: Is Leipzig’s media cluster a continuation of or a rupture with the past? Economic Geography, 79 (3), 265–293. Bathelt, H., & Cohendet, P. (2014). The creation of knowledge: Local building, global accessing and economic development—Toward an agenda. Journal of Economic Geography, 14(5), 869–882. Bathelt, H., & Glückler, J. (2014). Institutional change in economic geography. Progress in Human Geography, 38(3), 340–363.

20

1 Introduction

Bathelt, H., & Li, P.-F. (2014). Global cluster networks—Foreign direct investment flows from Canada to China. Journal of Economic Geography, 14(1), 45–71. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. Bernhofen, D., Upward, R., & Wang, Z. (2017). Quantity restrictions and price adjustment of Chinese textile exports to the us. The World Economy, 1. Binz, C., Truffer, B., & Coenen, L. (2016). Path creation as a process of resource alignment and anchoring: Industry formation for on-site water recycling in Beijing. Economic Geography, 92 (2), 172–200. Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional Studies, 39(1), 61–74. Boschma, R., & Capone, G. (2015). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, 44(10), 1902–1914. Boschma, R., & Capone, G. (2016). Relatedness and diversification in the European Union (Eu-27) and European Neighbourhood policy countries. Environment and Planning C: Government and Policy, 34(4), 617–637. Boschma, R., & Frenken, K. (2006). Why is economic geography not an evolutionary science? Towards an evolutionary economic geography. Journal of Economic Geography, 6(3), 273–302. Boschma, R., & Frenken, K. (2011). Technological relatedness and regional branching. In H. Bathelt, M. Feldman, & D. F. Kogler (Eds.), Beyond territory: Dynamic geographies of knowledge creation, diffusion and innovation (pp. 64–81). London: Routlege. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., & Martin, R. (2007). Editorial: Constructing an evolutionary economic geography. Journal of Economic Geography, 7(5), 537–548. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., Eriksson, R., & Lindgren, U. (2009). How does labour mobility affect the performance of plants? The importance of relatedness and geographical proximity. Journal of Economic Geography, 9(2), 169–190. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new Industries at the Regional Level in Spain: A proximity approach based on product relatedness. Economic Geography, 89(1), 29–51. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Cho, M., & Hassink, R. (2009). Limits to locking-out through restructuring: The textile industry in Daegu, South Korea. Regional Studies, 43(9), 1183–1198. Colombelli, A., Krafft, J., & Quatraro, F. (2014). The emergence of new technology-based sectors in European regions: A proximity-based analysis of nanotechnology. Research Policy, 43(10), 1681–1696. Cortinovis, N., Xiao, J., Boschma, R., & van Oort, F. (2016). Quality of government and social capital as drivers of regional diversification in Europe. Papers in evolutionary economic geography, 16.10, Utrecht University. Delgado, M., Porter, M. E., & Stern, S. (2016). Defining clusters of related industries. Journal of Economic Geography, 16(1), 1–38. Dollar, D. (1992). Outward-oriented developing economies really do grow more rapidly: Evidence from 95 Ldcs, 1976-1985. Economic Development and Cultural Change, 40(3), 523–544. Du, J., Lu, Y., & Tao, Z. (2008). Fdi location choice: Agglomeration vs institutions. International Journal of Finance & Economics, 13(1), 92–107.

References

21

Dunford, M. (2006). Industrial districts, magic circles, and the restructuring of the Italian textiles and clothing chain. Economic Geography, 82(1), 27–59. Dunford, M., & Greco, L. (2006). After the three Italies: Wealth, inequality and industrial change. Oxford: Blackwell. Dunford, M., & Greco, L. (2007). Geographies of growth, decline and restructuring: The rise and fall (Privatization) of the state-owned steel sector and the trajectories of steel localities in the Italian Mezzogiorno. European Urban and Regional Studies, 14(1), 27–53. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Gertler, M. (2005). Tacit knowledge, path dependency and local trajectories of growth. In G. Fuchs & P. Shapira (Eds.), Rethinking regional innovation and change: Path dependency or regional breakthrough? (pp. 23–42). New York: Springer. Gertler, M. S. (2010). Rules of the game: The place of institutions in regional economic change. Regional Studies, 44(1), 1–15. Goldberg, P. K., Khandelwal, A. K., Pavcnik, N., & Topalova, P. (2010). Imported intermediate inputs and domestic product growth: Evidence from India. The Quarterly Journal of Economics, 125(4), 1727–1767. Görg, H., & Greenaway, D. (2004). Much ado about nothing? Do domestic firms really benefit from foreign direct investment? The World Bank Research Observer, 19(2), 171–197. Grabher, G. (1993). The weakness of strong ties; the lock-in of regional development in the Ruhr area. In G. Grabher (Ed.), The embedded firm: On the socioeconomics of industrial networks (pp. 255–277). London: Routledge. Granovetter, M. (1985). Economic action and social structure: The problem of embeddedness. American Journal of Sociology, 91(3), 481–510. Grillitsch, M., & Trippl, M. (2014). Combining knowledge from different sources, channels and geographical scales. European Planning Studies, 22(11), 2305–2325. Guo, Q., & He, C. (2017). Production space and regional industrial evolution in China. GeoJournal, 82(2), 379–396. Han, S. S., & Pannell, C. W. (1999). The geography of privatization in China, 1978–1996. Economic Geography, 75(3), 272–296. Hassink, R. (2005). How to unlock regional economies from path dependency? From learning region to learning cluster. European Planning Studies, 13(4), 521–535. Hassink, R. (2007). The strength of weak lock-ins: The renewal of the Westmünsterland textile industry. Environment and Planning A, 39(5), 1147–1165. Hassink, R., & Shin, D. H. (2005). The restructuring of old industrial areas in Europe and Asia. Environment and Planning A, 37(4), 571–580. He, C., & Wang, J. (2012). Regional and sectoral differences in the spatial restructuring of Chinese manufacturing industries during the post-Wto period. GeoJournal, 77(3), 361–381. He, C., Wei, Y. H. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. Henning, M., Stam, E., & Wenting, R. (2013). Path dependence research in regional economic development: Cacophony or knowledge accumulation? Regional Studies, 47(8), 1348–1362. Hidalgo, C. A., Klinger, B., Barabási, A.-L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Hu, X., & Hassink, R. (2016). Place leadership with Chinese characteristics? A case study of the Zaozhuang coal-mining region in transition. Regional Studies, 1–12. Hu, X., & Hassink, R. (2017). Place leadership with Chinese characteristics? A case study of the Zaozhuang coal-mining region in transition. Regional Studies, 51(2), 224–234. Isard, W. (1960). Methods of regional analysis: An introduction to regional science. Cambridge, MA: MIT Press. Jiang, H., & Murmann, J. P. (2012). Regional institutions, ownership transformation, and migration of industrial leadership in China: The case of the Chinese synthetic dye industry, 1978–2008. Industrial and Corporate Change, 21(4), 933–970.

22

1 Introduction

Klepper, S. (2002). The capabilities of new firms and the evolution of the us automobile industry. Industrial and Corporate Change, 11(4), 645–666. Klepper, S., & Simons, K. L. (2000). Dominance by birthright: Entry of prior radio producers and competitive ramifications in the U.S. television receiver industry. Strategic Management Journal, 21(10–11), 997–1016. Lall, S., & Narula, R. (2004). Foreign direct investment and its role in economic development: Do we need a new agenda? The European Journal of Development Research, 16(3), 447–464. Lecocq, C., & Van Looy, B. (2016). What differentiates top regions in the field of biotechnology? An empirical study of the texture characteristics of biotech regions in North America, Europe, and Asia-Pacific. Industrial and Corporate Change, 25(4), 671–688. Li, P.-F., Bathelt, H., & Wang, J. (2012). Network dynamics and cluster evolution: Changing trajectories of the aluminium extrusion industry in Dali, China. Journal of Economic Geography, 12(1), 127–155. Lo Turco, A., & Maggioni, D. (2015). On firms’ product space evolution: The role of firm and local product relatedness. Journal of Economic Geography, 16(5), 975–1006. Marshall, A. (1920 [1890]). Principles of economics (8th ed.). London: Macmillan. Martin, R. (2010). Roepke lecture in economic geography—Rethinking regional path dependence: Beyond lock-in to evolution. Economic Geography, 86(1), 1–27. Martin, R., & Sunley, P. (2006). Path dependence and regional economic evolution. Journal of Economic Geography, 6(4), 395–437. Martin, R., & Sunley, P. (2015a). On the notion of regional economic resilience: Conceptualization and explanation. Journal of Economic Geography, 15(1), 1–42. Martin, R., & Sunley, P. (2015b). Towards a developmental turn in evolutionary economic geography? Regional Studies, 49(5), 712–732. Maskell, P. (2014). Accessing remote knowledge—The roles of trade fairs, pipelines, crowdsourcing and listening posts. Journal of Economic Geography, 14(5), 883–902. Maskell, P., & Malmberg, A. (1999). The competitiveness of firms and regions: ‘Ubiquitification’ and the importance of localized learning. European Urban and Regional Studies, 6(1), 9–25. Maskell, P., & Malmberg, A. (2007). Myopia, knowledge development and cluster evolution. Journal of Economic Geography, 7(5), 603–618. Maskell, P., Bathelt, H., & Malmberg, A. (2006). Building global knowledge pipelines: The role of temporary clusters. European Planning Studies, 14(8), 997–1013. Massey, D. (1984). Spatial divisions of labour. Social structures and the geography of production. London: Macmillan. Montinola, G., Qian, Y., & Weingast, B. R. (1995). Federalism, Chinese style: The political basis for economic success in China. World Politics, 48(1), 50–81. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Neffke, F. M. H., Henning, M., & Boschma, R. (2012). The impact of aging and technological relatedness on agglomeration externalities: A survival analysis. Journal of Economic Geography, 12(2), 485–517. Neffke, F., Hartog, M., Boschma, R., & Henning, M. (2014). Agents of structural change: The role of firms and entrepreneurs in regional diversification. Papers in evolutionary economic geography, 14.10, Utrecht University. Nooteboom, B. (2000). Learning and innovation in organizations and economies. Oxford: Oxford University Press. Oi, J. C. (1995). The role of the local state in China's transitional economy. The China Quarterly, 144, 1132–1149. Pan, F., Zhang, F., Zhu, S., & Wójcik, D. (2016). Developing by borrowing? Inter-jurisdictional competition, land finance and local debt accumulation in China. Urban Studies, 54(4), 897–916. Peterson, P. E. (1981). City limits. Chicago: University of Chicago Press.

References

23

Poncet, S., & Starosta de Waldemar, F. (2013). Export upgrading and growth: The prerequisite of domestic embeddedness. World Development, 51(0), 104–118. Poncet, S., & Starosta de Waldemar, F. (2015). Product relatedness and firm exports in China. World Bank Economic Review, 29(3), 579–605. Porter, M. (2003). The economic performance of regions. Regional Studies, 37(6–7), 545–546. Rigby, D. L., & Essletzbichler, J. (1997). Evolution, process variety, and regional trajectories of technological change in U.S. manufacturing. Economic Geography, 73(3), 269–284. Rodríguez-Pose, A. (2013). Do institutions matter for regional development? Regional Studies, 47 (7), 1034–1047. Rodríguez-Pose, A., & Di Cataldo, M. (2015). Quality of government and innovative performance in the regions of Europe. Journal of Economic Geography, 15(4), 673–706. Saxenian, A. (1994). Regional advantage: Culture and competition in Silicon Valley and route 128. Cambridge: Harvard University Press. Schamp, E. W. (2005). Decline of the district, renewal of firms: An evolutionary approach to footwear production in the Pirmasens area, Germany. Environment and Planning A, 37(4), 617–634. Schiff, M., & Wang, Y. (2006). North-South and South-South trade-related technology diffusion: An industry-level analysis of direct and indirect effects. Canadian Journal of Economics, 39(3), 831–844. Schumpeter, J. A. (1939). Business cycles: A theoretical, historical, and statistical analysis of the capitalist process. New York: McGraw-Hill. Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York: Harper & Row. Scott, A. (1988a). Metropolis: From the division of labor to urban form. Berkeley: University of California Press. Scott, A. (1988b). New industrial spaces: Flexible production organization and regional development in North America and Western Europe. London: Pion. Simmie, J. (2012). Path dependence and new technological path creation in the Danish wind power industry. European Planning Studies, 20(5), 753–772. Stam, E., & Martin, R. (2012). When high tech ceases to be high growth: The loss of dynamism of the Cambridgeshire region. Utrecht: Tjalling Koopmans Institute, Utrecht University School of Economics. Storper, M. (1997). The regional world: Territorial development in a global economy. New York: Guilford Press. Storper, M., & Walker, R. (1989). The capitalist imperative: Territory, technology and industrial growth. New York: Basil Blackwell. Sun, Q., & Tong, W. H. S. (2003). China share issue privatization: The extent of its success. Journal of Financial Economics, 70(2), 183–222. Sydow, J., Lerch, F., & Staber, U. (2010). Planning for path dependence? The case of a network in the Berlin-Brandenburg optics cluster. Economic Geography, 86(2), 173–195. Tanner, A. N. (2014). Regional branching reconsidered: Emergence of the fuel cell industry in European regions. Economic Geography, 90(4), 403–427. Tanner, A. N. (2016). The emergence of new technology-based industries: The case of fuel cells and its technological relatedness to regional knowledge bases. Journal of Economic Geography, 16 (3), 611–635. Van Grunsven, L., & Smakman, F. (2005). Industrial restructuring and early industry pathways in the Asian first-generation Nics: The Singapore garment industry. Environment and Planning A, 37(4), 657–680. Wei, Y. D., Li, W., & Wang, C. (2007). Restructuring industrial districts, scaling up regional development: A study of the Wenzhou model, China. Economic Geography, 83(4), 421–444. Wei, Y. H. D., Lu, Y., & Chen, W. (2009). Globalizing regional development in Sunan, China: Does Suzhou Industrial Park Fit a Neo-Marshallian District Model? Regional Studies, 43(3), 409–427.

24

1 Introduction

Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yang, C. (2012). Restructuring the export-oriented industrialization in the Pearl River Delta, China: Institutional evolution and emerging tension. Applied Geography, 32(1), 143–157. Yeung, H. W.-c. (2007). Remaking economic geography: Insights from East Asia. Economic Geography, 83(4), 339–348. Yu, X., Dosi, G., Lei, J., & Nuvolari, A. (2015). Institutional change and productivity growth in China’s manufacturing: The microeconomics of knowledge accumulation and “creative restructuring”. Industrial and Corporate Change, 24(3), 565–602. Yu, J., Zhou, L.-A., & Zhu, G. (2016). Strategic interaction in political competition: Evidence from spatial effects across Chinese cities. Regional Science and Urban Economics, 57, 23–37. Zhu, S., & Fu, X. (2013). Drivers of export upgrading. World Development, 51(0), 221–233. Zhu, S., & He, C. (2016). Global and local governance, industrial and geographical dynamics: A tale of two clusters. Environment and Planning C: Government and Policy, 34(8), 1453–1473. Zhu, S., He, C., & Liu, Y. (2014). Going green or going away: Environmental regulation, economic geography and firms’ strategies in China’s pollution-intensive industries. Geoforum, 55(55), 53–65.

Chapter 2

How Has Production Space Evolved in China?

2.1

Introduction

To answer the question of why some regions perform better than others has been one of major issues for economists and economic geographers. Recently, economic geographers inspired by the work of evolutionary economists (David 1985; Arthur 1989) have put forward evolutionary economic geography (EEG) by highlighting the evolutionary and dynamic process of regional development (Boschma and Frenken 2006; Frenken and Boschma 2007). In EEG, regional development is considered as an endogenous and self-reinforcing process, which is dependent on its historical trajectory and previous competencies such as technologies, institutions, labor skills, and industrial structure (Martin and Sunley 2006; Boschma and Frenken 2006; Boschma and Martin 2010). This process is called as “path dependence” in EEG (Boschma and Martin 2007, 2010). Most empirical studies confirm the existence of regional path dependence by using various measure of inter-industry technological relatedness, concluding that regions are more likely to follow their own industrial trajectory and develop new industries that are technologically related to those they have competitive advantages in. However, these studies use technological relatedness as a proxy of historical trajectory to quantitatively and statically examine the importance of path dependence (Frenken et al. 2007; Essletzbichler 2007; Bishop and Gripaios 2010; Boschma and Iammarino 2009; Boschma et al. 2012). Hidalgo et al. (2007) portray the “product space” on a global scale based on industrial relatedness and then explore the evolution of country’s position in the product space. Most developed and rich countries locate in the core of product space and have dense links with other parts, while most developing and poor countries locate in the periphery of product space

Modified article originally published in [Guo, Q. and C. He (2017) Production Space and Regional Industrial Evolution in China, GeoJournal, 82 (2), pp. 379–396]. Published with kind permission of © [Springer Nature, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_2

25

26

2 How Has Production Space Evolved in China?

and have sparse links with other parts. They conclude that developed countries with the core position and dense links have more opportunities to transition into new comparative advantages by developing goods close to their existing products, but peripheral countries in production network face more challenges to catch up with the core countries for lack of links with product space. In other words, the evolution of productive structure is subject to product space (Hausmann and Klinger 2007). Hidalgo et al. (2007) provide the evidence for path dependence from a dynamic and evolutionary perspective (Hidalgo et al. 2007; Neffke et al. 2011). If it is true, peripheral countries or regions would never catch up with the core countries or regions, and divergence between core regions and periphery regions would remain forever. Even though path dependence is viewed as a type of path creation because it attempts to answer how new industries or new path emerge, it is insufficient to explain the rise and catching-up of emerging economies over recent decades and to explain the successful renewal and rejuvenation of old industrial regions or lock-in regions. Therefore, economic geographers start to focus on the role of external factors or exogenous shocks in shaping new paths, for example, governmental policies. Following Hidalgo et al. (2007), this chapter calculates the industrial relatedness by adopting more accurate co-occurrence approach and represents the evolution of China’s production space and regional transformation in the production space. More importantly, the chapter empirically answers the question of how regional productive structure evolves in China and further testifies whether regional industrial evolution is influenced by industrial relatedness in the production space. Our results do not fully confirm regional path dependence. There are substantial regional differences in the regional evolution path. The evolution of regional productive structure in Coastal regions is significantly influenced by historical productive capability, while North West and South West seize the opportunity of “Western Development Strategy” to break the path-dependent trajectory and develop new sectors distant from their own production networks. The rest of the chapter is structured as follows. The next section will present literature review and analytical framework. The third section introduces the model specification. The fourth section shows how China’s production space and regional industrial structure changes using the co-occurrence approach put forward by Hidalgo et al. (2007). The fifth section empirically examines the process of regional path dependence and path creation. The last section will conclude the chapter with main findings and discussion.

2.2

Literature Review and Analytical Framework

The notion of path dependence is initially used by research work on management and institution to explain the importance of historical trajectory during the technological, industrial and institutional transformation. Since 1990s, the explanatory power of path dependence in the spatial or regional context has been developed by economic geography (Grabher 1993). The regional path dependence research, in

2.2 Literature Review and Analytical Framework

27

general, refers to two types of themes. In the early studies, case studies are widely used to study why some regions fail, but others not. They claim that the lock-in process in terms of technology, institutions and other aspects is one of major causes for a region’s or a industry’s failure (Glasmeier 1991; Grabher 1993; Bathelt 2001). The “lock-in” model is the canonical model in the path dependence research, which derives from the contribution of Evolutionary Economics (David 1985; Arthur 1994) but is empirically developed by “institutional turn” and “evolutionary turn” in economic geography, arguing that regional industrial path may be locked in the self-reinforcing mechanism because of continuation and conservation of old production modes (Glasmeier 1991; Grabher 1993). Compared with questions like why some regions fail, the idea of why and how region’s new paths emerge deserves more attention. There are two approaches to study regional path creation. The first one can be called regional branching process or path continuity, which emphasizes the endogenous process of new path creation. The process is sometimes also called as path dependence in some studies of evolutionary economic geography (Frenken and Boschma 2007; Neffke et al. 2011). They argue that regions are more likely to branch into industries that are technologically related to the existing industries in the region (Boschma and Frenken 2011; Neffke et al. 2011). The emergence of new industries in a region generally includes three situations: entrepreneurship, spin-off, and relocation of firms. First, entrepreneurship is, in some sense, the practice of a new idea or innovation in the market (Schumpeter 2000). Innovation often derives from the recombination of various existing knowledge, so Jacobs claims that the spillover of diversified knowledge in urban economies can provide fertile soil for innovation. However, recent studies have found that the efficiency of knowledge spillovers is subject to not only “spatial distance” but also “cognitive distance” (Nooteboom 2000), that is, either too much or too little cognitive proximity is less conductive to knowledge spillovers. Regions with a range of technologically related industries are more beneficial to the emergence of innovation and consequently entrepreneurship. This is one reason why new industries are related to existing industries. Second, spin-offs are highly related to their parent firms in terms of products, production technology, or skills of labor force in order to reduce R&D cost and minimize market risks (Klepper 2007; Wenting 2008), which can also make new industries appear highly related to existing industries. Third, a region with a range of related industries is a good choice for firm relocation because related industries not only can share similar knowledge, skilled labor, and institutions but also can promote innovation through knowledge spillovers. Therefore, new industries or new paths are more likely to emerge in a region with more related preexisting industries. In other words, a region with a wide range of industries which have dense links with other industries are more capable to develop new paths and new industries, while a region with less industries or with lots of industries independent of other industries is less promising. Empirically, recent studies have confirmed the existence of regional path dependence in the developed countries, for example, in Britain (Boschma and Wenting 2007), in Spain (Boschma et al. 2012), in the USA (Essletzbichler 2015), and in European regions (Colombelli et al. 2014). They use technological relatedness as a

28

2 How Has Production Space Evolved in China?

proxy of historical trajectory to quantitatively examine the importance of path dependence from a static perspective. From a dynamic perspective, Hidalgo et al. (2007) and Neffke et al. (2011) support the view of path dependence, concluding that history matters either at the country level or at the regional level. One of the goals of this chapter is to explore if the regional path dependence exists in China from a dynamic and evolutionary perspective. The other approach to study path creation stresses the exogenous process of new path creation and finds that the emergence of new industries sometimes is not dependent on the previous regional production competencies but is ascribed to external shocks or dramatic changes, for example, a technology revolution (Bathelt and Boggs 2003), a crisis (Meyer-Stamer 1998), or a governmental policy. An exogenous shock has been regarded as an opportunity to avoid falling into the “lock-in” situation. Either a technology revolution or a crisis is a dramatic change on a national or grander scale, so they may be less influential to path creation on a regional scale than governmental policies. Some case studies on regional innovation systems focus on the effect of specific regional innovation policies, but very little literature attempts to explore quantitatively the role of local supportive policies in shaping new paths, which is one of goals of this chapter. The above two approaches are both considered to be concerned with new path creation. However, whether the regional branching process is new path creation is questionable. Henning et al. (2013) notice the fuzzy border between path continuity and new path creation. Almost consistent with the notion of Henning et al. (2013), this chapter defines the situation when new industries are highly related to incumbent stable regional conditions as path continuity or path dependence and defines the situation when new industries are unrelated to existing regional conditions as new path creation. The rapid growth of China as one of developing countries seems a typical example to the path creation approach. Since economic reform and opening in 1978, China has rapidly transformed from a relatively poor country based on some agriculture and heavy industries into a world factory hosting a great variety of industries, rather than being locked in its previous trajectory. The outstanding performance of China as a whole during the last decades is mainly attributed to a series of national policies on reform and openness (Felipe et al. 2013). However, these favorable policies are only available to coastal regions rather than inland regions during the initial period, leading to rising regional economic inequality in China (Chen and Fleisher 1996; Wei 1999; Fan and Sun 2008; Fleisher et al. 2010). China’s governments started to propose a series of regional policies to narrow the regional gap at the end of the twentieth century. Can these policies open the way for inland region to catch up with coastal regions? What is the evolutionary process of regional development in China like? Most studies have focused on the magnitude and trend of inequality index (Kanbur and Zhang 2005; Lu and Wang 2002; Fujita and Hu 2001) and the impact of one or more factors on regional inequality in China (Zhang and Zhang 2003; Jones and Cheng 2003), but few studies demonstrate the evolution of regional productive structure to explore the regional economic development paths from an evolutionary perspective.

2.3 Data and Model

29

Overall, the contribution of this research is threefold. It is among the first to paint the China’s production space by using the co-occurrence approach and represent its evolution. It is among the first to explore the evolutionary process of regional productive structure based on China’s production space. It is among the first to empirically examine the process of regional path dependence and path creation.

2.3 2.3.1

Data and Model The Measurement of Industrial Relatedness

Technological relatedness between industries is one of most important concepts to study empirically path dependence and the evolutionary process of regional development. The quantitative studies in regional branching research have proliferated since the introduction of technological relatedness between industries. Its measure has been a big challenge for empirical studies in EEG. Initially, the most common way to measure technological relatedness is based on standard sector classifications. If two subsectors belong to the same sector classification, they are defined as related, otherwise, unrelated (Frenken et al. 2007; Boschma and Iammarino 2009). But this measure neglects the situation that high relatedness exists between subsectors that don’t belong to the same sector classification because of input-output linkage or knowledge spillovers between different sectors (Essletzbichler 2015). Another way to measure industrial relatedness is to use input-output tables to calculate similarity between sectors in the use of input factors (Farjoun 1994; Dumais et al. 2002). This measure assumes that if two sectors share similar input mixes, they apply similar production technology. However, there may be a large number of factors that influence the degree of relatedness between sectors, including institutions, infrastructure, combination of productive factors, and so forth. Beyond traditional approaches, Hidalgo et al. (2007) creatively put forward a co-occurrence approach to calculate the inter-product “proximity,” which is the conditional probability of two products co-exported by the same country. Its biggest contribution is to provide researchers a new idea to more accurately calculate the inter-sector “proximity” or “relatedness” at any sector level. In this chapter, we use Eqs. 1.3 and 1.4 to calculate relatedness (ϕi,j) between industry i and j, by using the ASIF dataset during 1998–2007. We define the production space as the set of all relatedness measures, which consist of a 424*424 matrix whose entries are the industrial relatedness between 4-digit manufacturing sectors. Every row and column of this matrix represents a particular sector. Thus, we achieve symmetric relatedness matrix.

30

2.3.2

2 How Has Production Space Evolved in China?

Model Specification and Variables

To accurately examine the effect of production space, we calculate the “density” of industry i in city c (densityi,c) as a measure of the link of one sector with regional productive structure by using Eq. 1.5. In order to investigate the path dependence of regional industrial evolution, we employ two methods: the kernel distribution and a probit model regression. The former visually displays the difference of density between transition and undeveloped sectors in kernel distribution. The latter statistically and accurately examine the effect of density on regional industrial evolution by estimating the econometric equation as follows. yict1 ¼ α0 þ α1 densityict0 þ δ1 City þ δ2 Sector þ εict1

ð2:1Þ

where the independent variable is “density” at the city-sector level in the initial year t0. yict1 is a binary variable, 1 for the sectors transitioning into advantage ones during t0 and t1 and 0 for undeveloped sectors during t0 and t1, so we estimate this equation by using a probit model, controlling for the heterogeneity of sectors (sector) and cities (city). A significant and positive α1 indicates that density can significantly increase the probability of sector i being one new emerging advantage sector in city c. That is, the evolution of regional industrial structure is path-dependent. According to prior theoretical and empirical studies, however, density is not the only one determinant of regional industrial evolution. The state and local policies may change local industrial path and especially help less developed regions “lock out” of the vicious circle and create a new path. To examine the effect of policies on path creation, we introduce policy variables into Eq. 2.1: yict1 ¼ α0 þ α1 densityict0 þ α2  I ict0 þ δ1 City þ δ2 Sector þ εict1

ð2:2Þ

where I ict0 is a vector of policy variables in the initial year, including the tax rate and the subsidy. The tax rate (tax) is measured as taxes levied on the 2-digit sectors divided by their gross output. The subsidy (subsidy) is measured as the subsidy 2-digit sectors receive at city c. A significant and positive α2 indicates that governmental policies can significantly increase the probability of sector i becoming one new emerging advantage sector in city c.

2.4 2.4.1

The Evolution of Production Network and Regional Path Dependence The Evolution of China’s Production Space

Figure 2.1 shows the distribution and statistic index of industrial relatedness in 1999 and 2007. The distribution of industrial relatedness is very left-skewed, suggesting that there are some strong links in the production space, but most links are rather

2.4 The Evolution of Production Network and Regional Path Dependence

31

Fig. 2.1 Distribution and statistic index of industry relatedness, 1999 and 2007 Table 2.1 Correlation coefficients of production relatedness during 1999–2007 2007 2006 2005 2004 2003 2002 2001 2000 1999

2007 1.00 0.93 0.89 0.85 0.78 0.72 0.70 0.68 0.66

2006

2005

2004

2003

2002

2001

2000

1999

1.00 0.93 0.88 0.80 0.74 0.72 0.70 0.68

1.00 0.91 0.82 0.75 0.73 0.71 0.69

1.00 0.82 0.76 0.74 0.72 0.70

1.00 0.85 0.82 0.80 0.78

1.00 0.91 0.86 0.83

1.00 0.90 0.87

1.00 0.92

1.00

Note: All of correlation coefficients are significant at the level of 1%

weak. The distribution characteristic is similar to most studies (Hidalgo et al. 2007; Boschma et al. 2012; Neffke et al. 2011). Over 70% of relatedness in 1999 and 2007 is below 0.2, and about 1% are above 0.35. These weak links are not significant, and they cannot even prove the existence of links between sectors, so it is necessary to define the links above a threshold as related. We choose 0.35 as threshold, which is more conservative than Boschma et al. (2012) and Neffke et al. (2011). That is, China’s production space is composed of industrial links whose proximity is equal or above 0.35. Thus, the 972 and 1264 links remain in the production space in 1999 and 2007, separately. There is very small difference between 1999 and 2007 in the distribution of production relatedness. Does it mean that production space has not changed? Table 2.1 shows the result of cross-year correlation of pairwise production relatedness. Most correlation coefficients between neighboring years are over 0.9 and decline gradually with distance between years. Taking 2007 as an example, its

32

2 How Has Production Space Evolved in China?

Node size Low gross output value high gross output value

Node color Chemical Electric apparatus, electronic and telecommunications equipment Food products and tobacco General machinery Metal products Non-metallic mineral products Other manufacturing products Pulp, paper and printing Textile and garment

Transport equipment Wooden products and furniture

Fig. 2.2 The production network of China, 1999. Note: (1) Every node is a 4-digit manufacturing sector. (2) The same color indicates that the 4-digit sectors belong to the same classification, whose criterion is based on China’s input-output tables (17 sectors). (3) The number of nodes and edges is 287 and 972, separately. (4) The threshold of relatedness is 0.35

correlation coefficient with 2006, 2005,. . ., and 1999 goes down from 0.93 to 0.66, suggesting that China’s production space evolves relatively rapidly in the 9 years despite of high correlation between neighboring years. To demonstrate the structure of production space, we draw the production networks with the cytoscape 3.2.1.1 Figures 2.2 and 2.3 show the production networks in 1999 and 2007. The number of nodes increases from 287 to 319, while the number of edges increases from 972 to 1264, which implies that more sectors and linkages are included in production network in 2007. Figure 2.4 further supports the result, by showing that the mean value of industrial relatedness fluctuates before 2002, but increases as a whole during 1999–2007. Figure 2.2 represents the core-periphery structure of China’s production space in 1999. There is obviously a major core in the production space, which is mainly based on electric apparatus and electronic and telecommunications equipment. In addition, there is a small cluster that consists of chemical and food products and non-metallic mineral products. The small cluster is linked with the core through some sparse linkages. Besides the core and sub-core cluster, the rest of production space are more emanative, and there are many peripheral linkages that seem very easy to split from production network (e.g., some subsectors in general machinery).

1 To make the network visualization clear, we adopt the edge-weighted spring-embedded layout which uses a force spring algorithm.

2.4 The Evolution of Production Network and Regional Path Dependence

33

.4 .39

.395

Relatedness

.405

.41

Fig. 2.3 The production network of China, 2007. Note: (1) The number of nodes and edges is 319 and 1264, separately. (2) The threshold of relatedness is 0.35

1998

2000

2002

Year

Total Between

2004

2006

2008

Within

Fig. 2.4 The mean value of industry relatedness for all links (Total), links within the same classification (Within), and between different ones (Between), 1999–2007

34

2 How Has Production Space Evolved in China?

The production space in 2007 is significantly different from that in 1999. The electric and electronic cluster remains one of important cores in 2007. The small cluster that was formed by chemical and food products and non-metallic mineral products in 1999 develops into another dense and important core of production space in 2007 (called as the second core). General machinery that located in the periphery in 1999 has played the important role in linking the two cores. The number of emanative linkages clearly declines, implying that industrial relatedness is increasingly stable. Furthermore, the core role of the two clusters is not so obvious since peripheral links become increasingly denser. In addition, we find by comparing Figs. 2.2 and 2.3 that the sectors which belong to the same classification (the same color) are increasingly close to each other, implying that the linkage within the same classification is stronger. Figure 2.4 provides more credible evidence to support the graphical results. The average linkage within the same classification is always much stronger than interclassification linkage during 1999–2007 (Fig. 2.4). The rise of industrial relatedness between sectors within the same classification is the main source of the rise of total production relatedness, suggesting that the likelihood of subsectors that belong to the same classification co-occurring in the same city goes up. That is, the level of urban production specification increases.

2.4.2

The Evolution of Regional Productive Structure

To compare the regional differences in production space and study the evolution of regions’ position in production space, we hold the production space in Fig. 2.2 fixed. Figure 2.5 shows the productive structure of eight different regions in China’s production space in 1999. In 1999, coastal regions except North Coast occupy the core of China’s production space and develop the some peripheral sectors at the same time. North Municipalities and South Coast are the most developed regions in electric apparatus and electronic and telecommunications equipment. In the coastal region, North region including North Municipalities and North Coast are better at developing heavy industries such as general machinery and chemical, but South Coast is better at hosting light industries such as textile and garment and food products and tobacco. Central Coast develops both types of industries. What is different from the coastal regions is that the advantage sectors of inland regions locate in the sub-core cluster and the periphery of production space are composed of general machinery, chemical and food products, and tobacco. The North East and Central China have competitive advantage in transportation equipment. To examine the magnitude of regional structure change in China during 1999–2007, we compute the correlation coefficients of regional advantage industries between 1999 and 2007 (Table 2.2). During the 9 years, China’s regions undergo substantial structural change, and the magnitude of the change varies across regions. Figure 2.6 further shows the evolution of regional productive structure during

2.4 The Evolution of Production Network and Regional Path Dependence

35

Fig. 2.5 Localization of the productive structure for different regions in the production space of China, 1999. Note: (1) Holding the production space in Fig. 2.2 fixed, a black solid circle represents a sector that a region has competitive advantage in (LQ>1), while a hollow circle represents the rest of sectors. (2) The top three advantage industry classifications in 1999 are shown below the corresponding regions. (3) Hereafter, North East contains Liaoning, Jilin, and Heilongjiang province; North West contains Xinjiang, Qinghai, Gansu, Ningxia, Shannxi, and Inner Mongolia province; North Municipalities contain Beijing and Tianjin; North Coast contains Hebei and Shandong province; Central Coast contains Shanghai, Jiangsu, and Zhejiang province; Central China contains Shanxi, Henan, Anhui, Hubei, Hunan, and Jiangxi province; South Coast contains Fujian, Guangdong, and Hainan province; South West contains Guangxi, Yunnan, Guizhou, Chongqing, Sichuan, and Tibet

1999–2007. Holding the production space in Fig. 2.3 fixed, a solid triangle represents a new advantage sector that a region has developed between the year 1999 and 2007, and a solid circle represents the preexisting advantage sectors in 1999, while a hollow circle represents the rest of sectors. The overall pattern of regional productive structure in 2007 has also changed. The coastal regions except North Coast still occupy the first core that is mainly composed of electric apparatus and electronic and telecommunications equipment, but they extend their advantage into periphery of

36

2 How Has Production Space Evolved in China?

Table 2.2 The correlation coefficients of regional advantage industries between 1999 and 2007 Region South Coast Central North West Central Coast North East North Coast South West North Municipalities

Correlation of A between 1999 and 2007 0.64 0.59 0.58 0.55 0.47 0.44 0.44 0.42

Note: A is equal to 1 if LQ>1

Fig. 2.6 The productive structure of different regions in the production space in China, 2007. Note: (1) Holding the production space in Fig. 2.3 fixed, a solid triangle represents a new advantage sector that a region has developed between the year 1999 and 2007, and a solid circle represents the preexisting advantage sectors in 1999, while a hollow circle represents the rest of sectors. (2) The top three new advantage industry classifications during 1999–2007 are shown below the corresponding regions

their links at the same time. The inland regions and North Coast occupy the second core that is composed of chemical and food products and non-metallic mineral products, but they also extend their advantage into the first core of production space. Further, the many new advantage sectors have links with the existing ones for all eight regions, suggesting that the regional industrial evolution, to some extent, is a

2.5 The Regional Path Dependence and the Effect of Institutions

37

path-dependent process. In other words, regional development can be subject to previous productive structure. Interestingly, inland regions and North Coast seem to be a little different from coastal regions in terms of industrial evolution. Coastal regions are more restricted by the production network, but inland regions successfully extend some new advantage sectors into those distant from their links. Does it imply that some regions can break the path dependence? To answer this question, the next section is to econometrically examine the effect of production space on regional industrial evolution.

2.5

The Regional Path Dependence and the Effect of Institutions

To examine how current transition is influenced by the previous density, we compare the kernel distribution of density in 1999 for “transition sectors” and “undeveloped sectors” in 2007 (Fig. 2.7). The former is defined as those with LQ rising from below 1 to above 1 during 1999–2007, while the latter is defined as those with LQ below 1 both in 1999 and 2007. Other observations are neglected. Contrary to our expectation, the distribution of undeveloped sectors is on the right side of distribution of transition sectors. That is, for high density levels, the probability of undeveloped sectors are higher than the probability of transition sectors, implying that transition

Fig. 2.7 Distribution of density for transition sectors and undeveloped sectors during 1999–2007

38

2 How Has Production Space Evolved in China?

sectors are not closer to regional productive structure than undeveloped sectors are. Moreover, the difference between two distributions is highly significant (reject the null hypothesis of ANOVA test which is that two distributions are statistically not different). As noted by Martin and Sunley (2006), “in many important aspects, path dependence and ‘lock-in’ are place-dependent processes, and as such require geographical explanation.” To explore why the density of undeveloped sector is higher, we decompose the overall distribution into different regions (Fig. 2.7). There exists a significant regional variation in the distribution of density. The density of either undeveloped sectors or transition sectors in coastal regions is much higher than the density of those in inland regions, leading to unexpected results for China as a whole. In North Coast, Central Coast, and North East, the density of transition sectors is significantly higher than the density of undeveloped sectors, implying that transition sectors are more dependent on the existing productive structure in these regions. A path-dependent process significantly influences industrial evolution of North Coast and Central Coast. However, North West and South West observe an opposite result. The density of transition sectors is significantly lower than that of undeveloped sectors, suggesting that industrial evolution of North West and South West is not subject to their previous productive structure. Distribution diagram cannot control for the heterogeneity of sectors, so we estimate a probit model (Eq. 2.1) with controlling for the heterogeneity of sectors and cities to accurately examine the effect of density on industrial evolution. During 1999–2007, China has been experiencing dramatic change in both production space and policies, especially policies that are always unstable in developing economies. If we take 1999 and 2007 as the initial year and observation year, respectively, some policies such as the tax rate and subsidy implemented in 1999 may have changed a lot in 2007. To shed light on their influence on industrial evolution more accurately, we divide the period into two stages: 1999–2003 and 2003–2007. Table 2.3 presents the effect of production space on industrial evolution; the overall statistical result in China in the first column reports that the influence of production space on industrial evolution is not significant (Table 2.3), which is consistent with the results in the above kernel distribution. There may be substantial variation across regions, so we need to estimate the regression models for different eight regions. The effect of density in the South West is negative and significant during 1999–2003, suggesting sectors that transition into advantage industries in West are not highly related to local industrial structure. The coefficient turns to be positive during 2003–2007, but its effect is insignificant. That is, industrial evolution in the South West has not followed the path-dependent process, which may be explained by “Western Development Strategy” proposed in 1999 in order to reduce regional disparity. A series of infrastructure construction and favorable policies are implemented to attract new industries to enter into western regions. These policies may change firms’ production condition in the West, which are conductive to development of new advantage industries technologically distant from local productive. Likewise, Central China and North West do not significantly follow the path-

0.515 0.763*** 11,795 309 3.233 0.072

2.625*** 1.199*** 13,706 301 56.79 0.000

Density Constant Observations Number of city LR chi2 Prob>chi2

Density Constant Observations Number of city LR chi2 Prob>chi2

20.082*** 8.892*** 372 2 20.23 0.000

16.723*** 7.767*** 346 2 13.12 0.000

N Municipalities

2.925*** 1.473*** 2147 28 14.60 0.000

2.016** 1.208*** 1716 28 6.827 0.008

N Coast

Central Coast 1999–2003 2.038* 1.397* 2433 25 1.934 0.164 2003–2007 8.057*** 3.158*** 3051 25 48.85 0.000 5.834*** 2.087*** 2234 31 39.52 0.000

6.637*** 2.267*** 1750 31 24.40 0.000

S Coast

Note: (1) Robust standard errors are in parentheses; (2) *** p < 0.01, ** p < 0.05, * p < 0.1

Total

Variables

Table 2.3 The effect of production space (density) on industrial evolution

1.086 0.810*** 3052 83 3.208 0.073

0.447 0.762*** 2736 83 0.531 0.466

Central

3.764*** 1.301*** 1198 33 8.105 0.004

2.370** 0.912*** 1107 35 3.493 0.061

Northeast

0.561 0.512** 567 46 0.143 0.705

0.367 0.334 662 52 0.0814 0.775

N West

1.499 0.850*** 1085 53 1.659 0.198

0.922* 0.403*** 1045 53 2.275 0.131

S West

2.5 The Regional Path Dependence and the Effect of Institutions 39

40

2 How Has Production Space Evolved in China?

dependent process. For other regions, industrial evolution has been always affected significantly by previous industrial structure, suggesting the path-dependent process exists in most developed regions of China. Industries are more likely to enter a region where they are highly related with their current industries, which is consistent with most empirical studies (Neffke et al. 2011). Further, it is worthwhile to point out that the effect of density has increased significantly from the first stage to the second one, indicating that the process of regional industrial evolution is more pathdependent. To examine if supportive policies have a significant influence on the pathdependent process, we estimate Eq. (2.2) with the policy variables. After adding to policy variables, the result of density in Table 2.4 is similar with that in Table 2.3. Table 2.4 shows that subsidy has a positive and significant sign for almost all regions throughout two stages, suggesting that high subsidy can increase the probabilities that new advantage sectors emerge. Particularly for North West and South West, the coefficient of subsidy is larger than other regions, implying that their new paths rely largely on governmental subsidies. However, comparing the difference of subsidies’ coefficients between two stages, we find that the importance of subsidies has been decreasing while the role of density has been increasingly important. It may be because regions learn and copy similar favorable policies with each other, and these policies are often not sticky and creative so that they are very easy to copy. In contrast, with the improvement of market institutions, the development of firms and industries is more dependent on knowledge, skilled labor, and production network, which are likely to be provided by clusters among technologically related industries. The tax rates have a negative sign in all regions during 1999–2003, and they are significant in four regions, indicating that lower tax rates are beneficial to the emergence of new advantage sectors. Comparing the results between two stages, the significance of tax rates has also been decreasing either for total samples or for eight regions, confirming the above conjecture on regional policies. In other words, favorable policies on tax rates can help regions develop new competitive advantage that is not technologically related to preexisting productive competencies. Therefore, it can be concluded that the results on two policy variables confirm the importance of governmental policies to the emergence of new paths and new industries, and, more notably, governmental policies can help regions break through path dependence at the initial stage, but when most regions adopt similar favorable policies, these policies will lose their appeal. As seen in the results above, industrial transition in the West is not subject to its existing production space but the process of path creation by branching into less related industries. However, we do not know whether the transition sectors with lower density are “good” or “bad.” Here, high productive sectors are considered as “good” ones, while low productive sectors are considered as “bad” ones. Figure 2.8 shows the relationship between the density and total factor productivity for the transition sectors. The horizontal line above the horizontal axis is the mean value of all sectors’ productivity, and the vertical line on the right side of the vertical axis is the mean value of density. Both intersecting lines divide the graphical space into four parts. We find that transition sectors in coastal regions such as North Municipalities

0.537 3.022** 0.020*** 0.764*** 11,795 309 54.34 0.000

2.661*** 0.834 0.008*** 1.209*** 13,706 301 92.99 0.000

Density Tax Subsidy Constant Observations Number of city LR chi2 Prob>chi2

Density Tax Subsidy Constant Observations Number of city LR chi2 Prob>chi2

20.385*** 22.239*** 0.007*** 8.994*** 372 2 22.09 0.000

20.417*** 35.935*** 0.012*** 9.258*** 346 2 19.68 0.000

N Municipalities

2.669*** 6.899* 0.034*** 1.420*** 2147 28 30.19 0.000

1.858** 7.770 0.137*** 1.190*** 1716 28 32.20 0.000

N Coast

Central Coast 1999–2003 3.282 19.921** 0.014** 1.730 2433 25 15.72 0.001 2003–2007 8.247*** 2.439 0.004*** 3.212*** 3051 25 54.42 0.000 6.069*** 2.283 0.020 2.135*** 2234 31 46.89 0.000

6.717*** 4.644 0.026 2.276*** 1750 31 28.48 0.000

S Coast

Note: (1) Robust standard errors are in parentheses; (2) *** p < 0.01, ** p < 0.05, * p < 0.1

Total

Variables

Table 2.4 The effect of supportive policies on path creation

0.988 0.218 0.058*** 0.819*** 3052 83 29.33 0.000

0.419 2.780* 0.075*** 0.761*** 2736 83 15.54 0.001

Central

3.366** 6.209 0.022* 1.228*** 1198 33 18.67 0.000

1.849 5.280 0.046** 0.813*** 1107 35 13.91 0.003

Northeast

0.361 8.792 0.105*** 0.582*** 567 46 18.58 0.000

0.292 0.463 0.148** 0.361* 662 52 13.38 0.004

N West

1.298 2.725 0.033*** 0.822*** 1085 53 10.90 0.012

1.061* 10.803** 0.158*** 0.373*** 1045 53 25.51 0.000

S West

2.5 The Regional Path Dependence and the Effect of Institutions 41

42

2 How Has Production Space Evolved in China?

10 8 6

TFP

2 0 .2

.4

0

.6

.2

.4

Density

North West

South West

10

Density

.6

6

TFP

4

0

0

2

2

4

6

8

8

10

0

TFP

Central Coast

4

6 4 0

2

TFP

8

10

North Municipalities

0

.2

.4

.6

Density

0

.2

.4

.6

Density

Fig. 2.8 The relationship between density and sectoral productivity for the transition sectors. Note: The horizontal line above the horizontal axis is the mean value of all sectors’ productivity (3.15), and the vertical line on the right side of the vertical axis is the mean value of density (0.14)

and Central Coast have higher density, but many transition sectors in North West and South West have lower density. Fortunately, many transition sectors in North West and South West are located in the top left area, implying that they are productive sectors though their density is very low. We believe new emerging high productive advantage sectors can be beneficial for regional economic growth. From Fig. 2.9 which is the enlarged view of the top left area in Fig. 2.8, we find that the transition sectors of North West and South West with low density and high productivity include general machinery, special equipment, automobile manufacturing, and so on. If these sectors as “good” mutations can branch into other “good” sectors through industrial relatedness, then North West and South West enter a virtuous cycle and have more opportunities to catch up with coastal regions.

2.6

Conclusion and Discussion

Economic development is not only a process of the output growth but also is a process of structural transformation. Hidalgo et al. (2007) recently put forward the notion “product space” to explore economic development paths from a dynamic perspective. They conclude that history matters. That is, regional economic growth is a path-dependent process and rooted in the preexisting economic structure

2.6 Conclusion and Discussion

43

Fig. 2.9 The transition sectors with low density and high productivity in North West and South West. Note: (1) This figure is the enlarged view of the top left area in Fig. 2.8. (2) The number in the graph is the code of SIC 4-digit sectors

(Boschma et al. 2012; Neffke et al. 2011). Regional structure evolves by attracting or branching into industries related to local productive advantages. To testify if regional development follows the path-dependent process, this chapter examines the impact of production space on regional industrial evolution in China. Based on firm-level data during 1999–2007, we find that China’s production space evolves rapidly and has changed from one-core to two-core structure. More sectors are embedded in the production space, and the magnitude and stability of the inter-sector relatedness are on the increase. The Coastal regions except North Coast occupy the core of China’s production space, composed of electric apparatus and electronic and telecommunications equipment, while inland regions occupy the sub-core cluster and periphery of production space, composed of general machinery, chemical and food products, and tobacco. Statistical analysis indicates that the evolution of regional productive structure in Coastal regions is significantly influenced by preexisting productive capability, while North West and South West break the path-dependent trajectory and transition into sectors distant from their own production network. The findings on the evolution of regional productive structure can help policymakers answer the questions of how regions should develop and what they can do to promote growth and avoid recession. The “Western Development Strategy” can be referred to as a shock for the western regions. A series of favorable policies improve their production conditions to attract “good” industries even though the industries are not related to the current productive structure. The good industries with low

44

2 How Has Production Space Evolved in China?

density can be considered as industrial “mutation” in the evolutionary procedure. If they as “good” mutations can branch into other “good” industries through industrial relatedness, North West and South West will enter a virtuous cycle and have more opportunities to catch up with coastal regions. If it is true, we believe that policy intervention can provide poor regions better opportunities to break the pathdependent process and improve their position in the production space. However, statistical results also show that governmental policies are more likely to help regions break through path dependence at the initial stage, but their influence will decay with time. The main reason is that favorable policies adopted by regions or cities are similar and easy to copy. When these policies such as subsidies and tax rates are widely used, their attraction for advantage industries will be lost. Therefore, not only should local governments depend on some favorable policies to attract new industries, but they should also establish a more stable institutional environment and foster more trustful business culture to develop local core competence.

References Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal, 99(394), 116–131. Arthur, W. B. (1994). Increasing returns and path dependence in the economy. Ann Arbor: University of Michigan Press. Bathelt, H. (2001). “Regional competence and economic recovery: Divergent growth paths”, in Boston’s high technology economy. Entrepreneurship & Regional Development, 13(4), 287–314. Bathelt, H., & Boggs, J. S. (2003). Toward a reconceptualization of regional development paths: Is Leipzig’s media cluster a continuation of or a rupture with the past? Economic Geography, 79 (3), 265–293. Bishop, P., & Gripaios, P. (2010). Spatial externalities, relatedness and sector employment growth in Great Britain. Regional Studies, 44(4), 443–454. Boschma, R., & Frenken, K. (2006). Why is economic geography not an evolutionary science? Towards an evolutionary economic geography. Journal of Economic Geography, 6(3), 273–302. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., & Frenken, K. (2011). The emerging empirics of evolutionary economic geography. Journal of Economic Geography, 11(2), 295–307. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., & Martin, R. (2007). Editorial: Constructing an evolutionary economic geography. Journal of Economic Geography, 7(5), 537–548. Boschma, R., & Martin, R. (Eds.). (2010). The handbook of evolutionary economic geography. Cheltenham: Edward Elgar Publishing. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Chen, J., & Fleisher, B. (1996). Regional income inequality and economic growth in China. Journal of Comparative Economics, 22(2), 141–164.

References

45

Colombelli, A., Krafft, J., & Quatraro, F. (2014). The emergence of new technology-based sectors in European regions: A proximity-based analysis of nanotechnology. Research Policy, 43(10), 1681–1696. David, P. A. (1985). Clio and the economics of qwerty. The American Economic Review, 75(2), 332–337. Dumais, G., Ellison, G., & Glaeser, E. L. (2002). Geographic concentration as a dynamic process. Review of Economics and Statistics, 84(2), 193–204. Essletzbichler, J. (2007). Diversity, stability and regional growth in the United States, 1975–2002. In K. Frenken (Ed.), Applied evolutionary economics and economic geography (pp. 203–229). Cheltenham: Edward Elgar. Essletzbichler, J. (2015). Relatedness, industrial branching and technological cohesion in Us Metropolitan areas. Regional Studies, 49(5), 752–766. Fan, C. C., & Sun, M. (2008). Regional inequality in China, 1978–2006. Eurasian Geography and Economics, 49(1), 1–17. Farjoun, M. (1994). Beyond industry boundaries: Human expertise, diversification and resourcerelated industry groups. Organization Science, 5(2), 185–199. Felipe, J., Kumar, U., Usui, N., & Abdon, A. (2013). Why has China succeeded? And why it will continue to do so. Cambridge Journal of Economics, 37(4), 791–818. Fleisher, B., Li, H., & Zhao, M. Q. (2010). Human capital, economic growth, and regional inequality in China. Journal of Development Economics, 92(2), 215–231. Frenken, K., & Boschma, R. A. (2007). A theoretical framework for evolutionary economic geography: Industrial dynamics and urban growth as a branching process. Journal of Economic Geography, 7(5), 635–649. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Fujita, M., & Hu, D. (2001). Regional disparity in China 1985–1994: The effects of globalization and economic liberalization. The Annals of Regional Science, 35(1), 3–37. Glasmeier, A. (1991). Technological discontinuities and flexible production networks: The case of Switzerland and the world watch industry. Research Policy, 20(5), 469–485. Grabher, G. (1993). The weakness of strong ties; the lock-in of regional development in the Ruhr area. In G. Grabher (Ed.), The embedded firm: On the socioeconomics of industrial networks (pp. 255–277). London: Routledge. Hausmann, R., & Klinger, B. (2007). The structure of the product space and the evolution of comparative advantage (Cid Working Papers, No. 146). Cambridge, MA: Center for International Development at Harvard University. Henning, M., Stam, E., & Wenting, R. (2013). Path dependence research in regional economic development: Cacophony or knowledge accumulation? Regional Studies, 47(8), 1348–1362. Hidalgo, C. A., Klinger, B., Barabási, A. L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Jones, D. C., & Cheng, L. (2003). Growth and regional inequality in China during the reform era. China Economic Review, 14(2), 186–200. Kanbur, R., & Zhang, X. (2005). Fifty years of regional inequality in China: A journey through central planning, reform, and openness. Review of Development Economics, 9(1), 87–106. Klepper, S. (2007). The evolution of geographic structures in new industries. In K. Frenken (Ed.), Applied evolutionary economics and economic geography (pp. 69–92). Cheltenham: Edward Elgar. Lu, M., & Wang, E. (2002). Forging ahead and falling behind: Changing regional inequalities in post-reform China. Growth and Change, 33(1), 42–71. Martin, R., & Sunley, P. (2006). Path dependence and regional economic evolution. Journal of Economic Geography, 6(4), 395–437. Meyer-Stamer, J. (1998). Path dependence in regional development: Persistence and change in three industrial clusters in Santa Catarina, Brazil. World Development, 26(8), 1495–1511.

46

2 How Has Production Space Evolved in China?

Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Nooteboom, B. (2000). Learning and innovation in organizations and economies. Oxford: Oxford University Press. Schumpeter, J. A. (2000). Entrepreneurship as innovation. In Entrepreneurship: The social science view (pp. 51–75). Wei, Y. D. (1999). Regional inequality in China. Progress in Human Geography, 23(1), 49–59. Wenting, R. (2008). Spinoff dynamics and the spatial formation of the fashion design industry, 1858–2005. Journal of Economic Geography, 8(5), 593–614. Zhang, X., & Zhang, K. H. (2003). How does globalization affect regional inequality within a developing country? Evidence from China. Journal of Development Studies, 39(4), 47–67.

Chapter 3

How Does Regional Industrial Structure Evolve in China?

3.1

Introduction

The integration of evolutionary economics into economic geography has revitalized research interests in the spatial emergence of new industries and industrial evolution of regions (Boschma and Lambooy 1999; Neffke et al. 2011; Boschma et al. 2012; Isaksen 2014). It is argued that regions often evolve through a process of creative destruction (Schumpeter 1934; Garud and Karnoe 2001). On the one hand, regional industrial evolution is path-dependent. The emergence of new industries is determined by a set of competences and assets accumulated at the local level, and previous experience may affect the emergence and performance of these new industries (Buenstorf and Klepper 2009; Klepper and Simons 2000). Recent studies show that new industries evolve out of preexisting regional industrial structures and technological relatedness among industries affects the ways in which regions create new industries over time (Boschma and Frenken 2011; Neffke et al. 2011). On the other hand, regional industrial evolution is path creative, and exogenous linkages and institutions have an impact on the regional industrial evolution. External linkages bring not only investments but also new knowledge into regions, creating new paths for regional industrial development (Zhang 2013). Institutions clearly play a crucial role in the process of regional branching, in which firm diversification, labor mobility, and social networking are important mechanisms (Boschma and Frenken 2009; Boschma and Capone 2015). Government intervention may cause the distortion of regional production away from patterns of comparative advantages (Young 2000; Poncet 2005; He et al. 2008) and downplay the importance of path dependence in regional industrial evolution. With regional

Modified article originally published in [He, C., Y. Yan and D. Rigby (2018) Regional Industrial Evolution in China, Papers in Regional Science, 97 (2), pp. 173–98.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_3

47

48

3 How Does Regional Industrial Structure Evolve in China?

decentralization, local governments however are able to take advantage of market forces to develop technologically related industries. Previous empirical evidence on the role of technological relatedness and the discussion about institutions have largely originated from European regions, which operate under the mature market economies (Neffke et al. 2011; Boschma et al. 2012; Boschma and Capone 2015). However, the effect of technological relatedness on industrial evolution and the role of institution in transitional economies have not been explored, where these industries face poor legal framework, imperfect markets, and changing policy regimes (Estrin and Prevezer 2010; Mocnik 2010). Given that transitional economies are undergoing economic, social, and political transformations, both market forces and government interference play roles in economic development (Park et al. 2006), and local forces interact intensively with global forces. As a transitional economy, China has experienced revolutionarily economic and institutional transformations during the past three decades. The economic transition process has resulted in the liberalization of economies, which facilitates knowledge spillovers, labor mobility, industrial linkages, and the creation of a large number of privately owned firms. In line with economic liberalization, China has widely opened its economy to the international market, actively participating in international trade, warmly inviting foreign investments, and bringing global capital, new knowledge, and human capital into Chinese local development. Meanwhile, China has promoted regional decentralization, giving local government authorities and incentives to intervene in industrial development, which has caused the fragmentation of domestic markets and the distortion of regional production away from patterns of comparative advantages (Young 2000; Poncet 2005; He et al. 2008). The institutional transformation has certainly influenced the spatial emergence of new industries and regional industrial revolution in China. This chapter takes China as an example to explore the role of technological relatedness, global linkages, market liberalization, and state involvement in regional industrial evolution. Based on firm-level data of China’s manufacturing industries during 1998–2008, this chapter finds that Chinese regions that branch into related industries and unrelated industries are more likely to fail. New industries are more likely to enter regions which are globalized, economically liberalized, and fiscally healthy. Global linkages, economic liberalization, and state involvement have also created favorable conditions to allow a larger role of technological relatedness in driving industrial dynamics in China. Local governments are capable of taking advantage of market forces such as relatedness to attract and sustain related industries. This chapter is structured as follows. Following the introduction, the second section provides the literature review to link technological relatedness, global linkages, and regional institutions to regional industrial evolution. The third section introduces data sources and methods and describes the spatial emergence and exit of industries. This chapter then conducts an econometric analysis to test the significance of key variables and finally concludes with a summary of empirical findings.

3.2 Understanding Regional Industrial Evolution

3.2 3.2.1

49

Understanding Regional Industrial Evolution Relatedness and Industrial Evolution of Regions

Regions are often subject to a process of creative destruction identified as the key driving force behind industrial development (Neffke et al. 2011). Regions develop new industries to compensate for the decline and death of other industries. The question is how new industries emerge and whether new industries are geographically embedded. Proposing the theory of Window of Locational Opportunity, Storper and Walker (1989) argue that localization of new industries is rather independent from preexisting industrial structures. However, Scott (2006) argues that understanding the economic landscape must point to the idea of path-dependent economic evolution and recursive interaction. Massey (1984) views regional economies as the historical product of the combination of layers of activities. Recently, Boschma and his co-authors suggest that the spatial emergence of new industries is not entirely an accidental process (Boschma and Martin 2007; Boschma and Frenken 2011). New industries are built on a set of generic, location-specific resources that have the potential to trigger the emergence of new industries. This idea of path dependency has been deployed in studies of the persistence of regional economic disparities, the lock-in of regions to particular economic specializations, the revival and reinvention of former local industrial configurations, and the emergence and self-reinforcing growth of high-tech clusters (Martin and Sunley 2006). The evolutionary turn in economic geography has proposed that regional industrial evolution is a path-dependent process, whereby new industries grow out of preexisting industrial structures through technologically related localized knowledge spillovers (Boschma and Frenken 2011). Since the important contribution of Boschma (2005), there has been an increasing awareness that cognitive proximity is more important than geographical proximity for knowledge spillovers. Technological relatedness only occurs when firms in a region operate within technologically related industries that have overlapping knowledge bases (Boschma and Frenken 2011). Local knowledge spillovers are more likely to occur within regions that host a large number of technologically related industries. Technological relatedness is therefore an important enabling factor for the creation of new industries and formation of new regional industrial paths (Boschma and Wenting 2007; Boschma and Iammarino 2009; Neffke et al. 2011). Considering the importance of technological relatedness, a new stream of literature has investigated the industrial evolution of regions and proposed the regional branching process (Neffke et al. 2011; Boschma et al. 2012; Essletzbichler 2015). They demonstrate that regions introduce new technologies, products, and industries through a process of creative destruction (Martin and Sunley 2006; Boschma and Frenken 2011; Essletzbichler 2015). Technological relatedness acts as the main driver of this diversification process, in which a new sector spawns from related sectors (Klepper and Simons 2000) or from the recombination of capabilities from multiple related sectors (Klepper 2002). Regions tend to expand and diversify in

50

3 How Does Regional Industrial Structure Evolve in China?

sectors that are strongly related to their current activities. Klepper and Simon (2000) show that successful television producers are experienced radio producers prior to entering the television industry, indicating a high level of complementarity in the core competences and routines between the two industries. Boschma and Wenting (2007) confirm that technological relatedness to the regional knowledge base plays a large role in the localization of the British car industry. In Sweden, industries are more likely to enter regions with technologically related industries and that existing industries are more likely to exit regions where other industries are not technologically related (Neffke et al. 2011). In Spain, regions tend to diversify into new industries that use similar capabilities as existing industries in the regions (Boschma et al. 2012). In the USA, technological relatedness is found positively related to the metropolitan industry portfolio membership and the industry entry and is negatively related to the industry exit (Essletzbichler 2015). Analyzing the emergence of nanotechnology-based sectors at the regional level, Colombelli et al. (2014) support the idea that the technological competences accumulated at the local level are more likely to shape the future patterns of technological diversification. The notion of technological relatedness has put a renewed vision of the role of agglomeration externalities on the performance of firms, industries, and regions. However, the empirical evidences largely come from the developed economies including European regions and the USA. Meanwhile, this approach of regional industrial evolution puts too much emphasis on the role of existing industries and the local knowledge bases. One can question the influence of technological relatedness in developing and transitional economies, which are seeking any development opportunities, but lacking perfect market systems.

3.2.2

Global Linkages, Regional Institutions, and Industrial Evolution of Regions in China

The path-dependent approach only concerns the endogenous factors as regions develop through technologically related diversification or through the combination of existing competence (Isaksen 2014). This is reasonable for mature and developed economies, which are supported by knowledge-intensive sectors with good market systems. However, for transitional economies, exogenous factors often play a crucial role for regional industrial development. There are several ways for regions to create new paths of development, including recombinant innovation based on existing industrial or technological diversity (Frenken et al. 2007), investment and technology transfer from outside the region (Bathelt et al. 2004), and technological change and endogenous transformation of firms in the region (Todtling and Tripple 2005). In this chapter, path creation refers to the emergence of new industries which are less technologically related to the existing industrial portfolio. Exogenous factors such as global linkages, regional institutions, and government intervention are likely to bring or encourage the entry of such unrelated industries, creating a new development

3.2 Understanding Regional Industrial Evolution

51

path. Meanwhile, exogenous factors would interact with the endogenous factors to facilitate regional industrial evolution.

3.2.2.1

Global Linkages

Global linkages often bring new knowledge and competence into a region, creating new industries and breaking regional lock-in (Bathelt et al. 2004). Therefore, new industries may not be related to the existing industries as a result of global linkages like foreign investments and international trade (Henning et al. 2013; Isaksen 2014). Moreover, regions lacking technology and knowledge can make up for the defect through bringing external knowledge and linkages (Neffke et al. 2014). Boschma and Iammarino (2009) further argue that the inflows of external knowledge should be related, to some extent, to the industrial structure of a region to exert a significant economic impact. They provide empirical evidence that extra-regional knowledge leads to employment growth in Italian regions when the knowledge originates from related sectors but not similar to the sectors that are already present in the regions. This is because knowledge spillover from one sector to another only occurs when the sectors are complementary in terms of shared competences (Bunnell and Coe 2001). Globalization is an important catalyst for transformation in transitional economies, providing financial resources, technologies, knowledge, management skills, and markets, which are necessary to transform and restructure the obsolete industrial systems inherited from the commanding economy (He et al. 2008). Global-local linkages can be established through many different channels, including trade exchanges, inward and outward FDI, and involvement in global production networks and technological alliances (Boschma and Iammarino 2009). Specifically, foreign investments and exports are two critical ways to channel new knowledge. On the one hand, foreign investments may come with new industries, creating a new regional industrial path. On the other hand, foreign investments have considerable knowledge spillover effects within and across industries through the effects of demonstration and competition, labor mobility, and business linkages (Blomström and Kokko 1998). Exporting activities can expose regions to the global market. Exporters are found not only to take advantage of industrial linkages through deeper division of labor to foster the formation of industrial clusters (Fujita and Hu 2001) but also to benefit from information spillovers derived from labor mobility, spatial agglomeration, technological imitation, and the diffusion of exporting experience (Trevor et al. 2007). In a word, regions heavily engaged into the globalization process are more likely to create new paths for regional industrial development. China has been successfully integrated into the global economy and actively involved into the process of globalization through utilizing foreign investments and trading with other countries. By now, China has been the top player in the global market. The global-local interaction has resulted in a larger number of foreign firms in Chinese cities, which bring new management and superior technology and would enhance positive spillover effects to local Chinese firms (Cheung and Lin 2004; Liu 2008). For instance, more than 30 Chinese cities can produce automobiles, which are

52

3 How Does Regional Industrial Structure Evolve in China?

often introduced by foreign joint ventures. It’s expected that Chinese regions would be more likely to develop new industries when they are heavily engaged into the globalization process.

3.2.2.2

Economic Liberalization

Institutional frameworks exert direct impacts on the type of industries which regions specialize in and on the role of technological relatedness (Hall and Soskice 2001). Provided that firm diversification, spin-offs, labor mobility, and social networking are important mechanisms for the process of regional branching (Boschma and Frenken 2011), regional institutions clearly play a crucial role for those mechanisms to work. Recent studies show that relatedness is a stronger driver of industrial emergence in the countries that focus more on “nonmarket” coordination in the domains of labor relations, corporate governance relations, product market relations, and interfirm relations (Boschma and Capone 2015). Hall and Soskice (2001) claim that firms and other actors in coordinated market economies are more willing to invest in specific and cospecific assets, while liberal market economies should invest more extensively in switchable assets. For transitional economies which are transforming from a command economy to a market-oriented economy, institutions are critical to allow the role of technological relatedness (He and Pan 2010). For instance, studying the emerging Internet industry in Beijing and Shanghai, Zhang (2013) argues that a region’s enduring political institutional embeddedness significantly affects the generation and evolution of their related varieties. Since the late 1970s, China has embarked on the road to economic reform with the goal of building a market-oriented economy that allows market forces into play. In the commanding economy, there were literally no well-functioning markets. Factor mobility was strictly limited, and there were no channels for knowledge spillovers. Most firms were owned by the state and firm leaders acted as both managers and government officials (Freund 2001; Qian 1996). Instead of pursuing profitability and efficiency, state-owned entrepreneurs (SOEs) were more likely to adopt strategies to fulfill the administrative tasks and pursue empire-building strategies at the expense of efficiency (Li and Xia 2008). Here administrative tasks were mostly heavy social responsibilities, including lifetime employment practices and employing urban residents not required by firms (Broadman 1996; Zhu 1999). Together with soft-budget constraints, SOEs tended to perform less efficiently (Li and Xia 2008). However, the liberalization and privatization have indeed corrected many of these inefficiencies in resource allocation and unlocked the productive potential of firms (Qian 2000). Liberalization of economy allows labor mobility, facilitates industrial linkages, and revitalizes entrepreneurship in China, generating favorable conditions to stimulate knowledge spillovers among firms and creation of new industries (He and Pan 2010). Meanwhile, many state-owned enterprises (SOEs) were wiped out, and non-state-owned sectors have developed quickly.

3.2 Understanding Regional Industrial Evolution

53

However, the process of marketization is regionally unbalanced in China (Han and Pannell 1999). Some coastal regions are more economically liberalized, with strong market forces in their economic systems. Empirical studies find strong interplant business linkages and business networks in economically liberalized regions such as Zhejiang, Jiangsu, and Guangdong (Yang and Liao 2010; Wei et al. 2010; Zhu and He 2013). The wide distribution of industrial clusters in the coastal provinces provides strong evidence that technological relatedness does play a critical role in regional industrial development. However, in the large inland regions with a significant share of state-owned sectors, market forces are very weak. In those regions, industrial linkages are relatively hard to establish, and knowledge spillover effects are marginalized. New industries are often introduced by personal networks or governmental interventions (Qiu 2005) and may not be related to the existing industrial structure. Consequently, the expected positive role of technological relatedness in regional industrial evolution depends on the extent of economic liberalization in Chinese regions.

3.2.2.3

State Involvement

State involvement can change the direction of industrial evolution. According to the dynamic comparative advantage theory, positive government intervention can promote industries with potential competitiveness in the future (Lin 2012). Stiglitz et al. (1996) take Eastern Asian countries as cases to illustrate the effectiveness of public policies in introducing new industries into regions. By providing investors a variety of fiscal, administrative, and other supports, local governments may attract new industries which are not related to the existing industries in that area. In transitional economies characterized with weak market forces, local governments play a more important role and tend to favor a certain industry. China’s economic transition has resulted in considerable regional decentralization, which gives more power and incentives to local governments. China has decentralized its fiscal system by introducing the tax-sharing system in 1994, which has fundamentally altered China’s central-local fiscal relations (Zhang 1999; Wong 2000). Fiscal decentralization has substantially raised the central share in revenue and reduced that of local governments (Lin and Yi 2011), so local governments have to largely selffinance their development. While local governments obtain economic decision power, the central government still makes decisions on the appointment and removal of local officials according to their economic performance (Li and Zhou 2005). Regional decentralization together with economic-oriented evaluation system has triggered fierce interregional competition in promoting economic growth (Xu 2011), which may have resulted in two effects. On the one hand, it may lead to local protectionism and rational imitation strategy of industrial policies (Zhao and Zhang 1999), which would cause the fragmentation of domestic markets and the distortion of regional production away from patterns of comparative advantages and technological competence (Young 2000; Poncet 2005). Therefore, decentralization may downplay the importance of technological relatedness in regional industrial

54

3 How Does Regional Industrial Structure Evolve in China?

evolution in China. On the other hand, local governments are more likely to provide subsidies including income tax breaks, rebates of value-added tax, and import duties for equipment purchases, low-priced land and firms located in special development zones, and cash payments to firms based on factors such as export performance (Barbieri et al. 2012) to attract new investment and industries. Under fiscal decentralization, regions with a bad fiscal condition have stronger incentives to attract industries without considering the existing competence and knowledge bases. To improve their competitiveness, fiscally healthy regions can provide a variety of subsidies to attract related industries, which are better to sustain rapid economic growth. To summarize, the evolutionary turn in economic geography has suggested that regions would diversify into industries related to the existing portfolio of industries and regional industrial evolution is a path-dependent process (Frenken and Boschma 2007; Boschma and Frenken 2011; Neffke et al. 2011). Technological relatedness is a crucial enabling factor for the creation of new industrial variety in developed economies. However, for transitional economies such as China, exogenous factors often play a crucial role for regional industrial development. Global linkages may bring new industries and new knowledge into Chinese regions, creating new regional paths. Economic liberalization would create conditions to enable the role of technological relatedness. State involvement can either bring unrelated industries or facilitate the entry of related industries. This chapter will investigate the entry of new industries and the exit of industries in Chinese prefectures and test the significance of technological relatedness, global linkages, economic liberalization, and state involvement.

3.3

Data and the Relatedness Indicator

In this chapter, we use the ASIF dataset. Based on the cleaned firm-level data, we compute 4-digit manufacturing industry data for all 337 prefectures using the number of firms, gross industrial output, and total employment. There are 424 4-digit manufacturing industries. No regions host all 424 industries. During 1998–2008, some prefectures gain new industries, while others lose industries. We define industry entry and exit as follows. If industry i is not in prefecture r in year t but in prefecture r in year t+c, then industry i is a new entry for prefecture r. Correspondingly, if industry i is in prefecture r in year t but disappears in year t+c, then industry i is an exit. There could be potential selection bias since the dataset does not cover small manufacturing firms. However, the included firms can account for more than 90% in terms of gross industrial output. Moreover, the official statistics of Chinese industries in a variety of statistical yearbooks are derived from this dataset. We believe that it is sufficient to investigate the regional industrial entry and exit using this dataset. The key variable in this chapter is inter-industry technological relatedness. Generally, there are three distinct approaches to measure inter-industry relatedness.

3.4 Descriptive Analysis

55

The first approach is derived from the hierarchical structure of the standard industrial classification system. However, this approach has been heavily questioned because it is based on the assumption that the hierarchical structure of industry classifications reflects the existence of scale economy (Neffke et al. 2011). The second approach utilizes resource-based indicators to capture the similarities of resources used in different industries, such as commodity flows, human capital, and technological resources (Fan and Lang 2000; Breschi et al. 2003). Since different resources play different roles in industries, a number of resource-based indicators are biased toward specific industries. The third approach is analyzing the co-occurrence of industries in portfolios. An advantage of this approach is that they combine the distributed information about scale economies that are strategically related to individual actors and could identify scope economies at even firm level. However, a major disadvantage is that co-occurrence approach is outcome based, which means this method first assumes that portfolios are coherent and then infers the implied relatedness (Neffke and Henning 2013). In this chapter, we follow the co-occurrence analysis by assessing the conditional probability of two industries located in the same location to measure the inter-industry relatedness at the prefecture level. The major reason to use co-occurrence approach is that we would like to calculate the relatedness between 4-digit manufacturing industries and other approaches would not achieve this requirement under the constraints of dataset in China. We use Eqs. 1.3 and 1.4 to calculate relatedness (ϕi,j) between industry i and j and the density indicator of industry i in city c (densityi,c). However, the co-occurrencebased measure of technological relatedness is not without limitation since other factors besides relatedness may partly determine the number of co-occurrences (Neffke et al. 2012). For example, if industries are very large, they are likely to be found in many firms, and, therefore, they will also more frequently co-occur with other industries. To reduce the impact of other factors, we use a higher standard for co-occurrence-based relatedness. At the 4-digit industry level, we are confident that two industries with higher probabilities of co-location are more likely to be technologically related.

3.4 3.4.1

Descriptive Analysis Industrial Entries and Exits in China

Based on 4-digit industry data, we compute the number of industry entries and exits for all Chinese prefectures. The average number of industrial entries and exits per prefecture is further derived and shown in Table 3.1. From the year 1998 to 2008, China has observed considerable industrial restructuring. There has been an industrial diversification process and more observations of industrial entries rather than exits in different years. For instance, in 2008, on average, there are 18.22 4-digit industries entering Chinese prefectures but 16.64 industries exiting. Notably, there is a spike in industrial entries and exits in the year 2004, in which the second national

56

3 How Does Regional Industrial Structure Evolve in China?

Table 3.1 Average number of new entries and exits in Chinese prefectures Year 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Number of industry entries 13.71 8.36 15.50 10.93 17.30 25.96 11.62 11.83 11.97 18.22

East 18.72 13.61 21.93 16.49 25.30 40.16 14.12 15.34 14.09 21.82

Mid 16.57 8.12 16.28 12.07 19.98 27.42 15.74 14.62 15.26 24.94

West 7.56 4.49 9.93 5.73 8.99 13.85 6.39 6.89 7.70 10.05

Number of industry exits 8.80 11.18 10.07 8.43 12.71 14.71 8.65 5.40 4.97 16.64

East 12.31 11.82 16.22 12.19 16.22 20.63 10.84 6.66 6.02 31.23

Mid 10.10 13.99 10.55 9.49 16.35 18.32 10.92 6.50 5.91 12.99

West 5.05 8.44 4.95 4.68 7.08 7.24 5.14 3.53 3.40 8.31

Fig. 3.1 Industrial diversification of Chinese prefectures (Left 1998, Right 2008)

economic census was conducted. More firms may be included in the dataset. However, this does not alter the empirical results. The coastal region is rather dynamic, seeing many more industrial entries and exits. The western regions experience slower industrial restructuring process, with fewer new entries and exits. Industrial diversification however differs in Chinese prefectures. Figure 3.1 shows the Theil index based on the employment of 4-digit manufacturing industries in 1998 and 2008 in Chinese prefectures. In 1998, prefectures were industrially diversified in the coastal region, especially the Jing-Jin-Ji area, Shandong, Jiangsu, Shanghai, and Zhejiang. Inland regions however are rather industrially specialized. Ten years later, the central region has considerably improved its diversification of industries, especially Jiangxi, Anhui, Henan, and Chongqing. The coastal region has also become more diversified, especially Liaoning, Fujian, Guangdong, and Guangxi provinces. The remarkable spatial variation in the industrial diversification is the result of industry entry and exit. To further understand the spatial industrial diversification process in China, we map the number of industry entries and industry exits during 1998–2003 and during

3.4 Descriptive Analysis

57

Fig. 3.2 Number of industry entries (Left 1998–2003, Right 2003–2008)

Fig. 3.3 Number of industry exits (Left 1998–2003, Right 2003–2008)

2003–2008 at the prefecture level (Figs. 3.2 and 3.3). An industry not present in 1998 but in 2003 is considered an entry, while an industry reported in 1998 but not in 2003 is an exit. The same logic is used for the period of 2003–2008. China entered the WTO in 2001 and has been actively engaged into the globalization process, signifying the new stage of economic development. Since then, a large number of foreign direct investments have flowed into China, and exports have seen exponent growth. The globalization may facilitate the industrial restructuring process. To see the effect of globalization, we divide the time period into two stages using the year 2003 as the critical year. New industries largely entered the coastal provinces, such as Shandong, Zhejiang, and Fujian provinces, during 1998–2003. Many new industries moved into the central region and the Northeast China during 2003–2008. On the one hand, China’s regional policies have shifted to help the revitalization of central region and the Northeast. On the other hand, the regional shift of industries can be accredited to the further development of globalization and marketization, which push the traditional industries into the inland region (He and Wang 2012).

58

3 How Does Regional Industrial Structure Evolve in China?

In terms of industry exit, the period of 1998–2003 saw more exits than the period of 2003–2008. During 1998–2003, many industries exit from the coastal and central provinces, particularly Hebei, Jiangsu, Hunan, Hubei, and Heilongjiang. The higher exit rate of industries in this period may be related to the entry of WTO, which has brought international competition into the domestic markets. Before the accession of WTO, industries in the central region were less exposed to the international competition. Less competitive industries are more likely to fail in front of foreign direct investments and imports. During 2003–2008, the coastal and central regions still observed more industry exits than the western regions. Industry exit has however slowed down. Overall, both industry entry and industry exit show strong spatial variations in the two time periods, indicating that localized forces play a role in industrial dynamics. There is also strong east-west divide, matching with the spatial divide of the globalization and marketization in China. Global forces and market forces can shed essential lights on regional industrial evolution in China.

3.4.2

Relatedness and Regional Industrial Evolution in China

Regions would branch into industries that are technologically related to the existing industrial structures. To examine the influence of relatedness on industry dynamics, we first examine the relationship between the average relatedness of entry and nonentry industries. Entry is defined as an industry that is not in a prefecture in the previous year but in a prefecture in the current year. Nonentry refers to an industry that is not in a prefecture in the previous and current years. We make the scattered plots, with the y-axis for the average relatedness of entries and the x-axis for the average relatedness of nonentries. The line indicates that y is equal to x. Each dot indicates one prefecture. If a dot is located in the left side of the line, then the average relatedness of entry is stronger than that of nonentry and vice versa. Figure 3.4 shows that almost all dots are located in the upper left of the line during 1999–2008, suggesting that entry has stronger relatedness than nonentry. To put it another way, industries with strong technological relatedness with the industry portfolio are more likely to enter a prefecture. In the same way, we examine the relationship between relatedness and industry exits. Exit is defined as an industry that is in a prefecture in the previous year but not in a prefecture in the current year. Non-exit refers to an industry that is in a prefecture in the previous and current years. In Fig. 3.5, the y-axis is the average relatedness of exiting industries, while the x-axis stands for the average relatedness of non-exits. The line indicates that y equals x. Most of the dots are now located in the right side of the line, indicating that exiting industries hold weaker relatedness with the existing industrial portfolio than non-exits. It is clear that industries are more likely to exit if they do not enjoy strong localized technological relatedness. Figures 3.4 and 3.5

3.5 Model Specifications and Empirical Findings

59

Fig. 3.4 Relationship between average relatedness of entry industries and nonentry industries

Fig. 3.5 Relationship between average relatedness of exit and non-exit industries

provide preliminary evidence to support the role of technological relatedness in regional industrial evolution in China.

3.5 3.5.1

Model Specifications and Empirical Findings Model Specifications

The descriptive analysis provides evidence to show that regions attract new industries that are technologically related to the existing industries while those less related are more likely to exit. We conduct an econometric analysis to test the importance of technological relatedness, global linkages, economic liberalization, and state involvement on industry entries and exits of Chinese regions. We define the following Logit model:

60

3 How Does Regional Industrial Structure Evolve in China?

ENTRYc,i,tþ5 ðor EXITc,i,tþ5 Þ
¼ β0 þ f 1 Densityr, i, t ; WDensityc, i, t þ f 2 ðFDIc, t ; TRADEc, t ; LIBc, t ; DEXP
 c, t ; DLANDc, t Þ þ f 3 Densityc, i, t ; WDensityc, i, t  ðFDIc, t ; TRADEc, t ; LIBc, t DEXPc, t Þ þ f 4 Control þ αc þ αi þ αt þ εc, i , t ð3:1Þ where t is 1998 and 2003 and ENTRYc, i, t + 5 is a dummy variable, 1 for industry entry (industry i enters prefecture c in year t+5). We also run the same model specification using the dependent variable EXITc, i, t + 5, which is also a dummy variable, 1 for industry exit (industry i exits prefecture c in year t+5). If industry i is not in prefecture c in 1998 (or 2003) but in prefecture c in 2003 (or 2008), then industry i is a new entry, and ENTRYr, i, t + 5 will be granted unity. If industry i is in prefecture c in 1998 (or 2003) but disappears in 2003 (or 2008), then industry i is an exit and EXITc, i, t + 5 will be given the unity. The key variable in the models is Densityc,i,t, defined as the technological relatedness of industry i with all other industries in prefecture c in year t. Densityc, i,t is expected to have a positive coefficient in the entry equation but a negative coefficient in the exit equation. We also expect cross-regional knowledge spillovers to influence industry dynamics. Technological relatedness in the neighboring prefectures may foster the entry of new industries. To test the spatial dependence of regional industrial evolution, we include the largest technological relatedness of industry i in the prefectures which share the common borders (WDensityc,i,t). As argued that global linkages would bring new industries and new knowledge into regions. Regions actively participating in the process of globalization would be more likely to develop new industries and expect fewer industry exits. To quantify the extent that a prefecture is engaged in the process of globalization, we compute the share of gross industrial output by foreign firms (FDIc, t) and share of exports in gross industrial output (TRADEc, t) to measure the globalization linkages of Chinese prefectures. Both variables are expected to have positive coefficients in the entry equation but negative coefficients in the exit equation. This chapter also argues that economic liberalization and state involvement condition the role of technological relatedness. Following the existing studies, we use the ratio of non-state-owned enterprises in the gross industrial output (LIBc,t) to quantify the extent of economic liberalization. We introduce the ratio of fiscal revenue and fiscal expenditure (DEXPc,t) and the ratio of land leasing fee and fiscal revenue (DLANDc,t) to measure the incentive of state involvement. The fiscal decentralization has substantially raised the central share in revenue and reduced that of local governments. With fiscal decentralization, local governments have strong incentives to secure extra-budget revenues. One of the major extra-budget revenues is the land leasing fee that is under the direct control of municipal and county governments (Lin 2007). We will introduce the interactions between LIBc,t

3.5 Model Specifications and Empirical Findings

61

Table 3.2 Definition of explanatory variables Variables Densityc,i,t WDensityc,i,t FDIc,t TRADEc,t LIBc,t DEXPc,t DLANDc,t LOCALc,i,t URBANc,t

Definitions Technological relatedness of industry i with all other industries in prefecture c in year t The maximum technological relatedness of industry i in neighboring prefectures in year t The share of gross industrial output by foreign firms in prefecture c in year t The share of gross industrial output by export firms in prefecture c in year t Ratio of non-SOEs in gross industrial output in prefecture c in year t Ratio of fiscal revenue and fiscal expenditure in prefecture c in year t Ratio of land leasing fee and fiscal revenue in prefecture c in year t Employment of two-digit industry i belongs to in prefecture c in year t Population in prefecture c in year t

(or DEXPc,t, DLANDc,t, FDIc,t, TRADEc,t) and Densityc,i,t in both equations to see the impacts of exogenous factors on the significance of technological relatedness. In addition, we control urbanization economy (LOCALc,i,t) and localization economy (URBANc,t) which will effect industry entry and exit. We also add a dummy for WTO entry and control the 2-digit industry dummies and province dummies. All variables are summarized in Table 3.2.

3.5.2

Empirical Results

Correlation analysis shows that explanatory variables are only moderately correlated (Table 3.3). There is no serious concern about collinearity issue in the model estimations. Since the dependent variable is a dummy, we apply the Logit model to identify the significance of the explanatory variables. There should be no existence of endogenous issue since we measure the explanatory variables in the initial year (1998 or 2003). Logit regression results for industry entry and exit equations are reported in Tables 3.4 and 3.5, respectively. All models are significant and perform well. The coefficients of technological relatedness (Density) are separately 9.54 and 9.21 and significant both in the industry entry model and exit model. It indicates that an industry is more likely to enter and less likely to exit a prefecture when it is strongly and technologically related to the existing industry structure. This finding evidently suggests that technological relatedness is an enabling factor to generate related variety in Chinese prefectures. The evidence of significance of technological relatedness from a transitional economy provides important complementarity to the existing literature based on American and European regional data in evolutionary economic geography (Neffke et al. 2011; Boschma et al. 2012; Essletzbichler 2015). Given the institutional framework with imperfect markets, weak legal environment, fierce interregional competition, government intervention, and changing policy

Density WDensity FDI TRADE LIB DEC DLAND LnLOCAL LnURBAN

Density 1 0.9768 0.0252 0.0071 0.0244 0.0418 0.0034 0.0449 0.061

1 0.0195 0.0135 0.0200 0.0451 0.0020 0.0549 0.0318

WDensity

1 0.3841 0.4758 0.2334 0.1008 0.1410 0.0856

FDI

Table 3.3 Correlation coefficients among independent variables

1 0.0194 0.4141 0.2041 0.2443 0.0723

TRADE

1 0.0235 0.3547 0.0847 0.1640

LIB

1 0.0436 0.3288 0.1671

DEC

1 0.0919 0.1279

DLAND

1 0.3375

LnLOCAL

1

LnURBAN

62 3 How Does Regional Industrial Structure Evolve in China?

0.30*** 0.41*** Included Included Included 7.89*** 22486.27 0.18 5.20E+04 149,553

Model1 9.54*** 1.12**

0.27*** 0.45*** Included Included Included 8.48*** 22946.21 0.18 5.20E+04 149,553

0.56***

Model2 9.20*** 1.54*** 1.01***

0.27*** 0.45*** Included Included Included 8.35*** 22961.26 0.18 5.20E+04 149,553

Model3 8.60*** 1.60*** 1.29*** 1.33* 0.14 2.03***

0.30*** 0.40*** Included Included Included 8.01*** 22541.18 0.18 5.20E+04 149,553

0.33***

Model4 9.72*** 0.92

Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density WDensity FDI Density*FDI TRADE Density*TRADE LIB Density*LIB DEXP Density*DEXP DLAND Density*DLAND LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.4 Logit regression results for entry equations

0.31*** 0.40*** Included Included Included 7.89*** 22547.76 0.18 5.20E+04 149,553

0.1 1.13**

Model5 9.07*** 0.95*

0.22*** 0.49*** Included Included Included 9.29*** 23152.09 0.19 5.00E+04 145,894

0.04

1.96***

Model6 11.87*** 0.81

1.85*** 0.71 0.44*** 2.46*** 0.22*** 0.50*** Included Included Included 9.06*** 23253.54 0.19 5.00E+04 145,894

Model7 10.47*** 0.61

0.21*** 0.50*** Included Included Included 9.57*** 23276.12 0.19 5.00E+04 145,894

0.02

1.81***

0.46***

0.33***

Model8 11.80*** 0.76 0.15*

Model9 9.89*** 0.49 0.33* 0.7 0.05 1.25** 0.36*** 0.44 1.73*** 0.55 0.42*** 2.27*** 0.21*** 0.51*** Included Included Included 9.24*** 23382.37 0.19 5.00E+04 145,894

3.5 Model Specifications and Empirical Findings 63

0.39*** 0.12*** Included Included Included 4.33*** 10772.85 0.15 3.20E+04 66,895

Model1 9.21*** 0.01

0.35*** 0.16*** Included Included Included 4.85*** 11011.67 0.15 3.10E+04 66,895

0.68***

Model2 9.11*** 0.32 0.75***

0.35*** 0.15*** Included Included Included 3.91*** 11187.9 0.15 3.10E+04 66,895

Model3 6.07*** 0.49 0.17 4.22*** 0.95*** 6.50***

0.39*** 0.12*** Included Included Included 4.41*** 10782.22 0.15 3.20E+04 66,895

0.19***

Model4 9.31*** 0.13

Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density WDensity FDI Density*FDI TRADE Density*TRADE LIB Density*LIB DEXP Density*DEXP DLAND Density*DLAND LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.5 Logit regression results for exit equations

0.39*** 0.12*** Included Included Included 3.84*** 10833.54 0.15 3.20E+04 66,895

0.83*** 4.09***

Model5 6.98*** 0.04

0.32*** 0.16*** Included Included Included 4.93*** 10896.11 0.15 3.10E+04 65,562

0.25***

1.25***

Model6 9.85*** 0.29

0.52*** 7.28*** 0.74*** 2.08*** 0.33*** 0.15*** Included Included Included 3.63*** 11077.13 0.15 31,000 65,562

Model7 4.74*** 0.03

0.31*** 0.18*** Included Included Included 5.20*** 11000.91 0.15 31,000 65,562

0.29***

0.85***

0.22***

0.57***

Model8 9.38*** 0.25 0.46***

Model9 4.02*** 0.59 0.28 3.67*** 0.37* 3.58*** 0.13 0.25 0.25 4.56*** 0.66*** 1.55*** 0.31*** 0.17*** Included Included Included 3.76*** 11235.42 0.16 31,000 65,562

64 3 How Does Regional Industrial Structure Evolve in China?

3.5 Model Specifications and Empirical Findings

65

regimes in China, relatedness still plays a significant role in regional diversification process. Our findings suggest that regional industrial development is a pathdependent process and is determined by historical layers of economic activities even in transitional economies (Massey 1984; Martin and Sunley 2006). Moreover, the importance of technological relatedness is not only just confined to regions themselves but also reflected through cross-regional knowledge spillovers in industrial development. The coefficient of technological relatedness in neighboring regions (WDensity) is 1.12 and significant in industry entry model, implying that cross-regional knowledge spillovers would stimulate the emergence of new industries. However, compared with the influence of direct relatedness in the localities, the effect of technological relatedness in neighboring regions is relatively much weaker. Meanwhile, the coefficient of WDensity is not significant in the exit models, which proves that cross-regional knowledge spillover effects only occur when new industries entry. There are several possible channels for the cross-regional spillover effects. First, Chinese regions are often engaged into an imitation behavior by copying the successful industrial development strategy in the neighboring regions. Thun (2004), for instance, has argued that local governments have incentives to duplicate industries that could rapidly improve local revenues or growth through a process of rational imitation. Second, technologically related industries often build regional business networks, which has been observed in the Pearl River Delta, Yangtze River Delta, and Jing-Jin-Ji area (Wei et al. 2010; Yang and Hsia 2007). Moreover, successful entrepreneurs may expand their business into neighboring regions by taking advantage of the geographical proximity. Cross-regional knowledge spillovers help create regional production networks, improving industrial productivity at the regional level. This will broaden the geographical scope of path dependence in industrial development. However, Chinese regions can boost new entries by establishing global linkages through utilizing foreign investments and participating in international trade. The coefficients of FDI and TRADE are 1.01 and 0.56 and significant in the entry model, and the coefficients of these two variables are 0.75 and 0.68 and significant in the exit model. These findings suggest that global linkages could nurture the development of new industries and discourage the industry exits when regions are deeply engaged in globalization and have more connections with foreign firms. On the one hand, foreign investment would bring advanced technology, human capital, and management skills according to the ownership advantage theory and have spillover effects to the local enterprises (Bathet et al. 2004; Boschma and Iammarino 2009). For instance, many former employees in foreign firms would start their own business related to activities of foreign firms. On the other hand, foreign firms may stimulate the entry of downstream and upstream industries to form the localized business networks and extend their business along the value chain (Yeung et al. 2006; He et al. 2011). Meanwhile, exporting activities would build international linkages, introducing new knowledge to facilitate the development of new industries. These external linkages can foster the emergence of new industries in Chinese regions. Moreover, the global linkage can moderate the role of technological relatedness in triggering industrial evolution. We add the interaction terms between

66

3 How Does Regional Industrial Structure Evolve in China?

technological relatedness (density) and foreign investment (FDI) and trade intensity (TRADE) in the models. The coefficient of term Density*FDI is 1.33 and significant in industry entry model, indicating that the effect of FDI on industrial entry will decline with the rising of the extent of technological relatedness in Chinese regions. It is acknowledged that FDI is mainly located in the coastal regions in China, which have more advanced industrial structure and relatively stronger inter-industry relatedness. For the regions with higher relatedness, which means that these regions would be more likely to have more FDI, the marginal spillover effect of FDI will be lower than regions with weaker relatedness. However, the coefficient of term Density*TRADE is 2.03 and significant in the entry model, which indicates that exporting activities would reinforce the role of technological relatedness. As widely known, Chinese exports significantly benefit from industrial clusters and the deep division of labor along the value chain (He et al. 2015). New industries would be attracted to regions with higher export intensity to benefit from technological relatedness. Notably, both terms of Density*FDI and Density*TRADE are significant and negative in exit model, which indicates that related globalization can discourage the industrial exit. This is consistent with Boschma and Iammarino (2009) which argue that the inflow of external knowledge should be related to some extent to the industrial structure of a region to affect industrial dynamics. As a transitional economy, regional institutions play a crucial role in regional industry development. Statistical results show that economic liberalization fosters the emergence of new industries in Chinese regions. Industries are also less likely to exit in economically liberalized regions. Economic liberalization allows market force to play its role in allocating resources, creating favorable conditions for the entrepreneurs to start new business. Moreover, the coefficient of term Density*LIB is 2.03 and significant, indicating that technological relatedness plays a larger role in encouraging industry entry in more economically liberalized regions. Related industries are more likely to enter the more economically liberalized regions. In transitional economies, market-based institutions allow a more crucial role of technological relatedness in regional industrial evolution, leading to path dependence of regional industrial development. This anchors the argument in He and Pan (2010) which report that economic transition has created conditions to allow a larger role of externalities in stimulating city-industry growth. In less liberalized regions, resources are not fully allocated by markets. The state controls a huge amount of resources such as land and bank loans and so on, and firms which have good connections with local officials would be favored. Moreover, as the main actors of markets, SOEs have their own internal labor market characterized with long-term employment, internal promotion, and deferred remuneration. Meanwhile, technology is highly protected through contracts with employees, and there is seldom technology spillover to other firms. Unrelated industries are more likely to enter the less liberalized regions. The new path however is not necessarily sustainable since new industries may not fit the pattern of comparative advantages and do not share the local knowledge bases and local competence. Moreover, when both LIB and Density*LIB are included in the exit model, Density*LIB has a negative coefficient, while LIB has a positive one, indicating that economic liberalization

3.5 Model Specifications and Empirical Findings

67

can only sustain related industries. The intensified market competition in economically liberalized regions would force unrelated industries to fail. State involvement has significant effects in industrial dynamics of Chinese regions. Regional decentralization has created great incentives for local governments to pursue revenues through the development of industries. The coefficients of DEXP in both entry and exit models are 1.96 and 1.25, which suggests that new industries are more likely to enter regions with better a fiscal situation, which also discourage related industries to exit. This indicates that pro-business state involvement would create favorable conditions to let technological relatedness play its role in industrial development. With the support of heathy fiscal situations, local governments can use subsidies, tax waiver, and other incentives to choose both related and unrelated new industries. For regions with a considerable fiscal deficit, local governments are hard to grant economic incentives to new businesses. They are not in the position to select new industries based on the existing knowledge and competence. Correspondingly, industries, particularly those that are technologically related to the existing industrial structure, are less likely to exit from the regions with healthy fiscal situation since they can provide subsidies to keep industries and even less productive firms (He and Yang 2016). Fiscal decentralization has triggered the interregional competition in attracting business. Fiscally healthy regions would have more chance to develop new industries, especially those to some extent linked to the current industries. Pathdependent process would occur only when local governments are in a better position to select industries based on their regional capabilities and competence. In the last decades, local governments have successfully mobilized land development to improve their fiscal situation. Land development has directly stimulated economic growth and has been used as a tool to attract foreign investments and to sustain infrastructure investments, indirectly triggering economic growth (He et al. 2014). The coefficient of DLAND is 0.25 and significant in exit model. Density*DLAND however has a positive coefficient in the entry model but a negative coefficient in the exit model. This suggests that more revenues from land development would help regions to attract and sustain technologically related industries, which provide further evidence that state involvement has taken advantage of market forces such as technological relatedness to diversify regional industrial structure.

3.5.3

Robustness Checks

We further check the robustness of technological relatedness. First, we define the industry entry and exit by 1 year rather than 5 years and create the annual data. Second, in the measurement of relatedness, we apply LQ ¼ 0.5 to indicate revealed comparative advantage. We then apply LQ ¼ 1.0 for industries with revealed comparative advantages to derive the measurement of relatedness. To reduce the impact of endogenous problem, we lagged all variables by 1 year. All checks confirm the importance of technological relatedness, global linkages, economic

68

3 How Does Regional Industrial Structure Evolve in China?

liberalization, and state involvement in regional industrial evolution. The interaction terms provide more robust evidence to support the moderating role of economic liberalization and regional decentralization. The estimated results in Tables 3.6, 3.7, 3.8, and 3.9 are not significantly different from Tables 3.4 and 3.5.

3.6

Summary and Discussion

EEG considers regions evolve through technologically related diversification, indicating that regional industrial development is path-dependent. Meanwhile, we argue that the path-dependent approach only concerns the endogenous factors but ignores the exogenous factors such as global linkages, regional institutions, and state involvement, which may bring unrelated industries, thus creating new path of regional industrial development. Based on firm-level data of China’s manufacturing industries during 1998–2008, this chapter examined the industrial diversification through the lens of entry and exit of 4-digit industries at the Chinese prefecture level. Chinese regions have undergone considerable industrial diversification, particularly in the coastal and central regions. During the study period, many industries enter and exit from the coastal and some part of the central regions. Statistical analysis suggests that Chinese regions will branch into new industries technologically related to the existing industries and related industries are more likely to survive in Chinese regions. Further analysis reveals that global linkages including utilizing foreign investment and international trade, economic liberalization, and state involvement have created favorable conditions to allow a larger role of technological relatedness in driving industrial dynamics in China. Local governments are able to take advantage of market forces to diversify regional industries. The significance of technological relatedness implies that regional industrial evolution in China is path-dependent. Meanwhile, exogenous factors have also generated opportunities for Chinese regions to introduce unrelated industries, creating new paths of industrial development. Our findings suggest that new industries are more likely to enter regions which are globalized, liberalized, and fiscally healthy. Global linkages could bring new industries and new knowledge to Chinese regions. Foreign investment and international trade can make up for the lack of technological relatedness through bringing external knowledge and linkages and foster the emergence of new industries. Economic liberalization inspires local entrepreneurship, encouraging the emergence of new industries. Fiscal decentralization creates economic incentives for local governments that help local development and attract new industries, which can quickly generate local revenues and promote economic growth. This chapter contributes to the literature of EEG in several ways. First, this chapter provides empirical evidence on the role of technological relatedness in regional industrial evolution from a transitional economy, generating additional support for the path dependence argument in evolutionary economic geography. Second, this chapter stresses that global linkages not only introduce new industries

0.78*** 0.27*** 0.37*** Included Included Included 8.31*** 41319.15 0.13 1.40E+05 753,895

Model1 10.64*** 1.80***

0.75*** 0.23*** 0.40*** Included Included Included 8.79*** 42166.65 0.13 1.40E+05 753,895

0.55***

Model2 10.42*** 2.22*** 0.75***

0.75*** 0.23*** 0.40*** Included Included Included 8.78*** 42229.78 0.13 1.40E+05 753,895

Model3 10.16*** 2.29*** 1.42*** 4.34*** 0.16** 2.46***

0.75*** 0.27*** 0.37*** Included Included Included 8.35*** 41334.77 0.13 1.40E+05 753,895

0.12***

Model4 10.72*** 1.70***

Note: Measurement of Density is based on LQ equal to 1.0 Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density WDensity FDI Density*FDI TRADE Density*TRADE LIB Density*LIB DEXP Density*DEXP DLAND Density*DLAND WTO LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.6 Logit regression results with annual data for entry equations (LQ ¼ 1.0)

0.75*** 0.27*** 0.37*** Included Included Included 8.53*** 41359.87 0.13 1.40E+05 753,895

0.41*** 1.84***

Model5 11.92*** 1.65***

0.86*** 0.18*** 0.46*** Included Included Included 9.59*** 43051.67 0.13 1.40E+05 736,867

0.26***

1.61***

Model6 13.93*** 0.74

1.20*** 2.65*** 0.22*** 0.27 0.87*** 0.18*** 0.46*** Included Included Included 9.34*** 43090.71 0.13 1.40E+05 736,867

Model7 12.37*** 0.63

0.75*** 0.18*** 0.47*** Included Included Included 9.86*** 43271.87 0.14 1.40E+05 736,867

0.24***

1.54***

0.40***

0.36***

Model8 13.99*** 0.8 0.03

Model9 12.71*** 0.64 0.47*** 2.72*** 0.25*** 0.74 0.56*** 0.93** 1.01*** 3.30*** 0.17*** 0.47* 0.75*** 0.18*** 0.47*** Included Included Included 9.68*** 43355.02 0.14 1.40E+05 736,867

3.6 Summary and Discussion 69

0.61*** 0.29*** 0.38*** Included Included Included 8.51*** 44616.58 0.14 1.40E+05 753,895

Model1 8.91*** 1.14***

0.58*** 0.25*** 0.40*** Included Included Included 8.95*** 45351.15 0.14 1.40E+05 753,895

0.48***

Model2 8.72*** 1.42*** 0.74***

0.57*** 0.25*** 0.40*** Included Included Included 8.91*** 45401.38 0.14 1.40E+05 753,895

Model3 8.40*** 1.45*** 1.32*** 2.62*** 0.05 1.94***

0.57*** 0.29*** 0.38*** Included Included Included 8.55*** 44635.1 0.14 1.40E+05 753,895

0.13***

Model4 8.97*** 1.05***

Note: Measurement of Density is based on LQ equal to 0.5 Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density WDensity FDI Density *FDI TRADE Density *TRADE LIB Density *LIB DEXP Density *DEXP DLAND Density *DLAND WTO LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.7 Logit regression results with annual data for entry equations (LQ ¼ 0.5)

0.57*** 0.29*** 0.38*** Included Included Included 8.73*** 44656.41 0.14 1.40E+05 753,895

0.41*** 1.27***

Model5 9.80*** 1.02***

0.68*** 0.21*** 0.46*** Included Included Included 9.73*** 46053.12 0.14 1.40E+05 736,867

0.25***

1.54***

Model6 11.13*** 0.68*

1.25*** 1.34*** 0.23*** 0.12 0.68*** 0.21*** 0.46*** Included Included Included 9.55*** 46071.79 0.14 1.40E+05 736,867

Model7 10.34*** 0.63

0.58*** 0.20*** 0.47*** Included Included Included 9.97*** 46240.56 0.14 1.40E+05 736,867

0.23***

1.49***

0.38***

0.29***

Model8 11.19*** 0.75* 0.05

Model9 10.38*** 0.67* 0.40*** 1.53*** 0.02 1.27*** 0.51*** 0.48 1.16*** 1.43*** 0.19*** 0.17 0.57*** 0.20*** 0.47*** Included Included Included 9.81*** 46294.34 0.14 1.40E+05 736,867

70 3 How Does Regional Industrial Structure Evolve in China?

0.31*** 0.27*** 0.09*** Included Included Included 1.52*** 24424.79 0.1 1.10E+05 358,273

Model1 8.07*** 0.64

0.32*** 0.24*** 0.11*** Included Included Included 1.87*** 24791.72 0.1 1.10E+05 358,273

0.47***

Model2 8.14*** 0.93 0.46***

0.32*** 0.24*** 0.10*** Included Included Included 1.24*** 25063.82 0.1 1.10E+05 358,273

Model3 5.43*** 1.15* 0.03 3.03*** 0.62*** 6.16***

0.35*** 0.27*** 0.09*** Included Included Included 1.59*** 24447.86 0.1 1.10E+05 358,273

0.16***

Model4 8.18*** 0.48

Note: Measurement of Density is based on LQ equal to 1.0 Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density WDensity FDI Density *FDI TRADE Density *TRADE LIB Density *LIB DEXP Density *DEXP DLAND Density *DLAND WTO LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.8 Logit regression results with annual data for exit equations (LQ ¼ 1.0)

0.35*** 0.27*** 0.09*** Included Included Included 1.13*** 24549.54 0.1 1.10E+05 358,273

0.58*** 4.12***

Model5 5.64*** 0.56

0.34*** 0.24*** 0.11*** Included Included Included 1.85*** 24582.35 0.1 1.10E+05 351,939

0.10***

0.56***

Model6 9.03*** 0.12

0.88*** 8.26*** 0.45*** 3.30*** 0.34*** 0.24*** 0.10*** Included Included Included 0.77*** 25095.2 0.1 1.10E+05 351,939

Model7 3.15*** 0.61

0.44*** 0.22*** 0.12*** Included Included Included 2.15*** 24849.63 0.1 1.10E+05 351,939

0.07***

0.36***

0.32***

0.51***

Model8 8.95*** 0.35 0.20***

Model9 2.15*** 0.9 0.28* 0.26 0.09 2.17*** 0.07 1.38*** 0.92*** 7.30*** 0.37*** 2.59*** 0.44*** 0.22*** 0.11*** Included Included Included 0.88*** 25400.96 0.11 1.10E+05 351,939

3.6 Summary and Discussion 71

0.47*** 0.28*** 0.07*** Included Included Included 1.62*** 25630.56 0.1 1.10E+05 358,273

Model1 6.90*** 0.22

0.49*** 0.26*** 0.08*** Included Included Included 1.85*** 25843.82 0.11 1.10E+05 358,273

0.33***

Model2 6.86*** 0.38 0.40***

0.49*** 0.26*** 0.07*** Included Included Included 1.13*** 26243.57 0.11 1.10E+05 358,273

Model3 4.50*** 0.51 0.51*** 4.08*** 0.81*** 4.66***

0.52*** 0.29*** 0.07*** Included Included Included 1.69*** 25657.93 0.1 1.10E+05 358,273

0.17***

Model4 6.99*** 0.1

Note: Measurement of density is based on LQ equal to 0.5 Note: ***significant at 1% level, **significant at 5%, and *significant at 10% level

Density Wdensity FDI Density *FDI TRADE Density *TRADE LIB Density *LIB DEXP Density *DEXP DLAND Density *DLAND WTO LnLOCAL LnURBAN Industry dummies Province dummies Year dummies Cons LR chi2 Pseudo R2 Log likelihood No. of observations

Table 3.9 Logit regression results with annual data for exit equations (LQ ¼ 0.5)

0.52*** 0.29*** 0.07*** Included Included Included 1.21*** 25778.87 0.11 1.10E+05 358,273

0.64*** 3.33***

Model5 4.94*** 0.14

0.52*** 0.26*** 0.08*** Included Included Included 1.78*** 25598.45 0.11 1.10E+05 351,939

0.08***

0.33***

Model6 7.28*** 0.01

1.00*** 5.54*** 0.41*** 2.09*** 0.51*** 0.26*** 0.08*** Included Included Included 0.81*** 26027.33 0.11 1.10E+05 351,939

Model7 3.30*** 0.23

0.61*** 0.25*** 0.09*** Included Included Included 2.01*** 25788.5 0.11 1.10E+05 351,939

0.05***

0.16***

0.28***

0.39***

Model8 7.18*** 0.12 0.23***

Model9 2.39*** 0.38 0.18 2.04*** 0.32*** 2.88*** 0.02 1.32*** 0.70*** 3.60*** 0.28*** 1.37*** 0.61*** 0.25*** 0.08*** Included Included Included 0.80*** 26344.86 0.11 1.10E+05 351,939

72 3 How Does Regional Industrial Structure Evolve in China?

References

73

directly but also reinforce the role of technological relatedness in facilitating new industries. Transitional and developing economies can substantially benefit from globalization. Both endogenous and exogenous factors and their interactions can create path dependence for regional industrial development. Third, this chapter assesses the role of market-oriented institutions and state involvement on the regional diversification process. Both economic liberalization and state involvement can help introduce new industries at the regional level, and local governments can take advantage of market forces to attract and sustain related industries. This chapter contributes to the theoretical development of EEG. There are a number of directions to expand the current research. First, this chapter only examined the impact of general institutions. More is demanded to analyze the impact of specific institutions on regional industrial evolution such as governmental efficiency, corruption, protection of intellectual property rights, contract enforcement, and industrial policies such as 5-year industrial plans, subsidies, and the establishment of industrial parks. Second, one would be interested in the agents of industrial structural diversification in Chinese regions. We could analyze the industrial attributes of new entries or exits and even the characteristics of entry and exit firms in different regions and study the responses of different industries and firms to relatedness and the regional institutions. Third, we could also compare the industrial diversification process for different regions. Regions differ in many aspects such as physical location, economic development, industrial structure, governmental intervene, and external linkages, which may shape the industrial diversification process. Fourth, we could open the black box of the spillover effects of FDI on industrial evolution in China and explore the hidden mechanisms such as competitive effect and modelling effect. Finally, we can compare the efficiency and sustainability of related and unrelated industrial diversification in Chinese regions. Unrelated industries which may be brought by external linkages or governmental intervention may not fit the local competence and comparative advantages. This may challenge the sustainable development of unrelated industrial diversification in China.

References Barbieri, E., Tommaso, D., Marco, R., & Bonnini, S. (2012). Industrial development policies and performance in Southern China: Beyond the specialized industrial cluster program. China Economic Review, 23(3), 613–625. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. Blomström, M., & Kokko, A. (1998). Multinational corporations and spillovers. Journal of Economic Surveys, 12, 247–277. Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional Studies, 39(1), 61–74. Boschma, R., & Capone, G. (2015). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, 44(10), 1902–1914. Boschma, R., & Frenken, K. (2009). Some notes on institutions in evolutionary economic geography. Economic Geography, 85(2), 151–158.

74

3 How Does Regional Industrial Structure Evolve in China?

Boschma, R., & Frenken, K. (2011). Technological relatedness and regional branching. In H. Bathelt, M. Feldman, & D. F. Kogler (Eds.), Beyond territory: Dynamic geographies of knowledge creation, diffusion and innovation (pp. 64–81). London: Routledge. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R. A., & Lambooy, J. G. (1999). Evolutionary economics and economic geography. Journal of Evolutionary Economics, 9(4), 411–429. Boschma, R., & Martin, R. (2007). Editorial: Constructing an evolutionary economic geography. Journal of Economic Geography, 7(5), 537–548. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Breschi, S., Lissoni, F., & Malerba, F. (2003). Knowledge-relatedness infirm technological diversification. Research Policy, 32, 69–87. Broadman, G. H. (1996). Policy options for reform of Chinese state-owned enterprises: Proceedings of a Symposium in Beijing, June 1995 (Vol. 335). World Bank Publications. Buenstorf, G., & Klepper, S. (2009). Heritage and agglomeration: The Akron tyre cluster revisited. Economic Journal, 119(537), 705–733. Bunnell, T. G., & Coe, N. M. (2001). Spaces and scales of innovation. Progress in Human Geography, 25(25), 569–589. Cheung, K. Y., & Lin, P. (2004). Spillover effects of FDI on innovation in China: Evidence from the provincial data. China Economic Review, 15(1), 25–44. Colombelli, A., Krafft, J., & Quatraro, F. (2014). The emergence of new technology-based sectors in European regions: A proximity-based analysis of nanotechnology. Research Policy, 43(10), 1681–1696. Essletzbichler, J. (2015). Relatedness, industrial branching and technological cohesion in US metropolitan areas. Regional Studies, 49(5), 752–766. Estrin, S., & Prevezer, M. (2010). A survey on institutions and new firm entry: How and why do entry rates differ in emerging markets? Economic Systems, 34(3), 289–308. Fan, J. P. H., & Lang, L. H. P. (2000). The measurement of relatedness: An application to corporate diversification. Journal of Business, 73(4), 629–660. Frenken, K., & Boschma, R. A. (2007). A theoretical framework for evolutionary economic geography: Industrial dynamics and urban growth as a branching process. Journal of Economic Geography, 7(5), 635–649. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Freund, E. M. (2001). Fizz, froth, flat: The challenge of converting China’s SOEs into shareholding corporations. Review of Policy Research, 18(1), 96–111. Fujita, M., & Hu, D. (2001). Regional disparity in China 1985–1994: The effects of globalization and economic liberalization. The Annals of Regional Science, 35(1), 3–37. Garud, R., & Karnøe, P. (2001). Path creation as a process of mindful deviation. In Path dependence and creation (pp. 1–38). Hall, P. A., & Soskice, D. (2001). Varieties of capitalism: The institutional foundations of comparative advantage. Oxford: Oxford University Press. Han, S. S., & Pannell, C. W. (1999). The geography of privatization in China, 1978–1996. Economic Geography, 75(3), 272–296. He, C., & Pan, F. (2010). Economic transition, dynamic externalities and city-industry growth in China. Urban Studies, 47(1), 121–144. He, C., & Wang, J. (2012). Regional and sectoral differences in the spatial restructuring of Chinese manufacturing industries during the post-Wto period. GeoJournal, 77(3), 361–381. He, C., & Yang, R. (2016). Determinants of firm failure: Empirical evidence from China. Growth Change, 47, 72–92.

References

75

He, C., Wei, Y. H. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. He, C., Zou, P., & Zhu, Y. (2011). Spatial organization of fortune global 500 corporations in China: An empirical study of multinationals from Japan, the United States and European Union. Eurasian Geography and Economics, 52(6), 809–834. He, C., Huang, Z., & Wang, R. (2014). Land use change and economic growth in urban China: A structural equation approach. Urban Studies, 51(13), 2880–2898. He, C., Guo, Q., & Ye, X. (2015). Geographical agglomeration and coagglomeration of exporters and nonexporters in China. GeoJournal, 81(6), 947–964. Henning, M., Stam, E., & Wenting, R. (2013). Path dependence research in regional economic development: Cacophony or knowledge accumulation? Regional Studies, 47(8), 1348–1362. Isaksen, A. (2014). Industrial development in thin regions: Trapped in path extension? Journal of Economic Geography, 15(3), 1–16. Klepper, S. (2002). The capabilities of new firms and the evolution of the us automobile industry. Industrial and Corporate Change, 11(4), 645–666. Klepper, S., & Simons, K. L. (2000). Dominance by birthright: Entry of prior radio producers and competitive ramifications in the U.S. television receiver industry. Strategic Management Journal, 21(10–11), 997–1016. Li, H., & Zhou, L. A. (2005). Political turnover and economic performance: The incentive role of personnel control in China. Journal of Public Economics, 89, 1743–1762. Li, S., & Xia, J. (2008). The roles and performance of state firms and non-state firms in China’s economic transition. World Development, 36(1), 39–54. Lin, G. C. S. (2007). Reproducing spaces of Chinese urbanization: New city-based and landcentered urban transformation. Urban Studies, 44(9), 1827–1855. Lin, J. Y. (2012). New structural economics: A framework for rethinking development and policy. Washington D.C: World Bank. Lin, G. C. S., & Yi, F. (2011). Urbanization of capital or capitalization on urban land? Land development and local public finance in urbanizing China. Urban Geography, 32(1), 50–79. Liu, Z. (2008). Foreign direct investment and technology spillovers: Theory and evidence. Journal of Development Economics, 85(1), 176–193. Martin, R., & Sunley, P. (2006). Path dependence and regional economic evolution. Journal of Economic Geography, 6(4), 395–437. Massey, D. (1984). Spatial divisions of labour. Social structures and the geography of production. London: Macmillan. Neffke, F., & Henning, M. (2013). Skill relatedness and firm diversification. Strategic Management Journal, 34(3), 297–316. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Neffke, F. M. H., Henning, M., & Boschma, R. (2012). The impact of aging and technological relatedness on agglomeration externalities: A survival analysis. Journal of Economic Geography, 12(2), 485–517. Neffke, F., Hartog, M., Boschma, R., & Henning, M. (2014). Agents of structural change: The role of firms and entrepreneurs in regional diversification. Papers in Evolutionary Economic Geography, October, Utrecht University. Park, S. H., Li, S., & David, K. T. (2006). Market liberalization and firm performance during China’s economic transition. Journal of International Business Studies, 37(1), 127–147. Poncet, S. (2005). A fragmented China: Measure and determinants of Chinese domestic market disintegration. Review of International Economics, 13(3), 409–430. Qian, Y. (1996). Enterprise reform in China: Agency problems and political control. The Economics of Transition, 4(2), 427–447.

76

3 How Does Regional Industrial Structure Evolve in China?

Qian, Y. (2000). The process of China’s market transition (1978–1998): The evolutionary, historical, and comparative perspectives. Journal of Institutional and Theoretical Economics, 1, 151–171. Qiu, Y. (2005). Personal networks, institutional involvement and foreign direct investment flows into China’s interior. Economic Geography, 81(3), 261–281. Schumpeter, J. A. (1934). The theory of economic development: An inquiry into profits, capital, credit, interest, and the business cycle (Vol. 55). Transaction publishers. New Brunswick/ London. Scott, A. J. (2006). Geography and economy. Oxford: Oxford University Press. Stiglitz, J. E. (1996). Some lessons from the East Asian miracle. The world Bank Research Observer, 11, 151–177. Storper, M., & Walker, R. (1989). The capitalist imperative: Territory, technology and industrial growth. Oxford: Blackwell. Thun, E. (2004). Keeping up with the Joneses: Decentralization, policy imitation and industrial development in China. World Development, 32(8), 1289–1308. Todtling, F., & Trippl, M. (2005). One size fits all? Towards a differentiated regional innovation policy approach. Research Policy, 34(8), 1203–1219. Trevor, B., Liu, X., Wei, Y., & Liu, X. (2007). The trade development path and export spillovers in China: A missing link. Management International Review, 47, 683–706. Wei, Y., Li, J., & Ning, Y. (2010). Corporate networks, value chains and spatial organization: A study of the computer industry in China. Urban Geography, 31(8), 1118–1140. Wong, C. P. W. (2000). Central-local relations revisited: The 1994 tax sharing reform and public expenditure management in China. China Perspectives, 2000(31), 52–63. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yang, Y. R., & Hsia, C. (2007). Spatial clustering and organizational dynamics of transborder production networks: A case study of Taiwanese information technology companies in The Greater Suzhou Area, China. Environment and Planning A, 39(6), 1346–1363. Yang, C., & Liao, H. (2010). Industrial agglomeration of Hong Kong and Taiwanese manufacturing investment in China: A town level analysis in Dongguan. Annals of Regional Science, 45(3), 487–517. Yeung, H., Liu, W., & Dicken, P. (2006). Transnational corporations and network effects of a local manufacturing cluster in mobile telecommunications equipment in China. World Development, 34(3), 520–540. Young, A. (2000). The Razor’s edge: Distortions and incremental reform in the People’s Republic of China. Quarterly Journal of Economics, 115(4), 1091–1135. Zhang, L. Y. (1999). Chinese central-provincial fiscal relationships, budgetary decline and the impact of the 1994 fiscal reform: An evaluation. The China Quarterly, 157(157), 115–141. Zhang, J. (2013). Related variety, global connectivity and institutional embeddedness: Internet development in Beijing and Shanghai compared. Regional Studies, 47(7), 1065–1081. Zhao, X., & Zhang, L. (1999). Decentralization reforms and regionalism in China: A review. International Regional Science Review, 22(3), 251–281. Zhu, T. (1999). China’s corporatization drive: An evaluation and policy implications. Contemporary Economic Policy, 17(4), 530–539. Zhu, S., & He, C. (2013). Geographical dynamics and industrial relocation: Spatial strategies of apparel firms in Ningbo, China. Eurasian Geography and Economics, 54(3), 342–362.

Chapter 4

What Matters for Regional Industrial Dynamics in China?

4.1

Introduction

Recent empirical research in EEG tends to focus excessively on technological relatedness across regional industries, combinatorial knowledge dynamics, and branching processes as key explanatory factors for regional industrial dynamics (Boschma and Frenken 2011; Neffke et al. 2011). Relatedness, it is argued, not only pushes forward the growth of existing industries through agglomeration externalities derived from related variety but is also responsible for the formation of new growth paths (Boschma and Capone 2015a; Boschma et al. 2013; Delgado et al. 2016; Neffke et al. 2011). New growth paths do not emerge from scratch but evolve out of preexisting regional industrial structures, because the set of competences and assets that a region possesses determine the new paths and new industries it can develop (Boschma et al. 2012; Hidalgo et al. 2007; Neffke et al. 2011). If a region already has most of the capabilities that a certain new industry requires, it will be easy for it to diversify into that new industry. If not, the barriers to entry could be too high for this region to overcome (Boschma et al. 2013). In short, regions are anticipated to branch into technologically related industries in path-dependent related diversification processes. Where new industries emerge is strongly contingent on (but not predetermined by) the preexisting regional industrial structure. This strand of literature has shed new light on how and why history matters for regional economic development. However, its conceptual and methodological approach has recently attracted several strands of criticism, particularly in economic geography, two of which will be examined in this chapter. First, EEG’s strong focus on technological relatedness downplays the influence of other region-specific assets that have been long emphasized in economic geography, including non-firm actors,

Modified article originally published in [He, C., S. Zhu and X. Yang (2017) What Matters for Regional Industrial Dynamics in a Transitional Economy?, Area Development and Policy, 2 (1), pp. 71–90.]. Published with kind permission of © [Taylor & Francis, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_4

77

78

4

What Matters for Regional Industrial Dynamics in China?

institutions, and public policy (Dawley 2014; Hassink et al. 2014; Martin 2010; Simmie 2012; Sydow et al. 2010; Tanner 2014). Boschma and Capone (2015b) have argued that these studies pay insufficient attention to the differences that industrial diversification process can display across regions and, more importantly, to the possible effect of regional institutions on the intensity and nature of the industrial diversification process. China’s manufacturing industry provides a rich context. Since the start of reform and opening-up, China has undergone dramatic economic growth and has experienced three fundamental transformations: (1) from a closed or partially closed to an open economy that is increasingly integrated into the global economy and oriented toward export markets, (2) from a centrally planned to a decentralized political and fiscal system, and (3) from a state- and collectively owned economy to one with growing market orientation and private ownership (He et al. 2008; Zhu and He 2013; Zhu and Pickles 2014). On the one hand, as the restrictions on factor mobility and commodity exchanges were gradually lifted and private entrepreneurship was increasingly encouraged (Chang and MacMillan 1991), technological relatedness started to play an increasingly crucial role in regional industrial evolution (Guo and He 2015). On the other hand, and more importantly, decentralization in China, also known as the regionally decentralized authoritarian system (Xu 2011), has not only created GDP-based interjurisdictional competition between local authorities that have strong incentives to intervene in regional economic development (Pan et al. 2016; Yu et al. 2016) but also allowed regional administrations to take different routes, resulting in a geographically uneven economic and institutional landscape (Zhu and He 2015). In other words, government intervention at the local level (e.g., public spending) must be taken into account to better understand China’s regional industrial dynamics. Second, by emphasizing organizational routines at the firm-level and knowledge spillovers through spin-offs, labor mobility, social networking, and/or firm diversification, the literature on technological relatedness tends to overlook the fact that industries are not equally affected by knowledge spillovers and historical industrial structures (Cainelli and Iacobucci 2016). The way technological relatedness and knowledge spillovers affect regional industrial dynamics may be contingent on industry characteristics. This chapter seeks to contribute to recent debates on the evolution of regional industrial structure, by bringing the regional institutional context to the forefront and comparing its role with that of technological relatedness. By using a firm-level dataset of China’s manufacturing industries in 1998–2008, we build on recent EEG studies but pay more attention to the ways in which regional industrial structure is constantly shaped by an assemblage of factors, including not only technological relatedness but also institutional contexts and industry characteristics. The next section will present an analytical framework. The third section introduces the data, variables, and model specifications. After presenting some descriptive analyses in the fourth section, Sect. 4.5 discusses the empirical results. The last section concludes and considers the policy implications of this research.

4.2 Regional Industrial Dynamics: Region-Specific and Industry-Specific Factors

4.2 4.2.1

79

Regional Industrial Dynamics: Region-Specific and Industry-Specific Factors Technological Relatedness and Regional Institutions

Regional industrial development and innovation depend a great deal on a variety of region-specific assets. Martin (2010), for example, pointed that innovation is often a highly localized phenomenon, heavily reliant on place-specific factors and conditions. However, in recent EEG literature, little attention has been paid to the potential effects of other region-specific assets (e.g., the institutional dimensions of industrial change) (Boschma and Capone 2015a). An overemphasis on technological relatedness has led to an underestimation of regional variations of institutional frameworks which can have a direct impact on innovation and regional industrial specialization (Coe et al. 2008; Hall and Soskice 2001; Smith et al. 2014), particularly in transitional economies like China where a triple process of decentralization, globalization, and marketization has resulted in enormous spatial variations in the economic and institutional landscape (Wei 2001). Since the late 1970s, China’s economic system has been transformed from a command economy toward a more market-oriented one, as privatization and market competition were progressively introduced (McMillan and Naughton 1992). As the restrictions on factor mobility and commodity exchange were gradually lifted and private entrepreneurship was increasingly encouraged (Chang and MacMillan 1991), firms became more motivated to exploit competitive advantages and more likely to locate close to specialized suppliers and customers in order to reduce not only logistics costs but also the costs of searching for matched suppliers and customers, generating a favorable ambience for knowledge spillovers. As a result, technological relatedness started to play an increasingly crucial role in regional industrial evolution (Guo and He 2015). This proposition leads to the following hypothesis: Hypothesis 4.1 Technological relatedness affects regional industrial diversification in Chinese regions. More importantly, state-led decentralization policies empowered local governments to get involved in shaping the regional economy, as planners, developers, and policy-makers. Some officials became heavy-handed, ever more convinced of the importance of their “steering” role (He et al. 2008; Wei 2001; Wei et al. 2007). Theories of fiscal federalism suggest that fiscal decentralization promotes growth by transferring powers and resources to subnational administrations: a better matching of policies with local preferences increases allocative efficiency, and interjurisdictional competition increases production efficiency (Peterson 1981). In China, decentralization is a much more complex process, encouraging local authorities to maximize their revenues by intervening in regional industrial development (Montinola et al. 1995; Oi 1995). A combination of political centralization and economic decentralization is also seen as driving a regionally decentralized

80

4

What Matters for Regional Industrial Dynamics in China?

authoritarian system (Xu 2011). A result is the so-called tournament competition, where officials occupying high-level positions in a hierarchical political system promote lower-level officials on the basis of their economic performance and local officials compete to boost regional economic development in order to maximize their chances of political promotion (Pan et al. 2016; Yu et al. 2016). All local governments adopt pro-business policies to attract and develop new industrial sectors. However, governments in China’s wealthy coastal regions are often more able to provide subsidies, financial and technological support, high-quality infrastructure, and public services than their counterparts in less developed inland China. Regional variations of institutional frameworks must be therefore considered to understand better China’s regional industrial evolution, leading to a second hypothesis: Hypothesis 4.2 Government intervention at the local level also affects regional industrial dynamics in China.

4.2.2

Industry-Specific Factors

The recent technological relatedness literature has paid even less attention to industry characteristics. Although it affords a renewed vision of the impact of agglomeration externalities over regional industrial dynamics, industries are not equally affected by knowledge spillover effects, technological relatedness, and historical industrial structure or indeed by regional institutional frameworks. This chapter thus seeks to incorporate this factor by asking what kinds of industries are regions with certain kinds of region-specific assets (e.g., institutional contexts and regional resources) more likely to attract and create. Based on learning processes in different industrial sectors, Pavitt (1984) has developed a taxonomy of industries. He distinguishes industries on the basis of their innovation modes and the intensity of use of technological knowledge for innovation. At the extreme ends of Pavitt’s spectrum are supplier-dominated and science-based industries. Supplier-dominated industries include firms in traditional, labor-intensive sectors (e.g., textiles and apparel, leather, footwear, and furniture). The technological intensity of these industries is low due to their narrow knowledge base. In contrast, science-based industries (e.g., electronics, biotechnology, and chemistry) are characterized by high levels of investment in R&D and innovation aimed at generating new products. Since entry barriers are lower in the former, it is hypothesized that it is much easier for a region, given its institutional context and preexisting industrial structure, to attract and/or create supplier-dominated industrial sectors, while the entry/exit of science-based industrial sectors is more influenced by regional institutional arrangements and technological relatedness. Furthermore, China’s economic reform has unleashed economic vitality, but its effectiveness is quite uneven across industries. In the early stages, the state first encouraged private enterprises to enter light and labor-intensive industries and then lowered the state-owned proportion of other manufacturing industries (Wei 2001).

4.3 Research Design

81

State-owned industries (SOEs) are still dominant in certain industries, such as tobacco processing and petroleum processing. Financial aid and tax incentives have been often disproportionately assigned to SOEs (Bai et al. 2004; Sun and Tong 2003). Some purportedly industry-wide policies have been applied preferentially to favor SOEs. The entry and exit of SOE-dominated industries may thus be reliant less on technological relatedness and knowledge spillovers, and on local government-provided public services (e.g., education, labor training, infrastructure, health care, etc.), but more on government aid, tax credits, and subsidies. The participation of developing country firms in international markets allows them to access more advanced and up-to-date technologies. However, fierce global competition forces them to develop capabilities to compete with foreign or even global lead firms for market share (Zhu and He 2015). Exporting also involves additional fixed costs (Melitz 2003), including the costs of establishing distribution networks and researching foreign markets to gain intelligence about consumers’ tastes, market structure, competitors, and regulations (Clerides et al. 1998; He et al. 2012; Lovely et al. 2005). Entry barriers in export-oriented industries are therefore high. Entry and exit may depend much more on technological relatedness and institutional context. The following sections test the influence of region- and industry-specific factors on industrial evolution at the prefectural city level in China by emphasizing the role of technological relatedness, regional institutions, and industrial characteristics, by evaluating the third hypothesis: Hypothesis 4.3 The relationship between institutional context, technological relatedness, and regional industrial dynamics may be contingent on industry characteristics.

4.3

Research Design

This chapter uses the ASIF dataset and focuses on 424 4-digit manufacturing industries and their entry into and exit from China’s 337 prefecture-level cities. Industry entry and exit is defined as follows. If industry i does not exist in city c in year t1 but exists in city c in year t2 (t2>t1), industry i is a new entry to city c. Likewise, if industry i is reported in city c in year t1 but disappears in year t2, it is assumed that industry i exited from city c during this time period. Based on Eqs. 1.3 and 1.4, a 424*424 matrix of proximity indicators for all 4-digit industries was estimated. We then use Eq. 1.5 to calculate the density indicator of industry i in city c, to measure the distance between new industries and region’s existing industrial structure. The following equation was estimated to investigate the ways in which industry entry and exit in China was affected by the articulation of region- and industryspecific factors:

82

4

What Matters for Regional Industrial Dynamics in China?

Entryi,c,t2 ðExit i,c,t2 Þ ¼ β0 þ f 1 ðRSi,c,t1 Þ þ f 2 ðISi,t1 Þ þ f 3 ½RSc,t1  ISi,t1  þ γX þ δY þ εi,c,t1 ð4:1Þ where t2>t1. The dependent variables, Entryi,c,t2 (Exiti,c,t2), take the value of 1 if industry i enters (exits) city c during the time period t1–t2 and zero otherwise. X represents control variables. Y is a vector of region-year and industry-year dummy variables, which was added to control for any time-varying region or industry characteristics. In Eq. (4.1), RS refers to region-specific assets. First, to investigate whether regions tend to attract and create new industries that are related to the preexisting regional industrial structure, the density indicator (density) was included. If an industry is related to a region’s preexisting industrial structure, the industry is more likely to enter and less likely to exit. Density is thus expected to have a positive coefficient in entry models but a negative coefficient in exit models. Another key region-specific variable is public spending by local governments. This variable captures the impact of government intervention under China’s regionally decentralized authoritarian system and local state support for local economic activity. Public spending (PUB) was measured as city c’s public fiscal expenditure per person. Government spending may affect regional economic development and industrial dynamics in many ways. For example, public expenditure on infrastructure, education, labor training, and health care can serve as unpriced inputs into production and therefore lower industry entry barriers (Fisher 1997), while the provision of public services and goods (e.g., highway system) may alter private consumption patterns, creating new demand for some industries while rendering others obsolete (Fisher 1997). The research on public spending generally found positive (negative) effects on industry entry (exit) (Van Cauwenberge et al. 2016). Industry-specific factors (IS) capture industry characteristics that affect industry entry and exit. In this research the proportion of SOEs’ output in industry i in the country (SOE) was used to reflect the extent to which industry i was dominated by SOEs. EXP is the export intensity—the ratio of exports to total output—of industry i in the whole country, measuring the degree of export orientation of industry i. R&D is calculated as the ratio of R&D investments to value-added in industry i in the whole country, to capture the extent to which industry i is science-based. Finally, to measure the extent to which industry i was labor-intensive, labor intensity (LAB) was defined as:
LABi ¼ Employmenti =Fixed Assetsi P i

Employment i =

P



ð4:2Þ

Fixed Assetsi

i

where Employmenti and Fixed Assetsi represent the number of employees and the value of fixed assets in industry i in the whole country, respectively. Three control variables were included. The GINI coefficient of industry i (GINI) was calculated to reflect the degree of market competition in industry i in China. Firms in highly concentrated markets are subject to fiercely aggressive behavior by

4.4 Regional Industrial Dynamics in China

83

rivals, which may frighten off new entrants. However, higher market concentration may also lead to higher price-cost margins in an industry, which may enable regions to create and attract it (Basile et al. 2016). Industry growth is measured by the annual growth rate of output of industry i in the entire country (GROW). Regions are expected to create and attract industries that are quickly expanding, and so it was hypothesized that this variable has positive (negative) effects on industry entry (exit). Population density (POP), the ratio of city c’s population to area, is included to control local market potential. Data on public spending and population density were derived from China’s City Statistical Yearbooks, while all other variables were derived from the ASIF. The time period 1998–2008 was divided into two stages: 1998–2003 and 2003–2008. These two 5-year periods are long enough for new industries to enter and for existing industries to exit. All models were also estimated using data in different years (e.g., 1999–2003 and 2003–2007) and using different threshold values to determine the industrial specialization in the calculation of the density indicator. These changes also produced only minor differences.1

4.4

Regional Industrial Dynamics in China

Figure 4.1a, b shows the relationship between the average density of the industries that were absent in a city in 1998 (or 2003) and the probability of this city creating or attracting a new industry (i.e., industry entry) by 2003 (or 2008). Each point on the scatter plot corresponds to one of China’s 337 prefectures. The graph displays a clear positive relationship between the average density of absent industries and the probability of industry entry in both stages. Likewise, the relationship between the average density of the industries already present in a city in 1998 (or 2003) and the probability of these industries exiting from this prefecture in 2003 (or 2008) was examined. As Fig. 4.2a, b shows, there is a strong negative relationship, suggesting that if the average density of a prefecture’s existing industries was low, it was more likely to face industry exit in subsequent periods. Indeed the density indicator is a synthetic measure of proximity, accounting for the degree of connectedness of each industry. The more regions can incorporate industries with high density in their portfolios, the higher their chances of keeping these industries and preventing them from exiting. The relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of this prefecture creating or attracting a new industry by 2003 (or 2008) (Fig. 4.3) and the relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of industries exiting from this prefecture by 2003 (or 2008) (Fig. 4.4) were also examined. The figures show that the probability

1 Due to space limitation, estimation results for these robustness checks are not reported here but available on request.

84

4

What Matters for Regional Industrial Dynamics in China?

Fig. 4.1 (a) (Left) and (b) (right) the relationship between the average density of the industries that were absent in a prefecture in 1998 (or 2003) and the probability of this prefecture creating or attracting a new industry in 2003 (or 2008)

Fig. 4.2 (a) (Left) and (b) (right) the relationship between the average density of the industries that a prefecture had in 1998 (or 2003) and the probability of these industries exiting from this prefecture in 2003 (or 2008)

of industry entry (exit) increases (decreases) with the region’s public expenditure, particularly during 1998–2003. Again, regions with higher levels of public spending may be able to provide better public goods and services and are thus more likely to not only attract and create new industries but also keep existing industries. However, public spending’s relationship with industry entry and exit is much weaker than the relationship between density and industry entry/exit. In addition, the former relationship becomes even weaker and less clear in the second stage. Overall, this result

4.4 Regional Industrial Dynamics in China

85

Fig. 4.3 (a) (Left) and (b) (right) the relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of this prefecture creating or attracting a new industry in 2003 (or 2008)

Fig. 4.4 (a) (Left) and (b) (right) the relationship between a prefecture’s public expenditure in 1999 (or 2003) and the probability of industries exiting from this prefecture in 2003 (or 2008)

indicates that government intervention under China’s regionally decentralized authoritarian system plays a less important role than technological relatedness in regional industry dynamics. However, Figs. 4.3 and 4.4 are only simple analyses of the role of government intervention at the city level. Its role in some industries may be much more evident than in others. In the following section, controls for other variables are added, and the relationship between government intervention and industry entry/exit at the 4-digit industry and city level is examined.

86

4.5

4

What Matters for Regional Industrial Dynamics in China?

Empirical Results

In the estimated equation, the logarithm of density, PUB, and POP was used. Lagged terms for independent and control variables were adopted, given that it takes time for industry entry and exit to be affected by the articulation of region-specific assets and industry characteristics. The time period was divided into two stages (1998–2003 and 2003–2008), and all the observations were pooled. The geographical unit of analysis was China’s prefecture-level cities. Correlation analysis indicated that the correlations between most independent variables were moderate or low, suggesting that there were no serious problems of multicollinearity (Table 4.1). Since the dependent variables are binary variables, a probit model was used. Before testing how regional industry dynamics are shaped by region- and industry-specific characteristics, the geographical restructuring of China’s manufacturing industries was examined, specifically to understand what kinds of industries were moving to or away from which parts of China. To do so, a distinction was made between cities in China’s three macro-regions (Western, Central, and Eastern China2), and two variables—EAST (if city c is located in Eastern China) and WEST (if city c is located in Western China)—were included in the models. Table 4.2 reports estimation results focusing on industry characteristics and two macro-regions. The dummy variable for industry (Industry in Table 4.2) was measured at the 2-digit level. In almost all models, control variables revealed a relationship with industry entry and exit that is consistent with the theoretical predictions. Regions were more likely to create and attract industries that were in the emergence and growth phase of their lifecycle, while declining industries were more likely to exit. The sign of logPOP’s coefficient suggests that regions with large market potential were more capable of attracting and creating new industries and less likely to lose industries. Although firms in highly concentrated markets were subject to fiercely aggressive behavior by rivals, high market concentration may also have led to higher price-cost margins, attracting new industry entry and reducing the probability of industry exit. As Table 4.2 shows, cities in more developed coastal regions witnessed higher levels of industry entry and lower levels of industry exit, whereas the opposite prevailed in Western China. This echoes recent studies that have shown that China’s manufacturing industries were increasingly concentrated in coastal regions, until the late 2000s and early 2010s when a reversal occurred and industries started to diffuse and to relocate from Eastern to Central and Western China (He et al. 2008; Zhu and He 2014).

2

The Eastern (Coastal) Region includes Liaoning, Hebei, Beijing, Tianjin, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian, Guangdong, Hainan, and Guangxi; Central Region includes Heilongjiang, Jilin, Shanxi, Neimenggu, Henan, Hubei, Hunan, Anhui, and Jiangxi; and Western Region includes Chongqing, Sichuan, Guizhou, Yunnan, Shaanxi, Xizang, Ningxia, Gansu, Qinghai, and Xinjiang.

logdensity logPUB SOE EXP LAB R&D logPOP GROW GINI

logdensity 1 0.13 0.04 0.17 0.11 0.08 0.54 0.01 0.04

Table 4.1 Pearson correlation matrix

1 0.19 0.04 0.01 0.09 0.06 0.34 0.02

logPUB

1 0.36 0.22 0.16 0.01 0.20 0.21

SOE

1 0.47 0.02 0.00 0.01 0.06

EXP

1 0.15 0.00 0.16 0.33

LAB

1 0.00 0.12 0.20

R&D

1 0.01 0.00

logPOP

1 0.00

GROW

1

GINI

4.5 Empirical Results 87

0.272*** 0.757*** 1.633*** Included Included 3.144*** 174,417 64491.467 2974.79

0.276*** 0.770*** 1.644*** Included Included 3.126*** 174,417 64335.599 3004.11

2 0.541*** 0.321*** 0.809*** 0.998*** 0.0136** 4.471*** 0.309*** 0.641*** 0.00159 2.359*** 0.583*** 0.520*** 0.00325 2.633*** 0.277*** 0.772*** 1.643*** Included Included 3.204*** 174,417 64365.938 2998.69

3 0.681*** 0.243*** 1.086*** 0.629*** 0.0135** 2.888***

4 0.548*** 0.306*** 1.022*** 0.921*** 0.0158** 3.869*** 0.102* 0.563*** 0.000625 1.758*** 0.521*** 0.230*** 0.00535 1.669** 0.278*** 0.776*** 1.649*** Included Included 3.148*** 174,417 64290.226 3009.4 0.287*** 0.656*** 0.897*** Included Included 0.920*** 64,295 32479.444 2022.24

Exit 5 0.924*** 0.0916 0.382*** 0.434*** 0.0584*** 7.516***

0.287*** 0.674*** 0.898*** Included Included 0.775*** 64,295 32352.562 2131.44

6 0.554*** 0.118 0.257*** 1.261*** 0.0755*** 6.482*** 0.182* 1.444*** 0.0364* 1.256 0.0531 0.866*** 0.145*** 0.962 0.291*** 0.656*** 0.902*** Included Included 1.014*** 64,295 32421.753 2060.94

7 0.908*** 0.248* 0.376*** 0.321*** 0.0364** 7.591***

8 0.614*** 0.0736 0.220** 1.313*** 0.0252 6.508*** 0.219* 1.498*** 0.0137 1.237 0.0917 0.131 0.154*** 0.0945 0.288*** 0.674*** 0.896*** Included Included 0.846*** 64,295 32337.28 2138.35

4

*p < 0.10, **p < 0.05, ***p < 0.01

EAST WEST SOE EXP LAB R&D EAST*SOE EAST*EXP EAST*LAB EAST*R&D WEST*SOE WEST*EXP WEST*LAB WEST*R&D logPOP GROW GINI Industry Province Constant N Log lik. Wald Chi-squared

Entry 1 0.668*** 0.321*** 0.943*** 0.735*** 0.0115** 3.420***

Table 4.2 Estimation results on industry characteristics and big region variables

88 What Matters for Regional Industrial Dynamics in China?

4.5 Empirical Results

89

During this time period, the spatial distribution of SOE-dominated industries was relatively stable, due probably to the fact that production by SOEs was predicated heavily on the state’s social, political, and military considerations (see also Ma and Wei 1997 for an example of the geographical dynamics of SOEs in China). Statistical results show that it was harder for regions to attract such industries and the probability of industry exit was low as well. Highly export-oriented industries were more likely to exit and less likely to enter, possibly because export-oriented industries in Chinese cities had already reached saturation point in the 2000s, and as labor costs rose, production geared toward domestic markets grew at the expense of export-oriented production (see also Zhu and Pickles 2015 for an example of how China’s manufacturing industries increasingly tapped into the booming domestic market in the 2000s). The spatial distribution of labor-intensive industries was relatively unstable during 1998–2008, as the coefficients of LAB in both entry and exit models are positive. In other words, labor-intensive industries were more likely to enter and exit, indicating they are quite footloose and always “racing to the bottom.” Since entry barriers in technology-intensive, science-based industries are high, it was difficult for Chinese regions to attract such industries. In addition, the probability of exit was higher, due possibly to the risks and uncertainties in such industries. The interaction terms between SOE and EAST/WEST have a negative/positive and significant sign in entry models, indicating that SOE-dominated industries have been relocating from Eastern to Western China. In contrast, the interaction terms between EXP and EAST/WEST show that export-oriented industries have been concentrating in China’s coastal regions. Eastern China attracted labor-intensive industries in this time period, as well as a large number of migrant workers moving from inland China to coastal provinces (see also Fan 2005). Finally, developed Eastern China was more successful in terms of creating and luring technology-intensive, science-based industries. Table 4.3 reports the empirical results showing how regional industry dynamics were shaped by region- and industry-specific factors. The key findings were as follows. First, the estimated parameters of all control variables and industry characteristic variables were mostly unaltered, although there were changes in significance in a few cases. Second, density had a positive (negative) effect on industry entry (exit), which is consistent with the theoretical predictions (Hypothesis 4.1). The parameter for public expenditure was also positive (negative) and significant in all entry (exit) models (Hypothesis 4.2). These findings resonate with those of Fisher (1997) showing how government spending that improves the quality and quantity of the available public goods and services may attract industry entry and reduce the probability of industry exit. Moving onto the results connected more closely with the central argument (Models 2–4 and 6–8 in Table 4.3), the interaction term between logdensity and SOE had negative (positive) and significant signs in all entry (exit) models, suggesting that the entry and exit of SOE-dominated industries was affected less by technological relatedness and preexisting regional industrial structures than that of non-SOE-dominated industries. Even if a region is not endowed with most of the

0.000696 0.704*** 1.824*** Included Included 0.438*** 163,526 53582.204 2530.77

0.000882 0.708*** 1.848*** Included Included 0.440*** 163,526 53532.451 2523.95

2 1.352*** 0.128*** 1.772*** 0.312** 0.0352* 7.964*** 0.314*** 0.0865 0.0269*** 6.310*** 0.628*** 0.0831* 0.0195** 2.399*** 0.00100 0.740*** 1.907*** Included Included 0.727*** 163,526 53504.49 2456.26

3 1.406*** 0.231*** 2.795*** 0.697*** 0.0276 1.570

4 1.371*** 0.240*** 3.015*** 0.539*** 0.0633** 11.15*** 0.104 0.101 0.0226** 6.188*** 0.632*** 0.0985** 0.0157* 1.505*** 0.000828 0.742*** 1.938*** Included Included 0.671*** 163,526 53466.009 2441.74 0.0548*** 0.683*** 1.217*** Included Included 3.174*** 63,738 30242.715 2839.58

Exit 5 1.567*** 0.0246 0.0776 0.231*** 0.0570*** 7.320***

0.0511*** 0.696*** 1.215*** Included Included 3.042*** 63,738 30228.658 2841.08

6 1.500*** 0.0233 0.146 0.747*** 0.102*** 10.77*** 0.0406 0.621*** 0.0302 2.235 1.026*** 0.0475 0.0230* 0.326 0.0530*** 0.677*** 1.282*** Included Included 3.680*** 63,738 30121.775 2780.17

7 1.615*** 0.191*** 2.409*** 0.347* 0.113*** 7.340***

8 1.427*** 0.216*** 1.553*** 0.489* 0.123*** 8.951*** 0.734*** 0.675*** 0.00958 1.215 1.142*** 0.149* 0.0219* 0.160 0.0507*** 0.676*** 1.339*** Included Included 3.391*** 63,738 30101.66 2763.76

4

*p < 0.10, **p < 0.05, ***p < 0.01

logdensity logPUB SOE EXP LAB R&D logdensity *SOE logdensity *EXP logdensity *LAB logdensity *R&D logPUB *SOE logPUB *EXP logPUB *LAB logPUB *R&D logPOP GROW GINI Industry Province Constant N Log lik. Wald Chi-squared

Entry 1 1.357*** 0.124*** 1.190*** 0.478*** 0.0180*** 3.689***

Table 4.3 Estimation results on region- and industry-specific factors

90 What Matters for Regional Industrial Dynamics in China?

4.5 Empirical Results

91

capabilities that a certain new industry requires, and even if the barriers to the development of this industry are high, SOEs may be able to overcome these barriers with the help of favorable policies, government aid, tax credits, etc. In addition, the interaction term, logPUB*SOE, had the same relationship with industry entry and exit, probably because SOE-dominated industries’ entry and exit depended less on public goods and services provided by local governments (e.g., education, labor training, infrastructure, health care, etc.) than on government aid, tax credits, and subsidies. The former measures the “facilitating” role of local governments, while the latter refers to direct government intervention capturing the extent to which local governments play an “intervening” and hands-on role in regional economic development. logdensity*EXP displays a relationship with industry entry that is consistent with the hypothesis, though it is not significant: the entry of export-oriented industries depended more on technological relatedness and regional historical industrial structure, as new, entering industries may benefit from related variety and gain knowledge about international markets, regulations, global competitors, etc. from their neighbors. In addition, export-oriented industries that were closely related to regional industrial structures were less likely to exit during this time period. On the other hand, the parameter of logPUB*EXP was negative (positive) and significant in entry (exit) models, indicating that export-oriented industries were unwelcome in regions with high levels of public expenditure. One possible explanation is that since the early 2000s, China’s wealthy regions with good fiscal performance, particularly those in coastal provinces, have increasingly abandoned export-processing industries and upgraded into more advanced industries (see also Zhu and Pickles 2014). Finally, the statistical results show that logdensity*R&D and logPUB*R&D had significant and positive signs in entry models, whereas logdensity*LAB and logPUB*LAB had significant and negative relationships. These relationships are consistent with our hypothesis. Entry barriers in labor-intensive industries are much lower than that in technology-intensive industries, and so it is easier for a region, given its institutional context and preexisting industrial structure, to attract and/or create labor-intensive industries. A science-based industrial sector’s entry to a region is, however, more dependent on public services and technological relatedness. Another reason for the positive relationship between logPUB*R&D and industry entry and for the negative relationship between logPUB*LAB and industry entry is that as labor costs rose in China, wealthy regions with a good fiscal performance gradually upgraded from labor-intensive, low-tech, and low-valueadded industries to high-tech, high-value-added, and science-based industries. In other words, more capable and developed local administrations have been abandoning the former while simultaneously embracing the latter.

92

4.6

4

What Matters for Regional Industrial Dynamics in China?

Conclusion

By examining the regional industrial dynamics of China’s manufacturing industries, this chapter has shed new light on debates about technological relatedness and industrial evolution. Technological relatedness, as a key driving force of regional economic development, not only affects the growth of existing industries through externalities derived from related variety but is also responsible for the entry of new industries. Recent studies emphasize the ways in which regional industrial diversification emerged as a path-dependent process, as regions branch into technologically related industries. This research pays too little attention to other factors, to the differences that the industrial diversification process can display across regions, and, more importantly, to the possible effect of regional institutions on the intensity and nature of industrial diversification processes. The recent technological relatedness literature has paid even less attention to industry characteristics. This chapter therefore seeks to bring regional institutional contexts and industry characteristics to the forefront in an investigation of the characteristics of industries that entered into or exited from with different region-specific assets (e.g., institutional contexts and regional resources). By using firm-level data of China’s manufacturing industries during 1998–2008, this chapter showed that SOE-dominated industries were relocating from China’s coastal provinces to inland regions during this time period, while export-oriented industries remained in the coastal region. Regions in relatively developed Eastern China were also more successful in creating and luring not only labor-intensive, low-tech industries but also high-tech, science-based industries during 1998–2008, resulting in massive migration flows from inland to coastal regions. More importantly, empirical results confirm the hypotheses concerning the co-shaping of regional industrial dynamics on the one hand and regional institutional contexts, technological relatedness, and industry characteristics on the other. Specifically, first, SOE-dominated industries’ entry and exit depended less on technological relatedness and knowledge spillovers, and less on public services provided by local governments (e.g., education, labor training, infrastructure, health care, etc.), but more on government aid, tax credits, and subsidies. Second, the entry and exit of export-oriented industries was more reliant on technological relatedness. However, statistical results also show that export-oriented industries were increasingly unwelcome in wealthy regions, as the latter sought to upgrade from low-value, low-end, and export-processing to high-value and high-end industries. Finally, local governments’ intention to upgrade their regional industrial structure can be also seen from the entry and exit pattern of labor-intensive and technology-intensive industries. More capable and developed local administrations have been abandoning the former while simultaneously embracing the latter. Furthermore, entry barriers are lower in the former than in the latter; it is thus much easier for a region, given its institutional context and preexisting industrial structure, to attract and/or create labor-intensive industries, while the entry/exit of science-based industrial sectors into/from a region is more reliant on regional institutional arrangements and technological relatedness.

References

93

Several policy implications can be drawn from the empirical findings. First, policy-makers should aim to foster a more market-oriented, less state-led business environment, as SOE-dominated industries tend to rely heavily on government aid, subsidies, and tax credits. What is worse is that their development is mostly isolated from the surrounding regional environment. The knowledge exchange between these industries and others is quite weak. Second, as China is upgrading from low-value, low-end, low-tech industries to more advanced and higher-tech production, local administrations can attract and generate new, high-tech industries by providing good public goods and services in order to lower entry barriers and costs for new entrants. However, the formation of technology-intensive industries also depends to a large extent on technological relatedness. Efforts by local administrations to create hightech industries may therefore be useless if the distance between new, high-tech industries and regional existing industrial structure is too great for the region to transcend. Some recent studies have pointed out the obsession of Chinese local governments with high-tech industries may be unrealistic and possibly even selfdefeating (Zhu and He 2015). Moreover, policies that imply a total abandonment of existing industrial sectors in favor of high-tech ones are sometimes developed without regard to existing capacities and the presence of parallel capabilities. It is time for Chinese local governments to rethink their fixation on high-tech industries. According to this research, policy-makers should pay more attention to high-tech industries that are related to preexisting regional industrial structures and lure them with good public goods and services.

References Bai, C. E., Du, Y., Tao, Z., & Tong, S. Y. (2004). Local protectionism and regional specialization: Evidence from China’s industries. Journal of International Economics, 63(2), 397–417. Basile, R., Pittiglio, R., & Reganati, F. (2016). Do agglomeration externalities affect firm survival? Regional Studies, 51(4), 1–16. Boschma, R., & Capone, G. (2015a). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, 44(10), 1902–1914. Boschma, R., & Capone, G. (2015b). Relatedness and diversification in the European Union (Eu-27) and European Neighbourhood Policy countries. Environment and Planning C: Government and Policy, 34(4), 617–637. Boschma, R., & Frenken, K. (2011). Technological relatedness and regional branching. In H. Bathelt, M. Feldman, & D. F. Kogler (Eds.), Beyond territory: Dynamic geographies of knowledge creation, diffusion and innovation (pp. 64–81). London: Routledge. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new industries at the regional level in Spain: A proximity approach based on product relatedness. Economic Geography, 89 (1), 29–51. Cainelli, G., & Iacobucci, D. (2016). Local variety and firm diversification: An evolutionary economic geography perspective. Journal of Economic Geography, 16(5), 1079–1100. Chang, W., & MacMillan, I. C. (1991). A review of entrepreneurial development in the People’s Republic of China. Journal of Business Venturing, 6(6), 375–379.

94

4

What Matters for Regional Industrial Dynamics in China?

Clerides, S. K., Lach, S., & Tybout, J. R. (1998). Is learning by exporting important? Microdynamic evidence from Colombia, Mexico, and Morocco. The Quarterly Journal of Economics, 113(3), 903–947. Coe, N., Dicken, P., & Hess, M. (2008). Global production networks: Realizing the potential. Journal of Economic Geography, 8(3), 271–295. Dawley, S. (2014). Creating new paths? Offshore wind, policy activism, and peripheral region development. Economic Geography, 90(1), 91–112. Delgado, M., Porter, M. E., & Stern, S. (2016). Defining clusters of related industries. Journal of Economic Geography, 16(1), 1–38. Fan, C. C. (2005). Interprovincial migration, population redistribution, and regional development in China: 1990 and 2000 census comparisons. The Professional Geographer, 57(2), 295–311. Fisher, R. C. (1997, March/April). Effects of state and local public services an economic development. New England Economic Review, 53–82. Guo, Q., & He, C. (2015). Production space and regional industrial evolution in China. GeoJournal, 80, 1–18. Hall, P. A., & Soskice, D. (2001). Varieties of capitalism: The institutional foundations of comparative advantage. Oxford: Oxford University Press. Hassink, R., Klaerding, C., & Marques, P. (2014). Advancing evolutionary economic geography by engaged pluralism. Regional Studies, 48(7), 1295–1307. He, C., Wei, Y. H. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. He, C., Ye, X., & Wang, J. (2012). Industrial agglomeration and exporting in China: What is the link? Regional Science Policy & Practice, 4(3), 317–333. Hidalgo, C. A., Klinger, B., Barabási, A.-L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Lovely, M. E., Rosenthal, S. S., & Sharma, S. (2005). Information, agglomeration, and the headquarters of U. S. exporters. Regional Science and Urban Economics, 35(2), 167–191. Ma, L. J. C., & Wei, Y. (1997). Determinants of state investment in China, 1953–1990. Tijdschrift voor Economische en Sociale Geografie, 88(3), 211–225. Martin, R. (2010). Roepke lecture in economic geography—rethinking regional path dependence: Beyond lock-in to evolution. Economic Geography, 86(1), 1–27. McMillan, J., & Naughton, B. (1992). How to reform a planned economy: Lessons from China. Oxford Review of Economic Policy, 8(1), 130–143. Melitz, M. J. (2003). The impact of trade on intra-industry reallocations and aggregate industry productivity. Econometrica, 71(6), 1695–1725. Montinola, G., Qian, Y., & Weingast, B. R. (1995). Federalism, Chinese style: The political basis for economic success in China. World Politics, 48(1), 50–81. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Oi, J. C. (1995). The role of the local state in China’s transitional economy. The China Quarterly, 144(144), 1132–1149. Pan, F., Zhang, F., Zhu, S., & Wójcik, D. (2016). Developing by borrowing? Inter-jurisdictional competition, land finance and local debt accumulation in China. Urban Studies, 54(4). Pavitt, K. (1984). Sectoral patterns of technical change: Towards a taxonomy and a theory. Research Policy, 13(6), 343–373. Peterson, P. E. (1981). City limits. Chicago: University of Chicago Press. Simmie, J. (2012). Path dependence and new technological path creation in the Danish wind power industry. European Planning Studies, 20(5), 753–772. Smith, A., Pickles, J., Buček, M., Pástor, R., & Begg, B. (2014). The political economy of global production networks: Regional industrial change and differential upgrading in the East European clothing industry. Journal of Economic Geography, 14(6), 1023–1051.

References

95

Sun, Q., & Tong, W. H. S. (2003). China share issue privatization: The extent of its success. Journal of Financial Economics, 70(2), 183–222. Sydow, J., Lerch, F., & Staber, U. (2010). Planning for path dependence? The case of a network in the Berlin-Brandenburg optics cluster. Economic Geography, 86(2), 173–195. Tanner, A. N. (2014). Regional branching reconsidered: Emergence of the fuel cell industry in European regions. Economic Geography, 90(4), 403–427. Van Cauwenberge, P., Beyne, P., & Vander Bauwhede, H. (2016). An empirical investigation of the influence of municipal fiscal policy on firm growth. Environment and Planning C: Government and Policy, 34(8), 1825–1842. Wei, Y. D. (2001). Decentralization, marketization, and globalization: The triple processes underlying regional development in China. Asian Geographer, 20(1–2), 7–23. Wei, Y. D., Li, W., & Wang, C. (2007). Restructuring industrial districts, scaling up regional development: A study of the Wenzhou Model, China. Economic Geography, 83(4), 421–444. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yu, J., Zhou, L. A., & Zhu, G. (2016). Strategic interaction in political competition: Evidence from spatial effects across Chinese cities. Regional Science and Urban Economics, 57, 23–37. Zhu, S., & He, C. (2013). Geographical dynamics and industrial relocation: Spatial strategies of apparel firms in Ningbo, China. Eurasian Geography and Economics, 54(3), 342–362. Zhu, S., & He, C. (2014). Global, regional and local: New firm formation and spatial restructuring in China’s apparel industry. GeoJournal, 79(2), 237–253. Zhu, S., & He, C. (2015). Global and local governance, industrial and geographical dynamics: A tale of two clusters. Environment and Planning C: Government and Policy, 34, 1453–1473. Zhu, S., & Pickles, J. (2014). Bring in, go up, go west, go out: Upgrading, regionalisation and delocalisation in China’s apparel production networks. Journal of Contemporary Asia, 44(1), 36–63. Zhu, S., & Pickles, J. (2015). Turkishization of a Chinese apparel firm: Fast fashion, regionalisation and the shift from global supplier to new end markets. Cambridge Journal of Regions, Economy and Society, 8(3), 537–553.

Chapter 5

What Facilitates New Firm Formation in China?

5.1

Introduction

New firm formation is essential for the continuing vitality of the modern economy. A higher entry rate of new businesses is a fundamental driver of sustainable economic growth (Djankov et al. 2002). The spatial distribution of new start-ups however is uneven and thus a significant explanation of regional economic disparity (Stam 2010). Silicon Valley appears more entrepreneurial than the declining cities of the Rust Belt in the USA (Glaeser and Kerr 2009). A distinct spatial difference of new start-ups exists in India (Ghani et al. 2014). Likewise, some coastal cities in China appear more active in new firm formation than inland cities. To explain why new firm formation is spatially uneven, previous studies have mainly focused on individual characteristics of potential entrepreneurs such as their location choice preference, age, sex, and educational level (Cooper and Folta 2000; Stam 2007; Michelacci and Silva 2007; Bates 1990; Armington and Acs 2002; Delfmann et al. 2014; Elert 2014). However, it would be difficult for talented entrepreneurs to start a business without favorable external environment (Stam 2010). Recently, there has been a growing debate on the role of regional economic environment, especially industrial clusters and agglomeration economies, in shaping new firm formation in developed countries (Porter 1990, 1998a, b; Saxenian 1994; Feldman 2001; Glaeser and Kerr 2009). They stress the importance of localization economies and urbanization economies to the birth of new businesses (Rosenthal and Strange 2003, 2004; Henderson 2003). However, it is far from enough in agglomeration studies to distinguish only localization economies from urbanization economies. Further studies on the internal structure of agglomeration economies are increasingly necessary. First, a rising body of literature in evolutionary economic

Modified article originally published in [Guo, Q., C. He and D. Li (2016) Entrepreneurship in China: The Role of Localisation and Urbanisation Economies, Urban Studies, 53 (12), pp. 2584–606.]. Published with kind permission of © [Sage, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_5

97

98

5 What Facilitates New Firm Formation in China?

geography introduces the notion of technological relatedness to redefine the boundary of knowledge spillovers and rethink the agglomeration effect. Knowledge spillovers are more likely to exist across technologically related industries rather than across unrelated industries and same industries (Boschma 2005), because interactive learning between firms is more efficient when there is some degree but not too much or too little cognitive proximity between firms (Timmermans and Boschma 2011). Technological relatedness is also considered as a key factor for new firm formation (Feldman et al. 2005; Glaeser and Kerr 2009). Although they refer to the importance of technological relatedness in explaining new firm formation, few studies measure accurately technological relatedness in such studies. Second, some attention has been paid to the role of small firms in the generation of agglomeration economies (Vernon 1960; Chinitz 1961; Piore and Sabel 1984; Saxenian 1994). The important role of small firm clusters is firstly put forward by Vernon (1960) and Chinitz (1961) and is then coined as “the Vernon-Chinitz effect” in the literature (Rosenthal and Strange 2010; Glaeser et al. 2010). In these studies, average firm size or Hirshmann-Herfindahl index is often measured to investigate the effect of small firms. However, the measurements cannot accurately reflect the real role of small firms. The main goal of this chapter is to examine the explanatory power of agglomeration economies and its internal structure in new firm formation in China. The gradually accelerating process of China’s economic reform has unleashed the emerging power of entrepreneurship, which is increasingly critical for China’s economic development. Institutional transformation such as the process of marketization, decentralization, and globalization (He et al. 2008) not only provides more market opportunities but also imposes more serious challenges for Chinese entrepreneurs. Facing complex and uncertain institutional environment, entrepreneurs are more inclined to locate in clusters with production networks and knowledge spillovers. This study is among the first to explore the effect of agglomeration economies on new firm formation in China by highlighting the role of technological relatedness and small firm clusters. Moreover, most variables in this chapter are measured at the 3-digit sector level and at the prefectural city level, so we can control the sectoral and spatial variations of new firm formation. The remainder of the chapter is structured as follows. Section 5.2 reviews the theoretical and empirical literature and develops the research hypotheses. Section 5.3 introduces data sources and describes the patterns of new firm formation. Section 5.4 gives more details on the measurement of variables. The empirical results are presented in Sect. 5.5. This chapter concludes with a summary of the findings and discussion on policy implications.

5.2 Literature Review and Hypothesis Development

5.2 5.2.1

99

Literature Review and Hypothesis Development The Impact of Localization and Urbanization Economies on New Firm Formation

Agglomeration economies are often divided into localization economies and urbanization economies in the prior literature (Moomaw 1988; Henderson 1997; Viladecans-Marsal 2004). Localization economies stem from the clustering of firms within the same industry in the local economy, which generates external economies available to local firms through labor market pooling, input-output linkage, and specialized knowledge spillovers (Marshall 1920). Compared with incumbent firms, new entries are more dependent on localization economies, since new firms are unfamiliar with local production network, and have not built strong ties with local suppliers and may be in shortage of local knowledge on workforce, business management, technology, and market (Stinchcombe 1965). So local supplier/customer linkages, specialized labor pooling, and knowledge spillovers could make it easier for potential entrepreneurs to overcome their initial liabilities (Ghani et al. 2014). Localization economies enable new firms to share local input suppliers (input sharing) and local clients with incumbents, cutting transportation costs and lowering entry barriers. Krugman (1991) also stresses that access to consumers could reduce shipping costs. Proximity to buyers and sellers could build the trust relationship during the economic transaction to improve translation efficiency and reduce transaction cost (Romero-Martínez and Montoro-Sánchez 2008). Porter (1990) further emphasizes that proximity to customers and suppliers could enhance innovation through knowledge spillovers. Moreover, a large number of suppliers and customers located in a specific area can help entrepreneurs find suppliers and buyers efficiently and thus reduce searching costs (Stuart 1979). Therefore, we would pay more attention to customer/supplier linkages in several mechanisms of agglomeration effects. This leads to the following research hypotheses. Hypothesis 5.1 Localization economies are beneficial to new firm formation. Hypothesis 5.2 Cities with more supplier/customer linkages provide more favorable environment for new firm formation in China. Urbanization economies arise when firms benefit from the concentration of various industries in the local economy. Jacobs (1969) argues that a wealth of industrial diversity in urban areas is the most important source of knowledge spillovers. The cross-fertilization incubated by industrial variety enhances innovation and entrepreneurial success (Duranton and Puga 2001). Jacobs externalities is coined by Glaeser et al. (1992) to capture the interindustry spillover benefits from local diversity. Stel and Suddle (2008) find that Jacobs externalities provide niche markets and more opportunities for success, leading to a higher rate of new firm formation.

100

5 What Facilitates New Firm Formation in China?

When comparing the significance of localization economies and Jacobs externalities, most studies (e.g., Henderson 1997; Combes et al. 2004) support the importance of localization argument, while the role of Jacobs externalities is less well established. Rosenthal and Strange (2003) report that localization effects are more important than urbanization effects in nearly all cases for both births and new firm employment. The role of Jacobs externalities is ambiguous in the prior empirical studies, but we still propose the following research hypotheses according to the above theoretical analysis. Hypothesis 5.3 Jacobs externalities may encourage new firm formation in China.

5.2.2

The Impact of Related Variety and Unrelated Variety

Knowledge spillovers among a great diversity of industries within a region are a major source of Jacobs externalities. But can knowledge flow efficiently between any industries? Since the important contribution of Boschma (2005), there has been an increasing awareness that relational proximity between economic agents is different from geographic proximity. Cognitive proximity is more important than geographical proximity for information spillover. Technological relatedness occurs when firms in a region operate within technologically related industries that have overlapping knowledge bases (Boschma and Frenken 2011). Boschma and Frenken (2011) argue that local knowledge spillovers are more likely to occur within regions hosting a large number of technologically related industries. Interactive learning between firms is more efficient when there is some degree but not too much cognitive proximity between firms (Timmermans and Boschma 2011). Frenken et al. (2007) find that the effect of Jacobs externalities is higher in regions with a related variety of sectors than in regions with an unrelated variety of sectors. Technological relatedness is an important driver of the birth and evolution of new technologies, new product, and even new industries and clusters (Boschma and Frenken 2006). Knowledge spillovers between related industries make it easier for potential entrepreneurs to find new technologies or markets and finally establish new firms. Moreover, specializing in technologically related industries can reduce a number of uncertainties that new firms are usually confronted with (Nelson and Winter 1982). An increasing number of studies have supported the influence of technological relatedness on industrial clustering (Boschma and Weterings 2007), plant survival (Neffke et al. 2012), and regional growth (Frenken et al. 2007) but rarely on new firm formation. This leads to the following research hypotheses. Hypothesis 5.4 Compared with unrelated variety, related variety has a larger effect on new firm formation.

5.2 Literature Review and Hypothesis Development

5.2.3

101

The Vernon-Chinitz Effect

Debates on the role of small firms in the generation of agglomeration economies have a long tradition. Vernon (1960) and Chinitz (1961) are the first to focus on the relationship between industrial organization and agglomeration economies. They notice that the larger effect of agglomeration economies arises in clusters with more small firms. The clustering of small firms is conducive to fostering supplier-customer linkages and to enhancing innovation by knowledge spillovers, which are very important source of agglomeration economies. The notion that small firms play a larger role in agglomeration economies is first put forward by Vernon (1960) and Chinitz (1961), so it is also coded as “The Vernon-Chinitz effect.” Other scholars also make important contributions on this effect (Jacobs 1969; Piore and Sabel 1984; Saxenian 1994). Previous studies have provided some important explanations for the VernonChinitz effect. First, small firms are likely to share local specialized suppliers with other local firms to reduce production cost, so specialized and independent suppliers are easier to survive in regions with the abundance of small firms, which would be favorable for birth of new firms (Vernon 1960; Chinitz 1961; Rosenthal and Strange 2003; Glaeser and Kerr 2009). From the perspective of spillover effect, networking among firms may be sparser in the region where large firms dominate, which lowers knowledge spillovers among firms in that local industry (Chinitz 1961; Saxenian 1994; Glaeser et al. 1992; Malmberg and Maskell 2002). Second, employees of small firms would be more likely to start a business because they not only engage in different types of tasks to accumulate a wealth of managerial experience about how to start and run small firms, but they also expose to network and business environment which is conductive to small firms (Johnson and Cathcart 1979; Dobrev and Barnett 2005; Gompers et al. 2005; Parker 2009). Third, wage which entrepreneurs work for before self-employment is the opportunity cost of self-employment (Acs and Armington 2006). Small firms generally pay lower salary than large firms do, so employees are more likely to quit small firms than in large ones to start their businesses. Fourth, a large presence of small firms signifies a friendly business environment that encourages new firm formation, because regions dominated by small firms have lower entry barriers for new firms (Santarelli and Sterlacchini 1994; Chen 2012). Moreover, small incumbent firms provide successful role models for potential entrepreneurs (Stuart and Ding 2006; Parker 2009). A few empirical studies have supported that small firms incubate effectively new firm formation (Johnson and Cathcart 1979; Reynolds et al. 1994; Acs and Armington 2006; Rosenthal and Strange 2003, 2010; Glaeser and Kerr 2009; Glaeser et al. 2010; Qian and Haynes 2014). The empirical evidence on the role of small firms is largely from the developed economies. Does the Vernon-effect exist in China? Does it increase the probabilities of new firm formation? This leads to the following research hypothesis. Hypothesis 5.5 The clustering of small firms has a larger effect on new firm formation than the clustering of large firms does in China.

102

5.3

5 What Facilitates New Firm Formation in China?

New Firm Formation in China

Data used in this chapter is the ASIF dataset. New firm formation, in a large number of empirical studies, is generally measured as the employment or number of new firms (Glaeser and Kerr 2009; Delgado et al. 2010) and entry rate which is defined as the number of new firms over the number of incumbents (Klapper et al. 2010). However, there are various kinds of new firms in transitional China, and they differ significantly in ownerships. Specifically, objective and motivation of establishing new firms differ across firm ownerships. Newly established foreign-owned firms, whose locational choice is dependent on the international strategy of parent firms, are not completely entrepreneurial. Newly established state-owned firms are mainly driven by national strategy or local governments rather than entrepreneurs. Newly established enterprise-owned firms should be also excluded since we focus on new firms that are independent from existing firms. Therefore, we only use private-owned start-ups to measure new firm formation. Most studies on new firm formation in China use individual or companies’ survey data. Although survey data can cover some important information on entrepreneurial traits, their sample size is often very limited. Though the AISF dataset has many advantages, we cannot ignore its obvious weakness that it does not cover non-state-owned enterprises with sales revenues below 5 million Yuan. Following Glaeser and Kerr (2009) and Delgado et al. (2010), we measure the number of start-ups in their starting years as a proxy of new firm formation. With the deepening reform of state-owned enterprises and development of market economy in China, the percentage of state-owned start-ups in employment and number decrease by 69.7% and 91.3%, respectively, during 2001–2007 (Table 5.1), while those of private-owned start-ups increase by 63.2% and 45.1%, respectively. These figures indicate that the influence of private-owned business on China’s manufacturing economy has proliferated in the early twenty-first century, and the vitality of the economy has been unleashed. For almost all industries except manufacture of chemical fibers, smelting and pressing of ferrous metals and manufacture of tobacco,1 the number of privateowned start-ups increases rapidly by 100% over the period 2001–2007 (Fig. 5.1). There is substantial industrial variation in the magnitude and change of new firm formation. Taking year 2007 as an example, the number of private-owned start-ups ranges from 17 for manufacture of chemical fibers to 776 for manufacture of nonmetallic mineral products. Most industries with a high level of new firm formation are those with low entry thresholds, such as manufacture of nonmetallic mineral products, manufacture of textile, processing of agricultural products, and so forth. During 2001–2007, the fast-growing industries in terms of new firm formation are not all labor-intensive or light industries. The level of new firm formation in manufacture of special purpose machinery, manufacture of rubber, and manufacture of general purpose machinery has also soared rapidly. 1

Manufacture of tobacco is fully constraint industries in China, so we exclude it in this study.

Year 2001 2002 2003 2004 2005 2006 2007 Year 2001 2002 2003 2004 2005 2006 2007

Empl. of start-ups 992,163 583,465 1,053,907 1,850,418 1,241,207 1,174,155 1,425,489 No. of Start-ups 4587 3101 6022 12,665 8981 9186 12,314

Private (%) 29.80 32.83 41.44 42.68 44.46 45.17 48.64 Private (%) 45.26 48.63 55.01 58.41 59.82 65.30 65.70

Foreign (%) 14.46 20.68 20.91 27.83 26.60 31.89 25.79 Foreign (%) 13.87 16.83 15.34 18.05 15.86 16.12 14.13

Table 5.1 The employment and number of all start-ups and start-ups of different ownership, 2001–2007 State (%) 5.52 2.96 2.38 1.16 3.42 0.77 1.67 State (%) 3.90 2.03 1.18 0.83 0.57 0.36 0.34

Others (%) 50.22 43.53 35.28 28.33 25.52 22.17 23.89 Others (%) 36.97 32.51 28.46 22.72 23.76 18.22 19.83

5.3 New Firm Formation in China 103

104

5 What Facilitates New Firm Formation in China?

Fig. 5.1 The number of private owned start-ups by 2-digit manufacturing industries, 2001 and 2007

We apply the Gini coefficient to illustrate the spatial distribution of new firm formation and its change in China between 2001 and 2007. All Gini coefficients are greater than 0.6, indicating that new firm formation in China is spatially concentrated. During 2001–2007, there see a trend of increasing concentration of new firm formation (Fig. 5.2). Specifically, the developed coastal regions and Chongqing, as the growth pole in the western regions, observe a high level of new firm formation. Cities in Shandong province, the inland of Fujian province, the north of Liaoning province, and the regions along the Yangtze River experience significant growing trend of new firm formation during the period. Shandong province is the fastest growing region of new firm formation, with 138 start-ups in 2001 and 2295 in 2007. To further explore the relationship between agglomeration economies and new firm formation,2 we overlay the spatial distribution of new firm formation with that of manufacturing density in 2007. Figure 5.3 shows that new firm formation seems higher in most cities with a high density of manufacturing, but there are some exceptions. Some cities in Heilongjiang province, the north of Hebei province, the south of Liaoning province, Henan province, and the coastal area of Fujian see a high level of manufacturing agglomeration but a low level of new firm formation. This may be related to regional industrial structure and the internal structure of agglomeration economies. The following section is to test the influence of agglomeration economies and its internal structure on new firm formation.

2

Agglomeration is measured by the number of all existing manufacturing firms in 2007.

5.3 New Firm Formation in China

105

Fig. 5.2 Spatial pattern of private-owned manufacturing start-ups, 2001, 2003, 2005, and 2007

Fig. 5.3 Spatial pattern of agglomeration and new firm formation in China, 2007

106

5.4 5.4.1

5 What Facilitates New Firm Formation in China?

Model Specification and Variables Dependent Variables and Model Specification

The above descriptive analysis indicates that new firm formation varies across cities and sectors. This section is to identify the model specification and measurement of variables in order to conduct a systematic analysis on the determinants of new firm formation. The dependent variable is defined as the number of private-owned startups at the 3-digit manufacturing sector level and at the prefectural city level. The number of start-ups is nonnegative integer, and the distribution of start-ups is generally skewed to the right. Moreover, it contains a large proportion of zeroes. We therefore estimate a negative binomial model with a city-sector panel to deal with the right-censored problem and simultaneously control for the heterogeneity of cities and sectors. The city-sector fixed effect is employed instead of the city-year fixed effect because sectoral variation is much larger than temporal variation. We would present the results in different years. The basic model (5.1) and one of extended model (5.2) are as follows. Pðnirt Þ ¼ βot þ β1 LOC irt2 þ β2 Dirt2 þ β3 ENTRY irt2 þ β4 GROWTH irt2 þ β5 SUBSIDY irt2 þ β6 γ 1 þ β7 γ 2 þ εirt

ð5:1Þ

Pðnirt Þ ¼ βot þ β1 SUPPLIERirt2 þ β2 CUSTOMERirt2 þβ3 Rirt2 þ β4 U irt2 þ β5 ENTRY irt2 þβ6 GROWTH irt2 þ β7 SUBSIDY irt2 þβ8 γ 1 þ β9 γ 2 þ εirt

ð5:2Þ

where i, r, and t denote sector, city, and time, respectively. All variables are aggregated ones at the city-sector level by using the data from the ASIF.

5.4.2

Localization Externalities and Supplier/Customer Linkages

Following Delgado et al. (2010), we use the location quotient (LOC) of employment at the 3-digit manufacturing sector level as a proxy of localization economies. According to Marshall (1920) and Krugman (1991), supplier/customer linkages are the major source of localization economies. So we introduce two variables— SUPPLIER and CUSTOMER—which are based on:

5.4 Model Specification and Variables

INPUTri ¼

107

X

inputi

k

 Emplrk

ð5:3Þ

k2I

OUTPUTri ¼

X

outputi!k  Emplrk

ð5:4Þ

k2I

where inputi k denotes the share of sector i’s inputs that come from sector k and is also regarded as the weight, ranging from zero (no input from sector k) to one (full dependency on sector k); Emplrk stands for the employment in sector k in city r; i indexes sectors. INPUTri represents the potential input relations provided by city r for new firms in sector i. Similarly, outputi ! k denotes the share of sector i’s outputs that are purchased by sector k, ranging from zero (no output goes to sector k) to one (all outputs go to sector k); OUTPUTri is the potential customer linkage provided by city r for new firms in sector i. Then the location quotient of INPUT and OUTPUT is used to measure comparative advantages of supplier linkage (SUPPLIER) and customer linkage (CUSTOMER), respectively. China’s 2002 input– output table with 122 sectors (National Bureau of Statistics 2006) is used to compute inputi k and outputi ! k in the formula of SUPPLIER and CUSTOMER.

5.4.3

Jacobs Externalities, Related Variety, and Unrelated Variety

In the literature, subsectors that belong to the same sector are defined as related, otherwise, unrelated (Frenken et al. 2007; Boschma and Iammarino 2009). But there may exist strong technological relatedness between subsectors that do not belong to the same sectors (Essletzbichler 2015). The other approach to measure technological relatedness is to calculate similarity between sectors in the use of input factors (Farjoun 1994; Dumais et al. 2002). As input mix reflects production technology, two sectors with highly similar input mixes mean a close “technological distance” (Frenken et al. 2007). Combining the above two methods, we measure technological relatedness between 3-digit manufacturing sectors as follows. First, technological relatedness between sectors captured the similarity between two sectors’ input mixes based on China’s 2002 input-output table with 122 sectors. Following Los (2000), the similarity is calculated using cosine distance, defined as the cosine between a pair of input coefficient vectors,

108

5 What Facilitates New Firm Formation in China?

P

αik  α jk k ωij ¼ sffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi  ffi P 2 P 2 αik  α jk k

ð5:5Þ

k

where αi and αj are the pair of input coefficient vectors, and k denotes the kth input. ωij would be closer to one, if two sectors show higher technological relatedness. Following Los (2000), two sectors are defined as related if cosine distance between two sectors is over 0.4. Second, each pair of 3-digit sectors within the same 2-digit sector is defined as related. Technological relatedness is a dummy variable in this study, 1 for related and 0 for unrelated. Therefore, we construct a matrix of technological relatedness among 162 3-digit sectors. To break down Jacobs externalities, we use the total employment in all other manufacturing sectors except sector i (D) as a proxy of Jacobs externalities that sector i in city r faced with. Related variety (R) is measured by the employment in the sectors related technologically with sector i in city r, while unrelated variety (U ) is measured by the employment in the sectors unrelated technologically with sector i in city r as follows. Rir ¼

X

E jr

U ir ¼

j2W , j6¼i

X

E jr

j= 2W , j6¼i

where W is a set of sectors technologically related with sector i. Ejr denotes the employment of sector j in city r. Hence, Jacobs externalities is the sum of Rir and Uir.

5.4.4

The Vernon-Chinitz Effect

To test the Vernon-Chinitz effect, we break down the agglomeration variables (including LOC, SUPPLIER, CUSTOMER, D, R, and U ) using employments of small firms (fewer than 50 employees), medium-sized firms (51–200 employees), and large firms (above 200 employees). The thresholds are higher than Rosenthal and Strange (2003, 2010). There are two reasons to adopt higher thresholds. One is that the manufacturing sector in China is more labor-intensive, leading to larger average size of firms; the other is that our classification makes every group account for about one thirds of the total, and the distribution is more even. As mentioned above, LOC are the location quotients as follows: LOC ¼

ei =e E i =E

ð5:6Þ

5.4 Model Specification and Variables

109

Table 5.2 Definition of independent variables Variable LOC

Definition Localization

Form Log

SUPPLIER

Suppliers

Log

CUSTOMER

Customers

Log

D

Diversification

Log

R

Log

_S _M _L ENTRY

Related variety Unrelated variety Subscript Subscript Subscript Entry rate

GROWTH

Growth rate

SUBSIDY

Subsidy rate

U

γ1 γ2

Log Log Log Log

Measurement The location quotient of employment at the sector/city level The sums of supply sectors’ employment weighted by the proportions of supply sectors’ inputs required by one sector at the city level The sums of demand sectors’ employment weighted by the share of one sector’s output sales that go to demand sector at the city level The sums of other manufacturing’s employment besides the sector itself at the city level The sums of other related manufacturing’s employment at the sector/city level The sums of other unrelated manufacturing’s employment at the city level The measurement calculated at small-sized firms The measurement calculated at medium-sized firms The measurement calculated at large-sized firms The share of the number of new firms in all existing firms at the sector/city level The growth rates at year t are the difference between employment at year t and t-1, divided by employment at year t at the sector/city level The ratio between the subsidies and gross output value at the sector/city level The sector fixed effect of panel model The city fixed effect of panel model

Note: Zero value in some variables such as LOC, SUPPLIER, and CUSTOMER was replaced with the minimum value of corresponding variables

where ei is local employment in sector i, e is total local manufacturing employment, Ei is national employment in sector i, and E is total national manufacturing employment. We decompose ei in the above formula into employments of small firms, medium-sized firms, and large firms. SUPPLIER and CUSTOMER are decomposed in the same way. When an entrepreneur decides to establish a new firm, local market prospects and the performance of incumbent firms are expected to affect the entrepreneur’s choice of whether to enter the city and the sector. We further include the entry rate (ENTRY) and growth rate (GROWTH) at the 3-digit sector at the city level in the model. In order to control the effect of industrial policies on new firm formation, we introduce the subsidy rate (SUBSIDY) as one of the control variables (Table 5.2). To solve the endogeneity problem of the agglomeration and other variables, all independent variables were lagged by 2 years. Most independent variables were in logarithmic. The definition and measurement of independent variables are summarized in Table 5.2.

110

5.5

5 What Facilitates New Firm Formation in China?

Empirical Results

We employ the city-sector fixed effect negative binomial model to estimate the impact of agglomeration economies, the Vernon-Chinitz effect, and the size effect of start-ups on the number of private-owned start-ups during 2001–2007. According to correlation analysis, the independent variables are not highly correlated (Table 5.3). However, suppliers, customers, industrial diversity, related variety, and unrelated variety at different sizes are strongly correlated (Tables 5.4 and 5.5), so we separately test their impacts on new firm formation. The results demonstrate that all models are significant and have strong explanatory power.

5.5.1

Impact of Agglomeration Economies

Statistical results about the effect of localization economies and Jacobs externalities on new firm formation are presented in Table 5.6. The coefficients of LOC are positive and significant in all models, which is consistent with the results of most studies (Henderson 1997; Combes et al. 2004; Glaeser et al. 2010). The results indicate that the variation of new firm formation in China can be explained by localization economies, much like the USA (Glaeser et al. 2010), but the effect magnitude in China is much smaller than 0.913 and 0.966 in the USA. There is still upside potential for the role of localization economies into play in China. Further, we replace LOC with SUPPLIER and CUSTOMER to test the Hypothesis 5.2. The results show that both variables hold expected signs and highly significant. Supplier/customer linkages have a significantly positive impact on new firm formation, indicating that industrial linkages are very important source of localization economies. Access to suppliers and customers can help new firms reduce transportation cost and search cost. Moreover, the coefficients of SUPPLIER are significantly much larger than those of CUSTOMER. This is consistent with the finding of Glaeser and Kerr (2009) that suppliers seem to be more important to new firm formation than customers do. For start-ups, search cost and production cost are crucial especially in the early stage. Proximity to a large number of suppliers not only help start-ups find suitable inputs quickly but also lower the price of inputs because of fierce competition between local suppliers. There is no evidence to support Hypothesis 5.3. The effect of industrial diversity is ambiguous since the coefficients of Jacobs externalities in all years except 2005 are insignificant, and the signs are not robust. Despite this, we still cannot deny the important role of industrial diversity. Local knowledge spillovers are more likely to occur among a large number of technologically related industries, rather among a large diversity of unrelated industries. Therefore, we decompose industrial diversity into related variety and unrelated variety. As expectedly, related variety does have a significant and positive coefficient during 2001–2007, while the ambiguity of Jacobs externalities is mainly caused by unrelated variety. The finding provides empirical

LOC SUPPLIER CUSTOMER D R U ENTRY GROWTH SUBSIDY

LOC 1.000 0.205 0.119 0.331 0.339 0.314 0.185 0.013 0.022

1.000 0.354 0.529 0.346 0.502 0.045 0.007 0.002

SUPPLIER

1.000 0.321 0.351 0.298 0.031 0.007 0.000

CUSTOMER

Table 5.3 Correlation coefficients of independent variables

1.000 0.514 0.993 0.042 0.010 0.003

D

1.000 0.453 0.054 0.013 0.005

R

1.000 0.040 0.008 0.002

U

1.000 0.006 0.000

ENTRY

1.000 0.004

GROWTH

1.000

SUBSIDY

5.5 Empirical Results 111

LOC_S LOC_M LOC_L SUPPLIER_S SUPPLIER_M SUPPLIER_L CUSTOMER_S CUSTOMER_M CUSTOMER_L

LOC_S 1.000 0.479 0.349 0.188 0.150 0.035 0.113 0.104 0.032

1.000 0.457 0.155 0.159 0.060 0.065 0.088 0.029

LOC_M

1.000 0.051 0.068 0.082 0.013 0.004 0.017

LOC_L

1.000 0.654 0.082 0.418 0.311 0.076

SUPPLIER_S

Table 5.4 Correlation coefficients of independent variables

1.000 0.202 0.285 0.338 0.096

SUPPLIER_M

1.000 0.024 0.030 0.050

SUPPLIER_L

1.000 0.637 0.144

CUSTOMER_S

1.000 0.310

CUSTOMER_M

1.000

CUSTOMER_L

112 5 What Facilitates New Firm Formation in China?

5.5 Empirical Results

113

Table 5.5 Correlation coefficients of independent variables D_S D_M D_L R_S R_M R_L U_S U_M U_L

D_S 1.000 0.931 0.793 0.595 0.529 0.446 0.989 0.926 0.789

D_M

D_L

R_S

R_M

R_L

U_S

U_M

U_L

1.000 0.906 0.537 0.538 0.477 0.919 0.993 0.899

1.000 0.444 0.472 0.484 0.780 0.899 0.991

1.000 0.802 0.709 0.521 0.482 0.398

1.000 0.801 0.471 0.474 0.421

1.000 0.393 0.421 0.419

1.000 0.923 0.783

1.000 0.902

1.000

support for Hypothesis 5.4. Entrepreneurs in China can benefit from knowledge spillovers between technologically related industries. However, unrelated variety may lead to fierce competition for local input factors such as electricity and infrastructure facilities. Similar findings are reported in the studies of Frenken et al. (2007) and Boschma and Frenken (2011). The negative impact of unrelated variety may mitigate the positive externality of related variety. The statistical results on localization economies, business linkages, and related variety clearly suggest that localized business network is crucial to facilitate new firm formation in China. Agglomeration externalities have been found in liberalized and globalized industries and regions to promote innovation, growth, and productivity (Pan and Zhang 2014; Xu 2009; He and Pan 2010; Ke 2010; Zhang et al. 2014). Our results provide additional evidence to support the effectiveness of agglomeration externalities in transitional China.

5.5.2

The Vernon-Chinitz Effect

To explore the Vernon-Chinitz effect on new firm formation, localization economies (LOC), supplier/customer linkages (SUPPLIER and CUSTOMER), Jacobs externalities (D), related variety (R), and unrelated variety (U ) are decomposed into the corresponding variables for small firms, medium-sized firms, and large firms. Because of serious collinearity issues (Tables 5.4 and 5.5), we introduce them in the models separately. Statistical results indicate that compared with the clustering of large firms, the clustering of small- and medium-sized firms are more important to new firm formation (Table 5.7). It is consistent with the view of Vernon-Chinitz effect. Because the coefficient of LOC_S and LOC_M is not much larger than that of LOC_L, we confirm the significant difference by performing the T test. That is, the city with more small- and medium-sized firms at the same sector provides more favorable environment for new firm formation because small firms are more likely to share independent suppliers, which are considered as one of major sources of the Vernon-

0.244 0.009 30.525* 1.825** 35,802 221 6024

0.048

2001 0.350***

0.108*** 0.127** 0.942*** 0.014 0.084 4.412*** 35,802 221 6024

0.497*** 0.214***

2001

0.574*** 0.001 6.931 0.982 39,528 244 8695

0.019

2003 0.266***

0.105*** 0.040 1.269*** 0.015 0.712 3.273*** 39,528 244 8695

0.447*** 0.158***

2003

Note: ***, **, and * indicate the significance level at 1%, 5%, and 10%, respectively

Variables LOC SUPPLIER CUSTOMER D R U ENTRY GROWTH SUBSIDY Constant Observations Number of city Log Lik 0.505*** 0.001 2.970 1.051* 40,986 253 11,655

0.153***

2005 0.275***

0.075*** 0.030 1.218*** 0.010 0.993 1.843*** 40,986 253 11,655

0.407*** 0.182***

2005

0.378*** 0.010 9.473 1.421*** 39,366 243 14,529

0.032

2007 0.272***

Table 5.6 The effect of agglomeration economies on new firm formation (hereafter, city-sector fixed effect negative binomial models)

0.090*** 0.086*** 0.988*** 0.053** 14.192*** 3.560*** 39,366 243 14,529

0.309*** 0.189***

2007

114 5 What Facilitates New Firm Formation in China?

39,366 243 12,890

(1) 0.118*** 0.176*** 0.104***

39,366 243 14,715

1.580***

(2)

39,366 243 14,719

0.302***

(3)

39,366 243 14,695

0.125***

(4)

39,366 243 14,718

1.244***

(5)

39,366 243 14,691

0.575***

(6)

0.149*** 39,366 243 14,705

(7)

Note: SUPPLIER and CUSTOMER at different sizes are put in the model separately because of collinearity, controlling for D, ENTRY, GROWTH, and SUBSIDY ***, **, and * indicate the significance level at 1%, 5%, and 10%, respectively

Variables LOC_S LOC_M LOC_L SUPPLIER_S SUPPLIER_M SUPPLIER_L CUSTOMER_S CUSTOMER_M CUSTOMER_L Observations Number of city Log Lik

Table 5.7 Decomposition of localization economies for the Vernon-Chinitz effect

5.5 Empirical Results 115

116

5 What Facilitates New Firm Formation in China?

Chinitz effect (Chinitz 1961). The further results show that the influence of small suppliers is indeed more important than that of medium size and large suppliers. The similar results are also found in developed economies (Glaeser and Kerr 2009). Moreover, access to small and independent customers is also critical to new firm formation because large customers who have the initiative in price negotiation will depress prices maliciously, thus narrowing profit margins of start-ups (Bain 1959; Scherer 1970). Even though large customers play a less role on new firm formation than small ones do, access to them still help entrepreneurs start new businesses. Decomposition of Jacobs externalities for the Vernon-Chinitz effect is shown in Table 5.8. Related variety and unrelated variety have opposite effects on new firm formation, which can explain the insignificance of Jacobs externalities (Table 5.8). There are some variations in the roles of different types of related variety. The role of small- and medium-sized firms within related variety is larger than that of large firms. It makes sense that large firms have higher motivation and capability to protect their knowledge and innovation from spilling over. Moreover, small-related firms are more likely to foster potential entrepreneurs who are familiar with the related tasks and can deal with all kinds of issues on new firm formation (Johnson and Cathcart 1979).

5.5.3

The Size Effect of Start-Ups

Jacobs (1969) argues that small firms benefit more from industrial agglomeration than large firms do because small firms are more dependent on external industrial environment. In order to examine the impact of agglomeration economies on startups of different sizes, we break down the dependent variable – the number of privateowned start-ups – into the number of small start-ups (fewer than 50 employees), medium-sized start-ups (51–200 employees), and large start-ups. Due to serious collinearity between independent variables, we introduce all variables in the models separately but combine the estimated results together in Table 5.9. Both small- and medium-sized start-ups can benefit from different types of agglomeration economies, but the magnitude of coefficients is significantly different according to the T tests. Smaller start-ups are more dependent on the clustering of small- and medium-sized firms. Small suppliers and small customers are the most important for small start-ups, which is consistent with the findings of Glaeser and kerr (2009) that small suppliers have the largest effect on smaller entrants. In terms of Jacobs externalities, smaller-related firms matter more to small start-ups than larger-related firms do. Therefore, cities with more small firms (whether at the same sector, suppliers, and customers or at the technologically related sectors) are more conducive to the birth of small firms, since small firms themselves lower the entry threshold through the development of independent suppliers, venture capitalists, and entrepreneurial culture. Medium-sized start-ups are more likely to locate with medium-sized firm clusters. Though large start-ups are positively attracted by the clusters of small- and medium-

39,366 243 13,453 0.000

(1) 0.011

39,366 243 13,453 0.000

0.000

(2)

39,366 243 13,452 0.000

0.043

(3)

39,366 243 13,422 0.000

0.067***

(4)

39,366 243 13,424 0.000

0.066***

(5)

39,366 243 13,436 0.000

0.041***

(6)

39,366 243 13,450 0.000

0.074***

(7)

39,366 243 13,448 0.000

0.097***

(8)

0.059* 39,366 243 13,451 0.000

(9)

Note: D, R, and U at different sizes are put in the model separately because of collinearity, controlling for variables such as LOC, ENTRY, GROWTH, and SUBSIDY ***, **, and * indicate the significance level at 1%, 5%, and 10%, respectively

Variables D_S D_M D_L R_S R_M R_L U_S U_M U_L Observations Number of city Log Lik Prob > chi2

Table 5.8 Decomposition of Jacobs externalities for the Vernon-Chinitz effect

5.5 Empirical Results 117

118

5 What Facilitates New Firm Formation in China?

Table 5.9 The VernonChinitz effect for start-ups with different sizes in 2007

LOC_S LOC_M LOC_L SUPPLIER_S SUPPLIER_M SUPPLIER_L CUSTOMER_S CUSTOMER_M CUSTOMER_L R_S R_M R_L

Small 0.167*** 0.167*** 0.083*** 1.707*** 0.253** 0.131*** 1.594*** 0.651*** 0.178*** 0.096*** 0.074*** 0.044***

Medium 0.099*** 0.192*** 0.122*** 1.468*** 0.339*** 0.132*** 0.634 0.541*** 0.152*** 0.055*** 0.071*** 0.048***

Large 0.034** 0.179*** 0.191*** 1.606 0.820*** 0.135*** 0.112 0.416** 0.079 0.004 0.005 0.003

Note: ***, **, and * indicate the significance level at 1%, 5%, and 10%, respectively

sized firms, they are most inclined to cluster with large firms. Compared with smalland medium-sized start-ups, large start-ups may not care about where small suppliers, customers, and related firms are located. That is, their location choice is more independent since most of them can source internally and thus are less dependent on external environment.

5.5.4

Robustness Check

To check the robustness of the above results, we replace the number of privateowned start-ups with the employment of private owned start-ups as dependent variables. Due to an excess of zero values, we estimate panel Tobit models and carry out the similar procedure mentioned above. The results hold for all models, so we will not show the estimated results here to save space.

5.6

Conclusion and Implications

Agglomeration is not as simple as it seems. Its internal structure determines its effectiveness. Localization and urbanization economies are traditional and rough classification of agglomeration economies, which is insufficient to explain the complex economic structure. This study introduces the supplier/customer linkages, related variety, and the Vernon-Chinitz effect in order to shed light on the internal structure of agglomeration economies. The supplier/customer linkages refer to

5.6 Conclusion and Implications

119

production networks within clusters, and related variety stresses knowledge spillovers. The Vernon-Chinitz effect emphasizes essentially the sharing of production and information network among small firms. The business network and sharing are the essence of agglomeration economies. This chapter answers the question why some regions are more entrepreneurial than others in China by stressing the importance of agglomeration economies. As we expect, localization economies predict new firm formation well, but the effect of Jacobs externalities is mixed. In terms of localization economies, supplier/customer linkages play a very important and positive role in shaping new firm formation. The mixed results of Jacobs externalities are mainly derived from the interweaving of related variety and unrelated variety. Related variety is conductive to new firm formation significantly, while unrelated variety in most cases discourages new firm formation. Traditional understanding of agglomeration economies highlights the role of size and density, but our findings reveal the critical role of technological relatedness and internal structure of agglomeration in facilitating new firm formation. The clustering of small- and medium-sized firms has a larger effect on new firm formation than that of large firms do. These results are very much in the spirit of Vernon (1960) and Chinitz (1961). The estimated results on the clustering of small suppliers and small-related firms provide more support for the Vernon-Chinitz effect. In addition, the above effects vary across start-ups of different sizes. This study is among the first to explore the impact of the internal structure of agglomeration economies on new firm formation in China. Due to data limitation, however, our sample does not cover non-state-owned enterprises with sale revenues below 5 million Yuan. There may be estimated bias. The average size of manufacturing firms is generally larger because of increasing returns to scale, so the estimated bias may be very small even if there were. Even though economic reform unleashes the vitality of the market in China, decentralization endows local governments with more power to intervene in local development. To improve local achievements as soon as possible or compete with neighboring areas, local governments prefer to attract foreign direct investment or big domestic enterprises via various preferential policies, rather than small firms. Once the foreign firms or large enterprises as the supports of the urban economy fail or relocate, the city would be faced with the risk of economic recession. Besides, local governments are engaged in establishing industrial parks, in which industries encouraged to move may be not technically related with each other or not linked with each other in terms of production. Our results suggest that clustering of small firms, related firms, and supplier/customer linkage can help construct favorable environment for new firm formation in China. Localized business networks are key for new firm formation. Therefore, local governments and financial institutions should pay more attention to the development of small firm clusters and provide them with more favorable policies and financial supports.

120

5 What Facilitates New Firm Formation in China?

References Acs, Z. J., & Armington, C. (2006). Entrepreneurship, geography, and American economic growth. Cambridge, MA: Cambridge University Press. Armington, C., & Acs, Z. J. (2002). The determinants of regional variation in new firm formation. Regional Studies, 36(1), 33–45. Bain, J. S. (1959). Industrial organization. New York: Wiley. Bates, T. (1990). Entrepreneur human capital inputs and small business longevity. The Review of Economics and Statistics, 72(4), 551–559. Boschma, R. (2005). Proximity and innovation: Acritical assessment. Regional studies, 39(1), 61–74. Boschma, R., & Frenken, K. (2006). Why is economic geography not an evolutionary science? Towards an evolutionary economic geography. Journal of Economic Geography, 6(3), 273–302. Boschma, R., & Frenken, K. (2011). Technological relatedness, related variety and economic geography. In P. Cooke (Ed.), Handbook of regional innovation and growth (pp. 187–197). Cheltenham: Edward Elgar. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., & Weterings, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industry and Corporate Change, 16(2), 213–238. Chen, J. (2012). The small firm effect and the quality of entrepreneurs. In The Academy of Management Annual Meeting. Chinitz, B. (1961). Contrasts in agglomeration: New York and Pittsburgh. American Economic Review: Papers and Proceedings, 51(2), 279–289. Combes, P. P., Magnac, T., & Robin, J. M. (2004). The dynamics of local employment in France. Journal of Urban Economics, 56(2), 217–243. Cooper, A. C., & Folta, T. B. (2000). Entrepreneurship and high technology clusters. In D. L. Sexton & H. Landstrom (Eds.), Handbook of entrepreneurship (pp. 348–367). Oxford: Blackwell. Delfmann, H., Koster, S., McCan, P., et al. (2014). Population change and new firm formation in urban and rural regions. Regional Studies, 48(6), 1034–1050. Delgado, M., Porter, M. E., & Stern, S. (2010). Clusters and entrepreneurship. Journal of Economic Geography, 10(10), 495–518. Djankov, S., La Porta, R., Lopezdesilanes, F., & Shleifer, A. (2002). The regulation of entry. Quarterly Journal of Economics, 117(1), 1–37. Dobrev, S. D., & Barnett, W. P. (2005). Organizational roles and transition to entrepreneurship. Academy of Management Journal, 48(3), 433–449. Dumais, G., Ellison, G., & Glaeser, E. L. (2002). Geographic concentration as a dynamic process. Review of Economics and Statistics, 84(2), 193–204. Duranton, G., & Puga, D. (2001). Nursery cities: Urban diversity, process innovation, and the life cycle of products. American Economic Review, 91(5), 1454–1477. Elert, N. (2014). What determines entry? Evidence from Sweden. The Annals of Regional Science, 53(1), 55–92. Essletzbichler, J. (2015). Relatedness, industrial branching and technological cohesion in US metropolitan areas. Regional Studies, 49(5), 752–766. Farjoun, M. (1994). Beyond industry boundaries: Human expertise, diversification and resourcerelated industry groups. Organization Science, 5(2), 185–199. Feldman, M. P. (2001). The entrepreneurial event revisited: An examination of new firm formation in the regional context. Industrial and Corporate Change, 10(4), 861–891. Feldman, M., Francis, J., & Bercovitz, J. (2005). Creating a cluster while building a firm: Entrepreneurs and the formation of industrial clusters. Regional Studies, 39(1), 129–141.

References

121

Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Ghani, E., Kerr, W. R., & O'connell, S. (2014). Spatial determinants of entrepreneurship in India. Regional Studies, 48(6), 1071–1089. Glaeser, E. L., & Kerr, W. R. (2009). Local industrial conditions and entrepreneurship: How much of the spatial distribution can we explain? Journal of Economics and Management Strategy, 18 (3), 623–663. Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100(6), 1126–1152. Glaeser, E. L., Kerr, W. R., & Ponzetto, G. A. M. (2010). Clusters of entrepreneurship. Journal of Urban Economics, 67(1), 150–168. Gompers, P., Lerner, J., & Scharfstein, D. (2005). Entrepreneurial spawning: Public corporations and the genesis of new ventures, 1986 to 1999. The Journal of Finance, 60(2), 577–614. He, C., & Pan, F. (2010). Economic transition, dynamic externalities and city-industry growth in China. Urban Studies, 47(1), 121–144. He, C., Wei, Y. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. Henderson, V. (1997). Externalities and industrial development. Journal of Urban Economics, 42 (3), 449–470. Henderson, J. V. (2003). Marshall’s scale economies. Journal of Urban Economics, 53(1), 1–28. Jacobs, J. (1969). The economy of cities. New York: Vintage Books. Johnson, P. S., & Cathcart, D. G. (1979). The founders of new manufacturing firms: A note on the size of incubator plants. Journal of Industrial Economics, 28(2), 219–224. Ke, S. (2010). Agglomeration productivity and spatial spillovers across Chinese cities. The Annals of Regional Science, 45(1), 157–179. Klapper, L. F., Amit, R., & Guillén, M. F. (2010). Entrepreneurship and firm formation across countries. In International differences in entrepreneurship (pp. 129–158). Chicago: University of Chicago Press. Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99 (3), 483–499. Los, B. (2000). The empirical performance of a new inter-industry technology spillover measure. In Technology and knowledge (pp. 118–151). Malmberg, A., & Maskell, P. (2002). The elusive concept of localization economies: Towards a knowledge-based theory of spatial clustering. Environment and Planning A, 34(3), 429–449. Marshall, A. (1920 [1890]). Principles of economics (8th edn). London: Macmillan. Michelacci, C., & Silva, O. (2007). Why so many local entrepreneurs? Review of Economics and Statistics, 89(4), 615–633. Moomaw, R. L. (1988). Agglomeration economies: Localization or urbanization? Urban Studies, 25(2), 150–161. National Bureau of Statistics. (2006). 2002 input–output table of China. Beijing: China Statistics Press. Neffke, F. M. H., Henning, M., & Boschma, R. (2012). The impact of aging and technological relatedness on agglomeration externalities: A survival analysis. Journal of Economic Geography, 12(2), 485–517. Nelson, R. R., & Winter, S. G. (1982). The Schumpeterian tradeoff revisited. The American Economic Review, 72(1), 114–132. Pan, Z., & Zhang, F. (2014). Urban productivity in China. Urban Studies, 39(12), 2267–2281. Parker, S. C. (2009). Why do small firms produce the entrepreneurs? The Journal of SocioEconomics, 38(3), 484–494. Piore, M. J., & Sabel, C. F. (1984). The second industrial divide: Possibilities for prosperity. New York: Basic Books. Porter, M. E. (1990). The competitive advantage of nations. Harvard Business Review, 68(2), 73–93.

122

5 What Facilitates New Firm Formation in China?

Porter, M. E. (1998a). Clusters and competition: New agendas for companies, governments, and institutions. Boston: Harvard Business School Press. Porter, M. E. (1998b). Competing across locations: Enhancing competitive advantage through a global strategy. Boston: Harvard Business School Press. Qian, H., & Haynes, K. E. (2014). Beyond innovation: The small business innovation research program as entrepreneurship policy. The Journal of Technology Transfer, 39(4), 524–543. Reynolds, P., Storey, D. J., & Westhead, P. (1994). Cross-national comparisons of the variation in new firm formation rates. Regional Studies, 28(4), 443–456. Romero-Martínez, A. M., & Montoro-Sánchez, A. (2008). How clusters can encourage entrepreneurship and venture creation, reasons and advantages. International Entrepreneurship and Management Journal, 4(3), 315–329. Rosenthal, S. S., & Strange, W. C. (2003). Geography, industrial organization, and agglomeration. Review of Economics and Statistics, 85(2), 377–393. Rosenthal, S. S., & Strange, W. C. (2004). Evidence on the nature and sources of agglomeration economies. Handbook of Regional and Urban Economics, 4(4), 2119–2171. Rosenthal, S. S., & Strange, W. C. (2010). Small establishments/big effects: Agglomeration, industrial organization and entrepreneurship. In Agglomeration economics. Chicago: University of Chicago Press. Santarelli, E., & Sterlacchini, A. (1994). New firm formation in Italian industry: 1985–89. Small Business Economics, 6(2), 95–106. Saxenian, A. (1994). Regional advantage: Culture and competition in Silicon Valley and Route 128. Cambridge, MA: Harvard University Press. Scherer, F. M. (1970). Industrial market structure and economic performance. Chicago: Rand McNally. Stam, E. (2007). Why butterflies don’t leave, locational behavior of entrepreneurial firms. Economic Geography, 83(1), 27–50. Stam, E. (2010). Entrepreneurship, evolution and geography. In R. Boschma & R. Martin (Eds.), The handbook of evolutionary economic geography (pp. 307–348). Cheltenham: Edward Elgar. Stel, A. V., & Suddle, K. (2008). The impact of new firm formation on regional development in the Netherlands. Small Business Economics, 30(1), 31–47. Stinchcombe, A. L. (1965). Organizations and social structure. Handbook of Organizations, 44(2), 142–193. Stuart, C. (1979). Search and the spatial organization of trading. In Studies in the economics of search (pp. 17–34). Stuart, T. E., & Ding, W. W. (2006). When do scientists become entrepreneurs? The social structural antecedents of commercial activity in the academic life sciences. American Journal of Sociology, 112(1), 97–144. Timmermans, B., & Boschma, R. (2011). Labour mobility and plant performance in Denmark, the significance of related inflows (Working Paper). Utrecht: Utrecht University. Vernon, R. (1960). Metropolis 1985. Cambridge, MA: Harvard University Press. Viladecans-Marsal, E. (2004). Agglomeration economies and industrial location: City-level evidence. Journal of Economic Geography, 4(5), 565–582. Xu, Z. (2009). Productivity and agglomeration economies in Chinese cities. Comparative Economic Studies, 51(3), 284–301. Zhang, P., He, C., & Sun, Y. (2014). Agglomeration economies and firm R&D efforts: An analysis of China’s electronics and telecommunications industries. The Annals of Regional Science, 53 (3), 671–701.

Chapter 6

Does Creative Destruction Work for Chinese Regions?

6.1

Introduction

Industrial renewal and restructuring take place through a dual process where on the one hand long periods of incremental innovation and technological changes gradually transform industrial structure, and on the other hand, radical technological paradigm shifts reshape economic landscape in a more fundamental way. Such a dual process has resulted in the rise and fall of regional economies and the restructuring of industrial areas, and we now have several examples of where such process has forced the restructuring of regional economies in Europe, North America, and Japan (Schamp 2005; Hassink 2007), the Asian newly industrialized economies since the mid-1990s (Grunsven and Smakman 2005), and emerging economies such as mainland China since the 2000s (Wei et al. 2009). Disturbances and shocks that emerge at the global, national, regional, and local level in the form of institutional transformation, policy changes, fluctuation of currency exchange rates, and technological shifts have a remarkable impact over industrial development and may result in decline, atrophy, or even shutdown of an entire industry in certain geographical areas (Martin and Sunley 2015). Resilience of regional economies in the face of disturbances and shocks has been valued as a key feature a region should possess in order to achieve sustainable development (Martin and Sunley 2015). One type of regional resilience is related to creative destruction (Schumpeter 1939, 1942), referring not only to the capability of local entrepreneurs to develop new product or processes that can replace the traditional ones and render the latter obsolete in the short term (Schumpeter 1942) but also to the capability of a certain geographical region to generate and attract new entrants to offset the destruction

Modified article originally published in [Zhou, Y., He, C., and Zhu, S. (2017), Does Creative Destruction Work for Chinese Regions? Growth and Change, 48: 274–296.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_6

123

124

6 Does Creative Destruction Work for Chinese Regions?

caused by firm exit and industrial decline. This process of firm exit and entry is dynamic and complex, underlining the broader industrial evolution and economic development. Many studies on creative destruction in developed economies tend to focus on the ways in which firm entry and exit as well as industrial restructuring have unfolded in capitalist market economy and on the key role of market economy in facilitating the process of creative destruction (Dunne et al. 1989; Hahn 2000; Audretsch and Fritsch 2002; Aghion et al. 2011; Knott and Posen 2005; Pe’er and Vertinsky 2008; Brixy 2014), whereas little attention has been directed toward the potential effect that nonmarket factors (e.g., institutional contexts) could have over the process of industrial evolution and even less attention to questions like what kind of geographical regions is able to attract more new entrants and therefore is more likely to maintain its competitiveness. China’s manufacturing industry provides a rich context. On the one hand, since the initiation of China’s reform and opening-up policies, China has undergone dramatic economic growth and has experienced three fundamental transformations: (1) from a planned to an increasingly market-based economy; (2) from a state-owned, collective economy to one with growing level of private ownership; and (3) from a partially closed economy to one oriented toward export markets (Wei 2001; He et al. 2008). On the other hand, decentralization in China has created opportunities for local authorities to take different routes, resulting in a geographically uneven economic and institutional landscape. Even though Brandt et al. (2012) work has pointed out the existence of creative destruction in China’s so-called socialist market economy, few has been done to understand the determinants of firm exit and entry and their interaction, the role of regional institution in the process of creative destruction, and the subsequent industrial restructuring in such a transitional economy. This research seeks to fill this gap. We contribute to the understanding of the role of firm exit in regional industrial renewal process and to the debate over whether firm exit is good (Knott and Posen 2005), by suggesting that firm exit may attract new entrants that are likely to become more productive through a redeployment of resources released by firm exit. This research echoes with Pe’er and Vertinsky (2008) findings based on firm exit and entry in Canada but pays more attention to the ways in which the abovementioned process is constantly shaped by an assemblage of various factors, including firm characteristics, industrial linkages, regional institutions, and geographical proximity. Our findings also highlight the bright side of firm exit and particularly its positive role in promoting regional productivity and therefore show that certain industrial policies that are designed to prevent inefficient firms from dying could be counterproductive in practice. The next section provides a literature review and proposes an analytical framework to understand firm exit and entry in transitional economy like China. In Sect. 6.3, we introduce data sources and describe the patterns of firm dynamics in China. After interpreting the model and variables in Sects. 6.4 and 6.5 reports and analyzes the statistical results. The last section concludes the chapter by summarizing the main findings and discussing policy implications.

6.2 Articulation Between Firm Entry and Exit

6.2

125

Articulation Between Firm Entry and Exit

Entrepreneurial activity is increasingly regarded as playing a primary role in shaping the regional economy and industrial renewal (Audretsch and Thurik 2001; Dejardin 2009; Fritsch and Mueller 2004), a process which is characterized by creative destruction that introduces superior products and technologies and renders old ones obsolete (Schumpeter 1942). While Schumpeter has focused on radical technological shifts and on how new entrants bring in new products and more advanced technologies, making existing technology regime and products of incumbents obsolete and forcing them to exit or catch up, Pe’er and Vertinsky (2008) have examined industrial renewal and restructuring through a different perspective that emphasizes a local creative destruction process through which technologies change incrementally. Unlike Schumpeter who examines the impact of the introduction of radical innovation and entry of new firms over exit of incumbents, Pe’er and Vertinsky (2008) show that firm exit actually creates a stimulus for entry of new firms, resulting in incremental innovation and productivity increase. In other words, both studies agree that firm exit and entry interact with one another but disagree on the direction of causality (Fig. 6.1). Exit of incumbent firms contributes to entry of new firms in many ways (Bates 2005; Deng 2009; Pe’er and Vertinsky 2008; Pe’er et al. 2008). First, the potential entrants can learn from the exit of incumbents in terms of why they fail to maintain their competitiveness (Cannon and Edmondson 2005; Amankwah–Amoah 2011; Madsen and Desai 2010). Second, new entrants are also able to recruit skilled and experienced employees released in the process of firm exit (Delacroix and Carroll 1983; Knott and Posen 2005). Third, tangible and intangible assets released in firm Schumpeterian creative destruction (Radical technological shifts): new entrants bring in radical innovation and new products, making incumbents’ products and technologies obsolete and force them to exit (or catch up).

Firm Entry

Firm Exit - Firm-specific factors - Industry-specific factors

Incremental creative destruction (Incremental technological shifts): exits of firms release resources and therefore contribute to entry of new firms.

Fig. 6.1 Two perspectives on the articulation of firm entry and exit

- Region-specific factors

126

6 Does Creative Destruction Work for Chinese Regions?

exit can be absorbed by new entrants particularly from the same and related industries, lowering the entry barrier and costs for new entrants (Clark and Wrigley 1997; Kessides 1991). Finally, firm exit also releases market for new entrants (Clark and Wrigley 1997; Kessides 1991; Knott and Posen 2005). This research builds on these studies but moves beyond and pays more attention to the ways in which the articulation between firm exit and firm entry has been constantly shaped by an assemblage of various factors—firm-specific, industry-specific, and region-specific (Fig. 6.1).

6.2.1

Firm-Specific Factors

Firm exit or failure is often related to its age and size (Dunne et al. 1989). Resources released by exiting firms are often determined by firm characteristics. For instance, the exit of young firms endowed with very limited amount of tangible and intangible assets releases less resources for firm entry than that of old, mature ones (Bates 2005; Harris and Hassaszadeh 2002). Exit of large firms that own a large number of workers and plenty of tangible and intangible assets may also release more resources and create more opportunities for entry (Haltiwanger et al. 2012). Small firms tend to employ more temporary, part-time and precarious workers, pay lower wages, and conduct more labor-intensive production (Brock and Evans 1989; Mayo and Murray 1991). In addition, they rely heavily on bank loans (Beck and Demirguc-Kunt 2006), which could dry up more quickly than large firms particularly during economic downturns (Fazzari, et al. 1988). Nevertheless, employees working in small firms may have a better chance to familiarize with a broader spectrum of operations (Storey 1982), and therefore small firms can also be seen as incubators, the exit of which may release some well-rounded workers (Cross 1981).

6.2.2

Industry-Specific Factors

Marshallian and Jacobian externalities are defined as the benefits arising from knowledge spillover and technology transfer within the same industry and across a variety of different industries, respectively. However, it is increasingly acknowledged that knowledge spillover does not exist across any industries; terms like cognitive proximity and technological relatedness have been used to explore the effectiveness of knowledge spillover and industrial evolution (Balland et al. 2014; Frenken et al. 2007; Hassink 2005; Neffke et al. 2011). There is a growing body of literature that suggests knowledge spillover occurs through sharing overlapping knowledge bases between related sectors rather than within a broad range of random sectors (Asheim et al. 2011; Frenken et al. 2007). Industries that were technologically related to the preexisting productive structure in a region have a higher probability of entering that region than do industries that are technologically unrelated to the region’s preexisting industries (Neffke et al. 2011). Technological

6.2 Articulation Between Firm Entry and Exit

127

relatedness plays a crucial role in industrial clustering (Boschma and Wenting 2007; Delgado et al. 2010), employment growth (Bishop and Gripaios 2010; Hartog et al. 2012), spinoff dynamics (Heebles and Boschma 2011), productivity growth (Quatraro 2010), and regional growth (Boschma and Iammarino 2009; Boschma et al. 2012; Frenken et al. 2007). In our case, the interaction between firm exit and firm entry is therefore expected to be shaped by industry-specific factors and technological relatedness. Our hypothesis is firm exit in one industry may be more attractive for entrants from the same or related industries that are more capable to utilize the released resources.

6.2.3

Region-Specific Factors

Research based on case studies in developed capitalist economies tends to emphasize the role of market economy that allows resources to flow relatively freely (Essletzbichler and Rigby 2005) but pay less attention to the impact of nonmarket factors (e.g., institution) over regional industrial evolution and to regional variation in terms of institutional and economic contexts. Regions can differ from each other drastically in terms of institutional contexts, particularly in developing and emerging economies like China where decentralization has empowered local authorities to participate directly in the development process as planners, developers, and policymakers, resulting in a geographically uneven institutional and economic landscape (Zhao and Zhang 1999). This process of decentralization may affect firm exit and entry in many ways. On the one hand, less developed regions in China are more likely to protect local economy, resulting in a high entry barrier. On the other hand, developed and rich regions particularly in China’s coastal areas are more active and capable in upgrading their industrial structure. Consequently, high level of firm exit in one industry in developed regions may indicate local authorities’ decision to abandon this industry as a whole and upgrade to more advanced industries. In this sense, firm exit does not necessarily lead to firm entry in the same industry. Marketization allows better labor mobility, and smoother knowledge spillover through industrial linkages, and boost entrepreneurship (He and Pan 2010), enabling new entrants to easily absorb resources released by firm exit. However, in a transitional economy, due to decentralization, marketization has been implemented to different extents in different regions, generating different institutional contexts that facilitate or hinder the process of creative destruction to different extents and shape the articulation between firm exit and entry in different ways (Han and Pannell 1999). Specifically, in more liberalized regions, market economy not only facilitates the mobility of released resources but also allows participants to quickly capture useful market information, such as information on firm exit and local authorities’ attitude toward specific industry. Decentralization and regional variation in terms of institutional contexts also result in high geographical and political barriers between different regions (Fan 2005; Wei 2001; Zhao and Zhang 1999). Even though the mobility of factors and

128

6 Does Creative Destruction Work for Chinese Regions?

resources has increased as marketization and globalization proceeded in China (Fan 2005; Fang and Dewen 2003; Luke 2005), it is still difficult for some resources (e.g., skilled workers and machinery) to become mobile enough to easily cross jurisdiction barriers derived from decentralization. The effect of firm exit over firm entry may thus be geographically bounded and exhibit a distance decay pattern.

6.3

Temporal and Spatial Changes of China’s Industrial Dynamics

In this chapter, we use the ASIF dataset. If firm i is reported in the ASIF in year t but not in year t
1, this is considered as a firm entry in year t. Likewise, if firm i is reported in the ASIF in year t
1 but not in year t, it is assumed that firm i exits in year t. Since ASIF dataset only includes non-state-owned enterprises with annual sales of five million RMB or more besides state-owned enterprises, firm exit is likely to be slightly overestimated due to the fact that non-state-owned enterprise that passes the threshold (annual sales of five million RMB or more) in year t but fails to do so in year t + 1 will be treated as an exiting firm. Nonetheless, this flaw only slightly affects research results; ASIF has been widely used to study firm exit and entry (Brandt et al. 2012). Given this issue, firm exit in our research is more like firm failure—a firm that is able to meet the threshold in year t fails to do so in year t + 1. Firm failure and firm exit have been also used interchangeably by Caves (1998) to describe firm discontinuance. Table 6.1 presents the temporal change of firm entry and exit in China. Firm entry rate is calculated as the ratio of the number of new firms to the number of all firms in a specific year, and firm exit rate is the share of the number of exiting firms. Year 2004 is China’s economic census year, and 2004 database includes many firms that have been omitted in non-census-years. Two 4-digit sectors are missing in 2008. As a result, statistics in 2004 and 2008 can be erratic and will be carefully treated. The number of firms increases from 134,063 in 1998 to 375,883 in 2008. The average entry and exit rate are 21.3% and 13.3%, respectively, throughout this time period. Firm exit rate decreases continuously from 14.98% in 1998 to 11.49% in 2007. Entry rate however fluctuates, ranging from 13.33% to 43.31% during this time period. We also compare exit rates of different types of firms (Table 6.1). Exit rate of young firms is lower than that of old firms, suggesting a negative relationship between firm age and firm exit. Young firms may be more energetic and prepared to adopt new technologies and products. Exit rate of small- and medium-sized firms is higher than that of large firms, which is consistent with other studies and indicates that small- and medium-sized firms are more dynamic (Faggio and Konings 2003). Large firms are more capable to tackle with challenges and therefore face lower turnover rate. Finally, firm ownership also has an impact over firm exit. While exit rate of foreign-owned firms is low and quite stable, that of state-owned and private firms is decreasing and increasing, respectively. Foreign-owned firms are often

Year All firms Firms entry Firm exit Entry rate (%) Failure rate (%) By age (exit rate, %) Above 4 years 1–3 years By firm size (exit rate, %) Small (200 employees) By ownership (exit rate, %) State owned Private Foreign owned

12.38 1.99

4.81 7.22 3.32

4.10 2.76 1.35

11.14 2.62

6.10 6.16 2.73

4.21 1.83 1.69

14.98

20,088

1999 131,507 17,532 20,198 13.33 15.36

1998 134,063

Table 6.1 Firm entry and exit in China (1998–2008)

4.80 3.65 1.37

4.63 8.10 5.14

15.06 1.72

2000 128,677 17,368 22,991 13.50 17.87

2.74 3.34 1.23

4.42 5.78 2.13

9.55 1.83

2001 141,899 36,213 17,498 25.52 12.33

2.51 3.89 1.13

4.48 5.66 2.02

9.35 1.90

2002 153,860 29,459 18,709 19.15 12.16

2.43 7.34 1.67

5.98 9.00 3.43

13.54 3.38

2003 170,305 35,154 31,335 20.64 18.40

2.10 6.93 1.77

6.80 6.49 1.87

9.37 4.17

2004 245,139 106,169 37,141 43.31 15.15

0.64 3.78 1.11

3.13 3.94 1.08

5.39 1.92

2005 240,877 32,879 19,625 13.65 8.15

0.92 3.40 1.00

3.14 3.51 0.94

5.56 1.27

2006 268,868 47,616 20,400 17.71 7.59

0.46 5.91 1.50

3.99 5.59 1.91

8.35 2.09

2007 303,876 55,408 34,915 18.23 11.49

28.45

2008 375,883 106,922

6.3 Temporal and Spatial Changes of China’s Industrial Dynamics 129

130

6 Does Creative Destruction Work for Chinese Regions?

subsidiaries of transnational corporations, and resources imported from parent firms may make these subsidiaries in China become more resilient. The decrease of exit rate of state-owned firm is due to the reform of state-owned firms in the 1990s in China (Walder et al. 2013). As marketization proceeded and entry barrier for private firms was lowered, private sector boomed, resulting in an increasing level of competition and exit rate in private sector. Figure 6.2 shows the geography of firm entry and exit in China in two periods: 1998–2003 and 2003–2008. During the first time period, firm entry rate is at a similar level as firm exit rate, except in a number of coastal provinces, such as Zhejiang, Fujian, and Shandong where entry rate is as high as 100% from 1998 to 2003. However, firm entry rate surpasses exit rate in the second time period, not only in the wealthy coastal regions (e.g., Liaoning, Jiangsu, Shandong) but also in central China (e.g., Anhui and Jiangxi) and West China (e.g., Sichuan and Chongqing). In some of the above notified provinces, entry rate reaches 200% during 2003–2008. To better understand the geographical and temporal change of China’s industrial dynamics, we follow Audretsch and Fritsch’s (2002) work and differentiate four types of growth regimes: entrepreneurial regime, revolving door regime, routinized regime, and downsizing regime. A region is defined as (1) an “entrepreneurial regime” region if in this region, firm entry rate exceeds the median value but the exit rate is below the median value; (2) a “revolving door regime” region if in this region, firm entry rate and exit rate both exceed the median value; (3) a “routinized regime” region if in this region, firm entry rate and exit rate are both below the median value; and (4) a “downsizing regime” region if in this region, firm entry rate is below the median value but exit rate exceeds the median value (Fig. 6.3). Figure 6.4 classifies China’s prefecture-level cities into four types of growth regimes in two time periods. There is a clear belt of entrepreneurial regime regions during 1998–2008, spanning from Shandong Peninsula to Yangtze River Delta. In addition, this belt becomes even wider and longer in 2003–2008 as more coastal areas become entrepreneurial regime regions. Entrepreneurial regime regions are also located in Pearl River Delta and some of China’s big inland cities along Yangtze River. Most of the revolving door regime regions are located in Northeastern, Inner Mongolia and part of Central and South China, where the reform of state-owned enterprises has fundamentally restructured local industrial structures in the 1990s and early 2000s. Routinized regime regions are concentrated in Central and Southwest China where industrial structures are relatively stable. Figure 6.4 also shows the industrial decline in West and Northwest China where the majority of downsized regime regions concentrates. We use Brandt et al. (2012) approach to calculate total factor productivity (TFP) of exiting and entering firms (Fig. 6.5a, b). First, exiting firms tend to be less productive than new entrants, suggesting that productivity is improved after new entrants redeploy resources released by exiting firms. This also resonates with conclusions made by Brandt et al. (2012) and Loecker and Konings (2004). Second, TFP of exiting and entering firms in East China is declining during 1999–2007, while that in Central and West China fluctuates in the same time period. Third, East China has the highest TFP in both exiting and entering firms.

6.3 Temporal and Spatial Changes of China’s Industrial Dynamics

Fig. 6.2 Firm exit and entry in China at provincial level in 1998–2003 and 2003–2008

131

132

6 Does Creative Destruction Work for Chinese Regions?

Fig. 6.3 Four types of growth regimes based on firm entry and exit rate

Fig. 6.4 Growth regimes in China at prefecture level in 1998–2003 and 2003–2008

6.4

Model Specification and Variables

The above statistical analysis indicates how firm entry and exit vary across space and time. This part will conduct a more systematic analysis to investigate the ways in which the articulation between firm entry and firm exit has been shaped by firmspecific, industry-specific, and region-specific factors in China’s manufacturing industries. The dependent variable (Entryi,j,t) is the number of employees of new entrants in a prefecture-level city i in year t and industrial sector j. A new entrant decides to locate in cities that guarantee the highest expected profits. The expected profits are not directly observable, but the new firms established in each city each year can be observed. In these circumstances, the data are censored, and the appropriate statistical model for estimating the firm entry in a city is the Tobit estimated by the maximum likelihood method (Tobin 1958). The Tobit model is defined as follows:
 Entryi, j, t ¼ β0 þ Exit i,t
1 þ f 1 FSi,t
1 ; ISi,j,t
1 ; RSi,t
1 
  þf 3 FSi,j,t
1 ; ISi,j,t
1  RSi,t
1 þ X i,t
1 þ w j þ vi þ ut þ εij

ð6:1Þ

where i, j, and t denote city, industry, and year, respectively. Exiti,t
1, measured as the number of employees of exiting firms in city i in year t
1 and in all industrial sectors, is included to test the impact of firm exit in all industrial sectors over firm entry in industry j.

6.4 Model Specification and Variables

133

Fig. 6.5 (a) TFP of exiting and entering firms in Central, East, and West China during 1999–2007 (left); (b) TFP of exiting and entering firms during 1999–2007 at provincial level (right)

134

6 Does Creative Destruction Work for Chinese Regions?

Firm-specific factors (FSi,t
1) capture the characteristics of exiting firms in all industrial sectors, such as firm age and size. Following (Pe’er and Vertinsky 2008), we consider an exiting firm as an old firm if its age is more than 4 years. Otherwise, it will be treated as a young exiting firm. Exiti,t
1 is therefore decomposed into Exit_oldi,t
1 and Exit_youngi,t
1, measured as the number of employees of old and young exiting firms in city i in year t
1 and in all industrial sectors, respectively, in order to investigate the different impacts of exit of old and young firms over firm entry. A considerable number of firm exits take place within the first 3 years after firm formation. Resources released by old and young exiting firms are often different from one another (Pe’er and Vertinsky 2008; Yang and He 2014). In addition, the exits of large-, medium-, and small-sized firms are likely to release different resources, tangible and intangible assets as well. We define Exit_largei,t
1, Exit_mediumi,t
1, and Exit_smalli,t
1 as the number of employees of large (more than 200 employees), medium (50–200 employees), and small (0–50 employees) exiting firms in city i in year t
1 and in all industrial sectors, respectively. ISi,j,t
1 refers to the industry-specific factors that affect firm entry. Tangible and intangible assets released by firm exit may be absorbed more easily by new entrants in the same and technologically related industries, than by those in unrelated industries. Accordingly, we define Exit_focali,j,t
1, Exit_focali,j,t
2, and Exit_focali,j,t
3 as the number of employees of exiting firms in city i in industrial sector j and in year t
1, t
2, and t
3, respectively, to explore the effects of firm exit over firm entry in the same industry. Furthermore, we adopt the co-occurrence approach developed by Hidalgo et al. (2007) to calculate technological relatedness between industries, measured as the minimum pairwise conditional probability of two industrial sectors located in the same region. Specifically, technological relatedness between industry j and k is calculated as: 

 Relatednessij ¼  min P RCAk jRCA j þ P RCA j jRCAk 1,LQ  0:5 Where, RCA ¼ 0,LQ < 0:5

ð6:2Þ

where LQ is the employment location quotient of industry j (or k) at the city level. We define revealed comparative advantage (RCA) as 0, if LQ is less than 0.5. Otherwise, RCA is 1. Based on this, firm exit in related industries (Exit_relatedi,j, t
1) is defined as the firm exit- weighted technological relatedness of industry j and k in city i and year t
1: Exit relatedi,j,t
1 ¼

X

Relatednessj, k  Exit i,k,t
1 . . .

ð6:3Þ

k

RSi,t
1 is the region-specific factors. This captures the impact of regional institutional contexts over the articulation between firm exit and entry and pays special attention to the role of marketization and fiscal decentralization in the context of China. State-owned enterprises (SOEs) are often considered as less productive and

6.5 Empirical Results

135

efficient than private enterprises in communist and postcommunist economies (Ahrend and Martins 2003), and a big proportion of SOEs reflects a low level of marketization. As a result, we use the proportion of non-SOEs’ output in the total in city i and year t
1 (Marketi,t
1) as a proxy of marketization. In addition, the ratio of fiscal revenue to fiscal expenditure in city i and year t
1 (Deci,t
1) captures the economic capability of local administration under China’s decentralization system. Xi,t
1 represents control variables. First, agglomeration externalities have been extensively seen as a key factor affecting firm entry (Fotopoulos 2014). Studies on localization economies argue that knowledge is predominantly industry specific, and therefore local specialization will foster economic growth and firm entry (Henderson 1994; Marshall 1920). In this chapter, localization economies (Locali,j,t
1) are measured as the density of total employments of industry j in city i and year t
1. The other hypothesis, proposed by Jacobs (1969), claims that regional diversity in economic activity will result in agglomeration externalities as knowledge developed by one industry can also be fruitfully applied in other industries, therefore increasing the attractiveness of region to new entrants. Likewise, urbanization economies (Urbani,t
1) are calculated as the density of total employment in city i and year t
1. Second, it is argued that diverse regional industrial structure is likely to attract more new entrants (Fotopoulos 2014; Gudgin 1978). This chapter uses Theil index (Theili,t
1) to measure regional industrial diversity.1 We also include the comparative advantage of industry j in city i and year t
1, measured by the location quotient (LQi,j,t
1). Finally, wj indicates the industry-specific effect, vi indicates the region-specific effect, and ut indicates the time-specific effect.

6.5 6.5.1

Empirical Results Empirical Results on Firm-Specific Factors

In the estimation equations, lagged terms of independent and control variables have been adopted, given the fact that it takes time for new entrants to be attracted by resources released by firm exit. The geographical unit of analysis is China’s prefecture-level city (excluding Taiwan, Hong Kong, and Macau). Correlation analysis indicates that correlations of most independent variables are moderate or low, suggesting no serious problem of multicollinearity. Variables of exits of different types of firms (in terms of age and size) are correlated with each other; we therefore put them into different models. Table 6.2 reports estimation results focusing on firm-specific factors. In Models (1)–(6), independent variables, including Exit, Exit_old, Exit_young, Exit_large, Exit_medium, and Exit_small, are calculated based on all 4-digit industrial sectors. For example, Exit_old is measured as

1

Please see Theil (1972) for the calculation of Theil index.

436.9*** 831.9***
0.683*** 0.350*** 0.015*** 0.283*** Included Included Included
744.2*** 1373.3*** 262,102 0.035
993,949 71703.0

*p < 0.05, **p < 0.01, ***p < 0.0

Exit Exit_old Exit_young Exit_large Exit_medium Exit_small Market Dec Urban Local LQ Theil Industry Province Year _cons Sigma _cons N Pseudo R2 Log lik. Chi-squared

426.4*** 766.4***
0.859*** 0.351*** 0.015*** 0.106 Included Included Included
822.7*** 1370.4*** 262,102 0.035
993,578 72445.4

All 4-digit sectors (1) (2) 0.005*** 0.009***

402.4*** 915.8***
0.389*** 0.349*** 0.015*** 1.178*** Included Included Included
510.6*** 1376.1*** 262,102 0.034
994,340 70920.4


0.005***

(3)

Table 6.2 Regression results on firm-specific factors (1999–2008)

428.2*** 772.7***
0.957*** 0.350*** 0.015*** 0.452*** Included Included Included
728.4*** 1369.8*** 262,102 0.035
993,491 72618.6

0.011***

(4)

406.4*** 921.4***
0.414*** 0.350*** 0.015*** 0.866*** Included Included Included
546.7*** 1376.3*** 262,102 0.034
994,363 70875.2

0.002***

(5)


0.018*** 396.4*** 929.6***
0.428*** 0.349*** 0.015*** 1.368*** Included Included Included
414.3*** 1375.9*** 262,102 0.034
994,297 71007.7

(6)

The focal 4-digit sector (7) (8) 0.002*** 0.002*** 0.625*** 0.517*** 0.507*** 1.373*** 0.428*** 435.3*** 418.5*** 788.6*** 800.2***
0.310***
0.159*** 0.260*** 0.233*** 0.012*** 0.0115*** 0.262*** 0.123* Included Included Included Included Included Included
733.1***
735.1*** 1286.2*** 1274.0*** 262,102 262,102 0.042 0.043
986,255
985142.4 87090.6 89315.9

136 6 Does Creative Destruction Work for Chinese Regions?

6.5 Empirical Results

137

the number of employees of old exiting firms in city i in year t
1 and in all industrial sectors. In Models (7) and (8), these independent variables are calculated based on the focal industry. For example, Exit_old is measured as the number of employees of old exiting firms in city i in year t
1 and in industry j. Model (1) shows that firm exit in all industrial sectors has a significant and positive impact on firm entry, indicating that firm exit creates opportunities for new entrants. Models (2), (3), and (7) suggest that exits of young and old firms in the focal industry and all industrial sectors have different effects over firm entry. Exit of old firms is likely to release more resources, for example, more skilled labor, and thereafter attract more new entrants (Models (2) and (7)). This is consistent with Pe’er and Vertinsky’s (2008) findings. Exit of young firms in the focal industry also has a positive and significant effect on firm entry, but its coefficient is smaller than that of old firms (See Model (7)), indicating a slightly weaker effect. However, exit of young firms in all industrial sectors exhibits a negative and significant effect over firm entry (Model (3)). This inconsistency can be reconciled by the following explanation. A high level of exit of young firms within all industrial sectors may indicate an unfriendly environment for new entrants and therefore give region a bad reputation that frightens new entrants away. Likewise, high level of exit of small firms within all industrial sectors may be due to the unsupportive local economic environment, which tends to frightens new entrants away (Model (6)). In contrast, exit of large, medium, and small firms in the focal industry all contribute to attracting new entrants. Exit of large firms releases plenty of tangible and intangible assets (Government of Canada Government of Canada 2013). However, employees working in small firms may have a better chance to familiarize with a broader spectrum of operations, and therefore small firms can also be seen as incubators, the exit of which may release some wellrounded workers (Loecker and Konings 2004). Model (8) shows that the coefficients of Exit_small and Exit_large are both smaller than that of Exit_medium, suggesting an inverted U-shaped relationship between firm size and the effect of firm exit over firm entry, probably due to the fact that medium firms are able to release both a considerable amount of resources and some well-rounded workers. In almost all models, control variables show a relationship with firm entry that is consistent with theoretical predictions. Regions with high localization economies are more attractive for new entrants. The sign of LQ’s coefficient suggests that new firms in an industrial sector are willing to enter regions which have comparative advantage in this specific industrial sector. The positive coefficient of Theil index indicates that cities with a diversity of industries are more capable to attract new entrants; this is consistent with Fotopoulos’s (2014) and Renski’s (2014) conclusion. However, urbanization economies have a negative and significant effect over firm entry probably because high level of urbanization economies leads to intensive local competition and high costs derived from congestion effects. The institutional context in China characterized by a dual process of marketization and decentralization also plays a critical role in regional industrial evolution. Statistical results show that high level of marketization not only generates better labor mobility and smoother knowledge spillover through industrial linkages but also boosts entrepreneurship, enabling

138

6 Does Creative Destruction Work for Chinese Regions?

new entrants to easily absorb resources released by firm exit. Finally, regions with better fiscal performance and higher level of fiscal revenue tend to have more new entrants, as such regions are capable to provide better infrastructure, more skilled and educated labor force, and more advanced supporting facilities and services.

6.5.2

Empirical Results on Industry- and Region-Specific Factors

To test our hypothesis that firm exit in one industry may be more attractive for entrants from the same or related industries as they are more capable to utilize the released resources, we develop a series of models (see Table 6.3). The estimated parameters of the control variables are mostly unaltered. In Models (1)–(3), we include 1-, 2-, and 3-year lagged terms of firm exit in the focal industry to investigate the time effect of creative destruction. All three have a significant and positive impact over firm entry, indicating that it takes time for resources released by firm exit to be fully absorbed by new entrants. In addition, the 1-year lagged term (Exit_focal) of course has the strongest effect. Resources released in one industry also attract new entrants in related industries (see Model (4)), however, to a lesser extent, as the parameter of Exit_related is much smaller than that of Exit_focal. This echoes with recent evolutionary economic geography literature on the role of technological relatedness in regional diversification (Boschma et al. 2013; Frenken et al. 2007; Neffke et al. 2011; Neffke and Henning 2008). In Models (5) and (6), we add four interaction terms between Exit_focal, Exit_related, Market, and Dec. First, the interactions terms, Exit_focal*Dec and Exit_related*Dec, both present significant and negative signs, while Dec exhibits positive and significant signs. This inconsistency could be reconciled by the following explanation. On the one hand, regions with better fiscal performance are capable to provide better infrastructure, more skilled and educated labor force, and more advanced supporting facilities and services, and therefore can attract more new entrants. On the other hand, in a transitional economy, developed and rich regions particularly in China’s coastal areas are more active and capable in upgrading their industrial structure. As a result, in regions with better fiscal performance and high level of firm exit in the focal industry or related industries may indicate local authorities’ decision to abandon this industry or this type of industries as a whole and upgrade to more advanced industries. For example, high level of firm exit in pollution-intensive industries in China’s developed coastal regions reflects an exodus of these industries and the intention of local governments in coastal regions to upgrade to cleaner and greener industries. Accordingly, local governments become less supportive for pollution-intensive industries; financial aid is likely to be directed more toward green industries. In this case, firm exit does not necessarily lead to firm entry, even though a large amount of resources may be released by firm exit.

419.0*** 836.2***
0.147*** 0.253*** 0.012*** 0.574*** Included Included Included
625.6*** 1283.7*** 262,102 0.042
986,102 87396.4

(1) 0.607***

*p < 0.05, **p < 0.01, and***p < 0.001

Exit_focal Exit_focalt
2 Exit_focalt
3 Exit_related Exit_focal*market Exit_focal*Dec Exit_related*market Exit_related*Dec Market Dec Urban Local LQ Theil Industry Province Year _cons Sigma _cons N Pseudo R2 Log lik Chi-squared 386.7*** 828.8***
0.259*** 0.297*** 0.013*** 0.798*** Included Included Included
554.2*** 1352.1*** 211,881 0.038
834,935 65743.9

0.446***

(2)

368.4*** 825.9***
0.248*** 0.292*** 0.015*** 0.858*** Included Included Included
557.4*** 1406.7*** 167,110 0.036
689,886 51962.5

0.476***

(3)

Table 6.3 Regression results on industry-specific and region-specific factors (1999–2008)

403.4*** 836.4***
0.599*** 0.349*** 0.015*** 0.645*** Included Included Included
695.5*** 1374.4*** 262,102 0.035
994,037 71526.7

0.013***

(4)

382.1*** 840.1***
0.120*** 0.239*** 0.012*** 0.507*** Included Included Included
598.0*** 1281.1*** 262,102 0.043
985,900 87799.8

0.320***
0.030**

(5) 0.440***


0.014***
0.013*** 464.8*** 855.2***
0.548*** 0.349*** 0.015*** 0.723*** Included Included Included
833.3*** 1374.4*** 262,102 0.035
993,986 71628.2

0.035***

(6)

6.5 Empirical Results 139

140

6 Does Creative Destruction Work for Chinese Regions?

Second, likewise, even though Market presents a positive and significant sign in Model (5) and (6), the interaction term between Exit_related and Market exhibits a negative and significant sign. High level of marketization not only allows better labor mobility and more efficient knowledge spillover, but also boosts entrepreneurship and therefore attracts more new entrants. However, regions with high level of marketization and high level of firm exit in one industry or related industries may suggest this industry or this type of industries are not able to survive in highly competitive markets due to factors (e.g., high land price and labor costs) that have industry-wide effects and therefore affect firms’ location decision. For example, high level of firm exit in labor-intensive industries in coastal regions reflects firms in these industries are not able to maintain their competitiveness and survive in highly competitive coastal regions where labor costs, land price and costs of other factor inputs have all increased drastically lately. Labor-intensive firms’ profit margins have been squeezed to an extent that they have to either upgrade or relocate their labor-intensive production to low-cost locations. Consequently, in this sense, resources released by firm exit do not necessarily attract new entrants.

6.5.3

Geographical Proximity in Creative Destruction

Finally, to test if the effect of firm exit over firm entry is geographically bounded, we include firm exit in the focal industry but in other cities within the same province, Exit_focal_province (Table 6.4). Model (1) shows that even though new entrants in one city benefit from resources released by exiting firms in the same industry in other cities within the same province, this effect is much weaker than that of firm exit in the same city over firm entry. In a word, it is not easy for resources to cross political borders, and therefore the effect of firm exit over firm entry exhibits a distance decay pattern. Furthermore, we add interaction terms between Exit_focal_province with RMarket and RDec, where RMarket (RDec) is the ratio of the average level of marketization (decentralization) in other cities within the same province to the level of marketization (decentralization) in the focal city, to measure the relative degree of marketization (decentralization) in other cities within the same province compared with the focal city. Exit_focal_province*RDec exhibits a positive and significant effect over firm entry in the focal city. High level of firm exit in other cities with relatively better fiscal performance may indicate that in order to upgrade their industrial structure, cities in the same province are deliberately ruling out certain industries, which are therefore forced to move to the focal cities. Likewise, Exit_focal_province*RMarket has a positive and significant effect over firm entry in the focal city. High level of firm exit in other cities within the same province where market is quite competitive may compel firms in certain industries that are unable to survive the market competition in other cities to relocate to the relatively less competitive focal city.

6.6 Conclusion

141

Table 6.4 Regression results on geography proximity (1999–2008) Exit_focal Exit_focal_province Exit_focal_ province*RMarket Exit_focal_ province *RDec RMarket RDec Market Dec Urban Local LQ Theil Industry Province Year _cons Sigma _cons N Pseudo R2 Log lik. Chi-squared

(1) 0.596*** 0.032***

426.4*** 851.1***
0.843* 0.243*** 0.0117*** 0.588*** Included Included Included
824.0*** 1178.9*** 262,025 0.036
1,033,752 76,358.2

(2)

(3)

0.0447***

214.1*** 586.8***


0.068*** 0.045*** 0.077*** 162.4*** 517.2***


0.287*** 0.333*** 0.0150*** 1.182*** Included Included Included
751.5*** 1246.1*** 262,025 0.029
1,040,310 63,241.4


0.287*** 0.318*** 0.0151*** 1.142*** Included Included Included
691.3*** 1241.9*** 262,025 0.030
1,039,953 63,957.3

*p < 0.05, **p < 0.01, ***p < 0.001

We also exclude 2004 and 2008 dataset and run the same models that we have in Tables 6.2, 6.3, and 6.4, in order to test the robustness of our estimation results. The reason that 2004 and 2008 dataset is excluded has been clarified in Sect. 6.3. Compared with the results presented above, these changes produce only minor effects.

6.6

Conclusion

Following Pe’er and Vertinsky’s (2008) work, we focus on industrial renewal and particularly on the capability of a certain geographical region to generate and attract new entrants to offset the destruction caused by firm exit. Our analytical framework emphasizes the ways in which firm exit creates a stimulus for firm entry, a process that is complementary to the process of technological change and industrial renewal articulated by Schumpeter who pays more attention towards how new entrants bring in radical innovation and new products, making incumbents’ products and technologies obsolete and force them to exit or catch up. Based on this analytical

142

6 Does Creative Destruction Work for Chinese Regions?

framework, this chapter seeks to argue that the articulation between firm exit and entry has been constantly shaped by an assemblage of various factors, including firm characteristics, industrial linkages, regional institutions, and geographical proximity. Using firm-level data of China’s industries during 1998–2008, we show that firm exit does stimulate firm entry, as new entrants are enticed by resources released by firm exit. First, characteristics of exiting firm have different impacts over firm entry. Specifically, exits of young and old firms in the focal industry both encourage firm entry, though the latter one has a larger effect since exit of old firms is likely to release more resources, for example, more skilled labor, and thereafter attract more new entrants. Likewise, the exits of large, medium, and small firms all attract new entrants, but the exit of medium firms has the strongest effect probably due to the fact that medium firms are able to release both a considerable amount of resources and some well-rounded workers. However, exit of young and small firms in all industries may indicate an unfriendly environment for new entrants and therefore give region a bad reputation that frightens new entrants away. Firm exit in one industry is more attractive for entrants in the same or related industries as they are more capable to utilize the released resources. The articulation between firm exit and entry is also shaped by region-specific factors, since in developing and transitional economies like China, decentralization has empowered local authorities to participate directly in the development process as planners, developers, and policy-makers, resulting in a geographically uneven institutional and economic landscape. In China, both marketization and decentralization are complex processes, which affect the articulation of firm entry and exit in different ways, suggesting that any analyses of industrial restructuring and renewal should be situated into a context where market economy has been implemented to different extents and local authorities have different capabilities in steering their economies. Finally, the effect of firm exit over firm entry is also geographically bounded and exhibits a distance decay pattern. Several policy implications can be derived from the empirical findings. First, localization economies are more important than urbanization economies. Industrial policies should aim to foster spatial sectoral specialization in certain industries, rather than investing in a broad range of industrial sectors. Second, policy-makers that seek to attract new entrants should pay more attention to new entrants from industrial sectors that their cities already have or at least are related with their existing industrial structure. Third, policy-makers must craft their policies with sensitive attention to the characteristics and processes of small geographical areas as the effect of creative destruction decays as geographical distance increases. Finally, policies that strive to maintain dying firms through subsidies or tax credits should be carefully evaluated, as firm exit stimulates firm entry. This process of creative destruction often results in productivity increase after new entrants redeploy resources released by exiting firms (see also Fig. 6.5a). As a result, local authorities should coordinate the exit of underperforming firms through regulatory and tax policies and facilitate the release of the valuable resources. Additional policies on educational and training programs would also enhance the recycling of released resources.

References

143

References Aghion, P., Blundell, R., Griffith, R., Howitt, P., & Prantl, S. (2011). Entry and productivity growth: Evidence from microlevel panel data. Journal of the European Economic Association, 2(2–3), 265–276. Ahrend, R., & Martins, J. O. (2003). Creative destruction or destructive perpetuation: The role of large state-owned enterprises and SMEs in Romania during transition. Post-Communist Economies, 15(3), 331–356. Amankwah–Amoah, J. (2011). Learning from the failures of others. Journal of Workplace Learning, 23(6), 358–375. Asheim, B. T., Roschma, R., & Cooke, P. (2011). Constructing regional advantage: Platform policies based on related variety and differentiated knowledge bases. Regional Studies, 45(7), 893–904. Audretsch, D. B., & Fritsch, M. (2002). Growth regimes over time and space. Regional Studies, 36 (2), 113–124. Audretsch, D. B., & Thurik, A. R. (2001). What’s new about the new economy? Sources of growth in the managed and entrepreneurial economies. Industrial and Corporate Change, 10(1), 267–315. Balland, P. A., Boschma, R. A., & Frenken, K. (2014). Proximity and innovation: From statics to dynamics. Regional Studies, 0(0), 1–14. Bates, T. (2005). Analysis of young, small firms that have closed: Delineating successful from unsuccessful closures. Journal of Business Venturing, 20(3), 343–358. Beck, T., & Demirguc-Kunt, A. (2006). Small and medium-size enterprises: Access to finance as a growth constraint. Journal of Banking & Finance, 30(11), 2931–2943. Bishop, P., & Gripaios, P. (2010). Spatial externalities, relatedness and sector employment growth in Great Britain. Regional Studies, 44(4), 443–454. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new industries at the regional level in Spain: A proximity approach based on product relatedness. Economic Geography, 89 (1), 29–51. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Brixy, U. (2014). The significance of entry and exit for regional productivity growth. Regional Studies, 48(6), 1051–1070. Brock, W. A., & Evans, D. S. (1989). Small business economics. Small Business Economics, 1(1), 7–20. Cannon, M. D., & Edmondson, A. C. (2005). Failing to learn and learning to fail (intelligently): How great organizations put failure to work to innovate and improve. Long Range Planning, 38 (3), 299–319. Caves, R. E. (1998). Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature, 36(4), 1947–1982. Clark, G. L., & Wrigley, N. (1997). Exit, the firm and sunk costs: Reconceptualizing the corporate geography of disinvestment and plant closure. Progress in Human Geography, 21(3), 338–358. Cross, M. (1981). New firm formation and regional development. Farnborough: Gower Publishing Company. Dejardin, M. (2009). Linking net entry to regional economic growth. Small Business Economics, 36 (4), 443–460.

144

6 Does Creative Destruction Work for Chinese Regions?

Delacroix, J., & Carroll, G. R. (1983). Organizational foundings: An ecological study of the newspaper industries of Argentina and Ireland. Administrative Science Quarterly, 28(2), 274–291. Delgado, M., Porter, M. E., & Stern, S. (2010). Clusters and entrepreneurship. Journal of Economic Geography, 10(10), 495–518. Deng, P. (2009). Why do Chinese firms tend to acquire strategic assets in international expansion? Journal of World Business, 44(1), 74–84. Dunne, T., Roberts, M. J., & Samuelson, L. (1989). The growth and failure of U. S. manufacturing plants. The Quarterly Journal of Economics, 104(4), 671–698. Essletzbichler, J., & Rigby, D. L. (2005). Technological evolution as creative destruction of process heterogeneity: Evidence from US plant-level data. Economic Systems Research, 17(1), 25–45. Faggio, G., & Konings, J. (2003). Job creation, job destruction and employment growth in transition countries in the 90s. Economic Systems, 27(2), 129–154. Fan, C. C. (2005). Interprovincial migration, population redistribution, and regional development in China: 1990 and 2000 census comparisons. The Professional Geographer, 57(2), 295–311. Fang, C., & Dewen, W. (2003). Migration as marketization: What can we learn from China’s 2000 census data? The China Review, 3(2), 73–93. Fazzari, S. M., Hubbard, R. G., Petersen, B. C., Blinder, A. S., & Poterba, J. M. (1988). Financing constraints and corporate investment. Brookings Papers on Economic Activity, 1988(1), 141–206. Fotopoulos, G. (2014). On the spatial stickiness of UK new firm formation rates. Journal of Economic Geography, 14(3), 651–679. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Fritsch, M., & Mueller, P. (2004). Effects of new business formation on regional development over time. Regional Studies, 38(8), 961–975. Government of Canada, Statistics Canada. (2013). The daily—Study: Canadian labour productivity differences across firm size classes, 2002 to 2008. Canadian Productivity Review Grunsven, L. V., & Smakman, F. (2005). Industrial restructuring and early industry pathways in the Asian 1st generation NICs: The Singapore Garment Industry (Papers in Evolutionary Economic Geography (PEEG) 0507). Utrecht University, Section of Economic Geography. Gudgin, G. (1978). Industrial location processes and regional employment growth. Farnborough: Saxon House. Hahn, C. H. (2000). Entry, exit, and aggregate productivity growth. In OECD economics department working papers. Paris: Organisation for Economic Co-operation and Development. Haltiwanger, J., Jarmin, R. S., & Miranda, J. (2012). Who creates jobs? Small versus large versus young. Review of Economics and Statistics, 95(2), 347–361. Han, S. S., & Pannell, C. W. (1999). The geography of privatization in China, 1978–1996. Economic Geography, 75(3), 272–296. Harris, R. I. D., & Hassaszadeh, P. (2002). The impact of ownership changes and age effects on plant exits in UK manufacturing, 1974–1995. Economics Letters, 75(3), 309–317. Hartog, M., Boschma, R., & Sotaruta, M. (2012). The impact of related variety on regional employment growth in Finland 1993–2006: High-tech versus medium/low-tech. Industry and Innovation, 19(6), 459–476. Hassink, R. (2005). How to unlock regional economies from path dependency? From learning region to learning cluster. European Planning Studies, 13(4), 521–535. Hassink, R. (2007). The strength of weak lock-ins: The renewal of the Westmünsterland textile industry. Environment and Planning A, 39(5), 1147–1165. He, C., & Pan, F. (2010). Economic transition, dynamic externalities and city-industry growth in China. Urban Studies, 47(1), 121–144. He, C., Wei, Y. H. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945.

References

145

Heebles, B., & Boschma, R. (2011). Performing in Dutch book publishing 1880–2008: The importance of entrepreneurial experience and the Amsterdam cluster. Journal of Economic Geography, 11(6), 1007–1029. Henderson, I. V. (1994). Where does an industry locate? Journal of Urban Economics, 35(1), 83–104. Hidalgo, C. A., Klinger, B., Barabási, A.-L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Jacobs, J. (1969). The economy of cities. New York: Vintage Books. Kessides, I. N. (1991). Entry and market contestability: The evidence from the United States. In P. Geroski & J. Schwalbach (Eds.), Entry and market contestability: An international comparison. Oxford: Blackwell. Knott, A. M., & Posen, H. E. (2005). Is failure good? Strategic Management Journal, 26(7), 617–641. Loecker, D. J., & Konings, J. (2004). Creative destruction and productivity growth in an emerging economy: Evidence from Slovenian manufacturing (SSRN Scholarly Paper ID 510106). Rochester/New York: Social Science Research Network. Luke, C. (2005). Capital and knowledge flows: Global higher education markets. Asia Pacific Journal of Education, 25(2), 159–174. Madsen, P. M., & Desai, V. (2010). Failing to learn? The effects of failure and success on organizational learning in the global orbital launch vehicle industry. Academy of Management Journal, 53(3), 451–476. Marshall, A. (1920 [1890]). Principles of economics (8th ed.). London: Macmillan. Martin, R., & Sunley, P. (2015). On the notion of regional economic resilience: Conceptualization and explanation. Journal of Economic Geography, 15(1), 1–42. Mayo, J. W., & Murray, M. N. (1991). Firm size, employment risk and wages: Further insights on a persistent puzzle. Applied Economics, 23(8), 1351–1360. Neffke, F., & Henning, M. S. (2008). Revealed relatedness: mapping industry space (Papers in Evolutionary Economic Geography (PEEG) 0819). Utrecht University, Section of Economic Geography. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Pe’er, A., & Vertinsky, I. (2008). Firm exits as a determinant of new entry: Is there evidence of local creative destruction? Journal of Business Venturing, 23(3), 280–306. Pe’er, A., Vertinsky, I., & King, A. (2008). Who enters, where and why? The influence of capabilities and initial resource endowments on the location choices of de novo enterprises. Strategic Organization, 6(2), 119–149. Quatraro, F. (2010). Knowledge coherence, variety and economic growth: Manufacturing evidence from Italian regions. Research Policy, 39(10), 1289–1302. Renski, H. (2014). The influence of industry mix on regional new firm formation in the United States. Regional Studies, 48(8), 1353–1370. Schamp, E. W. (2005). Decline of the district, renewal of firms: An evolutionary approach to footwear production in the Pirmasens area, Germany. Environment and Planning A, 37(4), 617–634. Schumpeter, J. A. (1939). Business cycles: A theoretical, historical, and statistical analysis of the capitalist process. New York: McGraw-Hill. Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York: Harper & Row. Storey, D. (1982). Entrepreneurship and the new firm. Taylor & Francis. Tobin, J. (1958). Estimation of relationships for limited dependent variables. Econometrica, 26(1), 24–36. Walder, A. G., Luo, T., & Wang, D. (2013). Social stratification in transitional economies: Property rights and the structure of markets. Theory and Society, 42(6), 561–588.

146

6 Does Creative Destruction Work for Chinese Regions?

Wei, Y. H. D. (2001). Decentralization, marketization and globalization: The triple processes underlying regional development in China. Asian Geographer, 20(1–2), 7–23. Wei, Y. H. D., Lu, Y., & Chen, W. (2009). Globalizing regional development in Sunan, China: Does Suzhou Industrial Park fit a neo-Marshallian district model? Regional Studies, 43(3), 409–427. Yang, R., & He, C. (2014). The productivity puzzle of Chinese exporters: Perspectives of local protection and spillover effects. Papers in Regional Science, 93(2), 367–384. Zhao, X., & Zhang, L. (1999). Decentralization reforms and regionalism in China: A review. International Regional Science Review, 22(3), 251–281.

Chapter 7

What Causes Firm Failure in China?

7.1

Introduction

An economy keeps its vitality through a continual process of firm entry, survival, and exit. Those are fundamental components of the structural changes taking place in industries. It is critical to investigate forces underlying firm dynamics to understand spatial economic changes (Nyström 2007; Buenstorf and Geissler 2011). Reflecting the importance of firm dynamics, there has been a large body of literature examining the determinants using firm-level data from developed economies (Siegfried and Evans 1994; Audretsch et al. 2000; Kangasharju 2000; Berglund and Brännäs 2001; Fotopoulos and Spence 2001; Acs et al. 2007; Nyström 2007; Carree et al. 2008; Cheng and Li 2011). Faced with extensive institutional changes, vibrant business climate, and governmental interventions, firms in developing and transitional economies are more difficult to survive (Tybout 2000; Bojnec and Xavier 2004; Männasoo 2008; Estrin and Prevezer 2010; Giarratana and Torrisi 2010; Močnik 2010). A systematic investigation of firm dynamics in those economies would significantly enrich the understanding of industrial development. During the past three decades, China has undergone economic transition, which has been conceptualized as a triple process of marketization, globalization, and decentralization (Wei 2001; He et al. 2008). The transition process has brought about the liberalization of prices, markets, and trade and the privatization of the stateowned sector, triggering market competition. Meanwhile, decentralization grants local governments authority to intervene in economic development, activating interregional competition (He 2006). Moreover, globalization brings international competition into local development. Chinese firms face substantial challenges from market competition and institutional uncertainties, resulting in remarkable business

Modified article originally published in [He, C. and Yang, R. (2016), Determinants of Firm Failure: Empirical Evidence from China. Growth and Change, 47: 72–92.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_7

147

148

7 What Causes Firm Failure in China?

failure. Surprisingly, firm failure has been extremely underexplored in China. Published in Chinese, Mao and Sheng (2013) is an exception. They find that both entry and exit rates of Chinese firms are fairly high; firm size matters for both entry and exit. However, business failure in China is driven by both state power and market forces and deserves further investigation. The objective of China’s gradual economic reform is to introduce market competition, which actually occurs at local, regional, and global levels. Firms which are now responsible for their own operations face fierce market competition. Gradual economic transition creates the opportunities for firms to learn. Inefficient firms are therefore more likely to fail due to competition effects while experienced old firms would be more likely to survive because of learning effects. The goal of economic reform is to develop the Chinese economy. Governments, from the central to the local level, now have strong incentives to provide support to firms, mitigating the competition effects and helping inefficient firms to survive. Based on the ASIF dataset, this chapter will investigate the pattern of firm failure and identify the role of market forces and governmental support. In particular, this study will first examine the relationship between firm total factor productivity (TFP) and firm failure. This study will then focus on the relationship between firm age and firm failure and, finally, test the direct and moderating role of local support. On average, failing firms are less productive than survivors in China. Older firms are more likely to fail, while firms with local support have more chance to survive. Statistical results indicate that competition effects do crowd out less productive firms. There is however an inverted U-shaped relationship between firm age and firm failure. In other words, when older firms overcome the market challenges, they can survive for a longer time. Local protectionism and supportive policies can certainly enhance the survival chance of some firms. Particularly, they can help older firms to mitigate the impact of competition effects. However, governmental intervention would generate negative externality, reducing the survival chance of firms without support. The findings indicate that governmental supports help some firms at the social costs of other firms. The chapter is structured as follows. The second section presents the literature review and develops the research hypotheses, followed by the introduction of data sources. The fourth section describes the patterns of firm failure. This chapter then discusses the results of linear probability models and concludes with a summary of empirical findings.

7.2

Literature Review and Research Hypothesis

There is a large body of literature exploring the determinants of firm survival and failure using data from developed economies. Much of the work has broadly argued that productive and efficient firms will survive and inefficient firms will fail (Siegfried and Evans 1994; Agarwal 1996; Cefis and Marsili 2006; Shiferaw 2009).

7.2 Literature Review and Research Hypothesis

149

Industrial organization literature examines firm-specific and industry-specific factors. At the firm level, studies report that larger firms are less likely to fail, consistent with the learning models of industry dynamics (Geroski 1995; Ericson and Pakes 1995; Mata and Portugal 1994; Dunne and Hughes 1994; Audretsch and Mahmood 1995; Wagner 1999). The effect of size is however not uniform and may be nonlinear (Mata and Portugal 1994; Esteve-Pérez and Llopis 2004). Those studies also report a negative relation between firm age and failure. As pointed out in Thompson (2005), theories that could explain the observed pattern include learning, financial frictions, and the fact that older firms are possibly active in a larger number of submarkets. A few studies find an inverted U-shaped relationship between firm age and firm exit (Agarwal et al. 2002). Both size and age effects differ across industries and firms (Hannan et al. 1998; López-García et al. 2007). Ownership is a major determinant of firm survival in the literature (Mata and Portugal, 1994). For instance, Agarwal and Gort (2002) and Agarwal et al. (2002) show that diversifying firms have lower failure rates than de novo entrants. Görg and Strobl (2003) conclude that foreign firms are more likely to exit than indigenous plants, while Mata and Portugal (2002) and Kimura and Fujii (2003) do not report a significant impact of foreign ownership. Another major concern is the impact of innovative activities on firm dynamics, confirming that firms with substantial innovation inputs and outputs are less likely to fail (Hall 1987; Audretsch 1995; Kimura and Fujii 2003; Fontana and Nesta 2009; Cefis and Marsili 2005; Wagner and Cockburn 2010). Industrial characteristics such as market size, growth rate, technology, market structure, entry rates and scale economies, and the life cycle consistently explain differences in survival rates across firms after controlling for firm-specific factors (Mata and Portugal 1994; Agarwal 1998; Agarwal and Audretsch 2001). Firms in high-technology industries are more likely to fail (Audretsch 1995; Agarwal and Audretsch 2001). High entry rates exert a positive effect on the likelihood of firm failure (Mata and Portugal 1994, 2002). Firms live longer in growing industries than in declining industries even controlling for industry turbulence, size, scale, type of entrant, and concentration (Mata and Portugal 1994; López-García et al. 2007). A few studies examine the impacts of local factors on firm dynamics. Stearns et al. (1995) find that new US firms in urban locations have significantly worse survival chances than those in rural locations. Fotopoulos and Louri (2000) show that firms located in greater urban areas have lower hazard rates than those located outside those areas. Industrial localization is also a key local variable to explain firm survival. Higher agglomeration is associated with higher firm mortality rate (Honjo 2000; Staber 2001; Folta et al. 2006; De Silva and McComb 2012). For instance, De Silva and McComb (2012) find that greater firm density within 1 mile of firms in the same industry increases mortality rates, while greater concentration over large distances reduces business mortality rates in Texas. In addition, there are reports about the impacts of local supportive programs. Almus (2016) studies business survival 5 years after creation in Eastern Germany and finds that firms’ failures are negatively related to the receipt of public subsidies. Crépon and Duguet (2003) explore the impacts of capital subsidies and banking loans on French firms created in

150

7 What Causes Firm Failure in China?

1994 during the subsequent 3 years and report a positive effect of this subsidy on survival of new firms created by both short-term and long-term unemployed people. Bank loans alone have no positive effect, but they reinforce the effect of public subsidies on firm survival. With the transition process, Chinese firms face substantial challenges from market competition and institutional uncertainties, causing remarkable business failure. Surprisingly, firm dynamics have been extremely underexplored. Mao and Sheng (2013) examine firm entry and exit in China and find that both entry and exit rates of Chinese firms are fairly high. They report that size matters for both firm entry and exit; less productive firms are more likely to exit. Business failure in China is a complicated phenomenon and is driven by multiple forces. We highlight the roles of competition effects, learning effects, and governmental support in firm dynamics. China has taken a gradualism approach to reform its economic system (Zhang 2000). The goal is to build a market-oriented economy by introducing market competition and opening to the international market. This competition not only occurs locally but also regionally and internationally and has generated significant impacts on industrial development in China. For instance, Girma and Gong (2008) find that market competition from foreign firms in the same sector and foreign firms in downstream sectors have a deleterious impact on the survival probability of SOEs in China. Facing intensive market competition, only productive firms can survive. This is the competition effect, which would force inefficient firms out of the market. Consequently, less productive firms are more likely to fail as reported in the studies based in data from developed economies (Siegfried and Evans 1994). During economic reform, China has gradually introduced market forces and marketoriented institutions (Zhang 2000). Both local governments and firms take time to learn how to experiment the market system and tackle the market forces. Firms gradually learn how to compete with their domestic and international competitors and how to interact with governments. As firms get older, they may accumulate rich experiences and develop their own core competence (Thompson 2005). Experiences and competence may help Chinese firms to survive in the versatile environment. This learning process may take a long time. Thereby, learning effects would reduce the failure of firms in markets for a longer time. Meanwhile, economic reform has significantly inspired Chinese entrepreneurship to create a large number of new firms annually, leading to growing competition (Huang 2008). Cheap labor, large market, and favorable FDI policy regime further attract many foreign firms to the Chinese market (Huang 2003). Meanwhile, firms created in the earlier times often suffer from institutional difficulties such as vague ownership and updated equipment (Huang 2008). Chinese firms are often accused of not investing in innovation activities and not developing core competences. Increasing competition and institutional challenges may dominate the learning effects, resulting in higher mortality rate of relatively older firms in China. We expect a nonlinear relationship between firm age and failure. With economic transition, decentralization has granted local governments greater authority over their economies. Local governments have a primary responsibility for economic development in their respective jurisdictions. Fiscal decentralization in particular has enhanced the importance of local budget constraints. More authorities

7.3 Data Source and Firm TFP Estimation

151

together with hardening budgets have led to the fierce interregional competition for regional development and also provided local governments with strong incentives to protect businesses by shielding local firms from interregional competition (Young 2000). Under the name of the assistance to local economies, local governments use their heightened administrative power in terms of trade and investment to implement multiform protection of firms under their authority (Zhao and Zhang 1999). It is a common practice for local governments in China to provide subsidies to firms, including income tax breaks for companies with foreign investment, located in special development zones or designated as having high technology; loans to encouraged industries from government-owned banks; rebates of value-added tax and import duties for equipment purchases; low-priced land for SOEs and companies located in special development zones and the provision of goods and services at below market prices by the government and SOEs; and cash payments to companies based on factors such as export performance (Barbieri et al. 2012). Local protectionism would shield firms from competition, downplaying competition effects and increasing the surviving chances of inefficient firms. Local supportive policies may allow the less productive firms in the markets at least in the short run. Based on the above analysis, we make the following two research hypotheses to guide our empirical analysis. Hypothesis 1 Less productive firms are more likely to fail due to competition effects, while governmental intervention will mitigate the competition effects, reducing the chance of firm failure. Hypothesis 2 There is an inverted U-shaped relationship between firm age and firm failure; local protectionism and support policies would reduce the likelihood that older firms fail.

7.3

Data Source and Firm TFP Estimation

Data used in this is the ASIF dataset. We construct the panel data as the following three steps. First, we match firms for two consecutive years using the legal person codes. The left firms will be matched using names of firms. If necessary, we will further use legal person code, county code or county code, telephone number, and starting year to match the left firms. We generate an unbalanced panel for two consecutive years by combining the matched and unmatched firms. Second, there may be some firms with missing information for several years. To match those firms in different years, we first apply for the similar methods to identify the corresponding firms. Consider the following situation: Firm A in the first year has no match with any firm in the second year but a good match for firm C in the third year. Firm C however can be matched with firm B in the second year. Consequently, firms A, B, and C in the three consecutive years can be treated as the same firm. By doing so, we construct the panel data for three consecutive years. Finally, based on the panel data for three consecutive years, we could expand the panel data to cover firms during 1998–2007.

152

7 What Causes Firm Failure in China?

There are 2,224,380 observations and 576,143 industrial enterprises in total during 1998–2007. Among the 576,143 enterprises, 26,135 enterprises report information throughout the 10 years, accounting for 5.14%. 130,921 enterprises are only reported for 1 year, accounting for 25.75%. The average survival year is 3.86 years. Recently, the ASIF dataset is widely applied in economic studies, generating some high-quality publications, including some published in American Economic Review (Song et al. 2011), Quarterly Journal of Economics (Hsieh and Klenow 2009), and Papers in Regional Science (Yang and He 2014), significantly improving economic research in China. To investigate industrial dynamics, we first define firm entry and firm failure. Assuming firmit is reported in the survey in year t, if there is no reported information for firmit in year t
1, then firmit is a new entry. If there is no reported information for firmit in year t + 1, then firmit fails. Failure rate is the ratio of number of failing firms to the total number of existing firms, while entry rate is the ratio of number of new entries to the total number of existing firms. Since the annual survey of industrial firms only includes those with sales revenues greater than five million Yuan, failing firms only mean that they fail to meet the standard in the dataset. Given this issue, we use the term of firm failure rather than firm exit. Table 7.1 reports the temporal patterns of firm entry and failure. Chinese industries are fairly dynamic, with annual firm entry rate of more than 20%. The year 2004 is the economic census year and may include more newly created firms, with a firm entry rate of 45.13%. Failure rate however fluctuates significantly, ranging from 11.82% to 24.88%. The aggregate analysis conceals the variations of firm dynamics, and it is critical to explore the industrial dynamics at the microlevel. In the statistical analysis, we will check the impacts of year 2004 and the five million Yuan as the entry standard on the model estimations. This study will link firm TFP to firm failure. The key task is to estimate firm TFP. TFP is the difference of output growth rate and the weighted average of growth rate of input factors. It assumes the contribution from technological progress or institutional changes. TFP is often estimated using the C-D production function with constant return to scale. This estimation however suffers from endogeneity issues derived from simultaneity and selection bias. Olley and Pakes (1996) develop a three-step regression model (the OP model) to solve the endogeneity issues. This study follows the OP model to estimate the TFP of Chinese firms.

7.4 7.4.1

Pattern of Firm Failure in China Firm TFP and Firm Failure

Based on the estimated firm TFP, we calculate the annual average firm TFP. The often-used method is to weigh firm TFP to compute the average TFP (Brandt et al. 2012; Hsieh and Klenow 2009). The weight can be the share of output or employment. This study applies the traditional method to compute the annual average TFP. This estimation does not weigh industries, which may differ in technology. This

Year Number of firms Firm entries Entry rate (%) Firm failure Failure rate (%)

25,193 19.83

1998 127,030

1999 132,439 30,602 23.11 29,867 22.55

2000 129,924 27,352 21.05 32,331 24.88

Table 7.1 Firm entry and failure in China during 1998–2007 2001 143,136 45,543 31.82 23,459 16.39

2002 151,526 31,849 21.02 24,771 16.35

2003 166,081 39,326 23.68 42,127 25.37

2004 225,906 101,952 45.13 42,224 18.69

2005 226,256 42,574 18.82 27,784 12.28

2006 253,685 55,213 21.76 29,993 11.82

2007 283,409 59,717 21.07

7.4 Pattern of Firm Failure in China 153

7 What Causes Firm Failure in China?

2.4 2

2.2

Average TFP

2.4 2.2

1.8

1.8

2

Weighted TFP

2.6

2.6

2.8

154

1998

2000

2002 Year failing firms

2004

2006

survival firms

1998

2000

2002 Year failing firms

2004

2006

survival firms

Fig. 7.1 Failing firms, surviving firms, and firm TFP

actually assumes that the relative importance of industries remains the same during 1998–2007, which is not true in reality. The estimation would underestimate firm TFP if input factors are reallocated from less productive industries to more productive ones. Comparing with weights by output and value added, the employment weight is better to reflect labor mobility. We examine the trend of employment weighted TFP for failing and surviving firms (Fig. 7.1). We also report the annual average firm TFP. Clearly, failing firms are significantly less productive than surviving firms. In 1998, the average TFP for failing firms and surviving firms are 1.84 and 2.67, respectively. In 2006, it turns to be 2.35 and 2.66, respectively. In other words, less productive firms are more likely to fail. Productivity seems a critical factor for firms to survive in the Chinese market. Interestingly, the productivity gap between failing and surviving firms has been narrowed down. As China entered the WTO, Chinese market is widely open to foreign investors, bringing international competition into domestic market. Meanwhile, the post-WTO period has also inspired the Chinese entrepreneurship to create many more firms, heightening market competition (Huang 2008). The increasing competition could force some more productive firms.

7.4.2

Firm Age and Firm Failure

We compute firm age based on the difference between the reporting year and the starting year. Figure 7.2 reports the average age of Chinese firms. During 1998–2006, the average age dropped from 14.3 years to 8.8 years. In the right figure, we drop SOEs and observe a lower average age of Chinese firms. SOEs are relative older in China. As the older SOEs failed, the average age difference between SOEs and other firms has converged. The declining average age of Chinese firms can

155

14 12 10

Average Age of Firms

12

8

10

Average Age of Firms

14

7.4 Pattern of Firm Failure in China

8

1998

1998

2000

2002 Year

2004

2000

2006

2002 Year all firms

2004

2006

excluding SOE

12

Average Age

10

.2

8

.18

Average failure rate

.22

14

.24

16

Fig. 7.2 Average age of Chinese firms during 1998–2006

.16

1998 0

10

20

30

40

50

Age

2000

2002 Year failing firms

2004

2006

survival firms

Fig. 7.3 Firm age and firm failure in China

be explained by the relationship between firm age and business failure. Figure 7.3 shows that older firms are more likely to fail. In other words, on average, failing firms are older than surviving firms. It is clear that firm age seems to have positive relationship with firm failure. Older firms may face more challenges to survive. Learning effects can be offset by competition effects.

7.4.3

Local Support and Firm Failure

We can identify direct subsidies from governments and loans from formal financial institutions for individual firms. Whether firms get subsidies and banking loans actually is more important than how much they get since both subsidies and loans represent governmental support. Figure 7.4 shows that the share of firms with subsidies is lower for failing firms than for surviving ones. Failing firms are also less likely to get banking loans. Subsidies and financial supports seem to

7 What Causes Firm Failure in China?

.7 .65 .6

Ratio of firms with loans

.55

.14 .12 .1

.5

.08

Ratio of firms with subsidy

.16

.75

156

1998

2000

2002 Year failing firms

2004

2006

1998

2000

survival firms

2002 Year failing firms

2004

2006

survival firms

14 12 10

Average Age of the Firms

16 14 12 8

8

10

Average Age of Firms

18

16

Fig. 7.4 Local supports and firm exit

1998

2000

2002 Year

firms with subsidy

2004

2006

firms without subsidy

1998

2000

2002 Year

firms with loans

2004

2006

firms without loans

Fig. 7.5 Firm age, local supports, and firm failure

reduce the chances of firm failure in China and keep some less productive firms in the markets. We further take a look at the relationship between local support and firm age. Figure 7.5 indicates that firms with subsidies are older than those without subsidies. Similarly, firms with banking loans are also older than those without loans. Local supports may help firms to survive for a longer time and mitigate the impacts of firm age on firm failure. Subsidies, banking loans, and firm TFP do not have a clear relationship (Fig. 7.6). In the late 1990s, less productive firms are more likely to have subsidies, and then more productive firms are subsidized for a couple of years. Recently both productivities of subsidized and nonsubsidized firms have been converging. Firms with banking loans on average are less productive. The findings indicate that TFP is just one of the many factors to influence local governments’ choice of favorable firms. SOEs are often easy to have subsidies and banking loans even they are less productive. Firms in local key industries are also often granted favorable supports.

157

2.5

2.6 2.55 2.5

Average TFP of Firms

Average TFP of Firms 2.55 2.65 2.6

2.7

2.65

7.5 Variables and Model Specification

1998

2000

2002 Year

firms with subsidy

2004

2006

1998

firms without subsidy

2000

2002 Year

firms with loans

2004

2006

firms without loans

Fig. 7.6 Local supports and firm TFP

7.5

Variables and Model Specification

The descriptive analysis suggests that firm TFP, age, subsidies, and banking loans are possible determinants for firm dynamics in China. We will conduct a systematic analysis to identify the determinants of firm failure. Based on the theoretical propositions, we particularly focus on two sets of explanatory variables. One set of variables includes firm TFP (TFP) and firm age (AGE). Firm TFP is estimated based on the model proposed by Olley and Pakes (1996). Firm age is the difference between the reporting year and the starting year of a firm. Productive firms are more competitive in the market, while inefficient firms would suffer from market competition (Sutton 1997; Caves 1998; Wagner 1994). Firms with high TFP are less likely to fail, and firm TFP will be expected to have a negative coefficient. The existing literature reports a negative influence of firm age on firm failure due to the learning effects. In China, competition effects may dominate learning effects, causing higher mortality rate of older firms. If that is the case, firm age will have a positive association with firm failure and vice versa. To test the nonlinear relationship between firm age and firm failure, we introduce its squared term (AGE*AGE) in the model. The other set of variables quantifies governmental supports. Both central and local governments have many different measures to support industrial development. Typically those policies work against with market forces. Firms in key industries are more likely to get favorable supports such as cheap land, subsidized electricity, and tax holidays from local governments. We measure the key industries as the location quotient of 3-digit industries at the prefecture-level cities (LQ). Local key industries will have a larger LQ. Meanwhile, firms in key industries may face intensive competition. If governmental support can help firm to overcome the competition effects, the LQ will have a negative coefficient. Otherwise, the LQ can have a positive coefficient. Typically, local governments provide subsidies to specific firms in industries and even request state-owned banks to provide loans to some firms in industries. We design two dummy variables to indicate firms with governmental subsidies and bank loans (SUBSIDY and LOAN) and expect negative

158

7 What Causes Firm Failure in China?

coefficients. To explore the moderating role of governmental interventions, we further introduce the interactions between the two sets of variables in the models. To make the estimations robust, we control a number of variables. First, firm size (SIZE) is included. We use the number of employees to measure firm size. Existing studies often report that large firms are less likely to fail (Siegfried and Evans 1994). Second, dummies for exporters (EXP) and foreign firms (FDI) are included. A number of studies found that being foreign firms or being an exporters is significantly correlated with firm survival (Görg and Strobl 2003; Bernard and Sjoholm 2003; Esteve-Pérez and Llopis 2004; Kimura and Kiyota 2006; Giovannetti et al. 2011). Third, the overall failing rate at the three-digit industry level in the prefecture level city is included (EXITRATE). A higher failing rate of an industry suggests that the industry may suffer from market competition, leading to more firm failure. Finally, we further control dummies for 2-digit industries, years, and provinces. Firm failure may vary across industries, provinces, and time. Variables are summarized in Table 7.2. The dependent variable (Exitstay) is a dummy variable, 1 indicating firms failing in the coming year and 0 for firms surviving the coming year. The data structure is unbalanced panel data. It is not proper to conduct survival analysis since firms are likely to enter or exit multiple times in this dataset. Both linear probability model (LPM) for panel data and probit model for panel data are proper methods. However, given the data structure in this study, there exist some theoretical problems regarding the interpretation and robustness of the average effects of the interaction items in the probit model estimations. With the support of a large sample, the estimation of LPM has no significant difference from those based on other discrete choice models such as probit models. Therefore, this study will apply LPM for panel data to estimate the coefficients. In addition, panel data are critical to mitigate the endogeneity associated with self-selection in the sample. However, panel data model only stresses the

Table 7.2 Definitions of dependent and explanatory variables Variables Exitstay TFP AGE LQ SUBSIDY LOAN SIZE EXP FDI EXITRATE Industry Province Year

Definitions 1 for firms failing in the coming year; 0 for firms surviving in the coming year Firm total factor productivity Firm age defined as differences between the reporting year and the starting year Location quotient of 3-digit industries at the prefecture city a firm is located 1 for firms with governmental subsidies; 0 for firms without subsidies 1 for firms with banking loans; 0 for firms without banking loans Log of number of employees 1 for exporters; 0 for non-exporters 1 for foreign firms; 0 for others Exit rate of 3-digit industry at the prefecture-level city Dummy for 2-digit industries that a firm belongs to Dummy for provinces where a firm is located Dummy for years (1998–2006)

7.6 Empirical Results

159

within-group variation instead of the between-group variation. We also apply probit model to check the robustness of the estimations of parameters. We define the following base model: exitstayit ¼ α þ β1 TFPit þ β2 Ageit þ δX it þ γ i þ λt þ εit

ð7:1Þ

where exitstay is the dependent variable and TFP and AGE are firm TFP and firm age. Xit represents other explanatory variables, including firm size (lnSIZE), dummy for exporters (EXP) and for foreign firms (FDI), and exit rate (EXITRATE), dummies for 2-digit industry and provinces. γ i is firm fixed effects, and λt stands for time fixed effects. To further investigate the role of governmental supports on firm failure, we introduce the interactions between policy proxies and TFP and AGE. Policy proxies include LQ, SUBSIDY, and LOAN. The expanded model is as follows: exitstayit ¼ α þ β1 TFPit þ β2 Ageit þ β3 policyit þ β4 policyit  TFPit þβ5 policy Ageit þ δX it þ γ i þ λt þ εit

7.6

ð7:2Þ

Empirical Results

The correlation analysis implies that explanatory variables are only weakly correlated (Table 7.3). There is no serious collinearity issue in the estimations. Table 7.4 reports the regression results from LMP for panel data. The results from the probit model for panel data do not show significant differences (to save space, we will not report the probit results). The first column presents the base model estimation. From columns 2 to 7, we introduce LQ, SUBSIDY, and LOAN and their interactions with TFP and AGE into the models. All models are statistically significant, indicating that our models have significant explanatory power. Statistical results suggest that firm TFP is a key determinant for firm survival. Firm TFP has a negative and significant coefficient in all models, indicating that less productive firms are more likely to fail. The results are consistent with the existing literature based in developed market economies (Siegfried and Evans 1994). The less productive firms suffer from low profitability and in turn a low level of investments. Competition would force inefficient firms of the market and redistribute resources to productive firms, improving industrial competitiveness and facilitating industrial upgrading in China. This is the positive role of market forces in industrial development. AGE is positive and significant, but AGE*AGE is negative and significant in all models, indicating that initially as firms get mature, they are more likely to fail in the coming year. When firms survive the market competition for a certain time period, they become more competitive and will be less likely to fail. There is an inverted U-shaped relationship between firm age and firm failure. In the existing literature,

Variable TFP AGE LQ SUBSIDY LOAN lnSIZE EXP FDI EXITRATE

TFP 1 0.051
0.0002 0.0244 0.011
0.007 0.089 0.0384
0.0987

LQ

1 0.0099 0.0348 0.1742 0.0779
0.0164
0.0587

AGE

1
0.034 0.0741 0.1173 0.2705
0.0205
0.1679 0.0444 1 0.0698 0.1442 0.0885 0.0079
0.0253

SUBSIDY

Table 7.3 Correlation coefficients among key explanatory variables

1 0.1679
0.0113
0.1268
0.0336

LOAN

1 0.2828 0.1403
0.0684

lnSIZE

1 0.3965
0.0602

EXP

1
0.0195

FDI

1

EXITRATE

160 7 What Causes Firm Failure in China?

(1) Exitstay
0.039*** 0.051***
0.023***
0.005***


0.071***
0.015*** 0.827***
0.013*** Y Y Y 0.164 1,555,981 0.258

(0) Exitstay
0.039*** 0.051***
0.023***


0.072***
0.015*** 0.826***
0.012*** Y Y Y 0.162 1,555,981 0.258

Note: ***p < 0.01, **p < 0.05, *p < 0.1

VARIABLES TFP AGE AGE*AGE LQ TFP*LQ AGE*LQ SUBSIDY TFP*SUBSIDY Age*SUBSIDY LOAN TFP*LOAN AGE*LOAN lnSIZE EXP EXITRATE FDI Industry Year Province Constant Observations R-squared

Table 7.4 Regression results from LPM for panel data


0.071***
0.015*** 0.827***
0.013*** Y Y Y 0.164 1,555,981 0.258

(2) Exitstay
0.039*** 0.051***
0.023***
0.004*** 0.001*** 0.000***


0.071***
0.014*** 0.826***
0.012*** Y Y Y 0.158 1,555,981 0.258


0.018*** 0.004***
0.001***


0.018***


0.071***
0.014*** 0.826***
0.012*** Y Y Y 0.160 1,555,981 0.258

(4) Exitstay
0.040*** 0.051***
0.023***

(3) Exitstay
0.039*** 0.051***
0.023***


0.071***
0.014*** 0.826***
0.012*** Y Y Y 0.168 1,555,981 0.258


0.013***

(5) Exitstay
0.039*** 0.051***
0.024***


0.004** 0.001**
0.001***
0.071***
0.014*** 0.826***
0.013*** Y Y Y 0.158 1,555,981 0.258

(6) Exitstay
0.040*** 0.052***
0.024***

7.6 Empirical Results 161

162

7 What Causes Firm Failure in China?

firm age is typically positively associated with firm survival due to the learning effects although a few studies report an inverted U-shaped age effect (Agarwal and Sarkar 2002). In the developed market economies, firms would accumulate experience and knowledge and develop their own core competence as they stay in the market longer, reducing their chances of failure. In the transitional China, firms in their first several years face substantial challenges. Rapid industrial upgrading and constantly adjusted institutional frameworks could make some Chinese firms quickly outdated. When firms survive the competitive markets for a long time, some firms may accumulate sufficient market knowledge or invest in innovative activities to develop core competence, reducing their chance of failure. This is still consistent with the learning model of industrial dynamics. The difference is that firms take longer time to learn in the transitional China. The three proxies for governmental supports (LQ, SUBSIDY, and LOAN) hold the expected negative coefficients in all models, indicating that firms in key industries are less likely to fail. Firms with subsidies and banking loans are more likely to survive the coming year. Studies in Germany and France report a positive role of subsidies in firm survival (Almus 2002; Crépon and Duguet 2003). In an undeveloped market economy, governmental supports are expected to be more important for firms to survive and allow some inefficient firms to survive. Overall, industry-specific and firm-specific supportive programs and policies are conducive to firm survival. Governmental intervention however has significant moderating roles in firm failure. TFP*LQ, TFP*SUBSIDY, and TFP*LOAN are all positive and significant. Given the presence of governmental support, firms with higher TFP are more likely to fail. In other words, given the same level of TFP, firms with governmental support such as subsidies and banking loans are hard to survive. The findings indicate that governmental support for less productive firms reduce the chance of productive firms. Governmental support generates negative externalities for firms without subsidies and banking loans. This is the dark side of governmental intervention in industrial development. Although government supports may be helpful to some firms, they have considerable social costs, which are transferred to those not favored by local governments. Governmental supports also moderate the impacts of firm age on firm failure. AGE*LQ is positive and significant. Older firms in key industries are more likely to fail. Literally, key industries mean important industries for a city and have a large number of firms. Local intra-industry competition among key industries is often very intensive, imposing substantial challenges for firms to accumulate knowledge and to learn. In those industries, new firms are often equipped with advanced technology and better organization structure, providing new firms the advantage of latecomers. In the key industries, competition effects dominate learning effects. It is reasonable that older firms in the key industries are hard to survive the market competition. However, AGE*SUBSIDY and AGE*LOAN are negative and significant, suggesting that subsidies and banking loans can reduce the failing chances of older firms. Figure 7.5 also shows that firms with subsidies and banking loans are older than those without. Subsidies and banking loans have favored older firms.

7.7 Summary

163

Firm-specific support policies rather than industry-targeted policies can mitigate the competition effects, increasing the business survival of older firms in China. The findings confirm our research hypotheses. The control variables are largely consistent with the theoretical expectation and the existing literature. lnSIZE has a negative and significant coefficient, indicating that large firms are easier to survive. Being an exporter or foreign firm is less likely to fail. Direct engagement into the globalization process is a critical factor to survive Chinese firms. EXITRATE has a significant and positive coefficient in all models, suggesting that firms are hard to survive when the city industry is suffering. We also control the dummies for 2-digit industries, Chinese provinces, and the reporting years to make the estimations robust. We now conduct several robust checks. The year 2004 is the economic census year and may include more newly created firms, with an entry rate of 45.13% and failing rate of 18.69%. The entry rate is significantly higher than other years. It may impose some influence on the estimations. We dropped the 2004 sample and re-estimated the parameters (Table 7.5). The results are largely consistent with the results derived from the full sample. The key variables remain significant with expected signs. There are some minor changes. The magnitude of the coefficient on TFP drops from 0.039 to 0.028, while the coefficient on AGE increases from 0.051 to 0.075. AGE*LQ, TFP*SUBSIDY, and TFP*LOAN turn insignificant. But overall, the 2004 sample does not cause significant influence on the effectiveness of the models. Furthermore, the dataset only includes firms with total sales revenues greater than five million Yuan, which is at the current year price. The five million Yuan entry standard may influence the model estimation. We then use six million Yuan and eight million Yuan to choose two subsamples to re-estimate the results. We also use the producer price index in 1998 to redefine the five million Yuan standard and make another subsample of firms during 1998–2007 and rerun the models. The main findings from the above exercises remain almost the same. Given the large sample of firms in our model estimations and the robust results from different estimations, we are confident that our empirical findings are convincing. The results indicate that both market forces and state power determine firm dynamics in China.

7.7

Summary

China has intended to build a market-oriented economy gradually since the late 1970s. The economic transition process can be coded as marketization, globalization, and decentralization. The triple process has fundamentally changed the institutional framework for firms to survive. Marketization and globalization have clearly introduced harsh market competition, but decentralization has granted local governments strong incentives to protect and support local businesses. This may provide institutional advantage for firm survival. We argue that competition

(1) Exitstay
0.028*** 0.075***
0.022***
0.006***


0.065***
0.014*** 0.726***
0.012*** Y Y Y
0.279 1,302,296 0.293

(0) Exitstay
0.028*** 0.075***
0.022***


0.066***
0.014*** 0.725***
0.011*** Y Y Y
0.282 1,302,296 0.293

Note: ***p < 0.01, **p < 0.05, *p < 0.1

VARIABLES TFP AGE AGE*AGE LQ TFP*LQ AGE*LQ SUBSIDY TFP*SUBSIDY AGE*SUBSIDY LOAN TFP*LOAN AGE*LOAN lnSIZE EXPORT EXITRATE FDI Industry Year Province Constant Observations R-squared
0.065***
0.014*** 0.726***
0.012*** Y Y Y
0.279 1,302,296 0.293

(2) Exitstay
0.028*** 0.075***
0.022***
0.004*** 0.001* 0.000


0.065***
0.014*** 0.725***
0.011*** Y Y Y
0.287 1,302,296 0.294


0.010*** 0.000
0.001***


0.021***


0.065***
0.014*** 0.725***
0.011*** Y Y Y
0.287 1,302,296 0.294

(4) Exitstay
0.028*** 0.075***
0.022***

(3) Exitstay
0.028*** 0.075***
0.022***

Table 7.5 Regression results from LPM for panel data (excluding the sample of year 2004)


0.065***
0.014*** 0.725***
0.011*** Y Y Y
0.279 1,302,296 0.294


0.011***

(5) Exitstay
0.028*** 0.075***
0.022***

0.003
0.000
0.001***
0.065***
0.014*** 0.725***
0.012*** Y Y Y
0.289 1,302,296 0.294

(6) Exitstay
0.027*** 0.076***
0.023***

164 7 What Causes Firm Failure in China?

References

165

effect, learning effect, and governmental support underpin firm survival and failure in China. Based on the ASIF dataset, this study explored the pattern of firm failure. It is reported that on average, failing firms are less productive than surviving ones. Older firms are more likely to fail, while firms with governmental supports have more chance to survive. More old firms have governmental supports. Productivity is just one of the many factors for governments to choose favorable firms. Statistical results provide strong evidence to support our research hypotheses. It is implied that competition effects crowd out less productive firms. When firms are in their young ages, competition dominates learning effects and imposes challenges on the survival of growing firms. However, when firms overcome the challenges and stay in the market for a certain time, they can survive for a longer time. There is an inverted U-shaped age effect. Local protectionism and supportive policies can certainly reduce the chance of firm failure. Particularly, they can help older firms to mitigate the impact of competition effects. However, governmental intervention would generate negative externality, reducing the survival chances of productive firms without supports. This is the social costs of governmental intervention in industrial development. This study contributes to the related literature from the following aspects. First, we process the annual survey of industrial firms in systematical ways and estimate firm TFP using the semi-parameter method, which can solve the selection bias and endogeneity issues. Second, this study is the first to explore the impacts of industrial policies on firm failure, and the findings make essential senses to understand industrial dynamics in China. Third, the findings enrich our understanding about the importance of state power and market force in the development of industries and local economies in the transitional China. Fourth, we realized some limitations in the data and definition of firm failure; we conduct several robust checks and confirm the results. This is just an initial effort to understand firm dynamics in China. Future research can extend to explore the industrial, spatial, and temporal variation of firm failure and its impacts on industrial upgrading and productivity improvement in Chinese industries and regions.

References Acs, Z. J., Armington, C., & Zhang, T. (2007). The determinants of new-firm survival across regional economies: The role of human capital stock and knowledge spillover. Papers in Regional Science, 86(3), 367–391. Agarwal, R. (1996). Technological activity and survival of firms. Economics Letters, 52(1), 101–108. Agarwal, R. (1998). Small firm survival and technological activity. Small Business Economics, 11 (3), 215–224. Agarwal, R., & Audretsch, D. B. (2001). Does entry size matter? The impact of the life cycle and technology on firm survival. Journal of Industrial Economics, 49(1), 21–43.

166

7 What Causes Firm Failure in China?

Agarwal, R., & Gort, M. (2002). Firm and product life cycles and firm survival. American Economic Review, 92(2), 184–190. Agarwal, R., & Sarkar, M. B. (2002). The conditioning effect of time on firm survival: An industry life cycle approach. Academy of Management Journal, 45(5), 971–994. Agarwal, R., Echambadi, R., Franco, A. M., & Sarkar, M. B. (2002). Knowledge transfer through inheritance: Spin-out generation, development, and survival. Academy of Management Journal, 47(4), 501–522. Almus, M. (2002). What characterizes a fast-growing firm? Applied Economics, 34(12), 1497–1508. Almus, M. (2016). Evaluating the impact of public start-up assistance-results from an econometric approach. Zew Discussion Papers, 5(1–23), 70–77. Audretsch, D. B. (1995). Innovation, growth and survival. International Journal of Industrial Organization, 13(4), 441–457. Audretsch, D. B., & Mahmood, T. (1995). New firm survival: New results using a hazard function. Review of Economics and Statistics, 77(1), 97–103. Audretsch, D. B., Houweling, P., & Thurik, R. (2000). Firm survival in the Netherlands. Review of Industrial Organization, 16(1), 1–11. Barbieri, E., Tommaso, D., Marco, R., & Bonnini, S. (2012). Industrial development policies and performance in southern China: Beyond the specialized industrial cluster program. China Economic Review, 23(3), 613–625. Berglund, E., & Brännäs, K. (2001). Plants’ entry and exit in Swedish municipalities. Annals of Regional Science, 35(3), 431–448. Bernard, A. B., & Sjoholm, F. (2003). Foreign owners and plant survival (National Bureau of Economic Research, No. W10039). Bojnec, Š., & Xavier, A. (2004). Entry and exit in transition economies: The Slovenian manufacturing sector. Post-Communist Economies, 16(2), 191–214. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Buenstorf, G., & Geissler, M. (2011). The origins of entrants and the geography of the German laser industry. Papers in Regional Science, 90(2), 251–270. Carree, M., Santarelli, E., & Verheul, I. (2008). Firm entry and exit in Italian provinces and the relationship with unemployment. International Entrepreneurship and Management Journal, 4 (2), 171–186. Caves, R. E. (1998). Industrial organization and new findings on the turnover and mobility of firms. Journal of Economic Literature, 36(4), 1947–1982. Cefis, E., & Marsili, O. (2005). A matter of life and death: Innovation and firm survival. Industrial and Corporate Change, 14(6), 1–26. Cefis, E., & Marsili, O. (2006). Survivor: The role of innovation in firm’s survival. Research Policy, 35(5), 626–641. Cheng, S., & Li, H. (2011). Spatially varying relationships of new firm formation in the United States. Regional Studies, 45(6), 773–789. Crépon, B., & Duguet, E. (2003). Bank loans, start-up subsidies and the survival of new firms: An econometric analysis at the entrepreneur level. Labor and Demography, 8(8), 2026–2034. Dunne, P., & Hughes, A. (1994). Age, size, growth and survival: UK companies in the 1980s. Journal of Industrial Economics, 42(2), 115–140. Ericson, R., & Pakes, A. (1995). Markov-perfect industry dynamics: A framework for empirical work. Review of Economic Studies, 62(1), 53–82. Esteve-Pérez, S., & Llopis, J. A. S. (2004). The determinants of survival of Spanish manufacturing firms. Review of Industrial Organization, 25(3), 251–273. Esteve-Pérez, S., Mánez-Castillejo, J. A., Rochina-Barrachina, M. E., Rochina-Barrachina, M. E., & Sanchis-Llopis, J. A. (2007). A survival analysis of manufacturing firms in export markets. In

References

167

J. Arauzo-Carod & M. C. Manjon-Antolin (Eds.), Entrepreneurship, industrial location and economic growth. Cheltenham: Elgar. Estrin, S., & Prevezer, M. (2010). A survey on institutions and new firm entry: How and why do entry rates differ in emerging markets? Economic Systems, 34(3), 289–308. Folta, T. B., Cooper, A. C., & Baik, Y. S. (2006). Geographic cluster size and firm performance. Journal of Business Venturing, 21(2), 217–242. Fontana, R., & Nesta, L. (2009). Product innovation and survival in a high-tech industry. Review of Industrial Organization, 34(4), 287–306. Fotopoulos, G., & Louri, H. (2000). Location and survival of new entry. Small Business Economics, 14(4), 311–321. Fotopoulos, G., & Spence, N. (2001). Regional variations of firm births, deaths and growth patterns in the UK, 1980–1991. Growth and Change, 32(2), 151–173. Geroski, P. (1995). Innovation and competitive advantage (OECD Economics Department Working Papers, p. 159). Paris: OECD Publishing. Giarratana, M. S., & Torrisi, S. (2010). Foreign entry and survival in a knowledge-intensive market: Emerging economy Countries' International linkages, technology competences, and firm experience. Strategic Entrepreneurship Journal, 4(1), 85–104. Giovannetti, G., Ricchiuti, G., & Velucchi, M. (2011). Size, innovation and internationalization: A survival analysis of Italian firms. Applied Economics, 43(12), 1511–1520. Girma, S., & Gong, Y. (2008). Putting people first? Chinese state-owned Enterprises’ adjustment to globalisation. International Journal of Industrial Organization, 26(2), 573–585. Görg, H., & Strobl, E. (2003). Multinational companies, technology spillovers and plant survival. Scandinavian Journal of Economics, 105(4), 581–595. Hall, B. H. (1987). The relationship between firm size and firm growth in the U.S. manufacturing sector. Journal of Industrial Economics, 35, 583–605. Hannan, M. T., Carroll, G. R., Dobrev, S. D., & Han, J. (1998). Organizational mortality in European and American automobile industries Part I: Revisiting the effects of age and size. European Sociological Review, 14(3), 279–302. He, C. (2006). Regional decentralisation and location of foreign direct Investment in China. PostCommunist Economies, 18(1), 33–50. He, C., Wei, Y., & Xie, X. (2008). Globalization, institutional change and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. Honjo, Y. (2000). Business failure of new software firms. Applied Economics Letters, 7(9), 575–579. Hsieh, C. T., & Klenow, P. J. (2009). Misallocation and manufacturing TFP in China and India. Quarterly Journal of Economics, 24(4), 1403–1448. Huang, Y. (2003). Selling China: Foreign direct investment during the reform era. Cambridge: Cambridge University Press. Huang, Y. (2008). Capitalism with Chinese characteristics: Entrepreneurship and the state. Cambridge: Cambridge University Press. Kangasharju, A. (2000). Growth of the smallest: Determinants of small firm growth during strong macroeconomic fluctuations. International Small Business Journal, 19(1), 28–43. Kimura, F., & Fujii, T. (2003). Globalizing activities and the rate of survival: Panel data analysis on Japanese firms. Journal of the Japanese and International Economies, 17(4), 538–560. Kimura, F., & Kiyota, K. (2006). Exports, FDI, and productivity: Dynamic evidence from Japanese firms. Review of World Economics, 142(4), 695–719. López-García, P., Puente, S., & Gómez, Á. L. (2007). Firm productivity dynamics in Spain. http:// ideas.repec.org/p/bde/wpaper/0739.html. Accessed Dec 2007. Männasoo, K. (2008). Patterns of firm survival in Estonia. Eastern European Economics, 46(4), 27–42. Mao, Q. L., & Sheng, B. (2013). China’s manufacturing firms’ entry-exit and dynamic evolution of TFP. Economic Research Journal, 4, 16–29. (in Chinese).

168

7 What Causes Firm Failure in China?

Mata, J., & Portugal, P. (1994). Life duration of new firms. Journal of Industrial Economics, 42(3), 227–245. Mata, J., & Portugal, P. (2002). The survival of new domestic and foreign-owned firms. Strategic Management Journal, 23(4), 323–343. Močnik, D. (2010). The impact of economic growth on the dynamics of enterprises: Empirical evidence for Slovenia’s non-agricultural sector. The Organ, 43(4), 154–162. Nyström, K. (2007). An industry disaggregated analysis of the determinants of regional entry and exit. Annals of Regional Science, 41(4), 877–896. Olley, G. S., & Pakes, A. (1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64(6), 1263–1297. Shiferaw, A. (2009). Survival of private sector manufacturing establishments in Africa: The role of productivity and ownership. World Development, 37(3), 572–584. Siegfried, J. J., & Evans, L. B. (1994). Empirical studies of entry and exit: A survey of the evidence. Review of Industrial Organization, 9(2), 121–155. Silva, D. G., & McComb, R. P. (2012). Geographic concentration and high tech firm survival. Regional Science and Urban Economics, 42(4), 691–701. Song, Z., Storesletten, K., & Zilibotti, F. (2011). Growing like China. American Economic Review, 101(1), 196–233. Staber, U. (2001). Spatial proximity and firm survival in a declining industrial district: The case of knitwear firms in Baden-Württemberg. Regional Studies, 35(4), 329–341. Stearns, T. M., Carter, N. M., Reynolds, P. D., & Williams, M. L. (1995). New firm survival: Industry, strategy, and location. Journal of Business Venturing, 10(1), 23–42. Sutton, J. (1997). Gibrat’s legacy. Journal of Economic Literature, 35, 40–59. Thompson, P. (2005). Selection and firm survival: Evidence from the shipbuilding industry, 1825–1914. Journal of Economics and Statistics, 87, 26–36. Tybout, J. R. (2000). Manufacturing firms in developing countries: How well do they do, and why? Journal of Economic Literature, 38(1), 11–44. Wagner, J. (1994). The post-entry performance of new small firms in German manufacturing industries. Journal of Industrial Economics, 17, 141–154. Wagner, J. (1999). The life history of cohorts of exits from German manufacturing. Small Business Economics, 13(1), 71–79. Wagner, S., & Cockburn, I. (2010). Patents and the survival of internet-related IPOs. Research Policy, 39(2), 214–228. Wei, Y. H. D. (2001). Decentralization, marketization and globalization: The triple processes underlying regional development in China. Asian Geographer, 20(1–2), 7–23. Yang, R., & He, C. (2014). The productivity puzzle of Chinese exporters: Perspectives of local protection and spillover effects. Papers in Regional Science, 93(2), 367–384. Young, A. (2000). The Razor’s edge: Distortions and incremental reform in the People’s Republic of China. Quarterly Journal of Economics, 115, 1091–1135. Zhang, W. (2000). Transforming China: Economic reform and its political implications. New York: St. Martin’s Press. Zhao, X., & Zhang, L. (1999). Decentralization reforms and regionalism in China: A review. International Regional Science Review, 22(3), 251–281.

Chapter 8

What Sustains Large Firms in China?

8.1

Introduction

Reflecting the importance of firm dynamics, a large body of literature has examined the determinants focusing on firm, industry, and regional factors using data from developed economies, including firm size, age, ownership, innovation, and agglomeration economies (Siegfried and Evans 1994; Audretsch et al. 2000; Fotopoulos and Spence 2001; Acs et al. 2007; Neffke et al. 2012; Silva and McComb 2012). In the existing literature, large firms are found to have more chance of sustaining. However, firms operate in rather different business environments in transitional economies (Estrin and Prevezer 2010; Mocnik 2010). A number of features are likely to challenge sustaining chance of businesses, such as a limited market for goods, weak infrastructure, weak legal systems, and contract enforcement (Tybout 2000). The transitional economies are undergoing economic transformation from planned economies to market-oriented economies with constant institutional development. Transitional economies therefore provide a unique opportunity to broaden the investigation of business dynamics by incorporating institutional and geographical factors. The purpose of this study is to investigate the sustaining opportunity of large firms in China. We define firm sustaining as the chance of large firms remaining large several years after birth. In line with the official statistics in China, large firms in this study refer to those with sales revenues greater than five Million RMB. We pay particular attention to the large firms since they contribute to more than 90% of gross industrial output in China. We answer the question of what sustains large firms several years down the line after birth. Since the late 1970s, China has been gradually transformed into a market-oriented economy from a

Modified article originally published in [He, C., Q. Guo and D. Rigby (2017) What Sustains Larger Firms? Evidence from Chinese Manufacturing Industries, The Annals of Regional Science, 58 (2), pp. 275–300.]. Published with kind permission of © [Springer Nature, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_8

169

170

8 What Sustains Large Firms in China?

commanding economy, creating both opportunities and challenges for business dynamics. This economic transition process has resulted in the liberalization of prices, markets, investment, and trade and privatization of selected state-owned sectors. In line with economic liberalization, China has widely opened its economy to the international market, warmly inviting foreign investments and bringing global forces into local development. Meanwhile, regional decentralization together with political centralization has triggered fierce interjurisdictional competition and local protectionism since subnational governments compete for economic growth and fiscal revenues (Oi 1992; Montinola et al. 1995) and local officials compete for promotions within the government (Blanchard and Shleifer 2001; Li and Zhou 2005). As a result, businesses in China survive and sustain in a regionally decentralized authoritarian system (Xu 2011). However, we still know very little about business dynamics in China, let alone the influence of regional and local conditions on such dynamics. This study is to examine what sustains large firms in the Chinese market, with a particular focus on the impact of institutional and geographical factors. Specifically, we examine why large firms remain large years after birth, assuming that businesses are influenced by global, provincial, and local forces. Using data on new large firms created during 1998–2005, this study largely confirms that global, provincial, and local forces are critical for business sustaining. The presence of foreign firms shows stronger competition effect at the prefecture city level than at the provincial level. Market-oriented institutions are key to sustain businesses in China. Non-state-owned enterprises are more dependent on market-oriented institutions than state-owned enterprises (SOE). Controlled for global and provincial forces, local factors are also effective to sustain local businesses. Firms sustain themselves by reaping the benefits from agglomeration economies and governmental supports. In addition, local factors work differently in the coastal, central, and western regions for sustaining businesses. This chapter contributes to the literature of industrial dynamics in several ways. First, this study provides evidence on business dynamics in a transitional economy, which is constantly developing its market-oriented institutions. Second, this study particularly highlights the critical role of institutional factors in sustaining business, revealing the deep root causes of business dynamics. Third, this study distinguishes the regional factors into different geographical levels, showing the impact of globallocal and regional-local interactions on sustaining firms. After all, this is among the first efforts to investigate business dynamics based on a large dataset in China. This chapter is structured as follows. After the introduction, this chapter provides a literature review and proposes three research hypotheses to understand business sustaining in transitional China. In the third part, this chapter introduces data sources and data processing in details and describes the patterns of firm sustaining. This chapter then applies the cox proportional hazard model to identify the determinants of sustaining businesses and discuss the statistical results. This chapter concludes with a summary of empirical findings and policy implications.

8.2 Literature Review and Analytical Framework

8.2

171

Literature Review and Analytical Framework

A regional economy keeps its vitality through a continual process of firm dynamics. Firm survival is considered the ultimate criterion of business effectiveness and can be defined as the probability that a firm will continue operations rather than exit a market (Hannan and Freeman 1988). Much effort has been expended in describing and explaining business survival and industrial dynamics, although chiefly in developed economies (Siegfried and Evans 1994; Cefis and Marsili 2005). Industrial organization literature and entrepreneurship studies have systematically examined firm and industry factors. This line of research follows in the tradition of Bain (1956), who pioneered the concept of structural barriers to entry to explain persistent differences in entry rates across industries. Small and young firms are less likely to survive, consistent with the learning models of industry dynamics (Ericson and Pakes 1995; Audretsch and Mahmood 1995; Wagner 1999). Size and age effect however may be nonlinear (Mata and Portugal 1994; Pérez et al. 2004; Agarwal and Gort 2002; Esteve-Pérez et al. 2007). In addition, ownership is another major determinant of firm survival (Mata and Portugal 1994; Agarwal and Gort 2002; Görg and Strobl 2003; Esteve-Pérez et al. 2007). Foreign ownership is positively related to business survival due to the competitive advantages of foreign firms. Some studies report a positive and precisely determined effect of exporting on firm survival using US, Spanish, and Japanese data, respectively (Bernard et al. 2005; Pérez et al. 2004; Kimura and Kiyota 2006). This is consistent with the learning by exporting arguments in literature. Innovative activities are shown to be beneficial for firm survival. Studies look at these innovation inputs such as R&D investments and technology uses (Audretsch 1995; Kimura and Fujii 2003; Fontana and Nesta 2009) and indicators of innovative performance (Cefis and Marsili 2005; Wagner and Cockburn 2010). In addition, industrial characteristics such as market size, growth rate, technology, market structure, entry rate and scale economies, and the life cycle consistently explain the differences in survival rates across firms (Mata and Portugal 1994; Audretsch 1995; Agarwal and Audretsch 2001; Audretsch and Mahmood 1995; Strotmann 2007). A relatively smaller number of studies have explored the impact of regional factors on firm survival. In particular, the link between agglomeration economies and firm survival has also been explored using direct measures of agglomeration. Some find that industrial agglomeration helps firm survival (Fritsch et al. 2006; Delgado et al. 2010; Wennberg and Lindqvist 2010). Recently, Neffke et al. (2012) confirm that technological relatedness substantially increases the survival rates of plants. Others conclude that industrial clustering is associated with higher mortality of businesses (Acs et al. 2007; Silva and Mccomb 2012). For instance, Acs et al. (2007) report negative impacts of both localization and urbanization economies on the survival of new service firms in the USA. Examining the evolution of the car industry in Britain during 1895 and 1968, Boschma and Wenting (2007) went on to find that the presence of previous related industries has a positive impact on the survival of automobile firms.

172

8 What Sustains Large Firms in China?

In summary, studies of firm survival in developed economies seem to have reached some consensus, confirming the impacts of firm size, age, innovation, and ownership and industry and regional factors. However, the literature has systematically ignored firm dynamics in transitional economies where firms face poor legal framework, imperfect markets, and changing policy regimes (Estrin and Prevezer 2010; Mocnik 2010). Given that transitional economies are undergoing economic, social, and political transformations, both the market and the state play their roles in economic development. Institutional factors are particularly relevant to firm dynamics in transitional economies. In the literature, the investigation of regional factors largely focuses on the role of agglomeration economics without considering the potential impacts of institutional factors. Moreover, the existing studies simply focus on regional factors at only one geographical scale without considering the interactions of global-regional-local forces.

8.2.1

Firm Sustaining in China: An Analytical Framework

The topic of firm dynamics is an extremely underexplored area in China although considerable industrial processes have been observed in the last decades (He and Yang, 2016). Firm dynamics in China is however better understood in the context of economic transition, which is the transformation from a commanding economy to a market-oriented economy. In transitional economies, institutional factors and geographical factors tend to play more fundamental roles in firm dynamics. Firm dynamics refer to a continual process of firm entry, survival, and exit. Due to the data availability, we focused on sustaining firms, which are the core part of firm survival. We define firm sustaining as the chance that firms remain a certain size after birth. Since the late 1970s, China has embarked on the road to economic reform. One of the fundamental institutional changes is the liberalization of prices, markets, investments, and trade and privatization of selected state-owned sectors. The liberalization and privatization have indeed corrected many of the inefficiencies in resource allocation and unlocked the productive potential of firms (Lin 1992; Qian 2000). The presence of market and market-oriented institutions is essential for business operation since firms can easily trade in markets. Meanwhile market competition could impose challenges for the sustaining of firms. The process of economic liberalization is however by no means regionally balanced (Han and Pannell 1999). Some regions, particularly the coastal region, are more economically liberalized, with strong market forces at work, while the Western China is less market oriented. Businesses in China sustain in regionally different institutions. In other words, the extent of economic liberalization would have a significant impact on the chance of sustaining firms in China.

8.2 Literature Review and Analytical Framework

8.2.2

173

Globalization, Global-Local Interaction, and Firm Sustaining

Accompanying economic liberalization, China has been deeply participated in the process of globalization. Right in the beginning of economic reform, China set up several special economic zones to experiment with a market economy but more importantly to attract foreign investments. A large number of foreign firms in Chinese cities are the result of global-local interaction. Based on their ownership advantages, foreign firms are able to reap from agglomeration economies, industrial linkage, favorable policies, and market potential and institutional advantages when they choose locations in China (He 2002, 2003; Du et al. 2008; Sun et al. 2002). The relatively rational locational choices provide foreign firms advantages over local firms. The presence of foreign firms in local industrial development would generate both competition effects and spillover effects (Chang and Xu 2008). On the one hand, foreign firms are likely to intensify local market competition and steal market shares of local businesses (Aitken and Harrison 1999; Caves 1974; Blomström et al. 2000). On the other hand, foreign firms may bring new management and superior technology and would enhance the productivity of local firms via knowledge spillovers (Blomström and Kokko 1998). Caves (1974) suggests that foreign entrants can increase the speed of technology spillovers to host industries by demonstrating their technological superiority and by transacting with domestic firms. Many other studies found contradictory evidence of technology spillovers (Aitken and Harrison 1999; Konings 2001; Backer and Sleuwaegen 2003). In the Chinese case, positive spillover effects from foreign firms to local firms are reported (Cheung and Lin 2004; Liu et al. 2002; Liu 2008), which indicate that presence of foreign firms may improve the sustaining chance of local firms. Others challenge this positive effect of FDI in China by arguing that the transference of technology from foreign firms to local firms, if any, has yielded only marginal positive influences (Chen et al. 1995; Wei 2002). Moreover, Jeon et al. (2013) found that foreign investments in the same industry are more likely to engender negative influences on local Chinese firms and the negative horizontal effects are particularly prominent in low technology sectors. The mixed results of foreign firms on local firms could be reconciled by considering the competition effects and spillover effects at different geographical levels. Local firms are likely to see the entry of foreign firms as a serious threat, especially when it is locally bound (Dawar and Frost 1999). Foreign firms typically enjoy technological superiority and strong ownership advantages and can secure preferential treatment from local governments due to their strong bargaining power (Kim 1988). In China, local governments are willing to provide foreign firms with favorable treatment. Correspondingly, the competition effects between foreign entrants and local firms will be stronger in a local market. On the contrary, spillover effects of foreign firms are likely to go beyond the narrowly defined local boundaries, such as cities, and become pronounced at the larger geographical levels (Keller 2002). The demonstration effect of foreign firms can take place across cities. The

174

8 What Sustains Large Firms in China?

interregional mobility of better-educated workers can facilitate the spillover of advanced knowledge and technologies across cities. Knowledge spillovers are also facilitated by cross-regional investments and the interregional business networks (Chang and Xu 2008). We define the prefecture-level city as local while province as regional. In line with the analysis, we propose the first hypothesis as follows. Hypothesis 1 Strong presence of foreign firms in prefectural cities would reduce the chance of sustaining firms due to competition effects, while the presence of foreign firms in provinces where prefectural cities are located would help sustain local businesses due to the spillover effects.

8.2.3

Regional Decentralization, Regional-Local Interactions, and Firm Sustaining

Economic reform has resulted in considerable regional decentralization, in which subnational governments run the economy while the central government has tight control over the local government personnel. The Chinese government consists of a region-based multilevel hierarchy. Below the central government, there are four levels of subnational governments: province, municipality (or prefecture), county, and township. China has decentralized its fiscal system by introducing the tax sharing system in 1994, which has substantially raised the central share in revenues and reduced that of local governments, forcing local governments to self-finance their economic development (Zhang 1999; Wong 2000). While subnational governments obtain economic decision power, the central government makes decisions on appointment and removal of provincial leaders, and most municipal leaders are directly controlled by corresponding provincial governments (Xu 2011). Economic regional decentralization with political centralization has triggered intensive interregional competition in propelling economic growth and is argued as the fundamental institution of China’s reform and development (Xu 2011). Two approaches have theorized the role of local governments in China’s economic growth, either viewing local officials as revenue maximizers (Oi 1992; Montinola et al. 1995) or assuming that local officials seek to maximize their chances of political promotion (Blanchard and Shleifer 2001; Li and Zhou 2005). Both approaches suggest that local governments would favor the pro-business policies and strategies. Local governments can directly provide supports to business and protect local firms away from competition (Bai et al. 2004; Barbieri et al. 2012). Local governments can take advantage of market forces and develop marketoriented institutions to help business operations (Xu 2011). The impact however depends on the status of local authorities in China’s political hierarchy. Provinces are granted more substantial power and autonomy to make and enforce laws, regulations, and policies. Cities and counties under their jurisdictions implement those laws and regulations. China is characterized by substantial provincial disparities in institutions such as de facto property rights protections, government intervention in

8.2 Literature Review and Analytical Framework

175

business operations, law enforcement capability, and contract enforcement (Du et al. 2008). Good institutions are critical for business operation and sustaining. For instance, market-oriented institutions support the effective function of markets, such that firms can engage in market transaction without incurring undue costs or risks (North 1990). In addition, compared with cities, provinces in China are sufficiently large in terms of market size so that they could build self-contained economies to successfully implement local protectionism policies (Bai et al. 2004; He et al. 2007). We propose the second research hypothesis. Hypothesis 2 Firms located in provinces with better market-oriented institutions and larger market potential are more likely to sustain. Municipalities and cities however directly work with businesses since they build industrial parks to develop clusters, lease land, and provide public services, infrastructure, and subsidies to businesses and collect taxes and fees from them. In a regionally decentralized authoritarian system, cities have strong fiscal and political incentives to create favorable environments for businesses using market forces and state power. On the one hand, firms would benefit from market-oriented forces such as industrial agglomeration, which include both localization economies and urbanization economies, which induce the reduction of production cost, transaction cost, and learning costs through sharing, matching, and learning (Pan and Zhang 2002; He and Wang 2012; Lin et al. 2012; Duranton and Puga 2004). Firms would have more chance to sustain or survive if they can take advantage of cost-saving activities through agglomeration externalities (Wennberg and Lindqvist 2010; Oort et al. 2012; Cainelli et al. 2014). On the other hand, municipal governments provide fiscal and administrative supports for businesses. It is a common practice for China’s local governments to provide subsidies to firms, including income tax breaks for foreign firms and high-tech firms; loans to encouraged industries from government-owned banks; rebates of value-added tax and import duties for equipment purchases; low-priced land for SOEs and firms located in special development zones; and cash payments to firms based on factors such as export performance (Barbieri et al. 2012). The governmental supports would further easy business operation and would sustain firms at least in the short run. In line with the argument, we propose the third research hypothesis. Hypothesis 3 Firms are more likely to sustain if they benefit from agglomeration economies and governmental supports at the municipal city level. With gradual ownership reforms, China now has varieties of firm ownership, including state, foreign, private ownership, and so force. Their business goals, information gathering, and processing capabilities might differ significantly. Firms with different ownership may face different institutional uncertainties and build different relationships with local governments (He and Wang 2012). State-owned firms enjoy institutional advantages, such as favorable policies granted from central and local governments (Bai et al. 2004), unique access to acquiring policy, and business information. Under the institutional environment, SOEs are less motivated to reduce production costs and transaction costs by agglomerating with other firms.

176

8 What Sustains Large Firms in China?

In contrast, foreign-owned firms may face substantial disadvantages due to the lack of local knowledge and would encounter institutional risks and uncertainties in China (Luo 1997; He 2002; He and Wang 2012), so they may reduce risks by following other foreign-owned firms. Besides, local governments prefer foreignowned firms because these firms can induce the fast growth of GDP through large investment projects and introducing global linkage. Privately owned firms are fully responsible for their own business operations and are more likely to survive in the market by relying on industrial agglomeration and FDI spillovers. Market forces largely influence their survival rate. Therefore, firms with different ownership may depend on different factors to varying degrees. We propose the fourth research hypothesis: Hypothesis 4 The impact of global, regional, and local forces on firm sustaining varies across different ownerships.

8.3

New Firm Sustaining in China During 1998–2005

We use the ASIF dataset in the chapter. This study focuses on manufacturing enterprises created during 1998–2005. We identify the newly created firms based on the year of birth. We track each firm from the year of its birth until it either exits the database (signifying a business failure) or sustains beyond a specific number of years. Following Brandt et al. (2012), we consider a firm as new entrant in year t, if firm i is created and reported in the ASIFs in year t but not in year t
c. Likewise, if firm i is created and reported in the ASIF in year t
c and still in year t, it is assumed that firm i sustains in year t. The purpose of this study is to examine the likelihood that large firms remain large (with sales revenue greater than five million) several years after birth. Based on the year of birth for each firm, we identify the number of newly created firms during 1998–2005. We then compute the chance of sustaining new firms for different years (Fig. 8.1). Since our data include firms created in 1998–2005, firms created in 1998 can be traced to sustain up to 8 years, while those established in 2005 can only be seen for 1 year. Not surprisingly, the chance of sustaining new firms declines as they live longer. Take firms created in 1998 as an example; the chance of sustaining for 1 year is 83% and drops to 46.55% in 2003 and furthers to 33.10% in 2006. Overall, there is considerable temporal change of sustaining chance for firms created in different years. To illustrate spatial variation of sustaining chance of firms, we compute the sustaining chance for new firms created in 1998 for different years in different regions (Fig. 8.2). Following some studies (Gustafsson and Shi 2002; Yao and Zhang 2001), 31 provinces can be divided into 3 regions: the coast, central, and west. Region classifications are shown in Fig. 8.2 (right). China has indeed seen the coastal and inland divide in many aspects such as GDP per capita, technology, exports, and FDI. Studies typically divide the whole China into the coastal, central,

177

.6 .2

.4

Survival Rate

.8

1

8.3 New Firm Sustaining in China During 1998–2005

1

2

3

4 5 Sruvival length _1998 _2000 _2002 _2004

6

7

8

_1999 _2001 _2003 _2005

.6 .2

.4

Survival Rate

.8

1

Fig. 8.1 New firms’ survival rates for different number of years (1998–2005)

1

2

3

4 5 Sruvival length COASTAL WESTERN

6

7

8

CENTRAL

Fig. 8.2 Survival rates of new firms born in 1998 across regions (left, survival rates; right, region classification of China) Data sources of left figure: Annual survey of industrial firms (ASIF)

and western regions based on the location of Chinese provinces. Firms in the coastal region are more likely to sustain than those in inland regions. In 1 year, chance of sustaining in the coastal region is 85.55%, while the number is 78.55% in the central region and 75.85% in the west. As firms in the central region stay longer, they face

178

8 What Sustains Large Firms in China?

more difficult situation and have the lowest chance to sustain. The sustaining chance for 8 years for the three regions is 36.19%, 25.76%, and 31.88%, respectively. In general, firms in the coastal region are more likely to be engaged into the globalization process and benefit from that engagement. Furthermore, they enjoy positive externalities derived from industrial clustering (Yang and Hsia 2007; He and Wang 2012). To further explore the spatial variation of firm sustaining, we map the provincial distribution of the sustaining rate of firms born in 1998 for 1 year, 3 years, 5 years, and 7 years (Fig. 8.3). There is remarkable provincial variation in the chance of sustaining in a shorter period. In the longer period, the sustaining rate observes a convergence across provinces. For instance, the 1-year rate for 1998 new firms is higher in most coastal provinces and in some inland provinces such as Hunan and Guizhou in the southwest and Xinjiang and Qinghai in the northwest. Other inland

Fig. 8.3 Provincial distribution of survival rate of firms born in 1998

8.4 Model Specification and Variables

179

provinces like Jilin in the northeast and Gansu in the northwest however see the lowest survival rate. The 5-year rate for 1998 new firms remains higher in the coastal provinces such as Tianjin, Fujian, Zhejiang, and Jiangsu but also in some western provinces such as Qinghai, Shaanxi, Guizhou, and Sichuan. The 7-year rate observes smaller interprovincial differences, but Qinghai, Shaanxi, Zhejiang, and Fujian still top the list.

8.4 8.4.1

Model Specification and Variables Cox Proportional Hazard Model

We treat estimate the semi-parametric Cox proportional hazard model that defines hazard rates as the probability that a firm exits the market at a certain time t conditional on its sustaining to that time and on a set of covariates Xit. This method is appropriate to handle right censoring in time-series data where the event of interest might not occur within the study period. In fixed-effect models of such events, observations for which the values of the dependent variable do not change are ignored. Such observations carry important information that researchers seek to exploit. The semi-parametric Cox proportional hazard model is a common method to investigate firm survival, but the data structure of firm sustaining is similar to that of firm survival. Their difference is that the threshold defining firm sustaining is above 500 million RMB sales but the threshold defining firm sustaining is above zero sales. The semi-parametric Cox proportional hazard model can be used to study firm sustaining. The basic cox proportional hazard model is defined as: hi ðt Þ ¼ h0 ðt Þexp ðX i βÞ

ð8:1Þ

where h0(t) is the baseline hazard function, X is a vector of independent variables, and β is a corresponding vector of coefficients. The subscript i denotes firm. This model is semi-parametric because the baseline hazard function h0(t) can be unspecified; the covariates enter the model linearly after taking the log form: log hi ðt Þ ¼ αðt Þ þ β1 X i1 þ β2 X i2 þ    þ βk X ik

ð8:2Þ

where α(t) ¼ log h0(t). The Cox model is estimated by the maximization of the partial likelihood function, developed by Cox (1972).

8.4.2

Explanatory Variables

Following the discussion from the conceptual framework, we will introduce variables at global, regional, and local levels. Foreign entrants represent global-local

180

8 What Sustains Large Firms in China?

interactions and generate both competition and spillover effects for firm sustaining. We expect that competition effects are stronger at the prefecture level while spillover effects could be expected at the provincial level. To test hypothesis 1 and examine the role of the global-local interactions, we introduce the ratio of industrial output by foreign enterprises in the 3-digit industry at the prefectural city and the province where a firm is located (CFDI and PFDI). CFDI is expected to have a negative coefficient, while PFDI has a possible positive effect. Economic transition results in economic liberalization and decentralization, causing substantial variations in institutional incentives and constraints for firms to sustain in China. At the provincial level, it is expected that market potential and institutional factors affect business operation. A large market potential would reduce the intensity of market competition and create opportunity for firms to exploit scale economies. Firms would have more chance to succeed when located in provinces with good market potential. Following Harris (1954), He (2003), and He and Wang (2012), a provincial market potential (PMARKET) is defined as the sum of its interactions with each province market (including itself). These interactions are proportional to the provincial population size and inversely related to the squared distance from that distance. To measure the institutional development in Chinese provinces, we adopt the economic liberalization index (ELI) as developed and annually updated by the National Economic Research Institute of China. The index is categorized into five dimensions, including (1) government and market relations, referring to the extent to which the economy is market driven; (2) growth of non-state sectors and reforms of state-owned enterprises; (3) interregional trade barriers, referring to price controls by bureaucratic departments; (4) factor-market development; and (5) legal frameworks, including the development of intermediate institutions and legal enforcement. The final index score for each province is equal to the weighted average of fivedimensional subindices. A higher score indicates a more developed market-oriented institutional framework. This index is the only official and most reliable measure of China’s provincial institutions, widely applied in recent literature (Du et al. 2008; Li et al. 2011; Wang and Chen 2014). The liberalization index and its five dimensions are expected to have positive impacts. At the prefectural city level, we expect agglomeration externalities and local supportive policies would help to sustain businesses. We distinguish agglomeration externalities into localization and urbanization economies. In accordance with Malmberg et al. (2000), Henderson (2003), and Pe’er and Vertinsky (2008), we also measure localization economies as the total number of firms of the 3-digit industry (LOC) in the prefecture-level city where a firm is located. Following the traditional measure (Brixy and Grotz 2007; Fritsch et al. 2006; Carree et al. 2011), urbanization economies are measured as the population density in the prefecture-level city where a firm is located. Both variables are expected to have a positive effect. To test the possible impacts of local supports, we further introduce two dummies for firms with subsidies (SUBSIDY) and for firms located in industrial parks

8.4 Model Specification and Variables

181

(PARK). We apply the firm address to identify whether a firm is located in industrial parks. They are all expected to have positive impacts. In addition, we measure the fiscal and political incentives for local governments to implement pro-business policies, which enhance business success. On the one hand, fiscal decentralization has encouraged local governments to secure extrabudget revenues from land conveyance fee that is under the direct control of municipal and county governments (Lin 2007). On the other hand, political centralization encourages local governments to mobilize land development to promote economic growth (He et al. 2015). Therefore, we apply the ratio of land conveyance fee in local revenues (FLAND) to measure fiscal and political incentives to create favorable institutions for businesses. A positive relationship between the chance of firm sustaining and the ratio is expected. Finally, following existing literature, we control firm-level variables, including employment size (SIZE), dummies for foreign firms (FDI), state-owned enterprises (SOE), and exporters (EXP). We introduce dummies for all 2-digit manufacturing industries to control unobserved industrial characteristics and year dummies to control time effects. All variables are summarized in Table 8.1. Table 8.1 Definitions of explanatory variables Variables CFDI PFDI lnPMARKET lnELI lnELI1 lnELI2 lnELI3 lnELI4 lnELI5 lnLOC lnURB SUBSIDY PARK FLAND lnSIZE FDI EXP SOE INDUSTRY PROVINCE

Definitions The ratio of industrial output by foreign enterprises in the 3-digit industry at the prefectural city where an enterprise is located The ratio of industrial output by foreign enterprises in the 3-digit industry at the province where an enterprise is located The weighted market potential in the province where an enterprise is located The economic liberalization index The government and market relations dimension The economic structure dimension The interregional trade barriers dimension The factor market development dimension The legal frameworks dimension The number of firms at the 3-digit industries at the prefectural city where an enterprise is located The population density at the prefectural city where an enterprise is located Dummy variable for enterprises with subsidies Dummy variable for enterprises located in industrial parks The ratio of land conveyance fee in local revenues The employment of a firm Dummy variable for foreign firms Dummy variable for exporters Dummy variable for state-owned enterprises Dummy variable for two-digit industries Dummy variable for provinces where an enterprise is located

182

8.5

8 What Sustains Large Firms in China?

Empirical Results

The correlation analysis indicates that some explanatory variables are strongly correlated, for example, between CFDI and PFDI (0.7), among five dimensions of economic liberalization index (0.7–0.8), between LnPMARKET and lnELI (0.4). To avoid the collinearity, we introduce these strongly correlated variables in separate model specifications to test their impacts. The regression results for the full sample are reported in Table 8.2. Statistical results provide partial support for our first research hypothesis, indicating that global-local interactions have significant impacts on the chance of sustaining firms. Particularly, foreign firms have generated negative externalities at the local level, implying that firms in a prefecture-level city with more output from foreign firms are less likely to sustain. Foreign firms enjoy strong ownership advantages derived from their parent companies and benefit from institutional advantages created by the local and central governments in China. Competition effects from foreign firms heighten the hazard rate of other firms. Our finding agrees with Girma et al. (2008) and Jeon et al. (2013), which report that market competition from foreign firms is more likely to engender negative influences on local Chinese firms. PFDI has insignificantly positive impact on firm sustaining at the provincial level when CFDI is included, but it turns significant once CFDI is excluded. The result may derive from collinearity, but it still can imply that not spillover effects but competition effects beyond the boundary of prefectures are observed even though its Table 8.2 Regression results of PH Cox models for all new firms Variables CFDI PFDI lnPMARKET lnELI lnLOC lnURB SUBSIDY PARK FLAND lnSIZE FDI EXP SOE Observations INDUSTRY PROVINCE LR Chi2 Prob > Chi2

(1) 0.132** 0.062
2.379*** 0.000
0.038***
0.114***
0.163***
0.151*** 0.016
0.266***
0.193***
0.109*** 0.285*** 30,226 YES YES 2670 0

Note: ***P < 0.01; **P < 0.05; *P < 0.10

(2) 0.151***
2.379*** 0.006
0.038***
0.114***
0.164***
0.151*** 0.016
0.266***
0.194***
0.108*** 0.284*** 30,226 YES YES 2669 0

(3) 0.139*
2.380***
0.001
0.035***
0.109***
0.162***
0.151*** 0.016
0.266***
0.167***
0.105*** 0.284*** 30,231 YES YES 2670 0

(4) 0.133** 0.064
2.750***
0.041***
0.107***
0.157**
0.138***
0.023
0.270***
0.182***
0.097** 0.331*** 30,226 YES YES 2424 0

8.5 Empirical Results

183

competition effects are less important than one at the city level. In a word, the globallocal interactions are more obvious at the city level. Results further reveal that direct global linkage helps sustain firms in China. Both foreign firms and exporters are more likely to remain. This is consistent with some existing studies such as Görg and Strobl (2003) and Alvarez and Görg (2005), which report positive impacts of foreign ownership on firm survival, and Bernard et al. (2005), Pérez et al. (2004), and Kimura and Kiyota (2006), which find a positive effect of exporting on firm survival. Foreign firms and exporters in China are more competitive in the markets; meanwhile they enjoy a variety of supports from governments, leading to higher chance of firm sustaining. We find strong evidence to support the second research hypothesis. Both market potential and market-oriented institutional factors at the province level are critical to sustain local businesses in China. Provincial market potential shows its expected role to enhancing the sustaining chance of firms. Firms in provinces with larger market potential face less intensive intra-provincial competition. Typically, large market potential allows firms to take advantage of scale economies and become more competitive (Ottaviano and Pinelli 2006; Holl 2012). Interestingly, economic liberalization index (lnELI) is insignificant when market potential is included, as shown in Models 1–3 in Table 8.2. However, lnELI and its five dimensions turn to be highly significant and negative when market potential is excluded (Tables 8.2 and 8.3). On the one hand, the significance of lnELI and its five dimensions without regard to influence of market potential implies that pro-business institutions are critical for business operations in China. Results from Model 4 in Table 8.2 and Models 6–10 in Table 8.3 suggest that firms in provinces with higher economic liberalization indices have more chance to sustain. Specifically, market-driven economies with developed non-state sectors enhance the chance of firms sustaining (Models 6 and 7 in Table 8.3); less interprovincial trade barriers and less price controls by bureaucratic department help sustain businesses (Model 8); development of factor market and high mobility of production factors sustain business (Model 9); and better legal framework and contract enforcement enhance business sustaining rates (Model 10). On the other hand, the insignificance of lnELI with inclusion of market potential may result from strong correlation between lnELI and lnPMARKET, but the result can still indicate that market potential is much more important than market-oriented institution framework to sustain businesses in China. With inclusion of market potential, only lnELI5—the legal frameworks dimension—is still significantly negative, implying that on condition that the level of market potential is fixed, the development of intermediate and legal enforcement still enhances the sustaining or survival rates of new firms, because the strong enforceability of contract in a sound and consistent legal system effectively punishes rule violators and ensures a fair and orderly market competition (Wang and Chen 2014; Du et al. 2008). China has taken a gradualism approach to reform its economic system, causing substantial spatial variation in market-oriented institutions. Firms supported with more favorable institutions would encounter much lower transaction costs and less market uncertainty, ensuring business operation. The findings of positive roles of institutions in business sustaining are an important complement to the existing literature, which has systematically ignored the importance of institutions.


0.038***
0.114***
0.164***
0.149*** 0.021
0.266***
0.193***
0.110*** 0.285*** 30,226 YES YES 2671 0

(1) 0.131** 0.060
2.491*** 0.218


0.038***
0.114***
0.166***
0.157*** 0.017
0.265***
0.191***
0.109*** 0.291*** 30,222 YES YES 2671 0

0.228

(2) 0.133** 0.055
2.644***

Note: ***P < 0.01; **P < 0.05; *P < 0.10

Variables CFDI PFDI lnPMARKET lnELI1 lnELI2 lnELI3 lnELI4 lnELI5 lnLOC lnURB SUBSIDY PARK FLAND lnSIZE FDI EXP SOE Observations INDUSTRY PROVINCE LR Chi2 Prob > Chi2


0.038***
0.114***
0.163***
0.152*** 0.017
0.266***
0.193***
0.109*** 0.286*** 30,226 YES YES 2670 0

0.149

(3) 0.132** 0.062
2.448***


0.038***
0.113***
0.164***
0.153*** 0.017
0.265***
0.194***
0.109*** 0.290*** 30,226 YES YES 2672 0

0.126

(4) 0.133** 0.058
2.473***


0.405***
0.035***
0.111***
0.161**
0.153*** 0.001
0.270***
0.188***
0.112*** 0.289*** 30,226 YES YES 2692 0

(5) 0.128** 0.072
1.997***

Table 8.3 Regression results of PH Cox models for decomposition of ELI


0.056***
0.106***
0.164***
0.177***
0.041
0.256***
0.189***
0.076** 0.402*** 30,226 YES YES 2153 0


3.170***

(6) 0.154**
0.005


0.050***
0.109***
0.148**
0.098** 0.016
0.267***
0.204***
0.094** 0.324*** 30,222 YES YES 2285 0


1.632***

(7) 0.132** 0.056


0.059***
0.093***
0.167***
0.131***
0.000
0.256***
0.184***
0.075* 0.393*** 30,226 YES YES 2080 0


3.305***

(8) 0.152**
0.040


0.065***
0.112***
0.159**
0.109** 0.020
0.252***
0.183***
0.082** 0.401*** 30,226 YES YES 1864 0


1.638***

(9) 0.134**
0.035


1.767***
0.042***
0.094***
0.157**
0.158***
0.044
0.271***
0.164***
0.109*** 0.407*** 30,226 YES YES 2299 0

(10) 0.128** 0.017

184 8 What Sustains Large Firms in China?

8.5 Empirical Results

185

With the inclusion of provincial market potential and institutional factors, agglomeration economies and governmental supports at the prefectural level are found to sustain businesses. On the one hand, both LOC and URB have negative coefficients in all model specifications, indicating that firms which are able to reap from localization and urbanization economies have lower hazard rates and experience higher sustaining or survival rates, which is consistent with recent studies (Wennberg and Lindqvist 2010; Oort et al. 2012; Cainelli et al. 2014). The provincial market-oriented institutions create favorable conditions to stimulate the role of agglomeration economies at the local level. On the other hand, statistical results show that firms located in industrial parks and with subsidies are more likely to sustain, indicating that local governmental supports are critical for businesses. Facing the intensive regional competition, local governments have an incentive to serve as a helping hand to sustain local business. The helping hand can then boost local economic performance and increase local revenues. The third research hypothesis is therefore confirmed. Finally, given the impacts of geographical factors at provincial and prefectural levels, large firms are found to have better chance to sustain. This is consistent with survival studies using data from developed economies (Ericson and Pakes 1995; Mata and Portugal 1994; Wagner 1999). Size effect is often explained by the learning model of industrial dynamics in the literature. However, in China, large firms are typically favored by both local governments and can gain substantial governmental support such as subsided electricity, cheap land, and market channels. The governmental supports for large firms are additional advantage to help them to sustain in the vibrant environment. In addition, giving the market-oriented institutions into play, SOEs are more likely to fail. SOEs make profits based on their institutional advantages. Governmental support and protection discourage SOEs to invest in innovation and improve their efficiencies. SOEs are difficult to sustain in market competition. In addition, during the studying time period, some SOEs may be privatized or changed into other ownerships such as foreign owned or privately owned or mixed ownership (Nolan and Wang 1999), causing the exit of SOEs. We further explore the impact of firm ownership in business sustaining and conduct the regression for the subsamples of SOEs, foreign firms, and private firms (Tables 8.4, 8.5 and 8.6). There are similarities and differences in the determinants of firm sustaining across ownership. The results confirm that marketoriented institutions are critical for the sustaining of all firms regardless of their ownerships. Both SOEs and non-SOEs are able to take advantage of market-oriented institutions in their business operation. The magnitude of the coefficients on LnELI in the models of foreign and private firms is much larger than that in the model of SOEs, implying that non-SOEs are more dependent on market-oriented institutional development to sustain. Beyond that, there are several other interesting observations. First, SOEs are very different from foreign and private firms in their behavior. They are not challenged by the entry of foreign firms and are not incentivized to reap the benefits from agglomeration economies and market potential. Larger SOEs are more likely to sustain in the markets since they enjoy political, institutional, and economic advantages in China. Large SOEs are often strongly favored by governments at all


1.648***


0.082
0.222
0.104 0.213
0.145
0.294*** 0.009 550 YES YES 145.6 4.12e-08


0.080
0.223
0.103 0.207
0.163
0.296*** 0.018 550 YES YES 145.8 5.96e-08

(2) 0.197 0.516

(1) 0.183 0.525 0.382
2.072*

Note: ***P < 0.01; **P < 0.05; *P < 0.10

Variables CFDI PFDI lnPMARKET lnELI lnELI1 lnELI2 lnELI3 lnELI4 lnELI5 lnLOC lnURB SUBSIDY PARK FLAND lnSIZE EXP Observations INDUSTRY PROVINCE LR Chi2 Prob > Chi2

Table 8.4 Regression results of PH Cox models for SOEs


0.090
0.209
0.127 0.073
0.290
0.283***
0.062 550 YES YES 135.6 6.86e-07


1.100**

(3) 0.235 0.362


0.086
0.207
0.111 0.211
0.161
0.298***
0.019 550 YES YES 142.6 9.80e-08


0.818***

(4) 0.219 0.453


0.093
0.217
0.130 0.200
0.173
0.288*** 0.007 550 YES YES 141 1.53e-07


1.765***

(5) 0.183 0.455


0.092
0.217
0.135 0.099
0.111
0.287*** 0.004 550 YES YES 139.6 9.92e-08


0.897***

(6) 0.144 0.426


1.339***
0.072
0.225
0.099 0.167
0.188
0.299*** 0.009 550 YES YES 148.6 1.72e-08

(7) 0.234 0.517

186 8 What Sustains Large Firms in China?


2.699***


0.037
0.169**
0.187
0.066
0.026
0.337***
0.030 4936 YES YES 397.3 0


0.031
0.176**
0.204
0.074
0.013
0.332***
0.047 4936 YES YES 418.3 0

(2) 0.170 0.284

(1) 0.169 0.270
2.058***
0.284

Note: ***P < 0.01; **P < 0.05; *P < 0.10

Variables CFDI PFDI lnPMARKET lnELI lnELI1 lnELI2 lnELI3 lnELI4 lnELI5 lnLOC lnURB SUBSIDY PARK FLAND lnSIZE EXP Observations INDUSTRY PROVINCE LR Chi2 Prob > Chi2
0.046
0.164**
0.210
0.099
0.080
0.331***
0.034 4936 YES YES 383 0


3.644***

(3) 0.161 0.220

Table 8.5 Regression results of PH Cox models for foreign firms


0.041
0.170**
0.205
0.022 0.001
0.334***
0.026 4936 YES YES 375.7 0


1.668***

(4) 0.147 0.312


0.049*
0.164**
0.222
0.047 0.002
0.326***
0.018 4936 YES YES 371.9 0


3.479***

(5) 0.194 0.217


0.071**
0.148*
0.186
0.018 0.002
0.336***
0.019 4936 YES YES 311 0


1.665***

(6) 0.134 0.259


1.787***
0.045
0.152*
0.155
0.092
0.041
0.338***
0.045 4936 YES YES 405.7 0

(7) 0.161 0.246

8.5 Empirical Results 187


2.880***


0.032**
0.123***
0.104
0.068 0.012
0.307***
0.120** 16,659 YES YES 1196 0


0.029*
0.111***
0.106
0.106 0.056*
0.299***
0.133** 16,659 YES YES 1371 0

(2) 0.143 0.011

(1) 0.132 0.002
2.967*** 0.622**

Note: ***P < 0.01; **P < 0.05; *P < 0.10

Variables CFDI PFDI lnPMARKET lnELI lnELI1 lnELI2 lnELI3 lnELI4 lnELI5 lnLOC lnURB SUBSIDY PARK FLAND lnSIZE EXP Observations INDUSTRY PROVINCE LR Chi2 Prob > Chi2
0.047***
0.109***
0.094
0.121* 0.002
0.297***
0.077 16,659 YES YES 1103 0


3.385***

(3) 0.146
0.035

Table 8.6 Regression results of PH Cox models for private firms


0.043***
0.118***
0.089
0.022 0.053
0.302***
0.108* 16,659 YES YES 1131 0


1.730***

(4) 0.135
0.012


0.050***
0.115***
0.105
0.059 0.054
0.294***
0.090 16,659 YES YES 1016 0


3.404***

(5) 0.170*
0.099


0.058***
0.119***
0.101
0.050 0.053
0.292***
0.089 16,659 YES YES 937.1 0


1.748***

(6) 0.156*
0.089


1.845***
0.033**
0.121***
0.102
0.085 0.003
0.304***
0.140** 16,659 YES YES 1135 0

(7) 0.154*
0.026

188 8 What Sustains Large Firms in China?

8.6 Conclusion and Implications

189

levels since they contribute to economic growth and revenues substantially. Second, foreign firms clearly benefit from market potential and agglomeration economies to enhance their chance to sustain. Foreign firms are not challenged by the entry of other foreign firms, suggesting that foreign firms may not compete with each other at the local level but compete for markets at the provincial level. This is consistent with the findings in locational studies of foreign firms in China (He 2003). Size matters for foreign firms, which can be interpreted as the learning effects. Third, private firms are the hardest to sustain in China. They are most likely to be crowded out by the entry of foreign firms. Their success highly depends on market-oriented institutions and their capabilities to reap from market potential and agglomeration economies. Fourth, statistical results strongly suggest that firm ownership is a critical factor for business operation in China. Ownership influences the chance of firm sustaining by creating different incentives and capabilities to utilize agglomeration externalities and reap from market-oriented institutions.

8.6

Conclusion and Implications

Transitional economies create unique institutional frameworks to sustain businesses, which face both intensive market competition and institutional uncertainties. Taking China as a case, this study made the first effort to investigate firm sustaining in a large transitional economy. We argue that firm sustaining in China is better understood in the context of economic transition, which has created a regionally decentralized authoritarian (RDA) system. The RDA system has generated both economic and political incentives to create pro-growth and pro-business regimes by utilizing market forces and by directly supporting businesses. Considering the nature of transitional economies, we particularly stress the importance of institutional factors and geographical factors, providing an important complement to the existing literature, which has systematically ignored the role of institutions and typically focuses on regional factors at only one geographical scale. Results from the Cox proportional hazard models largely confirm our hypotheses. We find that global-local interactions impose significant challenges to local firms, with strong competition effects at the prefecture level but no spillover effects at the province from the entry of foreign firms. Market-oriented institutions are critical for business operation in China. Non-SOEs however are more responsive to institutions than SOEs. Provincial market potential is also found to help sustain local businesses. The findings imply that as the most powerful subnational governments, provincial governments in China can sustain businesses in their jurisdiction by actively creating market-oriented institutions and by facilitating market integration across provinces. At the prefecture level, we find that firms able to reap from agglomeration economies and governmental supports are more likely to sustain. These findings provide empirical justification that local governments create a variety of industrial parks by concentrating firms in a certain geographical area. The few studies (Konings and Xavier 2002; Masso et al. 2007; Männasoo 2008) about firm dynamics in transitional

190

8 What Sustains Large Firms in China?

economies have not discovered major transition-specific anomalies, except relatively low firm survival rates in the early years of economic transition. Those studies however typically focus on firm- and industry-specific factors contributing to firm survival. Our empirical study shows that the spatially gradual market transition is critical to business operation. The interactions of institutions and market forces at different geographical scales can further predict the prospects for firm dynamics. The investigation of firm dynamics in transitional economies like China contributes importantly to the literature by brining institutional and geographical factors. Our findings confirm that both market forces and state power can play a substantial role in sustaining business in transitional economies. Our findings have important policy implications. First, efforts have to be made to reduce the unfair competition between foreign and domestic firms and prosper the spillover effects of foreign firms beyond the local boundaries and within the boundaries. Second, there are substantial variations in market-oriented institutions, and many provinces need to further reform their economic systems by creating pro-market institutional frameworks. Third, local governments can further pursue industrial specialization and nurture industrial clusters to help business success. This study however is not without limitations. Our data only cover SOEs and non-SOEs with sales revenues greater than five million RMB, which are relatively large and may enjoy favorable treatments. The focus on large firms has the risk to generate biased results. Moreover, our definition of firm exits may include cases in which firms are merged into other firms and firms fail to meet the criteria to be included in the dataset. However, those cases can indicate business failure. This dataset however is the only available comprehensive dataset to examine the very important topic of firm dynamics in China. Future research can be extended in several directions. First, it is urgent to build a full sample to include both large and small firms to systematically examine firm dynamics in China. Second, more works should be done to build link between regional decentralized authoritarian system and firm dynamics. For instance, the question of how fiscal incentives and political incentives help sustain businesses remains unanswered. Third, it will be beneficial to combine the institutional perspective and the evolutionary economic geography to understand China’s firm dynamics.

References Acs, Z. J., Armington, C., & Zhang, T. (2007). The determinants of new-firm survival across regional economies: The role of human capital stock and knowledge Spillover. Papers in Regional Science, 86(3), 367–391. Agarwal, R., & Audretsch, D. B. (2001). Does entry size matter? The impact of the life cycle and technology on firm survival. Journal of Industrial Economics, 49(1), 21–43. Agarwal, R., & Gort, M. (2002). Firm and product life cycles and firm survival. American Economic Review, 92(2), 184–190. Aitken, B. J., & Harrison, A. E. (1999). Do domestic firms benefit from direct foreign investment? Evidence from Venezuela. American Economic Review, 89(3), 605–618.

References

191

Alvarez, R., & Görg, H. (2005). Multinationals and plant exit: Evidence from Chile. International Review of Economics & Finance, 18(1), 45–51. Audretsch, D. B., & Mahmood, T. (1995). New firm survival: New results using a hazard function. Review of Economics and Statistics, 77(1), 97–103. Audretsch, D. B., (1995). Innovation, growth and survival. Int J Ind Org, 13(4):441–457 Audretsch, D. B., Houweling, P., & Thurik, A. R. (2000). Firm survival in the Netherlands. Review of Industrial Organization, 16(1), 1–11. Backer, D. K., & Sleuwaegen, L. (2003). Does foreign direct investment crowd out domestic entrepreneurship? Review of Industrial Organization, 22(1), 67–84. Bai, C. E., Du, Y., Tao, Z., & Tong, S. Y. (2004). Local protectionism and regional specialization: Evidence from China’s industries. Journal of International Economics, 63(2), 397–417. Bain, J. S. (1956). Barriers to new competition. Cambridge, MA: Harvard University Press. Barbieri, E., Tommaso, D., Marco, R., & Bonnini, S. (2012). Industrial development policies and performance in southern China: Beyond the specialized industrial cluster program. China Economic Review, 23(3), 613–625. Bernard, A. B., Jensen, J. B., & Schott, P. K. (2005). Assessing the impact of trade liberalization on importcompetingindustries in the Appalachian Region. Washington: Appalachian Regional Commission. Blanchard, O., & Shleifer, A. (2001). Federalism with and without political centralization: China versus Russia. IMF Staff Papers, 48(1), 171–179. Blomström, M., & Kokko, A. (1998). Multinational corporations and spillovers. Journal of Economic Surveys, 12(3), 247–277. Blomström, M., Kokko, A., & Zejan, M. (2000). Foreign direct investment: Firm and host country strategies. London: Macmillan. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Brixy, U., & Grotz, R. (2007). Regional patterns and determinants of birth and survival of new firms in Western Germany. Entrepreneurship and Regional Development, 19(4), 293–312. Cainelli, G., Montresor, S., & Marzetti, G. V. (2014). Spatial agglomeration and firm exit: A spatial dynamic analysis for Italian provinces. Small Business Economics, 43(1), 213–228. Carree, M. A., Verheul, I., & Santarelli, E. (2011). Sectoral patterns of firm exit in Italian provinces. Journal of Evolutionary Economics, 21(3), 499–517. Caves, R. E. (1974). Multinational firms, competition and productivity in host country markets. Economica, 41(162), 176–193. Cefis, E., & Marsili, O. (2005). A matter of life and death: Innovation and firm survival. Industrial and Corporate Change, 14(6), 1–26. Chang, S. J., & Xu, D. (2008). Spillovers and competition among foreign and local firms in China. Strategic Management Journal, 29(5), 495–518. Chen, C., Chang, L., & Zhang, Y. (1995). The role of foreign direct investment in China’s Post1978 economic development. World Development, 23(4), 691–703. Cheung, K. Y., & Lin, P. (2004). Spillover effects of FDI on innovation in China: Evidence from the provincial data. China Economic Review, 15(1), 25–44. Cox, D. R. (1972). Regression models and life-tables. Journal of the Royal Statistical Society, 34 (2), 187–220. Dawar, N., & Frost, T. (1999). Competing with giants: Survival strategies for local companies in emerging markets. Harvard Business Review, 77(2), 119–129. Delgado, M., Porter, M. E., & Stern, S. (2010). Clusters and entrepreneurship. Journal of Economic Geography, 10(10), 495–518. Du, J., Lu, Y., & Tao, Z. (2008). Economic institutions and FDI location choice: Evidence from US multinationals in China. Journal of Comparative Economics, 36(3), 412–429.

192

8 What Sustains Large Firms in China?

Duranton, G., & Puga, D. (2004). Micro-foundations of urban agglomeration economies. Handbook of Regional and Urban Economics, 4(4), 2063–2117. Ericson, R., & Pakes, A. (1995). Markov-perfect industry dynamics: A framework for empirical work. Review of Economic Studies, 62(1), 53–82. Esteve-Pérez, S., Mánez-Castillejo, J. A., Rochina-Barrachina, M. E., Rochina-Barrachina, M. E., & Sanchis-Llopis, J. A. (2007). A survival analysis of manufacturing firms in export markets. In J. M. Arauzo-Carod & M. C. Manjon-Antolin (Eds.), Entrepreneurship, industrial location and economic growth. Cheltenham: Elgar. Estrin, S., & Prevezer, M. (2010). A survey on institutions and new firm entry: How and why do entry rates differ in emerging markets? Economic Systems, 34(3), 289–308. Fontana, R., & Nesta, L. (2009). Product innovation and survival in a high-tech industry. Review of Industrial Organization, 34(4), 287–306. Fotopoulos, G., & Spence, N. (2001). Regional variations of firm births, deaths and growth patterns in the UK, 1980–1991. Growth and Change, 32(2), 151–173. Fritsch, M., Brixy, U., & Falck, O. (2006). The effect of industry, region and time on new business survival-a multi-dimensional analysis. Review of Industrial Organization, 28(3), 285–306. Girma, S., Gong, Y., & Görg, H. (2008). Foreign direct investment, access to finance, and innovation activity in Chinese enterprises. The World Bank Economic Review, 22(2), 367–382. Görg, H., & Strobl, E. (2003). Multinational companies, technology spillovers and plant survival. Scandinavian Journal of Economics, 105(4), 581–595. Gustafsson, B., & Shi, L. (2002). Income inequality within and across counties in rural China 1988 and 1995. Journal of Development Economics, 69(1), 179–204. Han, S. S., & Pannell, C. W. (1999). The geography of privatization in China, 1978–1996. Economic Geography, 75(3), 272–296. Hannan, M. T., & Freeman, J. (1988). The ecology of organizational mortality: American labor unions, 1836–1985. American Journal of Sociology, 94, 25–52. Harris, C. D. (1954). The market as a factor in the localization of industry in the United States. Annals of the Association of American Geographers, 44(4), 315–348. He, C. (2002). Information costs, agglomeration economies and the location of foreign direct Investment in China. Regional Studies, 36(9), 1029–1036. He, C. (2003). Entry mode and location of foreign manufacturing Enterprises in China. Eurasian Geography and Economics, 44(6), 399–417. He, C., & Wang, J. (2012). Regional and sectoral differences in the spatial restructuring of Chinese manufacturing industries during the post-Wto period. GeoJournal, 77(3), 361–381. He, C., & Yang, R. (2016). Determinants of firm failure: Empirical evidence from China. Growth and Change, 47(1), 72–92. He, C., Wei, Y., & Pan, F. (2007). Geographical concentration of Industries in China: The importance of spatial and industrial scale. Eurasian Geography and Economics, 48(5), 603–625. He, C., Zhou, Y., & Huang, Z. (2015). Fiscal decentralization, political centralization and land urbanization in China. Urban Geography, 37(3), 436–457. Henderson, J. V. (2003). Marshall’s scale economies. Journal of Urban Economics, 53(1), 1–28. Holl, A. (2012). Market potential and firm-level productivity in Spain. Journal of Economic Geography, 12(6), 1191–1215. Jeon, Y., Park, B., & Ghauri, P. N. (2013). Foreign direct investment effects in China: Are they different across industries with different technological levels? China Economic Review, 26(1), 105–117. Keller, W. (2002). Geographical localization and international technology diffusion. The American Economic Review, 92(1), 120–142. Kim, W. C. (1988). The effects of competition and corporate political responsiveness on multinational bargaining power. Strategic Management Journal, 9(3), 289–295. Kimura, F., & Fujii, T. (2003). Globalizing activities and the rate of survival: Panel data analysis on Japanese firms. Journal of the Japanese and International Economies, 17(4), 538–560.

References

193

Kimura, F., & Kiyota, K. (2006). Exports, FDI, and productivity: Dynamic evidence from Japanese firms. Review of World Economics, 142(4), 695–719. Konings, J. (2001). The effects of foreign direct investment on domestic firms: Evidence from firmlevel panel data in emerging economics. Economics of Transition, 9(3), 619–633. Konings, J., & Xavier, A. (2002). Firm growth and survival in a transition country: Micro evidence from Slovenia (p. 114). LICOS Center for Institutions and Economic Performance Discussion Paper. Li, K., Wang, T., Cheung, Y., & Liang, P. (2011). Privatization and risk sharing: Evidence from the Split share structure reform in China. Review of Financial Studies, 24(7), 2499–2525. Li, H., & Zhou, L. (2005). Political turnover and economic performance: The incentive role of personnel control in China. Journal of Public Economics, 89(9):1743–1762. Lin, J. Y. (1992). Rural reform and agricultural growth in China. American Economic Review, 82 (1), 34–52. Lin, G. C. S. (2007). Reproducing spaces of Chinese urbanization: New city-based and landcentered urban transformation. Urban Studies, 44(9), 1827–1855. Lin, S. H., Chen, Y. J., Wang, T. L., Wang, H. C., & Liu, Y. C. A. (2012). The decision behavior of Taiwan firms investing in China: Evidence from different industries. Review of Pacific Basin Financial Markets and Policies, 15(01), 2–27. Liu, Z. (2008). Foreign direct investment and technology Spillovers: Theory and evidence. Journal of Development Economics, 85, 176–193. Liu, X., Burridge, P., & Sinclair, P. J. (2002). Relationships between economic growth, foreign direct investment and trade: Evidence from China. Applied Economics, 34(11), 1433–1440. Luo, Y. (1997). Partner selection and venturing success: The case of joint ventures with firms in the People’s Republic of China. Organization Science, 8(6), 648–662. Malmberg, A., Malmberg, B., & Lundequist, P. (2000). Agglomeration and firm performance: Economies of scale, localisation, and urbanisation among Swedish export firms. Environment and Planning A, 32(2), 305–321. Männasoo, K. (2008). Patterns of firm survival in Estonia. Eastern European Economics, 46(4), 27–42. Masso, J., Eamets, R., & Philips, K. (2007). Creative destruction and transition: Evidence on firm demographics from Estonia. In J. M. Arauzo-Carod & M. Manjon-Antolin (Eds.), Entrepreneurship, industrial location and economic growth (pp. 81–103). Cheltenham: Edward Elgar. Mata, J., & Portugal, P. (1994). Life duration of new firms. Journal of Industrial Economics, 42(3), 227–245. Mocnik, D. (2010). Determinants of firm entries: Empirical evidence for Slovenia. Economic and Business Review, 12(2), 129–145. Montinola, G., Qian, Y., & Weingast, B. R. (1995). Federalism, Chinese style: The political basis for economic success in China. World Politics, 48(1), 50–81. Neffke, F. M. H., Henning, M., & Boschma, R. (2012). The impact of aging and technological relatedness on agglomeration externalities: A survival analysis. Journal of Economic Geography, 12(2), 485–517. Nolan, P., & Wang, X. (1999). Beyond privatization: Institutional innovation and growth in China’s large state-owned enterprises. World Development, 27(1), 169–200. North, D. C. (1990). Institutions, institutional change and economic performance. Cambridge: Cambridge University Press. Oi, J. C. (1992). Fiscal reform and the economic foundations of local state corporatism in China. World Politics, 45(1), 99–126. Oort, F. G. V., Burger, M. J., Knoben, J., & Raspe, O. (2012). Multilevel approaches and the firmagglomeration ambiguity in economic growth studies. Journal of Economic Surveys, 26(3), 468–491. Ottaviano, G. I. P., & Pinelli, D. (2006). Market potential and productivity: Evidence from Finnish regions. Regional Science and Urban Economics, 36(5), 636–657. Pan, Z., & Zhang, F. (2002). Urban productivity in China. Urban Studies, 39(12), 2267–2281.

194

8 What Sustains Large Firms in China?

Pe’er, A., & Vertinsky, I. (2008). Firm exits as a determinant of new entry: Is there evidence of local creative destruction? Journal of Business Venturing, 23(3), 280–306. Pérez, S., Esteve, L., Sanchis, A., & Llopis, J. A. S. (2004). The determinants of survival of Spanish manufacturing firms. Review of Industrial Organization, 25(3), 251–273. Qian, Y. (2000). The process of China’s market transition 1978–98: The evolutionary, historical, and comparative perspectives. Journal of Institutional and Theoretical Economics, 156(1), 151–171. Siegfried, J. J., & Evans, L. B. (1994). Empirical studies of entry and exit: A survey of the evidence. Review of Industrial Organization, 9(2), 121–155. Silva, D. G., & McComb, R. P. (2012). Geographic concentration and high tech firm survival. Regional Science and Urban Economics, 42(4), 691–701. Strotmann, H. (2007). Entrepreneurial survival. Small Business Economics, 28(1), 87–104. Sun, Q., Tong, W., & Yu, Q. (2002). Determinants of foreign direct investment across China. Journal of International Money and Finance, 21(1), 79–113. Tybout, J. R. (2000). Manufacturing firms in developing countries: How well do they do, and why? Journal of Economic Literature, 38(1), 11–44. Wagner, J. (1999). The life history of cohorts of exits from German manufacturing. Small Business Economics, 13(1), 71–79. Wagner, S., & Cockburn, I. (2010). Patents and the survival of internet-related IPOs. Research Policy, 39(2), 214–228. Wang, T., & Chen, Y. (2014). FDI, institutional development and environmental externalities: Evidence from China. Journal of Environmental Management, 135(4), 81–90. Wei, Y. (2002). Beyond the Sunan model: Trajectory and underlying factors of development in Kunshan, China. Environment and Planning A, 34(10), 1725–1747. Wennberg, K., & Lindqvist, G. (2010). The effect of clusters on the survival and performance of new firms. Small Business Economics, 34(3), 221–241. Wong, C. P. W. (2000). Central-local relations revisited: The 1994 tax sharing reform and public expenditure management in China. China Perspectives, 2000(31), 52–63. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yang, Y. R., & Hsia, C. (2007). Spatial clustering and organizational dynamics of transborder production networks: A case study of Taiwanese information technology companies in the Greater Suzhou Area, China. Environment and Planning A, 39(6), 1346–1363. Yao, S., & Zhang, Z. (2001). On regional inequality and diverging clubs: A case study of contemporary China. Journal of Comparative Economics, 29(3), 466–484. Zhang, L. Y. (1999). Chinese central-provincial fiscal relationships, budgetary decline and the impact of the 1994 fiscal reform: An evaluation. The China Quarterly, 157(157), 115–141.

Chapter 9

How Do Agglomeration Economies Contribute to Firm Survival in China?

9.1

Introduction

New firms are often seen as the catalyst of innovation and employment growth and are acknowledged for their role in promoting regional competitiveness. Yet new firms face a number of factors, or liabilities, including a lack of sufficient resources, higher vulnerability to external shocks, and greater likelihood to operate farther from the industry’s minimum efficient scale leading to cost disadvantages (Schutjens and Wever 2000). Due to such liabilities, less than half of all new start-ups are expected to survive more than 5 years after entry, regardless of country contexts (Cefis and Marsili 2011). Since their premature exit results in a loss of economic growth opportunities, understanding the factors that influence the survival chances of new firms is essential to promoting a vibrant economy. A well-developed body of literature rooted heavily in the tradition of industrial organization and strategic management attempts to reveal the key factors that may help mitigate new firms’ initial liabilities and improve their survival chances (Dunne et al. 1989). In the traditional survival literature, however, the regional dimension has been largely excluded from empirical analyses in spite of the fact that new firms tend to locate to existing clusters in order to reduce costs associated with learning about the local business environment (Stuart and Sorenson 2003). To incorporate a spatial perspective into the literature, economic geographers have increasingly focused their attention on studying the mechanisms of agglomeration and how externalities influence firm survival.1 A growing number of empirical

Modified article originally published in [Howell, A., He, C., Yang, R., and Fan, C. C. (2018) Agglomeration, (un)-related variety and new firm survival in China: Do local subsidies matter?. Papers in Regional Science, 97: 485–500.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. 1

For a good overview of the literature, see Frenken et al. (2015).

© Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_9

195

196

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

studies now exist that study the effects of spatial externalities on firm survival in countries ranging from Greece (Fotopoulos and Louri 2000) to the USA (Acs et al. 2007), Germany (Fritsch et al. 2006; Falck 2007), the UK (Boschma and Wenting 2007), and China (Howell 2015; He and Yang 2016). The findings from these studies, however, are mixed and sometimes directly contrast one another. Fotopoulos and Louri (2000), for instance, show that localization economies resulting from specialization positively influence new firm survival in Greece, while Folta et al. (2006) find that specialization reduces the survival chances of US firms. It is also common to observe mixed outcomes within the same study on firm survival depending on the dimension of agglomeration under scrutiny. In the US context, Acs et al. (2007) show that city size and diversity are important determinants of firm survival, while specialization is negatively related to the firm’s survival chances. Similarly in the German context, Staber (2001) studies the knitwear industry and finds that localization economies led to higher failure rates, while diversification reduced failure rates. In China, He and Yang (2016) find that specialization decreases the survival chances of Chinese firms, although the opposite effect is found for urbanization economies. It is clear from the aforementioned studies that the effects of agglomeration are country specific and depend on the type of agglomeration. While existing studies have contributed a great deal of insight into the nature of the relationship between agglomeration and firm survival, comparatively few studies exist that systematically take into account multiple dimensions of agglomeration on new firm survival in transitioning economy contexts. Moreover, while many existing studies show that the effect of agglomeration depends on firms’ characteristics, most studies focus on dimensions of the firm related to its age, size, and productivity or whether it is a single- or multiplant firm. By contrast, it is not very clear if access to state support, an important dimension of firm heterogeneity in China, influences the ability of new firms to benefit from different types of externalities. The focus of this paper, therefore, is to study the relationship between five different dimensions of agglomeration-specialization, city size, diversity, related variety and unrelated variety, and new firm survival. Considering the Chinese context, we also consider the effects of local state support (i.e., public subsidies) as a potential source of firm heterogeneity with respect to external economies (correcting for sample selection). The empirical strategy relies on a large sample of more than 135,000 new entrepreneurial firms in Chinese manufacturing from 1998 to 2007. In the next section, we provide a brief overview of the relevant literature. Section 9.3 introduces the data and key variables. Section 9.4 discusses the model estimation strategy. The empirical findings are presented in Sect. 9.5, and Sect. 9.6 concludes.

9.2 Literature Review

9.2

197

Literature Review

At least since Marshall (1890), the spatial concentration of economic activity is viewed as leading to performance-enhancing spillovers that are good for innovation, growth, and economic development. At the microlevel, firms that colocate together are expected to benefit from superior access to knowledge and cost-saving spillovers, which enables them to attain a comparative advantage in the market and survive longer (Tallman et al. 2004). However, it is recognized that if firms are unable to sufficiently exploit local externalities, they may become exposed to higher risks of failure due to extensive competition or having to pay higher rents typically associated with larger cities (Stuart and Sorenson 2003; Folta et al. 2006). Whether a firm benefits from or is harmed by the effects of agglomeration depends, in part, on the dimension of agglomeration. In the literature, agglomerations are generally distinguished into three different dimensions: localization economies, urbanization economies, and Jacobs externalities. Building on the early work of Marshall, localization economies are thought to arise from the spatial concentration of firms within the same industry. Firms potentially derive positive benefits resulting from a large pool of specialized labor, supplier-buyer linkages, and knowledge spillovers. Urbanization economies are associated with large cities, where all firms may potentially benefit from greater access to larger markets, higher-quality local amenities, better infrastructure, and public institutions, such as universities and research institutes. Jacobs (1969) externalities are thought to arise as firms benefit from many different industries within a region. Local industrial diversity is expected to lead to the combination of new ideas and more radical types of innovation due to spillovers that take place between industries (Glaeser et al. 1992). More recently, Jacobs externalities have been further decomposed into related variety and unrelated variety (Boschma and Wenting 2007). Viewed from the perspective of portfolio diversification strategy, as a region attracts a larger number of unrelated industries, it will become better protected from external demand shocks that may have a devastating impact on one or a few sectors (Frenken et al. 2007). The notion of related variety takes into account the role of cognitive or technological relatedness between industries (Boschma et al. 2012), whereby colocated firms in different but similar industries can more easily communicate with one another, which helps facilitate between-industry knowledge spillovers. From an evolutionary perspective, regions that develop and evolve into new related areas of economic activity, a process termed “regional branching,” are associated with higher economic growth (Neffke et al. 2011). Proponents of regional branching argue that policy support should be directed toward attracting and developing new industries that share a close technological proximity to the existing local industrial mix in order to facilitate knowledge spillovers, as opposed to supporting

198

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

leading industries that are already doing well. At the firm level, Howell et al. (2016) find that relatedness positively influences firms’ performance, although the effects are not the same for all firms. The authors find that technological proximity has a stronger positive effect on higher productive firms in China, leading to asymmetrical benefits depending on the firm’s own productive capabilities. As in other agglomeration studies, the findings in Howell et al. (2016) point to the importance of taking into account the role of firm heterogeneity in studying the effects of related variety on firm performance outcomes. In China, state subsidies are an important source of heterogeneity that may influence the incentives and ability of firms to interact with their local environment. The next section turns to a discussion of agglomeration and state interventionist policies in the Chinese context.

9.2.1

Agglomeration and the Role of State Support in China

In China, the central and local governments play a primary role in steering the direction, intensity, and location of economic activities. For instance, central and local governments have implemented various location-based policies to encourage firms to concentrate together in special economic zones (SEZs), industrial districts, and high-tech zones. Location policies have been relatively successful in China, playing an important role in accelerating its economic growth (Fan and Scott 2003). Hu et al. (2015), for instance, estimate that there have been more than 100 clusters designated by the state in over 60 cities since 1995, contributing 14% to China’s productivity growth between 1998 and 2007. Coinciding with the rise of clustering policies in China, the opening up of its economy to multinationals has led to intense local competition between domestic Chinese firms and foreign enterprises. Most Chinese domestic firms are technological laggards and cannot compete with the new foreign entrants, leading to their early exit from the market. The premature exit of Chinese firms due to unfair competition effects harms the Chinese economy and stifles opportunities for technological catchup and product upgrading. To help mitigate these negative selection effects, central and local governments have increasingly relied on place-based policy supports directed toward firms operating in SEZs, industrial districts, and high-tech zones in order to protect local profits and preserve state revenues (Howell 2016). Such policies include tax incentives, public subsidies, free or low-cost loans, subsidized energy, subsidized raw materials, and land and technology. He and Yang (2016), for instance, find that state support programs (proxied by a dummy variable for whether or not a firm received public subsidies) directly improve the chances of firm survival in China, indicating that subsidies are an effective tool to help buffer domestic Chinese firms from negative competition effects with foreign enterprises. Beyond their direct effects, however, the authors do not consider whether subsidies may also indirectly influence firms’ survival chances and subsequent performance depending on the local industrial structure.

9.3 Data and Variable Development

199

On the one hand, Chinese policy-makers channel public subsidies toward firms operating in agglomerated areas in the hopes of mitigating the negative selection effect so that domestic firms can stay in the market long enough to hopefully take advantage of positive externalities. Firms that otherwise would have exited the market may instead benefit from new sources of ideas and knowledge that are expected to spill over within the same industry or between (related) industries and, in turn, spur technological catch-up and product upgrading. On the other hand, firms that receive public financial support may become overreliant on the state for its survival, thereby lacking the incentives to pursue profit-maximizing strategies. In turn, subsidy-receiving firms may not have the appropriate incentives to seek out and benefit from place-based economies and may instead be exposed to increased risks of exit due to negative selection effects, thus necessitating perpetual future support from the state or face inevitable exit.

9.3

Data and Variable Development

Our study comprises more than 135,000 entrepreneurial firms within the first 5 years of operation. Besides their obvious policy importance, we focus on new firms to avoid capturing life cycle effects where firms after a certain age tend to be less likely to seek out external knowledge (Audretsch and Lehmann 2005). In addition, a clear benefit of focusing on new firms is that they are less constrained by previous decisions, that is, past capital installments, thereby reducing concerns of endogeneity. Our data are obtained from the Annual Report of Industrial Enterprise Statistics compiled by the State Statistical Bureau of China for the years 1998 through 2007. The data include all firms with sales above five million Renminbi (approximately US $600,000) and contain extensive information on productivity, sales, employment, geographic location, industry affiliation, and so forth. In total, our data captures over 90% of productivity in China. We build a panel by linking firms over time using the firm’s name, industry, address, etc. to assign unique numerical IDs.

9.3.1

Dependent Variable: Measuring and Interpreting Firm Survival

Firms’ survival span is developed using information on firm exit, along with entry and duration, which is obtained based on the firm’s unique numerical ID. The entry year of the firm is identified for the first year, t, that the firm is observed but not in any years prior to t. The exit year of the firm is defined as the last year, t, that the firm reported information but not in the year t + 1, t + 2, . . ., 2007. The duration of a firm is defined by counting the number of years the firm is in operation, excluding its initial year of operation.

200

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

Fig. 9.1 Average 3-year survival rates, 1998–2007

Firms may exit the sample for several reasons, including foreclosure, restructuring, or merger and acquisition (M&A), or the firm simply drops below the minimum sales threshold. Due to the minimum sales threshold, we cannot make the strong claim that firm exit is equated with firm foreclosure. Nevertheless, we do define firm exit as an indicator of serious financial distress leading to firm exit. We make this claim for three reasons. First, the minimum sales threshold is not strictly enforced. Over the time period of analysis, 5% of privately owned firms reported sales below the minimum threshold. This is good because it decreases the chances that the firm exits the sample solely due to dropping below the threshold. Second, we remove all firms that enter and exit the survey in the same year since these firms are most likely to be hovering around the sales threshold. Third, we assign new IDs to firms that undergo restructuring, merger, or acquisition, when possible. The fraction of firms in a year that can be linked to a firm in the previous year ranges from 84.5% in the years 1998–1999 up to 92.2%in the final 2 years (2006–2007). Overall, 95.9% of all year-to-year matches are constructed using firm IDs and 4.1% using other information on the firm. Figure 9.1 plots the average 3-year survival probabilities (in quartiles) across 333 cities for all firms from 1998 to 2007. Regions with higher survival probabilities tend to be quite diffused along the coastal regions of China. The 3-year average survival probabilities in large cities, such as Beijing, Tianjin, and Shenzhen, are lower than in the areas in and around Shanghai and Chongqing. In the western parts of the country, the average survival rates of the firm tend to be located along the bottom quartile at under 80%.

9.3 Data and Variable Development

9.3.2

201

Agglomeration Measures

We develop several proxies to measure different types of agglomerations: (i) external economies that arise from the spatial concentration of firms in the same industry, the so-called localization economies (LOC) (Glaeser et al. 1992); (ii) external economies arising from the urban size, population, and economic density, the so-called urbanization economies (city size); (iii) external economies that arise due to the spatial concentration of diverse industries (diversity) (Jacobs 1969); (iv) external economies that arise from the spatial concentration of different but related industries (related variety); and (v) external economies that arise from the spatial concentration of different and non-complimentary industries (unrelated variety) (Frenken et al. 2007). Following Delgado et al. (2010), we use location quotients of employment at the 3-digit manufacturing sector level to proxy for localization economies. We use labor density to control for urbanization economies typically found in larger cities. To proxy for industrial diversity, we use the total employment in all manufacturing sectors excluding the firm’s own sector for each year across all cities. We next distinguish between related variety and unrelated variety. Some earlier studies defined related industries as any subsectors that belong to the same sector (Frenken et al. 2007; Boschma and Iammarino 2009). This strategy, however, ignores the potential technological relatedness that may exist across subsectors (Essletzbichler 2013). Instead, we calculate similarity between subsectors on the basis of their shared use of input factors. The input mix reflects production technology implying that two subsectors that share similar input mixes also share close technological proximity (Frenken et al. 2007). We use China’s 2002 input-output tables with 122 sectors to capture similarity between two sector’s input mixes. Following Los (2000), we use the cosine distance to measure technological relatedness, defined as the cosine between a pair of input coefficient vectors: P k αmk  α jk wmj ¼ qffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffiffi P 2 P 2 ffi k αmk  k α jk

ð9:1Þ

where αm and αj indicate the pair of input coefficient vectors and k denotes the kth input. wmj ranges from 0 to 1, where a higher value indicates a higher degree of technological relatedness. As in Los (2000), we define two industries as being related if the value of wmj is over 0.4 and unrelated otherwise. We separately sum the total employment in industries that have a value of wmj over 0.4 and less than or equal to 0.4 for each year across all cities to obtain the respective measures for related variety and unrelated variety. Note that the sum of these two measures adds up to our proxy for Jacobs externalities.

202

9.3.3

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

Firm-Level Covariates

The main firm covariate of interest is the amount of subsidies it receives from central and local governments. We calculate the firm’s subsidy intensity as the ratio of production-related subsidies to the total number of the firm’s employees. In addition, we also include, as firm-level controls, the firm’s current size and its square, sales growth, and total factor productivity (TFP). Controlling for firm size and size squared addresses nonlinearities in the survival-size relationship found in the existing literature. Including firm sales growth helps to avoid capturing the “desperate” firm effects, where low growth firms may pursue a more risky survival strategy by switching products as a desperate measure to avoid imminent closure. TFP captures the firms’ productivity and is derived using the three-step procedure by Olley and Pakes (1996).

9.4

Model Specification

The general hazard function represents the probability of failure of a firm during t + Δt conditioned on the fact that the firm survives up to the time t. The hazard function is expressed as: hðt Þ ¼ lim

Δt!0

 P t  T < t þ Δt j T  t f ðt Þ f ðt Þ ¼ ¼ Δt 1  F ðt Þ Sðt Þ

ð9:2Þ

where f(t) is the density function, F(t) is the distribution function, and S(t) the Z is t survival function. The survival function is S(t) ¼ exp (Λ(t)), and Λðt Þ ¼ hð uÞ 0

du is the cumulative hazard function. While semi-parametric Cox’s hazard and discrete-time models are most often employed to study firm exit, they frequently violate the proportionality assumption when examining multiple cohorts. Moreover, it is not easy to take into account issues related to left truncation and right censoring. Instead, the accelerated failure time (AFT) model provides a more appropriate alternative as it has a time-scaling factor that allows us to avoid violating the proportionality assumption. In a AFT model, the survivor function at time t, S(t| x, α), is assumed to be of the following form: Sðtjx; αÞ ¼ S0 ðt=ψÞ

ð9:3Þ

where S0(t) is the baseline survival model associated with a set of time-varying firm covariates, x, time-varying industry covariates, and random effects α. The scaling factor ψ is expressed as follows:

9.4 Model Specification

203

  0 ψðx; αÞ ¼ expðηÞ ¼ exp w þ β Xþ

ð9:4Þ

where α ¼ exp(w) is assumed to have a gamma distribution with distribution function G(α) and η is the linear component of the model. The term α represents a frailty term with the mean of the distribution set to the value unity. Our data is left-truncated since firms are observed at different points in time during the observation period and subject to right censoring since not all firms exit the sample by the end reporting year in 2007. The likelihood function with lefttruncated and right-censored observations is given in general form as: L¼

1( g Z N Y Y i¼1

0

j¼1

 hð T Þ

c

Sð T Þ Sð E Þ

) dGðαÞ

ð9:5Þ

where Ei takes into account the left truncation, giving the first time a firm enters into the panel, and ci takes into account right censoring and takes the value of 1 for firms that fail and 0 for firms that are still active at the end of observation time. In order to estimate the hazard function, an appropriate underlying distribution must be specified. We consider the log-logistic distribution since it has a flexible form that allows for monotonous functional forms and other shapes as well. The hazard function with a log-logistic distribution is: ψ1=λ t ð1=λ1Þ i hðtjx, αÞ ¼ h λ 1 þ ðψt Þ1=λ

ð9:6Þ

The shape of the function is determined by λ. For λ > ¼ 1, the functional form is decreasing monotonously, and 0 < λ< 1 has a bell-shaped form. To obtain the survival probabilities of the firm, the hazard model in Eq. (9.6) can be equivalently expressed as a log-linear model expressed as: logðDit Þ ¼ β1 Subsit1 þ β2 Aggn þ β3 ðSubsit1  Aggn Þ þ β4 X it þ α þ μ þ σEi

ð9:7Þ

where Dit is a random variable based on the duration, in years, of the firm. Subsit1 is the amount of subsidies the firm receives in year t1, Aggn is a vector that contains each measure of agglomeration (outlined in Sect. 9.3.2), and β1 and β2 are the respective corresponding parameters; β3, a key parameter of interest, captures the effects of local state support given different local existing industrial structures on firm survival by multiplying the amount of subsidies the firm received in year t1 with each respective agglomeration measure, Aggn.

204

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

The vector, Xit, includes the following set of control variables: firm size and its square, firm sales growth, and a set of industry, region, and year dummies. As before, the term α represents a frailty term with the mean of the distribution set to the value unity. Finally, μ and σ are unknown location and scale parameters, and ?i has a distribution that determines ti.

9.4.1

Identification Issues

Two key issues arise when trying to estimate Eq. (9.7). First, it is difficult to interpret the parameter, β2, since the observed geographic influences on firm survival may be subject to issues related to selection bias in the location choice of the firm (Audretsch 1991). Second, the allocation of public subsidies is not likely to be randomly distributed to firms across different locations, thus making it difficult to interpret the parameters, β1 and β3. We will briefly discuss how we handle each of these estimation issues.

9.4.1.1

Firm Spatial Sorting

It is known that larger firms may optimize on their location decision by moving into more agglomerated regions in order to take advantages of anticipated positive externalities. If selection occurs, the survival estimation results would lead to artificially high coefficients on the agglomeration measures. While this is certainly a theoretical possibility, as a practical matter, such bias is likely to be negligent for new entrepreneurial firms in China for the following reasons. First, entrepreneurial firms do not typically engage in a multiregional site selection process; rather, they tend to be wherever the owner’s place of residence is located (Stam 2007). Second, the factors that influence firms’ entry do not necessarily have the same effect on their chances of survival (Stuart and Sorenson 2003; Renski 2011), which we would expect to be the case if firms engaged in selfselection behavior. Lastly, in the Chinese context, Au and Henderson (2006) argue that restrictions on migration (e.g., the Hukou system) obstruct individuals and firms from migrating into certain areas, thereby limiting self-sorting behavior. For these three reasons, the likelihood that Chinese firms optimize on their location decision is likely to be lower than in other country contexts and not be the main source driving the results. Nevertheless, as a simple sensitivity check, we remove all of the mobile firms that changed their industry or region during the time period from the sample. Then we re-estimate the survival model on only the subsample of stationary firms (see Table 9.4 below).

9.4 Model Specification

205

Table 9.1 Descriptive statistics Whole sample of firms Agglomeration measures Location quotients Diversity (Ln) Related variety (Ln) Unrelated variety (Ln) City size (Ln) Firm-level controls Size (Ln) Age Sales growth (Ln) TFP (Ln)

Receive subsidies No Yes

Difference

2.571 5.937 7.485 5.690 5.334

2.566 5.936 7.474 5.671 5.339

2.643 5.948 7.633 5.933 5.269

0.077*** 0.012* 0.160*** 0.262*** 0.070***

4.325 2.929 0.025 3.371

4.297 2.902 0.012 3.370

4.685 3.267 0.192 3.381

0.388*** 0.364*** 0.181*** 0.011

Note: ***p < 0.001, **p < 0.01, *p < 0.05

9.4.1.2

Subsidies Selection Bias

As an initial check for selection bias, Table 9.1 splits firms into respective groups according to whether or not they receive subsidies. The results show that subsidies tend to be channeled to firms that have higher local specialization and higher localrelated (unrelated) variety and that are located in smaller cities. At the firm level, the results further show that subsidies tend to go toward more successful firms, e.g., firms that tend to be larger in size, are more experienced, enjoy higher sales growth, and are more productive. To help remove selection bias, we adopt the following two-stage approach. In the first stage, we estimate the selection equation as follows:  SubsDit ¼

1 if SubsDit ¼ β1 Aggn þ β2 X it þ αt þ Eit > c 0 if SubsDit ¼ β1 Aggn þ β2 X it þ αt þ Eit  c

ð9:8Þ

where SubsDit is an (observable) indicator function that equals 1 if firm i receives subsidies and 0 otherwise. SubsD is a latent indicator function whereby the firm receives subsidies if these are above a given threshold c. As before, Aggn includes each measure of agglomeration, and Xit includes a set of control variables. Lastly, αi captures the unobserved firm heterogeneity, and Eit is an error term. We estimate Eq. (9.8) using a random effect probit model given the panel structure and the binary character of the dependent variable. The random effect structures assume that the errors are not correlated with the regressors, an unrealistic assumption in this case. To address this issue, the Mundlak (1978) specification is

206

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

used by including a vector of means of the time-varying regressors as control variables to allow for some correlation between the random effects and the regressors. Given the firm receives subsidies, in the second stage, we estimate the following specification:  SubsDit ¼

Subsit ¼ β1 Aggn þ β2 X it þ αi þ Eit if Subsit ¼ 1 Subsit ¼ 0 if Subsit ¼ 0

ð9:9Þ

where Subs is the unobserved latent variable representing the firm’s subsidy intensity. The parameters and variables have their same interpretations as above. Correcting for selection effects, Eq. (9.9) is estimated using the consistent estimator introduced in Wooldridge (1995). That is, a pooled OLS model is estimated including T inverse Mills ratios (interacted with time dummies), which are obtained from estimating T probit models (one for each year). Since inverse Mills ratios are used, we should find at least one exogenous variable that significantly affects the likelihood that the firm receives a subsidy, but is uncorrelated with the amount of subsidies given the subsidy allocation decision has already been made. We recognize that it is difficult to find a satisfactory exclusion restriction, and in theory, the factors that influence the decision to subsidize a firm may also be related to the intensity of that subsidy. In our case, we take the firm’s distance to the nearest port as the identification instrument. We motivate this choice theoretically: subsidies are more likely to be given to local Chinese firms that are closer to ports to help them compete against foreign firms which tend to be concentrated nearby shipping ports to export goods more cheaply. The main determinants of the subsidy amount, however, are based on firms’ productivity and overall performance and not their location. As an empirical check, the econometric results in Table 9.2 show that the distance measure, whose marginal effect is significant in the selection equation (column 1), does not affect firms’ subsidy intensity in the outcome equation (column 2). The results in Table 9.2 also show that firms tend to be more likely to receive subsidies and receive them more intensively if they are in areas with higher specialization, higher related variety, and higher unrelated variety. By contrast, firms in smaller cities tend to be more likely to receive subsidies, although given the firm receives a subsidy, more subsidies tend to go to firms in larger cities. In terms of firm’s characteristics, the results show that firms that are larger, are more experienced, experience higher growth, and are more productive tend to be more likely to receive subsidies and receive them more intensively. Although in columns (2) and (3), there is a non-linear effect on firm size, the coefficient on firm sales growth is statistically not significant.

9.4 Model Specification

207

Table 9.2 Sample selection model

Agglomeration measures LOC Related variety Unrelated variety City size Firm-level controls Size Size2 Age Sales growth TFP Distance to port Time dummies Industry dummies Region dummies IMRtime dummies Means Rho Corrected predictions Num. obs.

Receive subsidies (1¼yes, 0¼no) (1)

Subsidy-receiving firms only: subsidy in density (2) (3)

0.028*** (0.005) 0.086*** (0.008) 0.089*** (0.006) 0.021*** (0.005)

0.030*** (0.004) 0.018** (0.006) 0.043*** (0.005) 0.046*** (0.003)

0.032*** (0.004) 0.004 (0.006) 0.075*** (0.005) 0.026*** (0.003)

0.073* (0.032) 0.032*** (0.003) 0.024*** (0.004) 0.039*** (0.005) 0.007* (0.003) 0.027*** (0.006) Yes*** No No – Yes*** 0.567 78.13% 337,074

0.242*** (0.024) 0.042*** (0.003) 0.044*** (0.002) 0.005 (0.003) 0.010*** (0.003) 0.019 (0.015) No Yes*** Yes*** Yes*** Yes*** – – 25,382

0.264*** (0.024) 0.043*** (0.003) 0.039*** (0.002) 0.007 (0.004) 0.022*** (0.003)

No Yes*** Yes*** Yes*** Yes*** – – 25,382

Note: ***p < 0.001, **p < 0.01, *p < 0.05. Bootstrapped standard errors in parentheses. Column (1) is estimated with a RE probit model with the coefficients reported in marginal effects (at the means). Column (2) is estimated using the consistent estimator in Wooldridge (1995). Inverse Mills ratio (IMR)  time dummies is significant at 1% confirming the inclusion of a selection equation. Means of the time-varying regressors are included as control variables to allow for some correlation between the random effect and the regressors. Rho is the percentage of total variance contributed by the panel-level variance component (Rho >0 indicates that the panel estimator is different from the pooled estimator)

208

9.5

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

Effects of Agglomeration on New Firm Survival

Table 9.3 provides the estimates of the effect of agglomeration on new firm survival using the AFT survival models’ log-logistic distribution with random effects. Column (1) includes only the spatial agglomeration measures, column (2) adds the firmlevel covariates, column (3) decomposes the diversity measure into related variety and unrelated variety, and column (4) re-estimates the survival model on only the firms that received subsidies correcting for sample selection. Note that the agglomeration variables have been standardized for comparison purposes. The significance of the estimated effects is assessed by clustering standard errors at the city level to adjust for the potential correlation of errors among firms located within the same city. The results show that the relationship between agglomeration and firm survival varies depending on the dimension of agglomeration under scrutiny. The coefficients on LOC are positive and highly significant, while the coefficients on diversity do not influence firm survival chances in any significant way. The coefficient on city size is negative and statistically significant in column (1), but becomes statistically not significant with the inclusion of firm controls in column (2). In column (3), the coefficient on related variety is positive and significant, while the one on unrelated variety is statistically not significant. In terms of economic impacts, related variety exhibits the largest positive effect on firm survival, increasing the probability of survival by a factor of 1.9. The results on the firm-level covariates are largely consistent with the existing literature. The coefficient on subsidies is positive and statistically significant, even after correcting for sample selection in column (4). The coefficients on firm size are positive and significant in each specification although the effects are nonlinear. The coefficients on the firms’ sales growth and TFP are both positive and significant.

9.5.1

Alternative Model Specifications and Robustness Checks

Table 9.4 reports a series of alternative model specifications to check the validity of our initial findings between our agglomeration measures and new firm survival. Note that in each estimation, the individual terms remain included in all of the model estimations, but are not reported given their qualitatively unchanged values. Also, focus is placed on related variety and unrelated variety since the coefficient on Jacobs externality is statistically not significant in the baseline model. Note that a discrete hazard model is estimated in column (4) because it is more flexible in controlling for unobserved firm heterogeneity and addresses the issue of tied failure times.

9.5 Effects of Agglomeration on New Firm Survival

209

Table 9.3 Effects of agglomeration and subsidies on firm survival Survival models: accelerated failure time with random effects Agglomeration Agglomeration Related variety Subsidy-receiving firms measures only measures and firm and unrelated only (correcting for (1) controls (2) variety (3) sample selection) (4) Agglomeration measures LOC 0.031*** 0.013*** 0.013*** 0.016* (0.003) Diversity 0.002 0.005 (0.004) (0.004) Related 0.019* 0.057* variety (0.006) (0.019) Unrelated 0.010* 0.009* variety (0.005) (0.004) City size 0.005* 0.003 0.003 0.015 (0.002) (0.002) (0.002) (0.013) Firm-level controls Subs_Int 0.006** 0.005** 0.008** (0.002) (0.002) (0.003) Size 0.034*** 0.340*** 0.531*** (0.019) (0.019) (0.089) Size2 0.020*** 0.020*** 0.037*** (0.002) (0.002) (0.009) Sales 0.009*** 0.009*** 0.038** growth (0.003) (0.003) (0.015) TFP 0.012*** 0.011*** 0.018*** (0.002) (0.002) (0.004) RegionYes Yes Yes Yes fixed effects IndustryYes Yes Yes Yes fixed effects YearYes Yes Yes Yes fixed effects Log 182,804 178,396 178,195 10524.6 likelihood Num. obs. 337,074 332,529 332,529 25,382 Notes: ***p < 0.001, **p < 0.01, *p < 0.05. The t-statistics (in parentheses) are obtained using standard errors that are robust and clustered at the city level. The dependent variable is the duration period of the firm, and the model estimated is a frailty accelerated failure time models using the log-logistic distribution. The coefficients are presented as survival probabilities and represent the conditional probability of competing a survival spell

0.014* (0.007) 0.020*** (0.005) 0.006*** (0.002) 0.000 (0.003) Yes Yes Yes 179,793 332,529

Weibull distribution (2)

0.008** (0.003) 0.013* (0.007) 0.006** (0.003) 0.003 (0.003) Yes Yes Yes 103,150 304,236

$1 million threshold (3)

0.017** (0.006) 0.024** (0.008) 0.015** (0.003) 0.003 (0.003) Yes Yes Yes – 332,529

Exit Cox hazard (4)

0.029* (0.012) 0.004* (0.002) 0.013*** (0.006) 0.003 (0.006) Yes Yes Yes 198,118 332,529

Discrete-time hazard (5)

Notes: ***p < 0.001, **p < 0.01, *p < 0.05. The t-statistics (in parentheses) are obtained using standard errors that are robust and clustered at the city level. The dependent variable is the duration period of the firm in columns (1), (2), and (5) and firm exit in columns (3) and (4). Each specification includes all controls from Table 9.3. Note that excluding column (1), each specification is estimated on the full sample of firms

Agglomeration measures LOC 0.011* (0.005) Related variety 0.016** (0.007) Unrelated variety 0.012** (0.004) City size 0.006 (0.004) Region-fixed effects Yes Industry-fixed effects Yes Year-fixed effects Yes Log likelihood 94,241 Num. obs. 223,184

Survival Stationary firms only (1)

Table 9.4 Alternative model estimations

210 9 How Do Agglomeration Economies Contribute to Firm Survival in China?

9.5 Effects of Agglomeration on New Firm Survival

211

Table 9.5 Effects of local subsidies on firm survival Survival models: accelerated failure time with random effects All firms (without sample Only firms that receive subside selection correction) (1) (correction for sample selection) (2) Interaction terms Subs_IntLOC Subs_Intrelated variety Subs_Intunrelated variety Subs_Intcity size Region-fixed effects Industry-fixed effects Year-fixed effects Log likelihood Num. obs.

0.006** (0.002) 0.009** (0.003) 0.006** (0.002) 0.009** (0.004) Yes Yes

0.010* (0.005) 0.008* (0.004) 0.008** (0.003) 0.006** (0.002) Yes Yes

Yes 178,391 332,529

Yes 12,314 25,382

Notes: **p < 0.01, *p < 0.05. The t-statistics (in parentheses) are obtained using standard errors that are robust and clustered at the city level. The dependent variable is the duration period of the firm, and the model estimated is a frailty accelerated failure time models using the log-logistic distribution. The coefficients are presented as survival probabilities and represent the conditional probability of completing a survival spell. Each specification includes all controls from Table 9.3

In column (1) of Table 9.4, all of the mobile firms that changed their industry or region during the time period from the sample are removed, and the model is re-estimated on only the subsample of stationary firms. The results reveal that the direction of the relationship between each agglomeration measure and firm survival remains the same. Albeit the size of the coefficients tends to be slightly smaller in magnitude, an indication that the initial results may be influenced by selection effects, but certainly not driven purely by them. The results in columns (2)–(5) further show that the initial findings are neither sensitive to the minimum sales threshold or to the estimation strategy.

9.5.2

Effects of Local Subsidies on Firm Survival

Table 9.5 shows how increasing local state support influences new firms’ survival likelihood given different types of local existing industrial structures exist in those regions. Interaction terms are created that multiply the amount of subsidies the firm received in the previous year with each measure of agglomeration. Column (1) re-estimates the model on the entire sample of firms. Column (2) re-estimates the survival model on only the sample of firms that received subsidies (correcting for selection bias).

212

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

The coefficients on the interaction terms between subsidies and LOC, related variety, and city size are, respectively, negative and statistically significant. The results are largely consistent in both the uncorrected and corrected model, although the size of the coefficients changes. For instance, the size of the negative coefficient on the interaction term between subsidies and LOC is 40% larger in the sample corrected model. One reason for this is that in the uncorrected model, a greater amount of subsidies gets distributed to better performing firms that are local industry leaders (e.g., have a higher LOC value), which artificially reduces the negative selection effects associated with specialization in the uncorrected model. The main findings suggest that more heavily subsidized firms have a lower chance of survival than their less subsidized counterparts given higher values for LOC, related variety, and city size, respectively. One reason for these negative interaction effects is because receiving too many subsidies reduces the firms’ incentives to pursue profit-maximizing strategies, since they incorporate an expectation that the state will intervene on their behalf in times of duress. By contrast, agglomerated firms that receive a fewer amount of subsidies are unable to rely completely on the state for their survival and instead are motivated to exploit and benefit from localization economies and spillovers that are expected to take place within and between (related) industries. There is an important caveat with the present interpretation of these findings, however. In general, more heavily subsidized firms are unlikely to declare bankruptcy given the nature of state support. Our results therefore reveal that more heavily subsidized firms versus their less subsidized counterparts are more likely to encounter financial distress (e.g., sales revenue drops well below the minimum sales threshold) when operating in more agglomerated areas (e.g., higher values for LOC, related variety, and city size). In this regard, firms that become overreliant on external state support while also facing intense competition effects or other negative externalities experience a greater threat to their sustainability in that they are less likely to maintain revenues above a certain threshold. The only interaction term that returns a positive coefficient is the one between subsidies and unrelated variety. One reason why more heavily subsidized firms in areas with higher unrelated variety are more likely to survive than elsewhere may be because they do not have to compete against many local competitors. As a result, such firms can enjoy local monopolies that enable them to capitalize on profitmaking activities, which, in turn, help them sustain their sales revenue above the minimum threshold enforced on our data.

9.6

Concluding Remarks

Coinciding with China’s opening up and market reforms, new firms in China face fierce competitive pressures that often lead to their premature exit from the market. The ability of firms to seek out and benefit from agglomeration economies matters for their survival prospects. In particular, understanding the factors that incentivize new firms to seek out and successfully exploit externalities are key to implementing

References

213

successful place-based policies in China. This paper links five dimensions of agglomeration-specialization, diversity, related variety, unrelated variety, and city size to new firms’ survival prospects. Taking into account the Chinese context, consideration is given to studying how local state intervention influences the ability of firms to seek out and benefit from each dimension of agglomeration. Our study produces the following two key results. First, compared to other types of agglomeration, related variety has the largest positive effect on firm survival. Second, firms that become overreliant on external state support while also facing intense competition effects or other negative externalities experience a greater threat to their sustainability in that they are less likely to maintain revenues above a certain threshold. The implications of these findings for policy are as follows. First, in order to help new firms to mitigate their initial liabilities, local policy-makers should adopt a “regional branching” strategy (Neffke et al. 2011). That is, evolve the local industrial mix into new related industries in order to increase the supply of spillovers that are expected to take place between related industries. Second, the intensity and location of where subsidies get allocated matters. In order to promote regional competitiveness in China, allocating relatively fewer subsidies to new firms operating in regions with certain existing industrial structures may help motivate them to rely on agglomeration economies for their survival as opposed to becoming overreliant on external finance from the state.

References Acs, Z. J., Armington, C., & Zhang, T. (2007). The determinants of new-firm survival across regional economies: The role of human capital stock and knowledge spillover. Papers in Regional Science, 86, 367–391. Au, C., & Henderson, V. (2006). Are Chinese cities too small? Review of Economic Studies, 73(3), 549–576. Audretsch, D. (1991). New-firm survival and the technical regime. Review of Economics and Statistics, 73(3), 441–450. Audretsch, D. B., & Lehmann, E. E. (2005). Does the knowledge spillover theory of entrepreneurship hold for regions? Research Policy, 34(8), 1191–1202. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Cefis, E., & Marsili, O. (2011). Born to flip. Exit decisions of entrepreneurial firms in high-tech and low-tech industries. Journal of Evolutionary Economics, 21(3), 473–498. Delgado, M., Porter, M. E., & Stern, S. (2010). Clusters and entrepreneurship. Journal of Economic Geography, 10(10), 495–518. Dunne, T., Roberts, M. J., & Samuelson, L. (1989). The growth and failure of US manufacturing plants. The Quarterly Journal of Economics, 104(4), 671–698. Essletzbichler, J. (2013). Relatedness, industrial branching and technological cohesion in US metropolitan areas. Regional Studies, 49(5), 752–766.

214

9 How Do Agglomeration Economies Contribute to Firm Survival in China?

Falck, O. (2007). Survival chances of new businesses: Do regional conditions matter? Applied Economics, 39(16), 2039–2048. Fan, C. C., & Scott, A. J. (2003). Industrial agglomeration and development: A survey of spatial economic issues in East Asia and a statistical analysis of Chinese regions. Economic Geography, 79(3), 295–319. Folta, T. B., Cooper, A. C., & Baik, Y. (2006). Geographic cluster size and firm performance. Journal of Business Venturing, 21(2), 217–242. Fotopoulos, G., & Louri, H. (2000). Location and survival of new entry. Small Business Economics, 14(4), 311–321. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Frenken, K., Cefis, E., & Stam, E. (2015). Industrial dynamics and clusters: A survey. Regional Studies, 49(1), 10–27. Fritsch, M., Brixy, U., & Falck, O. (2006). The effect of industry, region, and time on new business survival: A multi-dimensional analysis. Review of Industrial Organization, 28(3), 285–306. Glaeser, E., Kallal, H., Scheinkman, J., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100, 1126–1152. He, C., & Yang, R. (2016). Determinants of firm failure: Empirical evidence from China. Growth and Change, 47(1), 72–92. Howell, A. (2015). ‘Indigenous’ innovation with heterogeneous risk and new firm survival in a transitioning Chinese economy. Research Policy, 44(1), 1866–1876. Howell, A. (2016). Firm R&D, Innovation and easing financial constraints in China: Does corporate tax reform matter? Research Policy, 45(10), 1996–2007. Howell, A., He, C., Rudai, Y., & Fan, C. (2016). Technological relatedness and asymmetrical firm productivity gains under market reforms in China, Cambridge Journal of Regions, Economy and Society, 9(3). https://doi.org/10.1093/cjres/rsw024 Hu, C., Xu, Z., & Yashiro, N. (2015). Agglomeration and productivity in China: Firm level evidence. China Economic Review, 33, 50–66. Jacobs, J. (1969). The economy of cities. New York: Vintage. Los, B. (2000). The empirical performance of a new interindustry technology spillover measure. In P. Saviotti, B. Noote-boom, & E. Elgar (Eds.), Technology and knowledge: From the firm to innovation systems. Cheltenham: Edward Elgar Publishing. Marshall, A. (1890). Principles of economics. London: Macmillan. Mundlak, Y. (1978). On the pooling of time series and cross-section data. Econometrica, 46(1), 69–85. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Olley, G. S., & Pakes, A. (1996). The dynamics of productivity in the telecommunication equipment industry. Econometrica, 64(6), 1263–1297. Renski, H. (2011). External economies of localization, urbanization and industrial diversity and new firm survival. Papers in Regional Science, 90(3), 473–502. Schutjens, V. A., & Wever, E. (2000). Determinants of new firm success. Papers in Regional Science, 79(2), 135–159. Staber, U. (2001). Spatial proximity and firm survival in a declining industrial district: The case of knitwear firms in Baden-Wurttemberg. Regional Studies, 35(4), 329–341. Stam, E. (2007). Why butterflies don’t leave: Locational behavior of entrepreneurial firms. Economic Geography, 83(1), 27–50. Stuart, T., & Sorenson, O. (2003). The geography of opportunity: Spatial heterogeneity in founding rates and the performance of biotechnology firms. Research Policy, 32(3), 229–253. Tallman, S., Jenkins, M., Henry, N., & Pinch, S. (2004). Knowledge, clusters, and competitive advantage. Academy of Management Review, 29(2), 258–271. Wooldridge, J. (1995). Selection corrections for panel data models under conditional mean independence assumptions. Journal of Econometrics, 68(1), 115–132.

Chapter 10

How Does Geese Fly Domestically? Firm Demography and Spatial Restructuring in China’s Apparel Industry

10.1

Introduction

Apparel manufacturing is one of the most globalized and footloose industries. The labor-intensive part of the production has been relocated from Europe and North America to Japan and the Four Asian Tigers to coastal China in the past decades (Evans and Smith 2006). This sequential pattern of industrial relocation to nations at different stages of development gave birth to the “flying geese” theory (Akamatsu 1962). Around 2003, when China “entered a new era of labor shortage” (Zhang et al. 2011, p. 542), soaring wage costs, alongside rising costs of raw materials, and increasing foreign trade disputes have resulted in competitive pressure on producers in the apparel industry, triggering industrial restructuring and spatial shifts. Scholars have begun to discuss industrial relocation away from coastal China. Some argue that the labor-intensive Chinese industries might follow the “flying geese” pattern and relocate to Southeast Asia or even sub-Saharan Africa (Ito 2013). Others contend the “domestic flying geese” model in which the labor-intensive industries relocate within China because of the large gap in economic development among different parts of the country (Cai et al. 2009). Extant studies determined that the tipping point for China’s textile and apparel industry from concentration in the coast to migration inland occurred around 2005 (Ruan and Zhang 2014). Existing research, however, has largely neglected that the spatial restructuring of China’s apparel industry is a multi-scalar process with dramatic changes both within and between provinces. The institution created in China’s transformation toward a market economy is summarized as a regionally decentralized authoritarian regime characterized by political centralization and economic regional decentralization (Xu 2011). On the one hand, while the central government formulated guidelines

Modified article originally published in [Shi, J., He, C. and Guo, Q. (2016), How did geese fly domestically? Firm demography and spatial restructuring in China’s apparel industry. Area, 48: 346–356.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_10

215

216

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

for the relocation of the apparel industry based upon the competitive advantages of different regions for the benefit of the whole country, provincial and city governments were eager to retain firms within their jurisdiction to avoid the potential erosion of their tax base and loss of jobs through ambitious local industrial policies. On the other hand, market forces drive firms to locations with lower cost of labor, emerging domestic markets, and the presence of related industries locally. The interaction of state involvement and market forces provided the driving force for the multi-scalar processes leading to the spatial restructuring in China’s apparel industry. Using a firm-level database from 1999 to 2008, this chapter aims to highlight the industrial dynamics of China’s apparel industry by means of “firm demography,” gross job flow analysis, and the use of the Theil index of spatial disparity. The multiscalar process of spatial restructuring identified, driven by start-up firms both within and between provinces, illustrates the extent of industrial relocation beyond the simple coast-to-inland picture.

10.2

Data and Method

10.2.1 The Firm-Level Dataset: The Annual Survey of Industrial Firms This chapter uses the ASIF dataset. Following a three-step procedure proposed by Brandt et al. (2012) and improved upon by Yang and He (2014), we constructed an unbalanced panel of Chinese industrial firms from 1999 to 2008 and extracted apparel enterprises from it.1 We then defined the firm-level dynamics as follows. Entrants were those that did not exist at time t but appeared in the panel for the first time at time t + 1. Exits were those in the panel at time t but not present at time t + 1. Continuing firms or incumbents were those operating in the same sector at time t and t + 1. Since we were dealing with firms in a particular 3-digit sector, two additional components of change were considered because apparel firms might start from firms previously operating in other sectors or end by switching to an alternative sector (Baldwin et al. 1998). Thus, “switch-ins” refer to firms operating in other sectors at time t but have changed to the apparel sector at time t + 1. Conversely, “switch-outs” were those operating in the apparel sector at time t but changed to another sector at time t + 1. The evolution of the panel of apparel enterprises is shown in Table 10.1. There were 28,858 apparel firms in total during 1999–2008, among which only 1127 firms (approximately 3.9%) persisted throughout the whole 10-year study period. The entry and exit of Chinese apparel firms was substantial, the average rate of entry and 1 Apparel firms are those in Sector 181 (clothes) out of Sector 18 (181 for clothes, 182 for shoes, and 183 for hat manufacturing).

Year 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008

Active firms 5359 5524 6556 7555 8554 10,972 10,826 11,999 13,638 16,358

Continuing firms 4370 4373 5673 6406 5682 9182 9691 10,905 10,848 Exits 921 886 757 870 1686 1681 1028 1019 1873

Switch-outs 68 265 126 279 1186 109 107 75 917

Entrants 1046 1991 1765 1946 4819 1410 2169 2578 4554

Incumbents 4370 4373 5673 6406 5682 9182 9691 10,905 10,848

Table 10.1 Evolution of the panel of apparel manufacturing enterprises in China (1999–2008)

108 192 117 202 471 234 139 155 956

Switch-ins

% of exit 17.2% 16.0% 11.5% 11.5% 19.7% 15.3% 9.5% 8.5% 13.7%

18.9% 30.4% 23.4% 22.7% 43.9% 13.0% 18.1% 18.9% 27.8%

% of entry

10.2 Data and Method 217

218

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

exit being 24.1% and 13.7%, respectively. The exceptionally high entry rate in 2004 and 2008 was partly due to the first and second economic census of companies during which many previously missing firms were included in the database. The impact of firms that switched into and out of the apparel industry was minimal for most years.2 The operational definitions of firm entry and exit derived from our unbalanced panel require a detailed explanation. Each record in the ASIF dataset is a “legal person” based on the registration of a business, which might be composed of several plants on the one hand and affiliated to an enterprise group on the other. Firms receive a new legal individual person code as a result of restructuring, merger, acquisition, or relocation, and firms are excluded from the dataset after a take-over. Consequently, firm entry in the dataset might be de novo start-up firms and establishments that have undergone restructuring, merger, acquisition, or relocation; firm exit might be de facto closures, relocation, takeovers, as well as non-stateowned enterprises whose sales have fallen below the threshold for inclusion.

10.2.2 Firm Demography and Gross Job Flow Analysis Firm demography is a framework proposed by Dijk and Pellenbarg (1999) which explores “the dynamics of the economy activity of a state, region, or city as a population of firms” (p. 2). Unlike the top-down explanation provided by a macroeconomic approach, firm demography attempts to provide a bottom-up view of the economy by looking through the “microscope.” Gross job flow analysis is closely associated with the firm demography approach, for entry/exit, and growth/decline of firms leading to job turnover in the labor market.3 Since the seminal work of Davis and Haltiwanger (1992), a set of indicators has been designed to reflect the process of job creation and destruction that may be hidden behind the net employment change. However, literature tends to examine the sectoral and temporal aspects of employment change and leave regional disparities within a country aside (Davis et al. 1996). Essletzbichler (2004) applied a spatial perspective to the gross job flow analysis for the first time, exploring the changing geography of job creation and destruction in the US manufacturing sector from 1967 to 1997.

Most of the “switch-ins/switch-outs” were from/into Sector 17 (textiles) and 19 (leather, fur, feathers, and its products) as well as Sector 182 (shoes) and Sector 183 (hats). These sectors are closely related to Sector 181 (clothes). Firms have considerable discretion in deciding the exact sector when reporting the sector to which it belongs in the ASIF dataset, especially when their activities cut across narrowly defined sectors. 3 Job turnover estimated from firm-level dynamics is the lower bound of the actual churning of employment because workers might choose to change jobs between different firms even if firms are still in operation. 2

10.2

Data and Method

219

Table 10.2 Components of employment change Component of employment change Net employment growth rate in region r Firm entry in region r

Variable NETr

Entryr

Incumbent growth in region r

Incgrr

Job creation in region r Exit in region r

JCr

Incumbent decline in region r

Incder

Job destruction in region r Switch-ins in region r

JDr

Switch-outs in region r

Outr

Exitr

Inr

Definition NET r ¼

t Etþ1 r E r E rt

P

¼ ðJC r þ Inr Þ  ðJDr þ Out r Þ

etþ1 fr

Entryr ¼

f 2ENT

Incgr r ¼

f 2INCGR

Ert

P

e tfr Þ ðetþ1 fr Ert

JCr ¼ Incgrr + Entryr P Exit r ¼

f 2EXIT E rt

e tfr

P Incder ¼

f 2INCDE

ðe tfr etþ1 fr Þ E rt

JDr ¼ Incder + Exitr P Inr ¼

f 2SWTIN Ert

etþ1 fr

P Out r ¼

f 2SWTOUT Ert

e tfr

Description Rate of change in total employment

Rate of job creation through firm entry Rate of job creation through the expansion of incumbent firms Rate of job creation Rate of job destruction through firm exit Rate of job destruction through the contraction of incumbent firms Rate of job destruction Rate of job restructuring through firms switching into the current sector Rate of job restructuring through firms switching out of the current sector

Notes 1. Definitions of the components of employment change are based upon Davis and Haltiwanger (1992), Baldwin et al. (1998), and Essletzbichler (2004) 2. ENT ¼ firm entrants; INCGR ¼ expanding incumbent firms; EXIT ¼ firm exits; INCDE ¼ shrinking incumbent firms; SWTIN ¼ firms that have switched into the apparel sector from other sectors; SWTOUT ¼ firms that have switched out of the apparel sector

Building upon previous methods and considering industry disaggregation, we decomposed the net employment change in Table 10.2. Job creation (JC) can be generated either from firm entry (Entry) or from incumbent growth (Incgr), whereas job destruction (JD) can be attributed to either firm exit (Exit) or incumbent decline

220

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

(Incde). Moreover, we added two additional indicators to account for the fact that firms might switch sectors during operation: (1) restructuring gain through firms switching into the apparel industry (In) and (2) restructuring loss through firms switching out of the apparel industry (Out).

10.2.3 Theil Index of Spatial Disparity and Its Decomposition The Theil index is most suitable for multi-scalar analysis of spatial disparity because it allows for a perfect and complete decomposition of the measure of inequality across groups (Conceição and Ferreira 2000). Here we structured the employment created by start-up firms into two hierarchical levels. First, we examined the distribution of start-ups across provinces, termed the “between province component.” Second, we focused on the distribution of start-ups within provinces, termed the “within province component.” Our basic unit of analysis was the prefecture-level city, into which we aggregated the employment created by each individual firm. Analogous to the measurement of inequality that summarizes how income is distributed among individuals, the Theil index measures the distribution of employment created by start-up firms across provinces and cities using employment of all firms in the apparel industry in provinces and cities as the benchmark. First of all, the Theil index for start-up firms at city level can be written as: X

Tt ¼

sct ln

c2CITY

where sct ¼ P

P

et f 2STAUP fc

c2CITY

P

et f 2STAUP fc

sct nt:c

ð10:1Þ

is the share of employment created by start-up firms

(STAUP) in city (CITY) c in year t to the total employment created by start-up firms Et in China in year t and nct ¼ P c E t is the share of employment in city (CITY) c in c2CITY c

year t to the total employment in China in year t. Subsequently, we decompose the Theil index for start-up firms at city level (Tt) t t into “between province component” (T bp ) and “within province component” (T wp ). That is: t t T t ¼ T bp þ T wp

ð10:2Þ

The “between province component” is calculated as: t T bp ¼

X p2PROV

spt ln

spt npt

ð10:3Þ

10.3

The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry

P

where spt ¼ P

et f 2STAUP fp

P

et f 2STAUP fp

p2PROV

221

is the share of employment created by start-up firms

(STAUP) in province (PROV) p in year t to the total employment created by startEt ups in China in year t and npt ¼ P p E t is the share of employment in province p2PROV p

(PROV) p in year t to the total employment in China in year t. The “within province component” is calculated as: t T wp

¼

X p2PROV

P

t ¼P where scp

X

spt

cp2CITY c IN PROV p

et f 2STAUP fcp

cp2CITY c IN PROV p

P

et f 2STAUP fcp

t scp

t scp ln t ncp

! ð10:4Þ

is the share of employment created by start-

up firms (STAUP) in city (CITY) c in province (PROV) p in year t to the total t ¼ employment created by start-up firms in province (PROV) p in year t and ncp P

t E cp

Et cp2CITY c IN PROV p cp

is the share of employment in city (CITY) c in province (PROV)

p in year t to the total employment in province (PROV) p in year t. Each province’s contribution is then weighted by spt and summed up to arrive at the “within province component.”

10.3

The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry

China’s apparel production has been concentrated in coastal regions since the very beginning of the reform and opening up of the Chinese economy. The uneven spatial distribution of the apparel industry has been largely determined by historical roots, open-up policies, market potential, and logistic infrastructure (Li & Fung Research Centre 2007). Around 2003, however, a new round of spatial restructuring of China’s apparel industry began as a response to a series of challenges, such as rising labor costs and turbulence in the global business environment (Ruan and Zhang 2014). The spatial shift was usually portrayed as a process of industrial relocation from coastal provinces to inland provinces, notably Henan and Anhui in Central China and Guangxi and Chongqing in West China (see the left half of Fig. 10.1), during which provincial governments intensively competed for new investment. Nevertheless, that broad picture failed to reveal the nature of industrial relocation within the provinces (see the right half of Fig. 10.1). As Zhu and Pickles (2014) demonstrated, Jiangsu province proposed a plan to relocate industries across the Yangtze River toward the underdeveloped north, whereas Guangdong province launched the

222

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

Fig. 10.1 Net employment change of China’s apparel industry by province and city for 1999–2003 and 2004–2008

“double relocation” program with 24 industrial parks established in the hinterland to attract investment and relocation projects from the Pearl River Delta. Therefore, a multi-scalar perspective would indeed further our understanding of the spatial restructuring in China’s apparel industry. We now turn to employment dynamics in China’s apparel industry, at province and city scale, to explore “how geese fly domestically” by uncovering the changing geography of job creation and destruction from 1999 to 2008.4

10.3.1 Province Scale: Domestically Flying Geese Driven by Differences in Job Creation Rate The provincial differences in job creation and destruction changed considerably over the 10-year period. Apparel manufacturing became increasingly concentrated in East

4

We chose 2004 as the break point of the two periods following the empirical evidence from Ruan and Zhang (2014). One additional factor came into play—the economic census in 2004 introduced an influx of apparel manufacturing firms including many previously not included. This gave rise to a clear distinction in the number of firm represented on the panel of firms. Balancing these considerations led us to select the following two time periods—1999–2003 and 2004–2008.

10.3

The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry

223

China from 1999 to 2003, with Zhejiang and Fujian provinces more than doubling their employment in the sector, while many provinces in Central and West China lost up to 30% of their jobs (Fig. 10.1a). Spatial variation in job destruction rates was relatively small overall; the rate of job destruction ranged from 5.1% in Hainan to 89.3% in Chongqing. The spatial variation in job creation rates, however, was much larger (Fig. 10.2a). Fujian took the lead in job creation with an increase of over 200%. Other traditional production centers in East China, such as Zhejiang, Shandong, and Jiangsu, also had more than 100% increase in jobs in 1999. The job creation rate in most of the provinces in Central and West China, however, fell below national average with a few exceptions. Qinghai performed worst in that no jobs were created. From 1999 to 2003, geographical differences in the rate of job creation, rather than the rate of job destruction, shaped the net employment change in China’s apparel industry. Although apparel manufacturing was concentrated in coastal regions of China from 1999 to 2003 in a self-reinforcing way, the geography of production began to change during 2004–2008 (see Fig. 10.1b for changes in net employment). East China lost its leading role in terms of job creation during this period (Fig. 10.2b). Chongqing and Guangxi in West China more than doubled their employment in the sector, with the rate of job creation in excess of 200%. Anhui and Henan in Central China came a close second with the rate of job creation as well as net employment increasing by more than 100%. In contrast, five provinces in East China, namely, Shandong, Jiangsu, Zhejiang, Fujian, and Guangdong, had lower rates of job creation, which were largely below the national average. Therefore, apparel manufacturing started to “fly” domestically around the year 2004. The domestic “flying geese” pattern during the second period was also largely driven by spatial disparities in the rate of job creation.

10.3.2 City Level: The Dominance of Firm Entry Through Start-Up Firms The job creation and destruction data at the city level was examined by the type of firms involved. The rate and share of four types of firm dynamics in relation to job creation and destruction were examined in relation to changes in net employment during 1999–2003 and 2004–2008, respectively (Table 10.3). In both periods, “firm entry” dominated job creation, and “firm exit” accounted for the majority of job destruction. Similar to findings at the provincial level, the rate of job creation was positively correlated with that of net employment change (Table 10.3). When job creation was considered, entry firms dominated the geography of the job generation process. Nevertheless, the effect of preexisting “incumbent” firm growth on net employment change increased during the 2004–2008 period (Table 10.3). As indicated above, firm entry does not necessarily mean that the company was a new start-up firm. The extent that the geography of start-ups indicates

224

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

Fig. 10.2 Job creation and destruction of China’s apparel industry by province for 1999–2003 and 2004–2008 Note: the dash line represents the mean value of the rate of job creation and destruction, respectively

10.3

The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry

225

Table 10.3 Employment dynamics of China’s apparel industry at the city level during 1999–2003 and 2004–2008

Avg. rate Share Corr. with NET

Avg. rate Share Corr. with NET

Job creation Entry 1999–2003 116.6% 63.0% 0.675*** 0.674*** 2004–2008 154.7% 71.8% 0.885*** 0.884***

Incgr 11.5% 6.2% 0.006 15.6% 7.2% 0.524***

Job destruction Exit

Incde

47.2% 25.5% 0.176** 0.193***

9.7% 5.2%

29.7% 13.8% 0.014 0.093

15.5% 7.2%

0.104

0.181***

Notes 1. Number of observation: 213 cities; correlation (Corr.) significance levels: *p < 0.1, **p < 0.05, ***p < 0.01 2. Average rate of job creation and destruction is mean value over 213 cities 3. NET ¼ net employment change; Entry ¼ firm entry; Incgr ¼ incumbent growth; Exit ¼ firm exit; Incde ¼ incumbent decline

entrepreneurship and economic development in a regional context (Fritsch and Storey 2014) requires further analysis. Since the ASIF dataset provides information on the year the firm started operation, we can effectively identify de novo start-ups from firm entry into the sector.5 From 1999 to 2008, the effect of start-up firms on the spatial restructuring of China’s apparel industry was moderate while that of non-start-ups decreased. The role played by start-up and non-start-up firms at the city level was different across 213 cities during both time periods (see their correlation coefficients in Table 10.4). This suggests that they were driven by different underlying socioeconomic forces during each period. Although the number of start-up firms only accounted for 28.0% and 34.9% of firm entry during the two periods, respectively, their total contribution to job creation was much greater. From 1999 to 2003, 42.3% of jobs created by firm entry could be attributed to start-up firms. During the period of 2004–2008, their contribution increased up to 64.1%, nearly double that of non-start-up firms. Both start-up and non-start-up firms were positively correlated with net employment change during the two time periods. Nevertheless, the influence of start-up firms grew, whereas that of non-start-ups declined (Table 10.4). Within firm entry category, start-up firms drove the changing geography of China’s apparel industry, especially during the 2004–2008 period.

5

See the supplementary document for technical details.

1.000 0.674***

1.000 0.228*** 0.983*** 0.870***

2004–2008 Start-ups 3734 34.9% 99.1% 64.1%

1.000 0.402*** 0.353***

Non-start-ups 6977 65.1% 55.5% 35.9%

1.000 0.884***

Entry 10,711 100.0% 154.7% 100.0%

Notes 1. Firm number is the sum of start-ups or entry each year from 2000 to 2003 and from 2005 to 2008 (entry is not defined in the first year of each period) 2. Number of observation: 213 cities; correlation (corr.) significance level: *p < 0.1, **p < 0.05, ***p < 0.01 2. Average rate of job creation through start-ups and entry is mean value over 213 cities NET ¼ net employment change

1.000 0.768*** 0.529***

1.000 0.143** 0.743*** 0.489***

Entry 6748 100.0% 116.6% 100.0%

10

Firm number Share of entry Avg. rate of job creation Share of job creation Corr. with Start-ups Non-start-ups Entry NET

Non-start-ups 4860 72.0% 67.3% 57.7%

1999–2003 Start-ups 1888 28.0% 49.3% 42.3%

Table 10.4 Start-ups vs. non-start-up firms within China’s apparel industry at the city level during 1999–2003 and 2004–2008

226 How Does Geese Fly Domestically? Firm Demography and Spatial. . .

10.3

The Multi-scalar Process of Spatial Restructuring in China’s Apparel Industry

227

Fig. 10.3 Decomposition of the Theil index for spatial disparity of start-ups in China’s apparel industry (1998–2008) Note: the basic spatial unit is the city, which is grouped by province

10.3.3 The Spatial Disparity of Start-Ups Between and Within Provinces Since start-up firms shaped the changing geography of China’s apparel industry, we examined their spatial distribution by calculating the Theil index and decomposing it into a “between province component” and “within province component.” In contrast to our usual emphasis on the provincial level of spatial restructuring (i.e., the “between province component”), analysis indicated that start-up firms within provinces (i.e., the “within province component”) accounted for more than two-thirds of total spatial disparity across cities for most of the 1999 to 2008 period (Fig. 10.3). This suggests that information was lost in the aggregate analysis at the provincial level. Then how did the spatial distribution of start-up firms within the provinces vary across China? The average values of the “within province component” of spatial disparity for each province for 1999–2003 and 2004–2008 are shown in Fig. 10.4. A similar spatial pattern occurred during both periods with coastal areas dominating followed by central provinces and finally the vast west. The spatial restructuring driven by start-up firms was most dynamic in the coastal provinces for two reasons. First, the geography of start-up firms within each coastal province was considerably different to that of incumbent firms, making the spatial shift observed significant and

228

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

Fig. 10.4 Theil index for spatial disparity of start-ups within provinces in China’s apparel industry during 1999–2003 and 2004–2008 Note: we mapped the average values of the “within province component” in the two periods, weighted by the share of employment created by start-up firms in each province to the total employment created by start-up firms in China

10.4

Conclusion

229

the “within province component” larger. Second, the total proportion of employment in start-up firms in coastal provinces was disproportionately higher than the rest of the country, leading to greater weightings in the calculations of the contribution of “within province components” to the Theil index of total spatial disparity. Therefore, despite the industrial relocation from coastal areas to inland China at the provincial level in terms of net employment change, cities in coastal provinces remained the most attractive locations for start-up firms within China’s apparel industry sector from 1999 to 2008.

10.4

Conclusion

Based upon the unbalanced panel representation of China’s apparel manufacturing firms from 1999 to 2008, this chapter advanced recent discussions regarding the “domestic flying geese” model of China’s labor-intensive industries by examining the multi-scalar process of spatial restructuring of firm demography. Gross job flow analysis demonstrated that the “domestic flying geese” model in China’s apparel industry was primarily driven by differences in the rate of job creation, especially through the entry of start-up firms. Interprovincial analysis confirmed previous findings, indicating that 2004 was a significant turning point when the apparel industry ceased to concentrate in the eastern coastal provinces, and started to relocate inland, especially to adjacent central provinces. Nevertheless, examination of the Theil index for start-up firms for within and between province patterns indicated that it was start-up firms within provinces that contributed more than two-thirds of total spatial disparity across cities for most of the years in 1999–2008. A further examination of the spatial pattern of the “within province component” indicated that cities in coastal provinces remained the most attractive locations for start-up firms even after 2004. The results suggested that the regionally decentralized authoritarian regime, which used market forces to shape the process of spatial restructuring, was pivotal to the understanding of the changing geography of China’s apparel industry. The case of China’s apparel industry illustrates the interaction of state involvement and market forces, which in turn influences company dynamics and their spatial distribution. The blueprint for central government to relocate the apparel industry across the country, based upon the competitive advantage of individual regions, is constantly modified by industrial policies of subnational governments with a myopic focus on the potential benefits for their jurisdictions. The regionally decentralized authoritarian regime mediates the effect of market forces on firms and the process of spatial restructuring in China’s apparel industry.

230

10

How Does Geese Fly Domestically? Firm Demography and Spatial. . .

References Akamatsu, K. (1962). A historical pattern of economic growth in developing countries. The Developing Economies, 1(s1), 3–25. Baldwin, J. R., Gorecki, P., Caves, R. E., Dunne, T., Haltiwanger, J. C., & Rafiquzzaman, M. (1998). The dynamics of industrial competition: A North American perspective. Cambridge: Cambridge University Press. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Cai, F., Wang, D., & Qu, Y. (2009). Flying geese within borders: How China sustains its laborintensive industries? Economic Research Journal (Jingji Yanjiu, In Chinese), 44, 4–14. Conceição, P., & Ferreira, P. (2000). The young person’s guide to the Theil index (Working Paper No. 14). Inequality project. Austin: University of Texas. Davis, S. J., & Haltiwanger, J. (1992). Gross job creation, gross job destruction, and employment reallocation. The Quarterly Journal of Economics, 107(3), 819–863. Davis, S. J., Haltiwanger, J. C., & Schuh, S. (1996). Job creation and destruction. Cambridge, MA: MIT Press. Dijk, J. V., & Pellenbarg, P. H. (1999). Demography of firms: Spatial dynamics of firm behaviour. Groningen: University of Groningen. Essletzbichler, J. (2004). The geography of job creation and destruction in the U.S. manufacturing sector, 1967–1997. Annals of the Association of American Geographers, 94(3), 602–619. Evans, Y., & Smith, A. (2006). Surviving at the margins? Deindustrialisation, the creative industries, and upgrading in London’s garment sector. Environment and Planning A, 38(12), 2253–2269. Fritsch, M., & Storey, D. J. (2014). Entrepreneurship in a regional context: Historical roots, recent developments and future challenges. Regional Studies, 48(6), 939–954. Ito, A. (2013). The end of the “workshop of the world”? New challenges for China and the global manufacturing equilibrium. Social Science Japan, 48, 22–26. Li & Fung Research Centre. (2007). Apparel production and cluster development in China (No. 10). Industrial series. Hong Kong: Li & Fung Group. Ruan, J., & Zhang, X. (2014). “Flying geese” in China: The textile and apparel industry’s pattern of migration. Journal of Asian Economics, 34, 79–91. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yang, R., & He, C. (2014). The productivity puzzle of Chinese exporters: Perspectives of local protection and spillover effects. Papers in Regional Science, 93(2), 367–384. Zhang, X., Yang, J., & Wang, S. (2011). China has reached the Lewis turning point. China Economic Review, 22(4), 542–554. Zhu, S., & Pickles, J. (2014). Bring in, go up, go west, go out: Upgrading, regionalisation and delocalisation in China’s apparel production networks. Journal of Contemporary Asia, 44(1), 36–63.

Chapter 11

How Do Environmental Regulations Affect Industrial Dynamics in China?

11.1

Introduction

One of the key challenges facing pollution-intensive firms is how to respond to environmental regulations (Tole and Koop 2008; List et al. 2003; Jeppesen et al. 2002; Yang and He 2015). The existing literatures on environmental regulations and industrial dynamics pay attention to firm behaviors, firm competitiveness, and their relationship with environmental regulations, based on either the pollution haven hypothesis (PHH) or the Porter hypothesis (PH) (Kearsley and Riddel 2010; Bommer 1999; Ambec et al. 2013). PHH suggests that uneven environmental regulations between countries/regions cause the relocation of pollution-intensive production to countries/regions where regulations are less strict (Birdsall and Wheeler 1993; Copeland and Taylor 2004; Tobey 1989). It is argued that high environmental standards may cause unemployment and disinvestment due to the additional costs incurred by environmental regulations (Golombek and Raknerud 1997). In contrast, PH claims that properly designed environmental regulations can catalyze innovations, which to some extent offset compliance costs (Porter 1991; Porter and van der Linde 1995). Such an “induced innovation” effect may lower the costs of complying with environmental standards on the one hand and generate new competitive advantages on the other (Palmer et al. 1995; Kumar and Managi 2009). According to the PH and the PHH, environmental regulations force firms to internalize their environmental costs and impact (Murty and Kumar 2003). This may either result in firms being less competitive in the market because of the additional costs required to comply with regulations or encourage firms to upgrade their production through innovations, as explained by the PH and its precursor—the

Modified article originally published in [Zhou, Y., S. Zhu and C. He (2017) How Do Environmental Regulations Affect Industrial Dynamics? Evidence from China’s Pollution-Intensive Industries, Habitat International, 60 pp. 10–8]. Published with kind permission of © [Elsevier, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_11

231

232

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

“induced innovation” hypothesis (Hicks 1932). However, existing studies tend to predicate on either the PH or the PHH and take a “black-or-white” attitude— testifying one hypothesis while implicitly overlooking or negating the other. Furthermore, most of the massive amount of researches done on environmental regulation and industrial restructuring have either used qualitative methods such as case studies and interviews (Zhu et al. 2014) or employed quantitative analyses analyzing industrial (re)location and upgrading at the industry level (Tole and Koop 2011). Findings based on qualitative methods could be biased and hard to generalize due to the limited sample size; the high level of aggregation in existing quantitative researches makes it hard to understand firm dynamics. This study thus contributes to these debates by stressing the PH and the PHH often coexist on the one hand and by examining the relationship between environmental regulation and industrial dynamics at the firm level on the other hand. Specifically, we focus on the restructuring of pollution-intensive firms in China and analyze four dimensions of firm dynamics: firm entry, exit, employment, and productivity change. Stringent environmental regulation incurs additional costs and thus may increase entry barriers, frighten off some firms, and lead to employment and productivity decline. However, innovation and upgrading induced by environmental regulations may enable incumbents to grow and attract newcomers. China has received much attention due to its dramatic economic growth and the subsequent environmental deterioration since its economic reform (He et al. 2012a). China’s high-growth, low-cost, resource-intensive development model has given rise to increasingly severe environmental pollution and degradation, particularly in its coastal regions where the reform first started (He et al. 2008). In addition to such a spatial variation of environmental pollution, the stringency of environmental standards as well as the enforcement of environmental laws also differs across geographical regions (Zhang and Fu 2008). With the deepening of economic reform, decentralization from the central to the local has granted local governments more autonomy, which started to take a primary responsibility for local economic development (He et al. 2008, 2016), resulting in the so-called decentralized authoritarianism in terms of economic development and, more importantly, of the enforcement of environmental laws (Zhu et al. 2014). Specifically, in some wealthy regions with more severe environmental issues and higher environmental awareness among stakeholders, local administrations may thus enforce environmental laws more wholeheartedly, whereas local states in less developed regions that face lower level of pollution may favor economic development and implement environmental policies less genuinely (Wang and Wheeler 2005). This chapter explores the relationship between environmental regulation and industrial dynamics in China—a country characterized by enormous spatial variation of environmental pollution on the one hand and variegated governance structure with respect to environmental regulations on the other hand. It does so by testifying the PH and the PHH at the firm level and by taking into account two key factors that have been largely overlooked in recent literature—firm heterogeneity and government intervention. The next section proposes an analytical framework. In Sect. 11.3, we introduce data and provide some descriptive analyses. After interpreting the

11.2

Environmental Regulation and Industrial Dynamics: The Role of Firm. . .

233

model and variables, Sect. 11.4 also analyzes empirical results. The last section summarizes the main findings.

11.2

Environmental Regulation and Industrial Dynamics: The Role of Firm Heterogeneity and Government Intervention

Based on either the PHH or the PH, plenty of researches have already explored the articulation between industrial dynamics and environmental regulations, but empirical evidence is at best mixed and inconclusive (Jeppesen and Folmer 2014; Murty and Kumar 2003; Zhu et al. 2014). One of the reasons is the role of firm heterogeneity has been largely overlooked. As argued by Zhu et al. (2014), even though the introduction of appropriate environmental policies may trigger industrial innovations and open up new market opportunities, both the PH and its precursor “induced innovation” hypothesis fail to include a discussion with respect to the underlying mechanisms of the introduction of innovations and to the actual availability of additional resources that innovations demand (e.g., investments, technological and technical know-how). In most studies on the PH and the PHH, there is an implicit assumption that firms are homogenous. Martin (2010) has, however, pointed out that regional economic systems are often complex, consisting of numerous heterogeneous firms with different competences, technologies, business models, and resources, though the firms may all belong to the same industry. Given this, it would be problematic to examine the relationship between environmental regulation and industrial dynamics without taking into account firm heterogeneity as well as whether some firms face greater pollution abatement costs or possess more resources to innovate than others. Firm characteristics may inflect the relationship between environmental regulations and industrial dynamics in many ways (Dean et al. 2000; Wang and Jin 2006). Specifically, some empirical studies have stressed firm size as a key explanatory factor in regional industrial dynamics under increasingly stringent environmental regulation (Heyes 2009; Dean et al. 2000). Dean et al. (2000) have argued that unit pollution abatement costs may be different for small and large firms due to compliance, enforcement, and statutory asymmetries. First, compliance asymmetries result from productive and administrative economies of scale in pollution abatement activities, even regulations are equally applied and enforced across small and large firms (Pashigian 1984). Since compliance is often capital-intensive and demands additional investments such as those on the installation of equipment, the optimal firm size tends to increase. Firms larger than the optimal size are more capable to take pollution reduction measures appropriate to their scales of operations, whereas small firms are not resourceful enough to deal with the technology-forcing aspects of environmental laws (Dean et al. 2000). Furthermore, the cost of interpreting and

234

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

discovering relevant laws and regulations, coping with regulatory organizations, and performing necessary paperwork could generate another type of fixed costs. Small firms again may be disadvantaged since they are not able to spread such administrative costs over high production volumes. Second, regulations may be enforced unequally across small and large firms, resulting in enforcement asymmetries. Theoretical arguments and empirical studies however generate mixed evidence. On the one hand, large firms, particularly those with brand names, have quickly become the targets not only for national and local governments but also for campaigns by activists and social groups that are intended to improve environmental conditions (Zhang et al. 2008; Vogel 2003). The rationale behind this is to identify the most profitable and visible branded firms in the market, not because the environmental standards adopted by such firms are the worst (actually, they are often relatively good) but rather because these firms have to protect their brand reputation with consumers (Walker et al. 2008; Zhang et al. 2008). On the other hand, other studies believe that enforcement asymmetries may occur in the opposite direction, favoring rather than penalizing large establishments (Bartel and Thomas 1987). Since enforcement agencies seek to maximize net political support, large firms often have an advantage in defending themselves with greater legal and political resources and are therefore subjected to less stringent enforcement. Third, statutory asymmetries may be due to differences in the stringency of legislation that small firms face compared with large ones. This type of asymmetries may favor small firms since legislators tend to shield small firms from regulations to minimize the potential disproportionate effects of regulations on small businesses. In short, the impact of environmental regulation is complicated and may vary across large and small firms. Another factor that has not received much attention in the academic literature is how local governments’ intervention affects the relationship between environmental regulation and industrial dynamics. Since the initiation of China’s Reform and Opening-Up Policies, China has undergone dramatic economic growth and has experienced three fundamental transformations: (1) from a state-owned, collective economy dominated by SOEs to one with growing level of private ownership and market orientation (marketization), (2) from a centrally planned to a decentralized economy (decentralization), and (3) from a closed or partially closed economy to one increasingly integrated into the global economy (globalization) (Wei 2001; He et al. 2008; Zhu and He 2013). These institutional transformations have profound influences on resource mobility and (re)allocation and thus on industrial dynamics. For instance, the state-led decentralization policies have enabled local governments to play a critical role in shaping regional industrial restructuring and economic development (He et al. 2008). High-level officials in the hierarchical political system promote low-level government officials based on their economic performance (Xu 2011). Even though environmental protection has been added into such a “cadre evaluation system” recently, local officials still value economic development more. Their main concern is excessively stringent environmental regulation may drive firms away from their territories and aggressive relocation to other regions would further weaken their own plans for local industrial and economic development

11.2

Environmental Regulation and Industrial Dynamics: The Role of Firm. . .

235

(Zhu et al. 2014). As a result, local governments seek to keep firms in situ, by offering a variety of subsidies, tax credits, and financial and political supports. As notified above, compliance with environmental regulations is capital-intensive and often requires the installation of equipment with high fixed costs. Government subsidies may alleviate the negative impact of compliance costs on firms and enable the latter to adopt expensive pollution abatement technologies (e.g., wastewater treatment plants and solvent recovery systems). Subsidies can also offset administrative costs derived from changing environmental regulations, such as costs related to the interpretation of relevant regulations and to necessary paperwork. Another type of local government effort to keep firms in situ can be seen in the establishment of industrial parks for pollution-intensive firms. Such industrial parks are often master-planned by governments with centralized waste treatment systems and other infrastructures. Wastes from all firms in the park can be collected and treated together, significantly lowering per unit treatment costs (Zhu et al. 2014). Furthermore, industrial parks are mostly located in unpopulated, remote areas with an abundance of cheap land and limited social exposure, causing little social discontents (Tao et al. 2014; Cao et al. 2008). Firms in parks enjoy a degree of anonymity and therefore may be able to avoid a portion of compliance costs. Finally, firm heterogeneity and local government intervention may interconnect with one another, co-shaping the articulation between environmental regulation and industrial dynamics. Chinese central government seeks to shift from a pro-growth, pro-business development model to one oriented more toward environmental sustainability, driven to some extent by the rising environmental awareness of the public (Clasen et al. 2012). However, as notified above, local authorities do not always share the concerns that incentivize national government policies. Instead, local governments often respond to central government’s ambition in different ways, resulting in increasing emphasis on environmental sustainability at the national level and segmented enforcement of environmental laws at the local level (Zhan et al. 2014). Specifically, in China, local environmental protection bureau (EPB) takes the primary responsibility for the enforcement of national environmental regulations and policies. Though supervised by the corresponding agency at higher administrative levels, local EPBs rely heavily on financial resources and political supports provided by local governments (Clasen et al. 2012). The enforcement of environmental laws thus could be compromised as local governments have other concerns. For instance, even after three decades of economic reform, China’s state-owned enterprises (SOEs) still play an important role and are systematically favored by various levels of governments (Huang 2003). Local governments, whose income and promotion opportunities are tied to the development of SOEs, have vested interests which oppose the dismantling of the inefficient state-owned sector. Local governments tend to resist the rationalization of SOEs within their jurisdiction through local protectionism (e.g., subsidies, discriminatory regulation enforcement, and intervention in the market) (Zhao and Zhang 1999; Poncet 2005). Since SOEs often have a close relationship with the state and possess greater political power, local protectionism is more widespread in SOEs-dominated industries. The strong

236

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

bargaining power SOEs have may lessen the deterrent impact of regulations on themselves, whereas non-SOEs are more closely scrutinized and face stricter enforcement (Wang and Jin 2002; Wang et al. 2003). Another characteristic of China’s post-reform institutional landscape is the systematic misallocation of financial resources. Private firms often face greater credit constraints, whereas SOEs have received the majority of government lending and investment funds. Similarly, driven by the “grasp the large, let go the small” policy, Chinese governments often favor large, key firms that have potentials to generate massive jobs and profits, whereas small enterprises, particularly those pollution-intensive ones, have been at times neglected by the central and local authorities (Sutherland 2003). Like SOEs, large firms may also enjoy reduced obligations to comply with regulations, have better access to finance and softer budget constraints, and therefore may be less affected by changing environmental policies.

11.3

Data Source and the Dynamics of China’s PollutionIntensive Industries

11.3.1 Data Source This chapter uses the ASIF dataset. We define firm exit and entry as follows. If firm i is included in the dataset in year t but not in year t
1, it is assumed that in year t firm i enters. Similarly, if firm i is reported in year t
1 but not in year t, it is seen as a firm exit. To calculate the pollution intensity of industries, we use another dataset—Key National Monitoring Sources of Polluting Firms (KNMSPF)—issued by China’s Ministry of Environmental Protection (MEP). The MEP has included over 80,000 firms in the KNMSPF list based on their major pollutant emissions— emissions of water pollutants (measured by ammonia emissions and industrial COD) and air pollutants (measured by industrial dust, soot, and industrial SO2)—in the previous year. All firms reported in the KNMSPF were monitored thereafter. In 2009, for instance, such firms accounted for 65% of total pollution emission. We match the 2009 KNMSPF list with the 2008 ASIF dataset by using legal person code and firm name. 73.6% of firms in the 2009 KNMSPF list are matched successfully.

11.3.2 Dynamics of China’s Pollution-Intensive Industries We use the percentage of industrial SO2 meeting standard for emission in a city to measure the degree of environmental regulation stringency and the effectiveness of the enforcement of regulations in a city. Environmental policies, such as the Air Pollution Prevention and Control Law and Two Control Zones designating an acid rain control zone and a sulfur dioxide pollution control zone, are often implemented

11.3

Data Source and the Dynamics of China’s Pollution-Intensive Industries

237

Fig. 11.1 The percentage of industrial SO2 meeting standard for emission at the city level in 2003 (left) and 2008 (right)

Fig. 11.2 Firm entry and exit rate of pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008

at the national level, according to which new firms that could not use low-sulfur coal were required to install desulfurization facilities or take other measures to control SO2 emissions and existing plants were encouraged to control SO2. However, the enforcement of such policies varied across cities (Hering and Poncet 2014). The percentage of industrial SO2 meeting standard for emission in a city is thus a good proxy of local governments’ attitude toward nationwide environmental policies and the stringency of environmental regulation at the local level. Figure 11.1 maps out the spatial variation of this indicator in 2003 and 2008. First, environmental regulation became more stringent in most Chinese cities and the average increased from 0.218 to 0.885 from 2003 to 2008. Second, a simple comparison between two maps shows that the degree of environmental regulation in inland cities is not significantly lower than that in coastal cities in 2003, whereas in 2008 the latter faced relatively stricter environmental regulations. We also examine the dynamics of China’s pollution-intensive industries at the 4-digit level. A 4-digit industry is considered as pollution-intensive, if there are firms in this industry being included in the KNMSPF. Figure 11.2 shows the geography of

238

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

Fig. 11.3 Employment change of incumbent firms in pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008

Fig. 11.4 TFP change of incumbent firms in pollution-intensive industries at the provincial level during 1998–2003 and 2003–2008

firm entry and exit in pollution-intensive industries in two periods: 1998–2003 and 2003–2008. Firm entry rate in both periods is dramatically higher than exit rate in all provinces. During the first period, provinces in China’s coastal region have witnessed higher entry rate, particularly Zhejiang, Shandong, Guangdong, and Jiangsu. During the second period, entry rate in a number of inland provinces, such as Anhui, Chongqing, Jiangxi, Inner Mongolia, and Sichuan, has increased significantly, with an entry rate as high as 180%, though entry rate in some coastal provinces remains high (e.g., Liaoning, Jiangsu, Shandong, and Zhejiang). These findings on the spatial and temporal dynamics of China’s pollution-intensive industries also echo what is shown in Fig. 11.1, suggesting that pollution-intensive production has been relocating from China’s coastal regions with stricter environmental standard to less developed inland provinces. The rapid interregional shifts of pollution-intensive industries are also seen in the spatial and temporal evolution of incumbent firms’ productivity and employment (Figs. 11.3 and 11.4). We adopt the algorithm developed by Olley and Pakes (1996)

11.4

Model Specification and Empirical Results

239

to calculate firms’ total factor productivity (TFP). During 1998–2003, pollutionintensive industries became increasingly concentrated in China’s coastal region (Fig. 11.3). Provinces in East China have witnessed high level of employment growth among incumbent firms, whereas most inland provinces experienced dramatic employment decline in pollution-intensive industries. However, during the second period, as environmental regulations became stricter and social pressures emerged particularly in coastal provinces where industrialization first started, the dominance of coastal cities in pollution-intensive industries was relatively weakened. Employment in a large number of inland provinces started to grow more rapidly (e.g., Shanxi, Shaanxi, and Qinghai), though employment remained high in some coastal provinces. Such a two-stage geographical and industrial transformation is also evident in firms’ productivity change (Fig. 11.4). During 1998–2003, provinces in East China and some in Central China outperformed others in terms of TFP improvement. As the stringency of environmental regulations increased particularly in East China, firms in coastal provinces had to pay higher level of compliance costs and set up additional resources to cope with new policies, whereas firms in inland provinces with relatively less stringent regulations were able to improve their productivity more quickly. These findings support the PHH, but within each province, the relationship between environmental regulations and firm dynamics is much more complicated, affected by firm heterogeneity and local government intervention. It is to this we now turn.

11.4

Model Specification and Empirical Results

11.4.1 Model Specification The following equation is estimated to investigate how firm heterogeneity and government intervention have affected the relationship between environmental regulation and industrial dynamics.
Y i, t ¼ β0 þ f 1 ERk,t
1  PI j þ f 2 ðFSi,t
1 ; IS
i,t
1 Þ þf 3 ðFSi,t
1 ; ISi,t
1 Þ  ERk,t
1  PI j þX j,k,t
1 þ w j þ vk þ ut

ð11:1Þ

where i, j, k, and t denote firm, industry, city, and year, respectively. Industry j represents a 4-digit industry. Four dependent variables are included. Entry (or Exit) takes the value of 1 if firm i is a new entrant (or exiting firm) in year t, and 0 otherwise. Employment and TFP are the change of employment and TFP of firm i from year t
1 to t. We use the percentage of industrial SO2 meeting standard for emission in city k in year t
1 (ER) to measure the degree of environmental regulation stringency. Data on this variable are derived from China’s City Statistical Yearbook. Pollution intensity

240

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

(PI) is the share of firms in industry j that have been included in the KNMSPF air pollution list, indicating intrinsic exposure of industry j to environmental regulations. The interaction term, ER*PI, thus shows to what extent industry j that firm i belongs to is affected by environmental regulations in city k. In Eq. 11.1, FS refers to firm-specific factors. The number of employees in a firm is used as a proxy of firm size (SIZE). Firm ownership structure also affects the relationship between firm dynamics and environmental regulations. Dummy variables, SOE, FOE, and POE, are included, taking the value of 1 if the largest share of a firm’s paid-in capital is state-owned, foreign-owned, and privately owned, respectively, and 0 otherwise. We further include two variables on local government intervention. The ratio of the value of government subsidies a firm has received to the firm’s total output value (SUBSIDY) is used to capture direct government intervention under China’s regionally decentralized authoritarian system that may affect firms’ capabilities in coping with environmental regulations. Finally, local governments often provide additional financial, technological, and political supports and centralized waste treatment system to firms located in government-designated industrial parks, which may lower firms’ compliance costs. PARK is a dummy variable, taking the value of 1 if a firm is located in industrial parks and 0 otherwise. X represents control variables. We include the comparative advantage of industry j in city k and year t
1, measured by the location quotient (LQ). Industry growth is measured as the annual growth rate of output of industry i in the entire country (GROW). The Theil Index of industry i in city k (Theil) is calculated to reflect local industrial diversity and the degree of market competition. The average labor cost of industry i and city k (LAB) is added as a proxy of production cost to control the latter’s impact on industrial dynamics. Finally, w indicates the industry-specific effect, v indicates the city-specific effect, and u indicates the time-specific effect. Variable definitions and descriptive statistics are shown in Table 11.1.

11.4.2 Empirical Results We estimate Eq. (11.1) with the linear probability model (LPM). Correlation analysis shows that correlations between variables are low, indicating no serious multicollinearity problem. Table 11.1 reports the econometric results. In Models (1)–(4) and (5)–(8), dependent variables are entry, exit, employment, and TFP, respectively. The dummy variable for industry (Industry in Table 11.2) is measured at the 2-digit level. Variable, LQ, has a negative and significant relationship with firm exit, indicating that the co-location of a large number of firms in the same industry generates agglomeration externalities and reduces the likelihood of firm exit. The sign of GROW’s coefficient suggests that it is common to see high level of firm entry and exit, and rapid productivity improvement in industries that are in the emergence and growth phase of their life cycle. Firms located in cities with high level of industry diversity and market competition (see the coefficients of Theil) are subject to fiercely aggressive behaviors by rivals, resulting in the coexistence of fewer entries and more exits. Intensive market competition may also have a negative

Control variables

Local government intervention

Firm characteristics

Environmental regulation

Variable Firm dynamics

LQ LAB Theil GROW

PARK

SUBSIDY

POE

FOE

SIZE SOE

PI

Employment TFP ER

Exit

Entry

The share of firms that have been included in the KNMSPF air pollution list The number of employees of a firm SOE takes the value of 1 if the largest share of a firm’s paid-in capital is state-owned and 0 otherwise FOE takes the value of 1 if the largest share of a firm’s paid-in capital is foreign-owned and 0 otherwise POE takes the value of 1 if the largest share of a firm’s paid-in capital is privately owned and 0 otherwise The ratio of the value of government subsidies a firm has received to the firm’s total output value Park takes the value of 1 if a firm is located in industrial parks and 0 otherwise An industrial sector’s location quotient The average labor cost in an industrial sector in a city Industrial diversity Annual growth rate of output of an industrial sector in the entire country

Definitions Entry takes the value of 1 if a firm is an entrant and 0 otherwise Exit takes the value of 1 if a firm exits and 0 otherwise Employment change of a firm TFP change of a firm The percentage of industrial SO2 meeting standard for emission

Table 11.1 Definition of variables and descriptive statistics

3.586 12.917 93.572 0.217

0.151

0.008

0.722

0.161

0.246 0.064

0.011

0.006 0.030 0.742

0.119

MEAN 0.327

2110.428 935 640.419 60.742

1.000

3728.000

1.000

1.000

569.67 1.000

1.000

512.203 16.424 0.996

1.000

MAX 1.000

0.000 0 0.539
0.856

15.724 12.410 113.394 0.289

0.358

3.425


4.337 0.000

0.448

0.368

1.309 0.244

0.000

0.000

0 0.000

0.085

0.769 0.949 0.324


522.371
18.543 0.000

0.000

0.324

SD 0.469

0.000

MIN 0.000

Model Specification and Empirical Results

ASIF ASIF

ASIF

ASIF

ASIF

ASIF

ASIF

ASIF ASIF

ASIF ASIF China Environmental Statistical Yearbook KNMSPF

ASIF

Data source ASIF

11.4 241


0.0677*** 0.020***


0.0327***
0.822*** 0.0001***
0.003*** 0.00003* Included Included Included Included Included Included 0.177*** 1,287,349 0.034


0.131***
0.0030

0.0050***
0.404
0.0002***
0.001*** 0.0017*** Included Included Included Included Included Included 0.680*** 1,415,402 0.195


0.0189***
0.0186*** 0.1189*** 0.1063***
0.0191 0.008**
0.138***
0.0017 Included Included Included Included Included Included
0.1887** 905,220 0.004

(3) Growth 0.0613***

0.0653***
0.141***
0.416***
0.001 0.097
0.0001*** 0.350*** 0.001** Included Included Included Included Included Included 1.291*** 674,761 0.239

(4) TFP
0.186*

0.00317***
0.430
0.0002***
0.001*** 0.002***
0.0474*** 0.0440*** 0.0941*** Included Included Included 0.609*** 1,415,426 0.197

(5) Entry
0.0710** 0.030*** 0.117***
0.0267
0.146*** 0.0878***
0.109
0.135***
0.003
0.0319***
0.813*** 0.0001***
0.003*** 0.00003** 0.0263***
0.0585***
0.0339*** Included Included Included 0.201*** 1,287,364 0.034

(6) Exit 0.0604*** 0.004*
0.0889***
0.0061
0.0361
0.0215 0.334***
0.0685*** 0.020***

(7) Growth
0.0379 0.0218***
0.4122***
0.1624***
0.0885** 0.0510
0.0920
0.0273***
0.0185*** 0.0119*** 0.0107***
0.300 0.008*
0.014***
0.0017
0.0208 0.0184*** 0.0049** Included Included Included
0.2009** 905,222 0.007

(8) TFP
0.402***
0.0030 0.494*** 0.617*** 0.190 0.0659
3.548*** 0.0644***
0.138***
0.415*** 0.0006 0.742
0.0001*** 0.400*** 0.0004**
0.140***
0.0332***
0.0003 Included Included Included 1.354*** 674,761 0.239

11

*p < 0.1, ** p < 0.05, *** p < 0.01

ER*PI ER*PI*SIZE ER*PI*SOE ER*PI*FOE ER*PI*POE ER*PI*PARK ER*PI*SUBSIDY SIZE SUBSIDY TFPt
1 PARK LQ Theil LAB GROW SOE FOE POE Industry Province Year _cons N R2

(2) Exit 0.0243***

(1) Entry
0.121***

Table 11.2 Empirical results

242 How Do Environmental Regulations Affect Industrial Dynamics in China?

11.4

Model Specification and Empirical Results

243

impact on individual firm’s innovation and productivity upgrading. High level of labor cost may frighten off new entrants. Existing firms in high-cost cities tend to downsize rather than exit. However, high labor cost is often associated with high labor quality, leading to high productivity growth. The interaction term, ER*PI, has a positive (negative) effect on firm exit (entry), which is consistent with theoretical prediction. This means that if a firm belongs to pollution-intensive industries, it is more likely to exit in cities with stringent environmental regulation and less likely to enter such cities, supporting the arguments made by the PHH. Firms tend to upsize in response to stringent environmental regulation (Model 3). One possible explanation is that firms do so to be more capable of spreading compliance and administrative costs over high production volumes. High compliance and administrative costs caused by stringent environmental regulation may also have some negative impacts on firms’ productivity growth (Model 4). Since we seek to examine if the relationship between environmental regulation and firm dynamics is affected by firm heterogeneity and local government intervention, we focus on the coefficients of interaction terms between ER*PI and firmspecific/city-specific variables in Models 5–8. The key findings are as follows. First, even though environmental regulations have a significant impact over firm dynamics in pollution-intensive industries, their influences over large firms are relatively weaker. Specifically, in pollution-intensive industries, large firms outperform small firms in terms of firm entry and employment growth in cities with strict environmental regulations, possibly due to large firms’ capabilities in spreading compliance and administrative costs over high production volumes and their strong political influence over local governments and EPBs. Second, firms with different ownership structure are also affected by environmental regulations differently, since they have different capabilities to absorb compliance costs and different bargaining power with the regulator. Pollution-intensive SOEs’ entry and exit are impacted by environmental regulations to a lesser extent than pollution-intensive POEs and FOEs. This resonates with a number of similar studies. Dean et al. (2009) suggest POEs have less bargaining power than SOEs and the latter are more capable to escape sanctions. Wang and Wheeler (2005) stress that state ownership is closely interconnected with the effective levy on firms’ air pollution in China. Furthermore, Models 7–8 show the other half of the story that has received less attention in recent studies. Even though pollution-intensive SOEs are more likely to downsize while facing stringent environmental regulations, what is interesting is they are more likely to innovate and upgrade than POEs. One possible explanation is SOEs have softer budget constraints and better access to financial resources, while POEs often suffer from great credit constraints (see Dollar and Wei 2007 on the systematic misallocation of financial resources in China). Likewise, FOEs are also more capable to innovate and increase productivity, due to financial and technological supports received from their mother companies. In short, these findings based on the dynamics of certain types of firms support the PH. Finally, another factor contributing to the coexistence of the PH and the PHH is related to the role of government intervention at the local level. Entry of pollution-

244

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

intensive firms located in government-designated industrial parks is affected by environmental regulations to a lesser extent, since infrastructure such as centralized waste treatment systems may lower firms’ compliance costs. In contrast, firms receiving government subsidies are more likely to exit and less likely to upgrade, indicating that such firms are less efficient and tend to rely on government subsidies rather than innovation and industrial upgrading. The reason that two types of government intervention—industrial parks and government subsidies—have different impacts over the articulation between environmental regulations and firm dynamics is possibly because the former reflect supportive government policies and capture the “facilitating” role of local governments, whereas the latter represent direct government intervention in the form of subsidies indicating to what extent local governments play an “intervening” and hands-on role in regional economic development. Industrial parks would help foster a nurturing environment for entrepreneurial activities and reinforce knowledge spillovers, while high level of direct subsidies may discourage firms from innovating and make them become increasingly reliant on government supports. As a robustness check, we re-estimate all models (1) by defining ER as the percentage of industrial wastewater meeting standard for discharge in city k in year t
1 and PI as the portion of firms in industry j that have been included in the KNMSPF water pollution list and (2) by using the number of staff in local EPB as a proxy of environmental regulation stringency in China (see Table 11.3).1 Compared with the results presented above, these changes produce only minor effects. Due to space limitation, these results are not reported here.

11.5

Conclusion

Recent studies on environmental regulation and industrial dynamics tend to predicate either on the PH or on the PHH and argue that environmental regulations may either render firms less competitive due to the additional costs required to comply with regulations and force them to exit or relocate or encourage firms to upgrade their production through industrial innovations. Existing studies thus fall short in uncovering the whole picture where various interconnected factors all have potentials to inflect the relationship between environmental regulations and industrial dynamics. In addition, most researches have either adopted quantitative methods analyzing industrial dynamics at the industry and regional level or employed qualitative analyses such as case studies. Findings from the former approach have limitations due to the high level of aggregation, while the latter researches could be biased due to the limited sample size. This chapter seeks to contribute to recent

1 This indicator has been used in some recent studies (He et al. 2012, 2013). However, data is only available at the provincial level. We hence prefer our current indicator to this one.


0.0565*** 0.0002***


0.0195***
0.0850*** 0.00009***
0.00002 0.0347*** Included Included Included Included Included Included 0.157*** 1,452,688 0.023


0.108***
0.0005

0.0065***
0.089***
0.0002***
0.001*** 0.122*** Included Included Included Included Included Included 0.214*** 1,174,871 0.033


0.0122***
0.0066 0.0119*** 0.0984*** 0.0010 0.0094**
0.0011*** 0.02044 Included Included Included Included Included Included
0.2731*** 850,862 0.003

(3) Growth 0.0308***

0.0572***
0.877***
0.412*** 0.0007
0.00411
0.0001*** 0.004*** 0.0342*** Included Included Included Included Included Included 1.266*** 621,655 0.239

(4) TFP
0.0324***

0.0048***
0.0855***
0.0002***
0.0001*** 0.119***
0.0369*** 0.0414*** 0.0895*** Included Included Included 0.113*** 1,174,892 0.035

(5) Entry 0.0117* 0.010***
0.0280***
0.0568***
0.0867*** 0.0364***
0.335***
0.142***
0.0005
0.0189***
0.0821*** 0.00009***
0.0003 0.0350*** 0.0470***
0.0365***
0.0207*** Included Included Included 0.125*** 1,452,712 0.023

(6) Exit 0.0452*** 0.004***
0.0901***
0.0161*
0.0304***
0.0241** 0.0727***
0.0664*** 0.0002***

(7) Growth
0.0234* 0.0904***
0.2859***
0.0214
0.0529 0.0970***
0.0469
0.0390***
0.00104 0.0120*** 0.0938*** 0.0049 0.0103**
0.0912***
0.0382 0.0120*** 0.0170*** 0.0029 Included Included Included
0.5927 850,864 0.012

(8) TFP
0.127***
0.003 0.225*** 0.352*** 0.0862***
0.0057
1.151*** 0.0587***
0.863***
0.412*** 0.0024
0.00623
0.0001*** 0.00407*** 0.0352***
0.138***
0.0387***
0.0058 Included Included Included 1.273*** 621,655 0.239

Conclusion

*p < 0.1, ** p < 0.05, *** p < 0.01

ER*PI ER*PI*SIZE ER*PI*SOE ER*PI*FOE ER*PI*POE ER*PI*PARK ER*PI*SUBSIDY SIZE SUBSIDY TFPt
1 PARK LQ Theil LAB GROW SOE FOE POE Industry Province Year _cons N R2

(2) Exit 0.0144***

(1) Entry
0.0460***

Table 11.3 Empirical results

11.5 245

246

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

studies by examining the relationship between environmental regulation and industrial dynamics at the firm level based on both the PHH and the PH. By combining a firm-level industrial dataset of China’s manufacturing industries and a dataset on polluting firms in China, we show the coexistence of the PHH and the PH in China, linking this story to two key factors that have been largely overlooked in existing studies—firm heterogeneity and government intervention. Empirical results confirm that large and politically influential firms are in a better position to cope with increasingly stringent environmental regulations in China as they face lower per unit compliance costs. Their strong bargaining power may also lessen the deterrent impact of regulations on themselves. Thanks to softer budget constraints, reduced obligations to comply with environmental regulations, and better access to financial resources, SOEs may maintain their business and even upgrade in some cases despite the issuance of new, stricter environmental laws, whereas POEs are compelled to adjust by downsizing their production due to increased costs. Finally, facilitating local governments play a more helpful role in the development of pollution-intensive industries than intervening ones. Theoretically, this research contributes to recent debates on the relationship between environmental regulations and industrial dynamics not only by emphasizing the coexistence of the PH and the PHH but also by bringing firm heterogeneity and local government intervention to the forefront and investigating how firms with different attributes in different cities act differently as the stringency of environmental regulations increases. Empirically, although in recent literature SOEs are often seen as having a bad reputation with respect to their environmental performance and lack of obligations to comply with regulations in China, our research shows they are actually investing in innovation in response to stringent environmental regulations. Large firms also play a critical role in China’s transition toward environmentfriendly development model. More importantly, local government should serve as a facilitator, whereas an “intervening” local government tends to have minor, if not negative, impacts over the sustainable development of pollution-intensive industries. In drawing from analyses of firm-level data from 1998 to 2008, we realize the potential limitations of our study. First, our firm-level dataset does not enable us to examine the industrial dynamics of small firms. Nonetheless, this flaw only slightly affects research results, because a comparison with the 2004 and 2008 full census of industrial firms reveals that firms in our dataset employed roughly 70% of the industrial workforce and generated more than 90% of output and exports. Second, with an increasing number of pollution incidences, such as those hazy and foggy days in Beijing since 2013, reported by the media, more attention has been directed toward reducing pollution and to promoting clean technology in the 2010s. However, this dataset only covers the years of 1998–2008. More recent data including both small and large firms is needed to better understand the interaction between changing environmental regulations and industrial dynamics. Future research could also explore the mechanisms underlying the interaction between environmental regulations, industry dynamics, firm heterogeneity, and government intervention by employing qualitative methods. Finally, this chapter only examines the role of industrial parks and subsidies—one facet of institutional context. Other facets of

References

247

regional institutions should also play a role, which we leave for future research as well.

References Ambec, S., Cohen, M. A., Elgie, S., & Lanoie, P. (2013). The Porter hypothesis at 20: Can environmental regulation enhance innovation and competitiveness? Review of Environmental Economics and Policy, 7(1), 2–22. Bartel, A. P., & Thomas, L. G. (1987). Predation through regulation: The wage and profit effects of the occupational safety and health administration and the environmental protection agency. The Journal of Law and Economics, 30(2), 239–264. Birdsall, N., & Wheeler, D. (1993). Trade policy and industrial pollution in Latin America: Where are the pollution havens? The Journal of Environment and Development, 2(1), 137–149. Bommer, R. (1999). Environmental policy and industrial competitiveness: The pollution-haven hypothesis reconsidered. Review of International Economics, 7(2), 342–355. Cao, G., Feng, C., & Tao, R. (2008). Local ‘land finance’ in China’s urban expansion: Challenges and solutions. China & World Economy, 16(2), 19–30. Clasen, S., Rempp, H., Schmidt, D., Schraml, C., Hoffmann, R., Claussen, C. D., et al. (2012). The local environmental regulatory regime in China: Changes in pro-environment orientation, institutional capacity, and external political support in Guangzhou. Environment and Planning A, 44(10), 2493–2511. Copeland, B. R., & Taylor, M. S. (2004). Trade, growth, and the environment. Journal of Economic Literature, 42(1), 7–71. Dean, T. J., Brown, R. L., & Stango, V. (2000). Environmental regulation as a barrier to the formation of small manufacturing establishments: A longitudinal examination. Journal of Environmental Economics and Management, 40(1), 56–75. Dean, J. M., Lovely, M. E., & Wang, H. (2009). Are foreign investors attracted to weak environmental regulations? Evaluating the evidence from China. Journal of Development Economics, 90(1), 1–13. Dollar, D., & Wei, S. (2007). Das (wasted) Kapital: Firm ownership and investment efficiency in China (Working Paper 13103). National Bureau of Economic Research. Golombek, R., & Raknerud, A. (1997). Do environmental standards harm manufacturing employment. Scandinavian Journal of Economics, 99(1), 29–44. He, C., Wei, Y., & Xie, X. (2008). Globalization, institutional change and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. He, C., Pan, F., & Yan, Y. (2012a). Is economic transition harmful to China’s urban environment? Evidence from industrial air pollution in Chinese cities. Urban Studies, 49(8), 1767–1790. He, C., Zhang, T., & Wang, R. (2012b). Air quality in urban China. Eurasian Geography and Economics, 53(6), 750–771. He, C., Huang, Z., & Ye, X. (2013). Spatial heterogeneity of economic development and industrial pollution in urban China. Stochastic Environmental Research and Risk Assessment, 28(4), 767–781. He, C., Chen, T., Mao, X., & Zhou, Y. (2016). Economic transition, urbanization and population redistribution in China. Habitat International, 51(2), 39–47. Hering, L., & Poncet, S. (2014). Environmental policy and exports: Evidence from Chinese cities. Journal of Environmental Economics and Management, 68(2), 296–318. Heyes, A. (2009). Is environmental regulation bad for competition? A survey. Journal of Regulatory Economics, 36(1), 1–28. Hicks, J. R. (1932). The theory of wages. London: Macmillan.

248

11

How Do Environmental Regulations Affect Industrial Dynamics in China?

Huang, Y. (2003). One country, two systems: Foreign-invested enterprises and domestic firms in China. China Economic Review, 14(4), 404–416. Jeppesen, T., & Folmer, H. (2014). The confusing relationship between environmental policy and location behaviour of firms: A methodological review of selected case studies. The Annals of Regional Science, 35(4), 523–546. Jeppesen, T., List, J. A., & Folmer, H. (2002). Environmental regulations and new plant location decisions: Evidence from a meta-analysis. Journal of Regional Science, 42(1), 19–49. Kearsley, A., & Riddel, M. (2010). A further inquiry into the pollution haven hypothesis and the environmental Kuznets curve, Ecological Economics, Special Section: Coevolutionary Ecological Economics: Theory and Applications, 69(4), 905–919. Kumar, S., & Managi, S. (2009). Win–win opportunities and environmental regulation: Test of the porter hypothesis. The Economics of Sustainable Development, natural resource management and policy 32 (pp. 157–66). Springer. List, J. A., Millimet, D. L., Fredriksson, P. G., & McHone, W. W. (2003). Effects of environmental regulations on manufacturing plant births: Evidence from a propensity score matching estimator. Review of Economics and Statistics, 85(4), 944–952. Martin, R. (2010). Roepke lecture in economic geography—Rethinking regional path dependence: Beyond lock-in to evolution. Economic Geography, 86(1), 1–27. Murty, M. N., & Kumar, S. (2003). Win–win opportunities and environmental regulation: Testing of porter hypothesis for Indian manufacturing industries. Journal of Environmental Management, 67(2), 139–144. Olley, G. S., & Pakes, A. (1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64(6), 1263–1297. Palmer, K., Oates, W. E., & Portney, P. R. (1995). Tightening environmental standards: The benefit-cost or the no-cost paradigm? The Journal of Economic Perspectives, 9(4), 119–132. Pashigian, B. P. (1984). The effect of environmental regulation on optimal plant size and factor shares. Journal of Law and Economics, 27(1), 1–28. Poncet, S. (2005). A fragmented China: Measure and determinants of Chinese domestic market disintegration. Review of International Economics, 13(3), 409–430. Porter, M. (1991). America’s green strategy. Scientific American, 264(4), 193–246. Porter, M. E., & van der Linde, C. (1995). Toward a new conception of the environmentcompetitiveness relationship. The Journal of Economic Perspectives, 9(4), 97–118. Sutherland, D. (2003). China’s large enterprises and the challenge of late industrialisation. London: Routledge. Tao, R., Su, F., Liu, M., & Cao, G. (2014). Land leasing and local public finance in China’s regional development: Evidence from prefecture-level cities. Urban Studies, 47(10), 2217–2236. Tobey, J. A. (1989). Economic development and environmental management in the third world. Habitat International, 13(4), 125–135. Tole, L., & Koop, G. (2008). Do environmental regulations affect the location decisions of multinational gold mining firms? International Conference on Distributed Computing Systems, 24(11), 2–9. Tole, L., & Koop, G. (2011, 1 January). Do environmental regulations affect the location decisions of multinational gold mining firms? Journal of Economic Geography, 11(1), 151–177. Vogel, D. (2003). The hare and the tortoise revisited: The new politics of consumer and environmental regulation in Europe. British Journal of Political Science, 33(4), 557–580. Walker, H., Sisto, L. D., & McBain, D. (2008). Drivers and barriers to environmental supply chain management practices: Lessons from the public and private sectors. Journal of Purchasing and Supply Management, Practice Makes Perfect: Special Issue of Best Papers of the 16th Annual IPSERA Conference 2007, 14(1), 69–85. Wang, H., & Jin, Y. (2002). Industrial ownership and environmental performance: Evidence from China. World Bank Publications. Wang, H., & Jin, Y. (2006). Industrial ownership and environmental performance: Evidence from China. Environmental and Resource Economics, 36(3), 255–273.

References

249

Wang, H., & Wheeler, D. (2005). Financial incentives and endogenous enforcement in China’s pollution levy system. Journal of Environmental Economics and Management, 49(1), 174–196. Wang, H., Mamingi, N., Laplante, B., & Dasgupta, S. (2003). Incomplete enforcement of pollution regulation: Bargaining power of Chinese factories. Environmental and Resource Economics, 24 (3), 245–262. Wei, Y. H. D. (2001). Decentralization, marketization, and globalization: The triple processes underlying regional development in China. Asian Geographer, 20(1–2), 7–23. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yang, X., & He, C. (2015). Do polluting plants locate in the borders of jurisdictions? Evidence from China. Habitat International, 50(12), 140–148. Zhan, X., Lo, W. H., & Tang, S. Y. (2014). Contextual changes and environmental policy implementation: A longitudinal study of street-level bureaucrats in Guangzhou, China. Journal of Public Administration Research and Theory, 24(4), 1005–1035. Zhang, J., & Fu, X. (2008). FDI and environmental regulations in China. Journal of the Asia Pacific Economy, 13(3), 332–353. Zhang, B., Bi, J., Yuan, Z., Ge, J., Liu, B., & Bu, M. (2008). Why do firms engage in environmental management? An empirical study in China. Journal of Cleaner Production, 16(10), 1036–1045. Zhao, X., & Zhang, L. (1999). Decentralization reforms and regionalism in China: A review. International Regional Science Review, 22(3), 251–281. Zhu, S., & He, C. (2013). Geographical dynamics and industrial relocation: Spatial strategies of apparel firms in Ningbo, China. Eurasian Geography and Economics, 54(3), 342–362. Zhu, S., He, C., & Liu, Y. (2014). Going green or going away: Environmental regulation, economic geography and firms’ strategies in China’s pollution-intensive industries. Geoforum, 55(55), 53–65.

Chapter 12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven Industry Space

12.1

Introduction

Some recent evolutionary economic geography (EEG) empirical studies tend to overly focus on technological relatedness as a key explanatory factor for regional development and argue that it not only pushes forward the growth of existing industries through agglomeration externalities derived from related variety but is also responsible for the formation of new growth paths (Neffke et al. 2011; Boschma et al. 2013; Boschma and Capone 2015a; Delgado et al. 2016). They further stress that new growth paths do not emerge from scratch, but evolve out of preexisting regional industrial structures, because the set of competences and assets that region possesses determines what new paths and new industries this region is able to develop (Hidalgo et al. 2007; Neffke et al. 2011; Boschma et al. 2012). If a region already has most of the competencies that a certain new industry requires, it is easy for this region to jump onto this new path. If not, the barrier to develop this industry could be too high for this region to overcome (Boschma et al. 2013). In short, regions tend to diversify into new industries that are related with preexisting regional industrial structure, and relatedness among industries affect the ways in which regions create new paths over time. By emphasizing industrial relatedness, Hidalgo and Hausmann (2008) have directed our attention away from a one-dimensional, GDP-centered view of development where a region’s development is measured by its upward movement on a GDP ladder (or ramp) regardless of the products and services the region produces, toward a two-dimensional network view of development. In the “network” metaphor, regions are thought as monkeys jumping among trees (i.e., industries) in a big,

Modified article originally published in [Zhu, S., C. He and Y. Zhou (2017) How to Jump Further and Catch Up? Path-Breaking in an Uneven Industry Space, Journal of Economic Geography, 17 (3), pp. 521–45.]. Published with kind permission of © [Oxford University Press, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_12

251

252

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

heterogeneous forest (i.e., a two-dimensional, uneven industry space/network where different industries are related with each other to different extents). Monkeys can only jump onto trees within certain distance (i.e., relatedness between two industries). In other words, regions are more likely to “jump” (or diversify) into industries that are closely related to their existing industries. Development of regions is thus shaped by technological relatedness among industries. In addition, the monkey’s position in the forest also affects its jumping trajectories. It would be easier for monkeys in dense areas to jump to neighboring trees than their counterparts in more deserted areas in the heterogeneous forest. Likewise, Hidalgo et al. (2007) have argued that developed countries that start from core, dense areas in the uneven industry space/network have more opportunities to jump to new related industries and therefore have more opportunities to sustain economic growth than do developing countries that jump from peripheral, deserted areas. Boschma et al. (2013) and Neffke et al. (2011) further pointed out that this process of related diversification and path-dependent economic development is more phenomenal at the regional level. Even though recent EEG studies do not exclude the possibility that developing countries/regions can enter core, dense areas in the uneven industry space/network, the future for them is not quite promising. In some extreme cases, developing countries/regions may only be able to wander around in peripheral areas of the industry space and have no capabilities to jump into core areas (see Hidalgo et al. (2007) for an example where the divergence between developing countries and developed countries persists forever and it is impossible for developing countries such as Chile to catch up with developed ones no matter how much time the former have). Overall, the development prospect for developing countries/regions is dimmed, if not hopeless, due not only to their peripheral starting points but also to the fact that rich countries are capable to jump further than poor ones (see Boschma and Capone (2015b) for an example where richer countries in Europe are more capable to diversify into less related industries). Such a pessimistic, deterministic conclusion stems from an overemphasis on technological relatedness and has recently attracted several stands of criticism, particularly in economic geography, two of which will be examined in this chapter. First, the role of other region-specific assets has been largely downplayed in EEG’s explanation of regional evolution. By focusing on technological relatedness, related variety, and path-dependent industrial diversification, EEG tends to pay insufficient attention to the role of firm agency, local social and institutional contexts, and policy-making in creating and/or renewing industrial development paths in a region (Martin 2010; Sydow et al. 2010; Simmie 2012; Dawley 2014; Hassink et al. 2014; Tanner 2014) and to their complicated interactions within the “local ecosystem” (Bathelt and Cohendet 2014). Second, by emphasizing endogenous regional branching and industrial diversification processes (Grillitsch and Trippl 2014; Tanner 2014), EEG risks embracing a regional fetishism (Martin and Sunley 2006; Binz et al. 2016), as it overlooks extra-regional linkages that may contribute to the renewal and restructuring of the regional resource base and the formation of specific regional growth paths (Bathelt et al. 2004; Maskell et al. 2006; Bathelt and Cohendet 2014; Maskell 2014).

12.2

Path Dependence and Path-Breaking

253

Based on these critiques, this chapter seeks to shed new light on regional development, by asking questions from a different angle: can developing countries/regions transcend the “confinement” of technological relatedness to catch up with developed ones? If yes, developing countries/regions have to jump further in the industry space. A follow-up question is: how to jump further in the industry space/network? In this chapter, we use a developing country with a high degree of regional disparity—China—as an example to explore whether developing regions/ countries can jump further and through what ways and what is the role of extraregional linkages, firm agency, institutions, and policy-making in this “catch-up” story. The next section will present an analytical framework and develop hypotheses. The third section introduces the data, variables, and specifications for empirical analysis. After presenting some descriptive analyses in the fourth section, the fifth section discusses the empirical results. The last section concludes the chapter and points out the theoretical and empirical significances of our central question.

12.2

Path Dependence and Path-Breaking

Before examining regional economic development and industrial restructuring through the lens of technological relatedness, it is useful to make some clarifications of the notion path dependence, which is arguably the core concept of the regional diversification or “regional branching” model (Neffke et al. 2011; Boschma et al. 2012, 2013; Boschma and Capone 2015b). In this chapter, we differentiate two types of new path creation (Fig. 12.1). The first one takes place when regions rely on the technological relatedness among industries (distance between trees) in order to jump into new industries. This type of new path creation is path-dependent, since regional diversification is determined not only by region’s position in the industry space and the density in the vicinity but also by the relatedness between industries. As regions jump according to technological relatedness, technologies evolve over time through cycles of long periods of incremental innovations, which enhance and institutionalize an existing productive structure. This path-dependent process means that there is some degree of cohesion in the industrial structure of a region (Neffke et al. 2011). When this type of new path creation dominates, it is difficult for developing countries/regions to catch up, since each region follows its own industrial trajectory (Rigby and Essletzbichler 1997). However, regional industrial trajectory is constantly redirected or redefined through the process of creative destruction (Schumpeter 1947). New paths can be created by technological discontinuities in which regions replace old, inferior technologies/industries with new, radically superior technologies/industries (Baum 1996). In this case, regions rely less on technological relatedness and jump directly to less related or even unrelated industries. While the first type is path-dependent where new industries are created based on tangible and intangible assets embodied in existing industrial structure, the second type of path creation is more path-breaking driven by technological discontinuities and radical innovations. Unlike recent

254

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Path Creation

Extra-regional linkages  

FDIs Imports Import rs

Internal innovation   

Firm agency Policy making Local social and insti t tu t ti t onal institutional contexts

Technological relatedness and incremental innovations

Technological discontinuities discontinu n ities and a d an radical innovat innovations a ions

Path dependence

Path breaking Pat a h breaki k ng

Hypothesis 1 Hypothesis 2 Regional economic development

Hypothesis 3

Fig. 12.1 Two types of new path creation and regional development

empirical EEG studies that focus heavily on the roles of relatedness and preexisting industrial structure in shaping regional diversification, we hypothesize that regions can jump further and regional development may depend less on technological relatedness (Hypothesis 1). The path-dependent regional diversification model has already been testified by recent empirical, quantitative studies based on export and plant-level data in developed countries (Neffke et al. 2011; Boschma et al. 2012, 2013). In this research, based on export data in a developing country, we seek to focus on the second type of path creation and examine how developing countries/ regions can diversify in a more path-breaking way, so that the “confinement” of technological relatedness can be transcended. Two sets of factors that have the potential to enable developing countries/regions to jump further are identified: extra-regional linkages and internal innovation (Fig. 12.1).

12.2.1 Extra-regional Linkages The standard canonical path dependence model portrays the regional industrial evolution as a four-phase development: (1) path creation, where historical accidents initiate a new path and have significant long-run effects on the technological, industrial, and institutional structure of a region; (2) path development, where emergence and development of local increasing returns and externalities assist the

12.2

Path Dependence and Path-Breaking

255

development of the path; (3) path rigidification, characterized by increasing rigidification of knowledge, networks, and structures of firms; and (4) path de-locking, where an exogenous shock disrupts or dislodges the regional economy resulting in an atrophy or industrial restructuring (David 1985; Arthur 1989, 1994; Martin and Sunley 2006; Martin 2010). After criticizing the standard canonical path dependence model’s overemphasis on continuity and stability, Martin (2010) has suggested a second type of trajectory which is more open and allows for constant endogenous change. His model diverges from the canonical one in the third step and proposes a new phase three where local industry is able to adapt and mutate constantly, which prevents it from being trapped in a stable, inflexible, and rigid state that only can be destabilized by an external shock. This idea has been further developed by Martin and Sunley (2011) when they employ a modified regional adaptive cycle model and introduce much more diverse trajectories of regional economic evolution. The gist of their argument for our present purposes is that apart from being stuck in a state of fixity and rigidification and waiting for an unpredictable external shock to set it free, a regional economy can evolve along another trajectory where firms in the region are able to innovate more or less continuously and the industrial structure constantly mutates and adapts. This second trajectory can be realized by keeping the regional economy relatively open (Hassink 2005). The openness of a region is partly supported by the diverse overlaps between organizations and institutions inside and outside the region and the subsequent information, technology, and knowledge exchange across regional borders (Sydow et al. 2010). The kind of understanding and learning that derives from participation in various kinds of links with others beyond the region is referred to as “pipelines” or extra-regional linkages (Bathelt et al. 2004; Boschma and Iammarino 2009). Extra-regional linkages that connect actors inside and outside the region may be important in enabling firms in the region to avert tendencies toward path dependence in the evolution of regional economy, thus enabling the region to remain innovative and competitive (Bathelt et al. 2004; Sydow et al. 2010; Bathelt and Li 2014). The idea that nurturing connections with distant actors may help prevent systematic industrial rigidification is also supported by Maskell and Malmberg (2007) who have argued that, from a microlevel perspective, localized learning and knowledge development often lead to overreliance on localized routines and over-embeddedness in the existing structure, what Maskell and Malmberg called “spatial myopia.” The potentially devastating long-run effects of spatial myopia may be avoided as long as some firms in the region actively invest in establishing linkages to extra-regional knowledge pools with dissimilar routines or institutional patterns. Rigidification and path dependence are therefore alleviated by rejuvenation processes where externally connected local firms are able to keep importing fresh knowledge and state-of-the-art technology. By deliberately investing in building extra-regional linkages to distant communities, regions may be able to increase the variety of knowledge, resources, and capabilities available to them and escape the potential limits stemming from their peripheral starting point in the industry space and the confinement of relatedness (Fig. 12.2). We thus hypothesize that since extra-regional linkages bring in fresh

256

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Firm B

Government

R&D

Regional institutional and social context

Firm A

Subsidiary of TNC

Firm C

R&D

Region 1

Region 2

Physical capital and infrastructure Human capital HQs and other branches of TNC

Financial Supports Government-firm interaction (government support) Intra-regional linkages

Region 3

Extra-regional linkages (Import and FDI linkages)

Fig. 12.2 Extra-regional linkages and internal innovation. (HQs, headquarters; TNC, transnational corporation; FDI, foreign direct investment)

know-how and technologies that are less related to region’s existing productive structure, it may enable regions to jump further (Hypothesis 2).

12.2.2 Internal Innovation Not only has the canonical path dependence model overlooked Martin’s second trajectory of cluster evolution, but it also rarely pays attention to the role of individual agency in affecting path-dependent processes (David 1985; Arthur

12.2

Path Dependence and Path-Breaking

257

1989, 1994; Martin 2010; Henning et al. 2013). Once a regional economy enters into the third phase in the traditional path dependence model, it is assumed that it will remain or be trapped in path-dependent development until it is disturbed or liberated by some unpredictable and unexpected exogenous shock (Martin 2010; Martin and Sunley 2011). This focus on exogenous impacts has meant that the role of individual agency in reworking or disrupting forms of path dependence and creating new pathways is much less well developed over against the much stronger analytical focus on the role of external shocks in dislodging stable, inflexible, or rigid state of regions. To counter this lacuna, Martin and Sunley (2011) have suggested that, while the path dependence of firms, networks, and structures in a region may limit the vitality and adaptability of the region resulting in an atrophy, it may also encourage or enable a reorganization of resources and greater opportunities for surviving firms, or it may force purposeful actions by individual actors in a region (firms, industry associations, and governments) who are deliberately trying to de-lock themselves. In this way, the entire region can dislodge itself from an old path and create a new one (Sydow et al. 2010). As a result, there are reasons to consider path-dependent process of regional economic evolution as possibly being shaped by purposeful actions of actors. In this chapter, we build on these insights to evaluate the relative roles of purposeful actions of actors within the region in shaping path-dependent processes as they inflect regional economic evolution through internal innovation. Here by purposeful actions of actors, we are referring not only to the corporate strategies and innovation implemented by enterprises but also to various supporting facilities and services and financial, political, and technological aids provided by regional governments. In the Chinese case, in particular, it is vital that the evolution of regional economy should be examined in ways that recognize the strategic intents of both regional governments and enterprises to innovate and adopt new technologies/ products. Li et al. (2012) have demonstrated how new regional paths can be created through collective mobilization of local agents in a low-tech industrial cluster, while Wang et al. (2010) and Lin et al. (2011) have focused on the high-tech ICT industry in China and concluded that most manufacturers obtained their core technology largely through internal R&D activities. Actions of local actors and internal innovation are further embedded in regional social and institutional contexts that include political regimes, market conditions, norms, value and belief systems, and local conventions and culture (Storper and Walker 1989; Li et al. 2012). Such contexts affect entrepreneurship spirit and are thus closely related to regional industrial diversification. Context can both constrain and enable actions of local actors, as it has an impact on the ways in which local actors search for and adopt information and knowledge. The effects of context on actions of local actors are not predetermined as they can be positive or negative (Storper 2009). On the negative side, parochial and conservative social-institutional environment may inhibit internal innovations and the adoption of fresh knowledge, whereas, on the positive side, contexts may facilitate such processes by creating an open-minded, supportive social and business atmosphere.

258

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Ideas developed in this section lead to the following hypothesis: internal innovations, which are not only driven by purposeful actions of local actors but also embedded in regional social and institutional contexts (Fig. 12.2), may also generate radical technological shifts and enable regions to break old pathways and jump to less related industries (Hypothesis 3).

12.3

Research Design

12.3.1 Model Specification The proximity indicator between industries is computed using Eq. 1.1 and 1.2. This chapter uses the CCTS dataset. In this research, we focus on 4-digit level industries (1080 industries in the dataset1). The geographical unit of analysis is China’s prefecture-level cities. A matrix of proximity indicators among all 4-digit industries can be estimated. This 1080*1080 matrix therefore defines the industry space. We then use Eq. 1.5 to calculate the density indicator of industry i in city c in year t (di,c,t). The following econometric equation is estimated: xi,c,t2 ¼ α0 þ α1 xi,c,t1 þ α2 di,c,t1 þ β1 EXT i,c,t1 þβ2 INT i,c,t1 þ γ 1 xi,c,t1 d i,c,t1 EXT i,c,t1 þγ 2 ð1  xi,c,t1 Þdi,c,t1 EXT i,c,t1 þ γ 3 xi,c,t1 di,c,t1 INT i,c,t1 þγ 4 ð1  xi,c,t1 Þdi,c,t1 INT i,c,t1 þ δX þ πY þ εi,c,t1

ð12:1Þ

where t2 > t1 and EXT and INT represent extra-regional linkage and internal innovation variables, respectively. Y represents control variables. X is a vector of city-year and industry-year dummy variables, which is added to control any timevarying city or industry characteristics. Dependent variable takes the value of 1 if city c has a comparative advantage in industry i in year t2 and zero otherwise. Following Hausmann and Klinger (2007) and Boschma et al. (2013), we distinguish between the effect of independent variables on keeping a comparative advantage in current industries that are already part of the existing industrial structure of a city and their contribution in developing a comparative advantage in new industries that were not part of the preexisting industrial structure of a city. γ 1 and γ 3 capture the impact of specific independent variables in keeping a comparative advantage in industry i, while γ 2 and γ 4 capture the impact of specific independent variables in developing a comparative advantage in new industries. High level of di,c,t indicates that the distance between industry i and city c’s existing industrial structure is small, and the positive effect of density thus means that regions jump into related industries and regional economic development is pathdependent. Our hypothesis is that two sets of variables (EXT and INT) may change

1

We focus on the secondary industry and therefore exclude data on agriculture.

12.3

Research Design

259

regions’ jumping capability, reducing (or strengthening) regions’ reliance on relatedness among industries while jumping. Therefore, the impact of density should vary across regions as the latter are different from one another in terms of extraregional linkages and internal innovation. To test our hypothesis, we follow Boschma and Capone (2015a) and include the interaction terms between the density indicator and EXT/INT variables in Eq. (12.1). A positive and significant sign of the interaction term indicates that a specific EXT/INT variable enhances regions’ reliance on density while jumping, whereas a negative and significant sign means a weaker effect of density (Boschma and Capone 2015a, b). In the latter case, regions’ certain characteristics in terms of extra-regional linkage and internal innovation reduce the confinement of relatedness, enable regions to jump further to less related industries, and finally allow regions to become path-breaking. A nonsignificant sign is also possible, suggesting the impact of density does not vary across regions because regions have different levels of extra-regional linkages or internal innovation.

12.3.2 Variables First, we seek to test the effects of extra-regional linkages that breathe new life into a region through a diversified set of import sectors and through foreign direct investment (FDI). On the one hand, imports expand the set of inputs available in the economy and thus increase regions’ productivity (Amighini and Sanfilippo 2014). The rising availability of inputs may encourage the creation of new domestic varieties (Goldberg et al. 2010). Imports can also provide more sophisticated inputs that enable regions to upgrade their production and export. On the other hand, more importantly, there is a certain degree of new knowledge embedded in imported products, which could translate into new learning opportunities involved in the use of new products (Dollar 1992; Schiff and Wang 2006). As a result, IMPORTi,c,t is included, measured as the import value of city c in industry i and year t. FDI also plays a critical role in promoting regional economic development through a variety of channels, such as the formation of forward and backward linkages, the existence of competitive and demonstration effects, the possibility for domestic firms to recruit more experienced and skilled workers that are released from foreign-owned firms, and finally the knowledge spillover effect between domestic and foreign-owned firms (Görg and Greenaway 2004; Lall and Narula 2004; Poncet and Starosta de Waldemar 2013; Zhu and Fu 2013). Foreign-owned firms are important for regional economy, as they not only contribute to productivity increase in existing industries but, more importantly for our present purpose, they also bring new knowledge and ideas that may enable regions to break their old paths and jump into less related industries (Amighini and Sanfilippo, 2014). Variable FDIi, c,t is calculated as the share of output of foreign-owned firms in industry i, city c, and year t. For this indicator, we use data from China’s Annual Survey of Industrial Firms (ASIF). The problem is that this dataset uses the Standard Industrial Classification (SIC) system, while Chinese customs data is compiled in the Harmonized

260

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

System (HS) code. We concord FDI indicators for all 4-digit SIC industries (525 industries) at the city level to 4-digit HS industries (1080 industries)2. Not only can new knowledge and ideas that enable regions to jump further be acquired through access, transfer, and assimilation of external knowledge, but regions can also create new knowledge through indigenous innovation. Localized capabilities that include tangibles (e.g., R&D, human capital, physical capital/infrastructure, and government supports) and intangibles (e.g., social and institutional contexts) play a critical role in indigenous knowledge creation (Maskell and Malmberg 1999; Zhu and Fu 2013; Boschma and Capone 2015b). Regions with different types of R&D, human and physical capital, social and institutional contexts, and government policies are likely to have different jumping capabilities, and this will affect the process of regional diversification as well as region’s reliance on relatedness (Boschma and Capone 2015b). Human capital (HCAPc,t) is measured as the percentage of population with more than secondary schooling in city c and year t. We use the length of highway over land area in city c and year t as a proxy of physical capital and infrastructure (PCAPc, t). Investments in infrastructure and human capital would foster the formulation of a favorable environment for entrepreneurial activities and lower entry barriers and costs for new entrants (Lo Turco and Maggioni 2015). Data on human and physical capital are derived from China’s Population Census and China’s City Statistical Yearbook, respectively. R&Di,c,t is calculated as the R&D investment by enterprises in industry i, city c, and year t. Like FDI indicator, R&D indicator is also calculated by using the ASIF dataset. Government supports in different industries are computed as the difference between export tax rebates and tax rates in industry i and year t (REBATEi,t). Data for this indicator is taken from the Chinese customs data on export tax rebates and tax rates. Open-minded social-institutional contexts may also encourage internal innovations and the adoption of fresh knowledge. We use the number of KFC and McDonald stores per 10,000 people in city c and year t to measure a Chinese region’s capability to embrace new things and fresh ideas—i.e., social openness (SOc,t). Economic openness (EOc,t), or economic liberalization, is calculated as the proportion of non-state-owned enterprises’ output in the total in city c and year t. Finally, this chapter uses control variable—Theil index (Theilc,t)—to capture the diversity of a region’s preexisting industrial structure3.

2

As there are more 4-digit HS industries than 4-digit SIC industries, a SIC industry is often bigger than a HS industry, and the latter is often a subset of the former. In some cases, we have to use the FDI value in a big SIC industry as a proxy of that in a small HS industry. We admit this concordance is not precise. However, due to the unavailability of FDI data compiled in the HS code, we have to use data compiled in the SIC system. 3 Please see Theil (1972) for the calculation of Theil Index.

12.4

12.4

The Relationship Between Density and the Emergence of New Industries. . .

261

The Relationship Between Density and the Emergence of New Industries in Chinese Cities

We divide the time period 2002–2011 into two stages, 2002–2006 and 2007–2011, for two reasons. First, the HS code used by Chinese customs dataset was slightly adjusted in 2007. Second, these two stages both cover 4 years, which are long enough for new industries to emerge and for density to have a strong impact (Boschma et al. 2013). As notified above, city c is considered as having a comparative advantage in industry i if RCAi,c is above 1. Industry i is considered as a developed industry in city c. Otherwise, it is an undeveloped industry. If industry i is an undeveloped industry in 2002 (or 2007) and becomes a developed industry in 2006 (2011), it is named as a transition industry. Figure 12.3 shows the proportion of transition industries (4-digit) in five 2-digit industrial sectors in China’s five big regions: East, Central, Southwest, Northeast, and Northwest China. During 2002–2006, China’s five regions have converted around 5% 4-digit undeveloped industries into developed industries in five 2-digit industries, whereas during 2007–2011, transition industries have accounted for around 20% in East China, 10% in Central and Northeast China, and 7% in Northwest and Southwest China. This indicates a dramatic and vibrant process of industrial diversification in the entire country, particularly in Northeast, Central, and East China, where the share of transition industries has increased 2–3 times from the first stage to the second one. Such a rapid industrial diversification is evident not only in labor-intensive industries like textile but more importantly in technology- and capital-intensive industries. For example, in HS18, the share of transition industries in all five big regions has increased 3–4 times from the first stage to the second one, while that in laborintensive HS16 has increased 2–3 times. We then analyze the relationship between the average density of the industries without a comparative advantage in a city in 2002 (or 2007) and the probability of this city developing a comparative advantage in a new industry in 2006 (or 2011). As can be seen in Fig. 12.4a, on average, cities in relatively developed, wealthy East China have higher level of average density of the industries without a comparative advantage in 2002 and higher level of probability of developing a comparative advantage in a new industry in 2006 than do cities in the rest of China, while cities in Northwest China have the lowest scores on both aspects. There is a clear positive relationship between the average density of undeveloped industries in a city in 2002 and the probability of this city developing a transition industry in 2006. For example, the three dots in the upper, rightmost area are Shanghai, Nanjing, and Beijing from East China that have the highest level of average density of undeveloped industries in 2002 (0.375, 0.299, and 0.322, respectively) and the highest probability of developing a transition industry in 2006 (0.162, 0.116, and 0.112, respectively). In contrast, Guyuan and Longnan in Northwest China have the lowest level of average density of undeveloped industries in 2002 (0.0018 and 0.0022, respectively) and the lowest level of probability of developing a transition industry in 2006 (0.00092 and 0.00093, respectively). In Fig. 12.4a, all dots are close to the fitted regression line, R2

262

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Fig. 12.3 The share of transition industries in China’s five big regions during 2002–2006 and 2007–2011

of which is as high as 0.762, suggesting that Chinese regions were heavily reliant on industrial relatedness while jumping into new industries during 2001–2006. As is shown in Fig. 12.4b, during 2007–2011, even though the average density of undeveloped industries in a city in 2007 is still positively related with the probability of this city developing a transition industry in 2011, this relationship has changed. Dots become more scattered, and a considerable number of dots are located further away from the fitted regression line than they do in Fig. 12.4a. Although, for the entire country, R2 of the 2007–2011 fitted regression line (0.805) is higher than that of the 2002–2006 line (0.762), it is the opposite in four out of five big regions, East,

12.4

The Relationship Between Density and the Emergence of New Industries. . .

263

Fig. 12.4 (a) (top) and (b) (bottom) Relationship between the average density of the industries without a comparative advantage in a city in 2002 (or 2007) and the probability of this city developing a comparative advantage in a new industry in 2006 (or 2011)

264

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Central, Southwest, and Northeast China, where R2 has decreased from 0.738 to 0.636, 0.795 to 0.692, 0.762 to 0.707, and 0.814 to 0.796 from the 2002–2006 stage to the 2007–2011 stage, respectively. This means some other factors have reduced regions’ reliance on density in the process of industrial diversification, enhanced regions’ jumping capabilities, and further weakened the relationship between density and new path creation. This is particularly evident in the abovementioned four big regions, as some cities with low level of density in 2007 still have high probability of developing transition industries in 2011. In short, the evolution of regional industrial structure should not be understood solely based on industrial relatedness as well as the distance between new industries and regions’ existing industrial structure, since the interaction between industrial relatedness and regional diversification is increasingly inflected by other factors that enable regions to jump further.

12.5

Empirical Results

We follow Boschma et al. (2013) and estimate Eq. (12.1) with the linear probability model (LPM)4. Some highly correlated terms are separated into different models (Table 12.1). Table 12.2 reports the econometric results. First, density has a positive effect, which is consistent with theoretical prediction. The parameter of xi,c,t1 is also positive and significant, suggesting that having a comparative advantage at the beginning of the period raises the probability of having a comparative advantage at the end of the period. This echoes with the findings of Boschma et al. (2013). Second, although the coefficients of some are insignificant, most independent variables in Models 2, 3, 8, and 9 have positive and significant effects, indicating the important role of extra-regional linkages and internal innovation in developing or maintaining a comparative advantage in an industry. Control variable, Theil, has a negative impact in most models, suggesting that the more diverse a region’s preexisting industrial structure is, the fewer undeveloped industries there are for the region to jump to, and the harder this region can develop a comparative advantage in a new industry. Moving onto the results connected more closely with the central argument (Models 4–6 and 10–12), as clarified in Sect. 12.3.2, we distinguish between the effect of independent variables on keeping a comparative advantage in current industries that are already part of the existing industrial structure of a city (“on current” variables in Table 12.2) and their contribution in developing a comparative advantage in new industries that were not part of the preexisting industrial structure of a city (“on new” variables). We first focus on the role of the interaction terms

4 LPM does not need any distributional assumption to model unobserved heterogeneity and in general delivers good estimates of the partial effects on the response probability near the center of the distribution of the regressor. Nevertheless, probit model has also been estimated as a robustness check (see Table 11.5).

Density xi,c,t1 FDI Import R&D HCAP PCAP Rebate EO SO Theil

Density 1 0.408 0.611 0.067 0.050 0.517 0.339 0.054 0.297 0.460 0.802

0.177 0.025 0.009 0.152 0.099 0.018 0.087 0.134 0.232

xi,c,t1 1

Table 12.1 Correlation matrix

1 0.060 0.060 0.325 0.380 0.013 0.163 0.518 0.545

FDI

1 0.013 0.047 0.036 0.004 0.021 0.081 0.040

Import

1 0.114 0.019 0.039 0.036 0.043 0.058

R&D

0.260 0.207 0.045 0.431 0.448

1

HCAP

1 0.003 0.164 0.463 0.317

PCAP

1 0.156 0.001 0.135

Rebate

1 0.103 0.331

EO

1 0.310

SO

1

Theil

12.5 Empirical Results 265

0.000001

0.000001

0.111***

0.00005

0.0029***

0.0187***

R&D

HCAP

PCAP

Rebate

EO

on new

R&D*density

on current

R&D*density

on new

Import*density

on current

Import*density

new

FDI*density on

current

FDI*density on

SO

10***

1.34e-

10***

Import

0.0162***

0.009***

0.003***

1.34e-

0.00484

FDI

0.495***

0.494***

1.789***

xi,c,t1

Density

(6)

0.00003**

0.00003***

1.11e-09***

6.26e-10***

0.199***

0.0412

0.000003*

0.000003**

0.00003***

0.00003***

1.03e-09***

7.15e-10***

0.0015***

0.0132

0.00002*

0.00003***

1.01e-09***

6.61e-10***

0.0207*

0.0195***

0.00004

0.000002

8.54e-11

0.0445***

0.0892***

0.0048***

0.001***

0.321***

0.000001

1.52e-11**

9.81e-11*

1.063***

9.98e-11*

0.426***

0.330***

0.198***

0.451***

1.173***

(8)

0.0255***

0.437***

0.716***

(5)

(7)

(4)

2007–2011

(3)

(1)

(2)

2002–2006

Table 12.2 Determinants of having developed industries in China (LPM)

0.184***

0.00890**

0.00476***

0.000001

1.49e-11**

0.0483***

(9)

0.000004

0.00001***

7.27e-11

9.90e-11

0.501***

0.729***

0.000001

2.25e-11

0.147***

0.0179***

1.358***

(10)

0.00001*

0.00001***

3.37e-12

2.98e-11

0.00213***

0.0943***

0.000001

7.82e-13

0.0372***

1.184***

(11)

0.00001**

0.00002***

8.99e-11

1.11e-10

0.207***

0.0360***

0.0014***

0.000001*

2.89e-11

0.0432***

1.427***

(12)

381.4

F

*p < 0.10, **p < 0.05, ***p < 0.01

264,384

0.139

R2

0.0075

0.0882***

_cons

N

Included

Included

HS

841.5

0.276

263,227

Included

Included

Province

0.0192***

0.0540***

Theil

new

SO*density on

current

SO*density on

new

EO*density on

current

EO*density on

on new

Rebate*density

on current

Rebate*density

on new

PCAP*density

on current

PCAP*density

on new

HCAP*density

on current

HCAP*density

854.9

0.275

263,227

0.0373***

Included

Included

0.0223***

851.2

0.282

264,384

0.0142

Included

Included

0.0101***

831.0

0.285

261,165

0.0729***

Included

Included

0.0137***

0.0147***

0.0245***

1.730***

1.765***

815.5

0.283

264,384

0.0067

Included

Included

0.00784***

0.0970

0.309***

0.467***

0.738***

0.0001

0.003**

263.9

0.100

265,156

0.0176

Included

Included

0.0263***

221.6

0.090

266,986

0.134***

Included

Included

0.0149***

209.5

0.084

266,986

0.0720***

Included

Included

0.0285***

247.7

0.102

265,156

0.00785

Included

Included

0.0368***

240.8

0.102

265,156

0.0224

Included

Included

0.0299***

0.00709***

0.0151***

0.0585

0.540***

236.3

0.102

265,156

0.0301*

Included

Included

0.0341***

0.567***

0.448***

0.243***

0.521***

0.00245*

0.003**

268

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

between density and EXT/INT variables in developing a comparative advantage in new industries (“on new” variables). The interaction term between FDI and density presents a negative and significant sign in both stages, indicating that FDI has brought into path-breaking knowledge that can further lead to radical innovation. Knowledge spillovers from foreign-owned firms have enabled Chinese economy to jump further into less related industries. In contrast, the interaction term between density and the other EXT variable, import, has a positive and statistically significant sign in the first stage and an insignificant sign in the second stage, suggesting that Chinese firms have not obtained path-breaking knowledge from imports. Rather, knowledge embedded in imported products have been closely related to China’s preexisting industrial structure and thus reinforced Chinese region’s reliance on industrial relatedness. The interaction term between R&D and density has a negative and significant sign in both stages. This means Chinese cities have benefited from radical innovations during 2002–2011 that generated path-breaking, new knowledge. High levels of human capital and investment in infrastructure are likely to enable regions to achieve more radical innovation and jump further into less related industries, which is consistent with our expectation. Government supports in the form of export tax rebate only play a supportive and secondary role in the creation of radical innovation and of course could not generate path-breaking knowledge on its own. Finally, culturally open-minded and economically liberalized regions are more willing to embrace new ideas and therefore have better jumping capability, particularly during 2007–2011. We move onto the “on current” variables. Another problem plaguing developing countries/regions is even if they manage to jump further and develop a comparative advantage in new industries that are not close to their preexisting industrial structure, it is still difficult for them to maintain a comparative advantage in these technologically distant and less related industries (Neffke et al. 2011). Table 12.2 shows that regions can overcome this problem by improving local physical and human capital, investing in R&D activities, and learning from foreign firms, in order to obtain pathbreaking knowledge and stand firm in technologically distant industries. Regions should also seek to cultivate a tolerant, inclusive, liberal attitude toward new ideas and different beliefs in order to maintain a comparative advantage in less related industries. In short, extra-regional linkages and internal innovation have the potential to reduce regions’ reliance on industrial relatedness in the process of developing a comparative advantage in new industries and keeping a comparative advantage in current industries. It is also worthwhile to point out that the absolute values of the coefficients of “PCAP*density on new,” “SO*density on new,” and “FDI*density on new” have increased dramatically from the first stage to the second one, indicating that these factors have played a greater role in the second stage. This may contribute to the temporal change of the relationship between density and the probability of developing transition industries, as presented in Fig. 12.4. The fact that the coefficient of density has decreased from the first stage to the second one (Models 1 and 7 in

12.5

Empirical Results

269

Table 12.2) also shows this relationship is weakening. In other words, Chinese regions have become less reliant on relatedness and more path-breaking. To better understand the temporal change shown in Fig. 12.4, we compare estimation results in East and Northeast China to see how such a temporal change varies across China’s big regions5 (Table 12.3). FDI and physical capital have contributed to developing a comparative advantage in new industries and keeping a comparative advantage in current industries in both East and Northeast China; however, their effects are much stronger in Northeast China, particularly during 2007–2011. Northeast China has also managed to seize the learning opportunity provided by new knowledge embedded in imported products and to achieve pathbreaking regional development during 2002–2006 (Models 7–9). East China that already has high level of FDI, investment in infrastructure, and imports should pay more attention to increasing its R&D activities and economic and social openness in order to develop in a more path-breaking way (Model 6). Coefficients of HCAP also support this finding regarding regional variation. In East China human capital has played a critical role in the first stage but only a minor role in the second stage probably because human capital has reached a saturation point in developed East China during 2007–2011. In contrast, in less developed Northeast China, human capital has had an increasingly significant effect in regional path-breaking development. In summary, both East and Northeast China managed to enhance their jumping capabilities and jump further, particularly during the second stage, but they did so by employing different strategies. Table 12.4 reports the empirical results on whether the impact of the articulation between density and EXT/INT variables varies across industries. We here focus on two industries: HS11 (textile and apparel) and HS16 (mechanical and electrical equipment). FDI in both industries has a considerable impact over regional pathbreaking development, indicating that Chinese firms in these two industries have been able to achieve radical innovation and upgrading through learning from foreign-owned firms. During 2002–2006, import was particularly crucial in pushing forward path-breaking regional diversification in the textile industry, as global buyers increasingly outsourced higher-value-added and high-end functions (e.g., original design manufacturing) to their Chinese suppliers (see also Zhu and Pickles (2015)). However, such effect of import faded away in the second stage as Chinese suppliers matured. On the other hand, in technology-intensive HS16, import only has minor effects in regional path-breaking. Instead, import reinforced the effect of density during 2002–2006, showing that imported products only brought in knowledge closely related to preexisting industrial structure and triggered path-dependent evolution. R&D played a key role in maintaining a comparative advantage in technologically distant and less related industries in technology-intensive mechanical and electrical industry, while in labor-intensive textile industry, R&D investment had

5 Due to space limitation, we only compare East and Northeast China. Estimation results on other regions are available on request.

0.00001

R&D

on current

HCAP*density

on new

R&D*density

on current

R&D*density

on new

Import*density

on current

Import*density

new

FDI*density on

current

0.0001**

0.0001***

1.21e-09***

7.09e-10***

0.0605

2.123***

0.0001**

0.0001***

1.30e-09***

8.20e-10***

0.0001**

0.0001***

1.24e-09***

7.71e-10***

0.00522

FDI*density on

0.0581***

0.000647

0.00001*

-1.36e-10**

EO

0.00216**

0.0818*

0.00001*

1.54e  10**

SO

Rebate

PCAP

0.0932

-1.66e-11

1.30e  10*

Import

HCAP

0.153***

0.00001

0.00002**

5.53e  11

8.43e-11

0.133*

0.353***

0.000002

0.0188***

0.0581***

0.476***

FDI

0.493***

1.077***

0.488***

0.123*

xi,c,t1

1.250***

0.710***

Density

0.363*

0.00001

0.00002**

3.61e  11

6.39e-11

0.00363***

0.134**

0.000003

9.30e-12

0.0321***

1.084***

(5)

0.00001

0.00003***

9.52e-11

1.19e-10

0.273***

0.0817***

0.00254***

0.000003

2.99e-11

0.0592***

1.558***

(6)

0.00002

0.00003***

-1.12e-08***

09***

9.26e-

0.0216

0.767***

0.000002

2.15e-09***

0.0700***

0.356***

0.613***

(7)

(4)

(1)

(3)

2002–2006

(2)

Northeast China 2002–2006

2007–2011

East China

Table 12.3 Determinants of having developed industries in East and Northeast China

0.299

0.00002

0.00003***

-1.04e-08***

09***

8.33e-

0.00109

0.00212

0.000002

1.98e-09***

0.357***

0.892***

(8)

0.00001

0.00002***

-1.04e-08***

09***

8.47e-

0.0558

0.0332**

0.00272***

0.000001

1.98e-09***

0.357***

0.268**

(9)

0.00001

0.00003

2.92e-10

7.59e-11

1.200***

1.593***

0.000001

5.43e  11

0.179***

0.0153*

2.214***

(10)

2007–2011

2.820***

0.000003

0.00004

2.96e-10

6.57e-10

0.00131

0.307***

0.000002

1.08e-10

0.0298***

2.438***

(11)

0.00002

0.00004

1.32e-10

2.22e  10

0.226***

0.0429**

0.0041***

0.0000004

-1.46e-11

0.0499***

1.889***

(12)

0.309

384.0

R2

F

*p < 0.10, **p < 0.05, ***p < 0.01

0.0304*

87,572

Included

HS

N

Included

Province

_cons

0.019***

Theil

new

SO*density on

current

SO*density on

new

EO*density on

current

EO*density on

on new

Rebate*density

on current

Rebate*density

on new

PCAP*density

on current

PCAP*density

on new

HCAP*density

374.8

0.313

86,507

0.116***

Included

Included

0.026***

0.0126**

0.0263***

1.813***

364.7

0.311

87,572

0.121***

Included

Included

0.022***

0.205**

0.363***

0.804***

0.850***

0.00362

0.00109

71.35

0.077

87,826

0.0996***

Included

Included

0.051***

69.09

0.076

87,826

0.145***

Included

Included

0.054***

0.00171

0.00913***

0.0263

68.27

0.078

87,826

0.0406

Included

Included

0.059***

0.670***

0.570***

0.355***

0.693***

0.00458*

0.00317

94.63

0.224

90.47

0.224

30,775

0.0195

0.0190 31,156

Included

Included

0.003

Included

Included

0.008***

0.0176*

0.0141

0.977*

89.26

0.225

31,156

0.0313

Included

Included

0.008***

0.544

0.122

0.679***

0.754***

0.0058

0.0216*

39.23

0.107

31,243

0.0914***

Included

Included

0.058***

37.81

0.106

31,243

0.0501

Included

Included

0.053***

0.0173***

0.0171**

2.003***

36.81

0.107

31,243

0.117***

Included

Included

0.055***

0.655

0.334

0.518***

0.0286

0.0449***

0.0512***

on current

HCAP*density

on new

R&D*density

on current

R&D*density

on new

Import*density

on current

Import*density

new

FDI*density on

current

FDI*density on

0.0597

0.0002

0.0006**

3.95e-09*

5.31e-10

0.250*

2.059***

0.0004

0.0002

09**

4.16e-

3.20e-10

0.868***

0.0003

0.0004*

09**

4.68e-

1.19e-11

0.00008

0.00002

3.48e-09

4.71e-09

0.594***

1.060***

0.00007

0.00002

4.95e-09

6.26e-09*

0.00006

0.00001

7.46e-10

3.01e-09

0.0468*** 0.369***

0.0002

0.062*

0.001

0.230***

0.000004

1.88e-09

0.0625***

1.395***

0.004

0.0007*

0.00001

3.21e-09**

0.0419***

1.415***

SO

0.00351*

0.165***

0.00001

2.79e-09*

0.186***

0.0230***

1.457***

EO

Rebate

PCAP

HCAP

0.00003

R&D

1.64e-09***

1.65e-09***

Import

0.00002

0.0421**

FDI

0.395***

0.362***

0.000005

0.399***

0.413***

xi,c,t1

1.51e-09***

1.296***

0.750***

Density

0.0001

0.0002

9.96e-10**

4.95e-10**

0.157

0.409***

1.338***

0.0003

0.0004

7.10e-10*

4.97e-10***

0.0006

0.0648

0.000002

0.0001

0.0002

9.31e-10**

4.64e-10**

0.0258

0.0260**

0.0003

0.00001

1.18e  10

0.00005

0.0001***

1.32e-10

8.98e-11

0.977***

1.003***

0.00001

-1.84e-11

0.000002

8.40e  11

0.000647

1.647***

1.27e-10*

0.407***

0.647***

0.210***

0.442***

1.200***

0.0214

0.419***

0.793***

2007–2011

0.984***

0.0001

0.0001***

4.68e-11

2.96e-11

0.007**

0.247***

0.00001

6.19e-12

0.0247**

1.357***

HS16. Machinery and mechanical appliances and electrical equipment 2002–2006

2002–2006

2007–2011

HS11. Textile and textile articles

Table 12.4 Determinants of having developed industries in different industries

0.0001

0.0001***

1.27e-10

8.20e-11

0.266***

0.0515**

0.0028***

0.00001

-1.88e-11

0.0303***

1.705***

0.00945

Included

0.0718***

37,934

0.302

302.9

HS

_cons

N

R2

F

*p < 0.10, **p < 0.05, ***p < 0.01

Included

Included

288.4

0.304

37,668

Included

0.0121***

Province

0.0124***

0.0718***

0.00313

1.156***

Theil

new

SO*density on

current

SO*density on

new

EO*density on

current

EO*density on

on new

Rebate*density

on current

Rebate*density

on new

PCAP*density

on current

PCAP*density

on new

HCAP*density

274.7

0.303

37,934

0.0701**

Included

Included

0.0105***

0.0729

0.588***

0.494***

0.682***

0.00462

0.00362

87.88

0.111

37,934

0.0103

Included

Included

0.0415***

82.53

0.110

37,934

0.0268

Included

Included

0.0363***

0.00355

0.0344***

0.770***

80.06

0.113

37,934

0.0249

Included

Included

0.0374***

1.324***

0.861***

0.292**

0.589***

0.0144***

0.00546

280.6

0.303

27,159

0.0427*

Included

Included

0.0105***

262.5

0.303

27,159

0.0285

Included

Included

0.0159***

0.0321**

0.00746

0.706*

246.4

0.304

27,159

0.0736**

Included

Included

0.0090***

0.311

0.309*

0.379***

0.664***

0.00758*

0.0104**

52.16

0.075

27,159

0.0286

Included

Included

0.0408***

47.73

0.073

27,159

0.0572

Included

Included

0.0299***

0.0297

0.0372*

0.591*

45.51

0.075

27,159

0.135***

Included

Included

0.0352***

0.551**

0.357*

0.372***

0.618***

0.0142***

0.0122***

274

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

an insignificant effect, particularly during 2007–2011. Infrastructure allows regions to jump further, especially in heavy industries like HS16. Human capital also reduces regions’ reliance on density and relatedness in both labor-intensive textile industry and technology-intensive machinery industry. One reason for this is that the human capital indicator we used is measured as the share of people with more than secondary schooling, including both white-collar workers engaging in knowledgeand technology-intensive industries and blue-collar labor force trained in vocational and technical schools in China. In addition, although China’s textile industry is traditionally seen as based on low-skilled or semiskilled labor, a growing number of white-collar jobs are being created, as the industry upgrades from low-end to medium- and high-end production. Finally, the estimated parameters of economic and social openness variables are mostly unaltered, not only reducing relatedness’s effect in generating new transition industries but also enabling regions to maintain a comparative advantage in technologically distant and less related industries. As a robustness check, all models are also estimated by the probit model (Table 12.5), by using data in different years (e.g., 2002–2005 and 2007–2010), and by using different threshold values (0.8 and 1.2) to determine a comparative advantage. We also identify intermediary firms in our dataset based on Chinese characters that have the English-equivalent meaning “importer,” “exporter,” and/or “trading” in the firm’s name and re-estimate our models after excluding these firms. Compared with the results presented above, these changes produce only minor effects.

12.6

Discussion and Conclusion

Although traditional economic geography literature has attributed regional development to a wide range of factors, some recent empirical EEG studies tend to overly focus on technological relatedness as a key driving force and emphasize that regions often branch into industries that are technologically related to their preexisting industrial structure. Such a path-dependent regional diversification can be thought as regions/countries jumping in a heterogeneous and uneven industry space where they are only allowed to jump certain distance. This distance is determined by technological relatedness among industries. It is thereafter argued that developed countries that start from core, dense areas in the uneven industry space have more opportunities to jump to new related industries and therefore have more opportunities to sustain economic growth than do developing countries that jump from peripheral, deserted areas. Even though researches based on industrial relatedness do not exclude the possibility that developing countries can reach core areas in the industry space, empirical studies are mostly centered on regional diversification in developed countries and highlight the key role of industrial relatedness in the process of regional economic development. In some extreme cases, it is impossible for developing countries to enter core areas; the divergence between developed and

0.000003

0.699***

0.0008

0.0332***

0.0940***

R&D

HCAP

PCAP

Rebate

EO

0.0345

0.0002** 0.0002

0.0002*** 0.0002

R&D*density on current

R&D*density on new

8.028*** 6.064***

7.366***

HCAP*density on new

0.000001

0.00001

(continued)

0.00002

0.00007***

1.04e-10

3.49e-10

11.80***

0.000004

0.00003

1.66e-10

1.61e-10

1.036***

0.278***

0.0122***

0.00001

2.84e-11

0.276***

7.286***

2.09e-10

0.0340***

1.775***

0.000002

1.20e-10

0.208***

5.405***

HCAP*density on current

0.0001

0.0002***

4.41e-09**

4.68e-09**

3.77e-09*

2.60e-10

Import*density on new

4.10e-09***

4.52e-09***

Import*density on current

0.0000004

5.491***

0.493***

0.0437**

0.0298***

0.000004**

0.177

0.308***

0.0301***

0.0017***

1.389***

0.0000004

FDI*density on new

0.106

0.158***

0.0040***

0.00002

4.48e-11

1.334***

0.0793***

7.318***

6.556***

5.13e09***

0.0398***

0.234

0.00002

4.74e-11

0.166***

2.873***

0.0284

0.00002*

4.59e-10

0.149*** 3.42e-11

6.01e-10

3.960***

2007–2011

3.89e-10

1.802***

2.071*** 0.813***

1.726***

5.525***

0.164***

1.670***

3.886***

FDI*density on current

SO

0.000003

4.28e10***

Import

0.0331***

4.47e10***

0.0370

FDI

1.687***

1.684***

7.695***

2002–2006

xi,c,t1

Density

Table 12.5 Determinants of having developed industries in China (probit)

0.220***

Included

2.305***

Included

Included

HS

0.245***

44408.4

44455.8

27445.4

Chi-squared

*p < 0.1, **p < 0.05, ***p < 0.01

53947.8

53924.1

62955.6

Log lik.

2.138***

263,227

263,227

2.299***

264,384

_cons

Included

Included

N

Included

0.205*** 0.00260

45628.4

53864.0

264,384

2.208***

Included

Included

45869.2

53001.6

261,165

1.665***

Included

Included

0.0143

0.0690***

25320.6

90489.2

53757.0 45842.5

265,156

1.757***

Included

Included

264,384

2.148***

Included

Included

0.000219

0.110***

24303.2

91260.4

266,986

2.491***

Included

Included

0.182***

22747.4

92038.3

266,986

1.567***

Included

Included

0.176***

26901.5

89698.8

265,156

1.880***

Included

Included

26373.2

89962.9

265,156

1.589***

Included

Included

0.118***

26572.2

89863.4

265,156

2.154***

Included

Included

0.146***

2.937***

1.266**

SO*density on new

Province

2.219***

0.400

SO*density on current

Theil

2.145***

3.564***

EO*density on new

3.955***

1.057***

0.0837***

0.108***

Rebate*density on new

EO*density on current

0.0205***

0.0574***

0.0350***

0.0126

PCAP*density on new

Rebate*density on current

0.0296***

2007–2011 0.0340***

PCAP*density on current

2002–2006

Table 12.5 (continued)

12.6

Discussion and Conclusion

277

developing countries persists due to the dominance of path dependence. Such a conclusion is pessimistic, particularly for developing countries/regions, and deterministic, since it predicates too much on an assumption that regional diversification is affected or even confined by relatedness among industries, but pays less attention to whether or not countries/regions’ jumping capabilities in the uneven industry space can be changed. In this chapter, we differentiate two types of new growth path creation—pathdependent and path-breaking—and focus on the second type in order to examine how developing countries/regions can jump further in a more path-breaking way, so that the “confinement” of technological relatedness can be transcended. In doing so, we argue that the capacity of classical evolutionary ideas centered on technological relatedness to explain regional economic development is obviously limited and potentially misleading. Rather than focusing on path-dependent regional diversification and evolution, analyses, particularly in the context of less developed countries/regions, should pay more attention to the role of firm agency, local social and institutional contexts, policy-making, and extra-regional linkages in reducing regions’ reliance on relatedness and enabling regions to jump further. Not only do these factors promote regions’ jumping capability, but they also contribute to regions’ capability of maintaining a comparative advantage in technologically distant and less related industries. Regional and industrial differences should not be overlooked as well, since the effects of extra-regional linkages and internal innovation vary across regions and industries. Empirically, this research seeks to find a brighter future for developing countries/ regions in an increasingly competitive global economy by pointing out that the seemingly dominant path-dependent development trajectories can be broken through continuously improving infrastructure and education, providing government supports, investing in R&D, and fostering an open-minded social and institutional context. In doing so, less developed regions may be able to catch up and jump from peripheral, deserted areas further into core, dense areas in the uneven industry space/network. Theoretical studies in EEG have long pointed out extra-regional linkages and internal innovation have the potential to bring in and/or generate fresh know-how that are less related to region’s existing industrial structure and therefore may enable regions to dislodge path dependence and develop in a more path-breaking way. Empirical works based on case studies also testify this argument. This chapter thus provides a quantitative research that is complementary to existing theoretical and qualitative studies. In addition, our research is based on recent studies (Neffke et al. 2011; Boschma et al. 2013; Boschma and Capone 2015a) but asks questions from a different angle and includes some economic and institutional factors that have been left out in previous studies. Not only have we investigated the role of economic and institutional factors at the regional level, but we also explore how their impacts over path-breaking regional diversification vary across industries and regions. Our research also speaks with debates on regional resilience and cluster life cycles that have recently emerged in the literature on regional development in economic geography. Scholars have advocated an evolutionary, out-of-equilibrium

278

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

approach to regional resilience in which the focus is on the long-term capacity of regions to reconfigure their socioeconomic structure rather than on the ability of regions to return to a stable equilibrium state after a shock (Crespo et al. 2014; Boschma 2015; Martin and Sunley 2015). Resilience then depends on the ability of regions to create new growth paths, in order to avert the devastating impact of decline and stagnation (Boschma 2015). This research contributes to this strand of literature by developing a better understanding of the tensions and conflicts in adaptive systems between connectedness, relatedness, and increasing order, on the one side, and the resilience properties on the other side. While some recent studies focus on the first ones and highlight there is some degree of cohesion and stabilization in the industrial structure of a region (path-dependent new path creation), this chapter pays more attention to the second ones that require a high level of flexibility and restructuring capabilities (path-breaking new path creation). Regions’ preexisting industrial structure sets limits to regional resilience, but opportunities also exist as regions can jump further. China’s path-breaking regional diversification may be better understood by relating to the country’s recent efforts toward pushing and encouraging low-value manufacturers to relocate from the high-cost coastal regions to release space and resources for higher-value production while simultaneously encouraging economic development in less developed inland regions. Future research can employ qualitative case studies to provide a more in-depth interpretation of regional path-breaking development in both coastal and inland China. Due to data unavailability, this chapter does not take into account the role of knowledge transfer from other regions in China (i.e., interregional linkages within China), which we leave for future research.

References Amighini, A., & Sanfilippo, M. (2014). Impact of south–south Fdi and trade on the export upgrading of African economies. World Development, 64(0), 1–17. Arthur, W. B. (1989). Competing technologies, increasing returns, and lock-in by historical events. The Economic Journal, 99(394), 116–131. Arthur, W. B. (1994). Increasing returns and path dependence in the economy. Ann Arbor: University of Michigan Press. Bathelt, H., & Cohendet, P. (2014). The creation of knowledge: Local building, global accessing and economic development—Toward an agenda. Journal of Economic Geography, 14(5), 869–882. Bathelt, H., & Li, P.-F. (2014). Global cluster networks—Foreign direct investment flows from Canada to China. Journal of Economic Geography, 14(1), 45–71. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. Baum, J. A. C. (1996). Organizational ecology. In S. R. Clegg, C. Hardy, T. Lawrence, & W. R. Nord (Eds.), Handbook of organization studies (pp. 77–114). London: Sage. Binz, C., Truffer, B., & Coenen, L. (2016). Path creation as a process of resource alignment and anchoring: Industry formation for on-site water recycling in Beijing. Economic Geography, 92 (2), 172–200.

References

279

Boschma, R. (2015). Towards an evolutionary perspective on regional resilience. Regional Studies, 49(5), 733–751. Boschma, R., & Capone, G. (2015a). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, 44(10), 1902–1914. Boschma, R., & Capone, G. (2015b). Relatedness and diversification in the European Union (Eu-27) and European Neighbourhood policy countries. Environment and Planning C: Government and Policy, 34(4), 617–637. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new Industries at the Regional Level in Spain: A proximity approach based on product relatedness. Economic Geography, 89(1), 29–51. Crespo, J., Suire, R., & Vicente, J. (2014). Lock-in or lock-out? How structural properties of knowledge networks affect regional resilience. Journal of Economic Geography, 14(1), 199–219. David, P. A. (1985). Clio and the economics of qwerty. The American Economic Review, 75(2), 332–337. Dawley, S. (2014). Creating new paths? Offshore wind, policy activism, and peripheral region development. Economic Geography, 90(1), 91–112. Delgado, M., Porter, M. E., & Stern, S. (2016). Defining clusters of related industries. Journal of Economic Geography, 16(1), 1–38. Dollar, D. (1992). Outward-oriented developing economies really do grow more rapidly: Evidence from 95 Ldcs, 1976–1985. Economic Development and Cultural Change, 40(3), 523–544. Goldberg, P. K., Khandelwal, A. K., Pavcnik, N., & Topalova, P. (2010). Imported intermediate inputs and domestic product growth: Evidence from India. The Quarterly Journal of Economics, 125(4), 1727–1767. Görg, H., & Greenaway, D. (2004). Much ado about nothing? Do domestic firms really benefit from foreign direct investment? The World Bank Research Observer, 19(2), 171–197. Grillitsch, M., & Trippl, M. (2014). Combining knowledge from different sources, channels and geographical scales. European Planning Studies, 22(11), 2305–2325. Hassink, R. (2005). How to unlock regional economies from path dependency? From learning region to learning cluster. European Planning Studies, 13(4), 521–535. Hassink, R., Klaerding, C., & Marques, P. (2014). Advancing evolutionary economic geography by engaged pluralism. Regional Studies, 48(7), 1295–1307. Hausmann, R., & Klinger, B. (2007). The structure of the product space and the evolution of comparative advantage (Cid Working Papers, No. 146). Cambridge, MA: Center for International Development at Harvard University. Henning, M., Stam, E., & Wenting, R. (2013). Path dependence research in regional economic development: Cacophony or knowledge accumulation? Regional Studies, 47(8), 1348–1362. Hidalgo, C. A., & Hausmann, R. (2008). A network view of economic development. Developing Alternatives, 12, 5–10. Hidalgo, C. A., Klinger, B., Barabási, A.-L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Lall, S., & Narula, R. (2004). Foreign direct investment and its role in economic development: Do we need a new agenda? The European Journal of Development Research, 16(3), 447–464. Li, P.-F., Bathelt, H., & Wang, J. (2012). Network dynamics and cluster evolution: Changing trajectories of the Aluminium extrusion industry in Dali, China. Journal of Economic Geography, 12(1), 127–155. Lin, G. C. S., Wang, C. C., Zhou, Y., Sun, Y., & Wei, T. H. D. (2011). Placing technological innovation in Globalising China: Production linkage, knowledge exchange and innovative performance of the ICT Industry in a Developing Economy. Urban Studies, 48(14), 2999–3018.

280

12

How to Jump Further? Path Dependence and Path-Breaking in an Uneven. . .

Lo Turco, A., & Maggioni, D. (2015). On firms’ product space evolution: The role of firm and local product relatedness. Journal of Economic Geography. Martin, R. (2010). Roepke lecture in economic geography—Rethinking regional path dependence: Beyond lock-in to evolution. Economic Geography, 86(1), 1–27. Martin, R., & Sunley, P. (2006). Path dependence and regional economic evolution. Journal of Economic Geography, 6(4), 395–437. Martin, R., & Sunley, P. (2011). Conceptualizing cluster evolution: Beyond the life cycle model? Regional Studies, 45(10), 1299–1318. Martin, R., & Sunley, P. (2015). On the notion of regional economic resilience: Conceptualization and explanation. Journal of Economic Geography, 15(1), 1–42. Maskell, P. (2014). Accessing remote knowledge—The roles of trade fairs, pipelines, crowdsourcing and listening posts. Journal of Economic Geography, 14(5), 883–902. Maskell, P., & Malmberg, A. (1999). The competitiveness of firms and regions: ‘Ubiquitification’ and the importance of localized learning. European Urban and Regional Studies, 6(1), 9–25. Maskell, P., & Malmberg, A. (2007). Myopia, knowledge development and cluster evolution. Journal of Economic Geography, 7(5), 603–618. Maskell, P., Bathelt, H., & Malmberg, A. (2006). Building global knowledge pipelines: The role of temporary clusters. European Planning Studies, 14(8), 997–1013. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Poncet, S., & Starosta De Waldemar, F. (2013). Export upgrading and growth: The prerequisite of domestic embeddedness. World Development, 51(0), 104–118. Rigby, D. L., & Essletzbichler, J. (1997). Evolution, process variety, and regional trajectories of technological change in U.S. manufacturing. Economic Geography, 73(3), 269–284. Schiff, M., & Wang, Y. (2006). North-south and south-south trade-related technology diffusion: An industry-level analysis of direct and indirect effects. Canadian Journal of Economics, 39(3), 831–844. Schumpeter, J. A. (1947). The creative response in economic history. The Journal of Economic History, 7(2), 149–159. Simmie, J. (2012). Path dependence and new technological path creation in the Danish wind power industry. European Planning Studies, 20(5), 753–772. Storper, M. (2009). The economics of context, location and trade: Another great transformation? In G. Becattini, M. Bellandi, & L. De Propis (Eds.), The handbook of industrial districts (pp. 141–157). Cheltenham: Edward Elgar. Storper, M., & Walker, R. (1989). The capitalist imperative: Territory, technology and industrial growth. New York: Basil Blackwell. Sydow, J., Lerch, F., & Staber, U. (2010). Planning for path dependence? The case of a network in the Berlin-Brandenburg optics cluster. Economic Geography, 86(2), 173–195. Tanner, A. N. (2014). Regional branching reconsidered: Emergence of the fuel cell industry in European regions. Economic Geography, 90(4), 403–427. Theil, H. (1972). Statistical decomposition analysis: With applications in the social and administrative sciences. Amsterdam: North-Holland. Wang, C., Lin, G. C. S., & Li, G. (2010). Industrial clustering and technological innovation in China: New evidence from the ICT Industry in Shenzhen. Environment and Planning A, 42(8), 1987–2010. Zhu, S., & Fu, X. (2013). Drivers of export upgrading. World Development, 51(0), 221–233. Zhu, S., & Pickles, J. (2015). Turkishization of a Chinese apparel firm: Fast fashion, regionalisation and the shift from global supplier to new end markets. Cambridge Journal of Regions, Economy and Society, 8(3), 537–553.

Chapter 13

What Drives the Evolution of Export Product Space in China?

13.1

Introduction

The links between agglomeration externalities and the performance of firms, industries, and regions have been studied extensively. Marshallian externalities are seen as based on industrial specialization, while researches on Jacobs externalities accredit urban growth to the clustering of local firms in a variety of sectors by emphasizing the role of knowledge spillovers in a diversified economic structure. Both Marshallian and Jacobs externalities contribute to economic growth, productivity gains, innovation, and firm survival (Glaeser et al. 1992; Henderson 1997). Empirical evidence on the effects of Marshallian and Jacobs externalities over regional and firm performance, however, remains mixed (Beaudry and Schiffauerova 2009). One of the reasons is that, as Nooteboom (2000) argued, studies on both types of externalities tend to stress the importance of geographical proximity in knowledge spillovers while paying insufficient attention to the role of cognitive distance. Based on this critique, evolutionary economic geographers have moved beyond existing externality literature and proposed the concept of cognitive proximity as well as other types of proximity such as institutional and organizational proximity. Relevant empirical studies have also confirmed that knowledge spillovers tend to occur among industries that are technologically related (Boschma and Frenken 2011; Frenken et al. 2007; Essletzbichler 2007; Boschma and Iammarino 2009). Technological relatedness is also a significant driving force for the evolution of regional productive structure (Hidalgo et al. 2007; Neffke et al. 2011). Regions develop new technologies, products, and industries through a process of creative destruction and branch into technologically related industries (Martin and Sunley

Modified article originally published in [He, and Zhu (2018), Evolution of Export Product Space in China: Technological Relatedness, National/Local Governance and Regional Industrial Diversification. Tijds. voor econ. en soc. geog, 109: 575–593.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_13

281

282

13 What Drives the Evolution of Export Product Space in China?

2006; Boschma and Frenken 2011; Essletzbichler 2015). Hidalgo et al. (2007) coin the term “product space” to characterize the network of relatedness between products at the global level. Products manufactured in a country reflect the latter’s specific knowledge and capacities, which cannot be easily transferred across countries (Hidalgo and Hausmann 2009). This implies that a country’s future productive structure is to some extent shaped by the history—the country’s past productive structure, through a path-dependent process underpinned by technological relatedness (Hidalgo et al. 2007). At the subnational/regional level, empirical studies also show that regions tend to diversify into new industries that demand similar resources and capabilities to those required by regions’ existing industries (Klepper and Simon 2000; Boschma and Wenting 2007; Neffke et al. 2011; Boschma et al. 2012; Essletzbichler 2015). In the past few decades, Chinese regions have transformed their industrial structures considerably. However, few studies have thoroughly explored such a process of regional industrial evolution and the underlying mechanisms, despite a small number of exceptions (Poncet and Waldemar 2013; Guo and He 2017). It is still unclear whether regional industrial evolution in China is driven by technological relatedness. Moreover, recent literature on technological relatedness and regional industrial diversification has largely neglected the role of institutional factors that may disturb or reinforce the path-dependent process of regional industrial evolution and have quite an effect on the intensity and nature of the diversification process (Boschma and Capone 2015). This could be questionable particularly in China where regional economic development has been heavily shaped by both government intervention and market coordination. Institutional arrangements exert direct impacts on the type of industries in which regions specialize and more importantly inflect the ways in which technological relatedness shapes regional industrial development. Since the 1980s, China has issued plenty of industrial policies to sustain economic growth and promote industrial restructuring, resulting in a complex and relatively comprehensive institutional framework made up by policies in different forms (e.g., export rebates and tax credits) and issued by various levels of governments (Xiang and Zhang 2013). One fundamental transformation that China has undergone after the reform is from a centrally planned to a decentralized political and fiscal system, also known as the regionally decentralized authoritarian system (He et al. 2008; Xu 2011). Decentralization has resulted in a GDP-based interjurisdictional competition between local authorities that have strong incentives to intervene in regional economic development (Pan et al. 2016; Yu et al. 2016), but also allowed regional administrations to take routes sometimes different from the central government’s initial plan (Zhu and He 2016). In other words, government intervention at both the national and local level must be taken into account to better understand China’s regional industrial diversification. Specifically, based on the CCTS dataset, this study contributes to the evolutionary economic geography (EEG) literature by exploring the evolution of export product space in a transitional economy where regional industrial evolution has been affected by both market forces and state power. The existing EEG studies tend to overstress the importance of technological relatedness. This study, however,

13.2

Theoretical Framework and Research Hypotheses

283

shows that other factors such as industrial and regional policies at both the national and local level may have direct and indirect impacts over regional industrial diversification, and more importantly institutional arrangements set up by various levels of governments interact with technological relatedness, co-shaping regional development trajectories and industrial restructuring. The next section presents the theoretical framework and develops research hypotheses. Section 13.3 introduces data sources and the measurement of technological relatedness. After some descriptive analyses on the evolution of China’s export product space in the fourth section, we further analyze the results of econometric analysis in Sect. 13.5. The last section concludes the chapter.

13.2

Theoretical Framework and Research Hypotheses

13.2.1 Path-Dependent Evolution of Productive Structure Literature on Marshallian externalities proposes that regional growth is driven by the clustering of firms in the same industry within a locality, through local labor market pooling, input-output linkage, and specialized knowledge spillovers (Marshall 1920), whereas proponents of Jacobs externalities believe that knowledge may easily spill over across different sectors as ideas developed by one sector can be applied in another (Jacobs 1969). Both types of externalities stress the importance of knowledge spillovers to a region’s economic growth, productivity increase, and innovation. However, as Nooteboom (2000) argued, both strands of literature tend to pay too much attention to the role of geographical proximity, but fail to acknowledge the fact that for knowledge spillovers to enhance firm performance, there needs to be cognitive proximity, technological relatedness, or complementarity between firms. If cognitive distance is too large, there is no common knowledge base for inter-industry communication to take place effectively, while cognitive “lock-in” may emerge if cognitive distance is too little (Boschma 2005). In recent EEG literature, cognitive proximity is often seen as playing a greater role in the process of knowledge spillovers than geographical proximity (Boschma 2005). It is argued that firms are more likely to learn from each other when they are technologically related and operating in related industries that have cognitive proximity (Boschma and Frenken 2011). Knowledge spillovers are likely to take place effectively in regions hosting a large number of technologically related industries with shared competences. Technological relatedness is a key enabling factor for the emergence of new technologies, new products, and new industries in a region (Boschma and Frenken 2011)—i.e., the formation of new regional industrial paths (Neffke et al. 2011). Empirical studies already confirm the positive impacts of technological relatedness over industrial clustering (Boschma and Wenting 2007), employment growth (Bishop and Gripaios 2010; Hartog et al. 2012), spin-off dynamics (Heebels and Boschma 2011), and regional growth (Frenken et al. 2007).

284

13 What Drives the Evolution of Export Product Space in China?

Technological relatedness also acts as a driving force for the evolution of regional productive structure (Hidalgo et al. 2007; Neffke et al. 2011). Hidalgo et al. (2007) define product space as a network that formalizes the idea of relatedness between products traded in the global economy, based on which Hidalgo et al. (2007, Hidalgo and Hausmann 2009) further explicate how technological relatedness influences the evolution of a country’s productive structure. Distance in product space represents technological relatedness between products, indicating how difficult it is for a country to move from one product to another. Products in which a country specializes reflect the country’s specific knowledge and capacities, which cannot be easily transferred across countries (Hidalgo and Hausmann 2009). It is much easier for a country to diversify into new products related to its existing products, since such new products demand resources, knowledge, and capacities similar to what the country already possesses (Boschma and Capone 2016). The current productive structure of a country is thus affected by its own historical productive structure through a path-dependent process underpinned by the relatedness between products (Hidalgo et al. 2007). Based on the data of China’s industrial firms, Guo and He (2017) have confirmed that as the restrictions on factor mobility and commodity exchange were gradually lifted in China after the reform, technological relatedness started to play an increasingly crucial role in regional industrial evolution. Evolution of export products in China is also expected to be governed by the relatedness of exports and imports. Knowledge spillovers from exports and imports have already been widely studied (Redding 2011; Wagner 2012; Boschma and Iammarino 2009). Lileeva and Trefler (2010) argue that exporters often have higher levels of productivity due in part to the self-selection of better firms into exporting activities. Others propose the effect of learning by exporting (Alcacer and Oxley 2014; Boschma and Capone 2016). In the case of imports, most studies have paid attention to indirect learning effects (Castellani et al. 2010). Importers can exploit the availability of more varieties in inputs, and absorb more advanced knowledge embedded in high-quality, high-end imported products (Boschma and Capone 2016; Cohen and Levinthal 1990). Furthermore, massive imports in a sector indicate a high relevance of this sector to the country’s existing productive structure, leading to strong tendencies for the country to diversify into this sector (Boschma and Capone 2016). Related knowledge may be also transferred from one region to another through business networks and imitation behaviors between firms. However, the spread of knowledge is heavily constrained by geographical distance: knowledge spillovers are more likely to occur between regions that are geographically close to each other (Jaffe et al. 1993). For instance, Bahar et al. (2014) report that a country has a higher probability to develop a comparative advantage in a new industry if a neighbor country has a comparative advantage in that industry, resulting in some sort of convergence of export baskets between neighboring countries. Boschma et al. (2017) further confirm the effect of knowledge spillovers between neighboring regions based on the evolution of regional productive structure in the USA. To sum up, ideas developed in this section lead to the following hypotheses;

13.2

Theoretical Framework and Research Hypotheses

285

Hypothesis 13.1 Technological relatedness affects the evolution of regional export products in China. Hypothesis 13.2 Knowledge spillovers from neighbor regions affect regional industrial diversification.

13.2.2 State Power and Evolution of Productive Structure The path-dependent approach of regional industrial evolution tends to focus excessively on knowledge spillovers and technological relatedness (Isaksen 2014). Boschma and Capone (2015) have criticized recent EEG literature’s overlook of the potential effects of the institutional dimensions of industrial change. Regional variations of institutional frameworks can have a direct impact on innovation and regional industrial specialization, particularly in transitional economies like China where a triple process of decentralization, globalization, and marketization has resulted in enormous spatial variations in the economic and institutional landscape (He et al. 2016a). He et al. (2016b) have further pointed out that Chinese regional industrial development is a path-dependent process constantly shaped by government intervention at the local level (e.g., public spending). Since the 1980s, industrial policies in various forms have been issued by different levels of governments in order to promote efficient and sustainable economic development (Xiang and Zhang 2013). In this chapter, we differentiate two types of government intervention: national and local governance. National governance can be seen as nationwide industrial policies issued by the central government, particularly the State Council of China. Since the early 2000s, China has become more integrated into the global economy (He et al. 2016a). In order to forge global-local linkages, first, the central government has provided tax rebates to boost exports. Furthermore, in 1995, it has issued the Catalogue for the Guidance of Foreign Investment Industries to attract foreign direct investments (FDIs) in certain industries while prohibiting and limiting the development of FDIs in some other industries. This catalogue has been modified several times thereafter according to changing national strategic and developmental needs. Exports and FDIs represent two key channels to bring new knowledge/industries to regions and to push forward new path creation and industrial upgrading (Bathelt et al. 2004). Exporting activities could strengthen the connections between the Chinese economy and the global economy due to the possibility to develop deeper division of labor between those two based on industrial linkages (Fujita and Hu 2001). FDIs may either introduce new industries directly or influence regional industrial evolution indirectly by facilitating knowledge spillovers through the effects of demonstration and competition, labor mobility, and business linkages between foreign and domestic firms (Cheung and Lin 2004). In short, export tax rebates and FDI-related policies have been designed to restructure regional industrial structure in China and create new industries in Chinese regions.

286

13 What Drives the Evolution of Export Product Space in China?

Hypothesis 13.3 National governance may affect the emergence of new export products in Chinese regions. In China, local governments are now largely responsible for economic development within their jurisdictions and have strong incentives to intervene in local development (Zhao and Zhang 1999). High-level officials in the hierarchical political system promote low-level government officials based on their economic performance, and local officials compete to boost regional economic development in order to maximize their chances of political promotion (Yu et al. 2016). To attract industries and promote economic growth, Chinese governments often establish a variety of development zones such as economic and technological development zones, high-tech industrial parks, exporting promotion zones, and others. Those development zones could be at the national, provincial, and local level. Firms located in development zones often have access to a number of favorable policies such as income tax credits, rebates of value-added tax, export tax rebates, import duties for equipment purchases, low-price land, and cash rewards to firms with better performance (Barbieri et al. 2012). We consider development zones at all levels as different forms of local governance, since relevant policies are often implemented at the local level and the development of such zones are highly localized. Such policies are expected to attract new industries. Since local governments can better match policies to local needs and preferences, local governance may be more likely to take advantage of technological relatedness while creating new industries. In short, we anticipate that, Hypothesis 13.4 Local governance would be likely to introduce new products that are technologically related to regional existing export portfolio. In a country with massive population and continental size, China’s leaders have had to balance carefully central control and local discretion. The balance is difficult to achieve (Donaldson 2016). On the one hand, excessive centralization leads to inflexible implementation of government policies, resulting in unnecessary local problems. On the other hand, too much local power could create centrifugal forces. Since the reform, the distribution of fiscal and administrative power among China’s central and local governments has changed. Even though decentralization is not a zero-sum game, central and local governments are sometimes in conflict (Filippetti and Sacchi 2016). One of the most distinctive features of the Chinese current regionally decentralized authoritarian system is the central government’s longlasting struggle to ensure local compliance with national policy. Despite enormous changes in China’s political and economic system in the last few decades, local governments do not always share the concerns that motivate the central government, and at times they evade or subvert central policies (Zhu and Pickles 2014). The gap between central and local interests, as well as the constraint this gap places on the state’s ability to govern effectively, remains a fundamental concern for China’s leaders today and must have a fundamental impact on regional industrial dynamics. Negative effects on economic efficiency might arise as a result of conflicts between national and local governments when decentralization is pursued without

13.3

Evolution of Export Product Space in China

287

establishing a clear division of power and competences among them (Filippetti and Cerulli 2015). This proposition leads to the following hypothesis: Hypothesis 13.5 National and local governments may play conflicting roles in regional industrial dynamics.

13.3

Evolution of Export Product Space in China

This chapter uses the CCTS dataset. The geographical unit of analysis is China’s 31 provincial-level administrative divisions (22 provinces, 4 municipalities, and 5 autonomous regions), and we focus on 1271 4-digit products. We further adjust the product HS code based on the 1996 version.1 Records with import/export value less than US$5000 are excluded in this research. The export product space is a 1271*1271 symmetric matrix. Each row and column represents a particular 4-digit product, and the value of each entry is the relatedness between two products. The proximity indicator between industry i and j in year t is computed using Eqs. 1.1 and 1.2 (ϕi,j,t). We then use Eq. 1.5 to calculate the density indicator of industry i in city c in year t (di,c,t). Similarly, we compute the import density indicator. The formula is as follows, where Ij,c,t takes the value of 1 if province c has a comparative advantage in import product j in year t and 0 otherwise. P idi, c, t ¼

j

I j, c, t ϕi, j, t P ϕi, j, t

ð13:1Þ

j

Figure 13.1 shows the distributions of technological relatedness between China’s export products in 2000 and 2011. The result is similar to findings in other studies (Hidalgo et al. 2007; Boschma et al. 2012; Guo and He 2017), but is more leftskewed—fewer strong links and more weak links between products in terms of technological relatedness. Over 60% of product relatedness values in 2000 and 2011 are below 0.2, whereas only about 1% is above 0.6. Boschma et al. (2012) have argued that it is difficult to determine what level of relatedness is needed to consider two products are related. We have taken a conservative position and have considered that two products are related if their relatedness is equal or above 0.6.2 After deleting all weak product links with a relatedness indicator lower than 0.6, 579 and 1291 links are kept in the product space in 2000 and 2011, respectively. 1

4-digit HS code has changed slightly in 1996, 2002, 2007, and 2012. Different threshold values (e.g., 0.4 and 0.5) have been also used to determine product relatedness. Such changes only produce minor differences. Furthermore, this threshold value is only used in the visualization of product space (Fig. 12.2), but not in the calculation of the density indicator. In the latter scenario, all product pairs are taken into account. In other words, our choice of the threshold value only affects Fig. 12.2 and has no impacts on other figures, tables, and all econometric results. 2

288

13 What Drives the Evolution of Export Product Space in China?

Fig. 13.1 Distribution of China’s export product relatedness in 2000 and 2011

Although the distributions of product relatedness in 2000 and 2011 are similar to each other (Fig. 13.1), it does not mean China’s export product space has only changed slightly. Table 13.1 shows the Pearson’s correlation coefficient between the product spaces generated with data from different years. Most correlation coefficients between adjacent years are around 0.7. The further 2 years are from each other, the lower the correlation between these 2 years is. Taking 2011 as an example, its correlation coefficient with 2010, 2009,. . ., and 2000 goes down from 0.84 to 0.41, indicating that China’s product space has undergone a dramatic transformation in this time period. However, China’s productive structure has been also stabilizing, as the correlation between relatedness indicators in two adjacent years has increased from 0.70 (between 2000 and 2001) to 0.84 (between 2010 and 2011). To demonstrate the structure of export product space, we visualize the product space by using cytoscape 3.3.0. Figure 13.2a, b shows China’s export product space in 2000 and 2011, respectively. The number of nodes increases from 255 to 320, while the number of edges increases from 579 to 1291, implying that more products and linkages are included in the product space of 2011. Export products are increasingly related with each other, and the export product space has become denser during 2000–2011. Figure 13.2a shows a clear core-periphery structure of China’s export product space in 2000. The major core is in the bottom right corner of the product space, consisting of textile, chemical, and metal products. In addition, there is a sub-core made up by electric apparatus, electronic and telecommunication equipment in the upper left corner, which is loosely connected with the main core. Besides the main core and sub-core, the rest of the products are in the periphery area of the product space with weak linkages with the main core and sub-core (e.g., shoes, hats, and other manufacturing products).

2011 1.00 0.84 0.79 0.70 0.65 0.62 0.59 0.55 0.52 0.50 0.47 0.41

1.00 0.83 0.73 0.67 0.64 0.61 0.56 0.53 0.51 0.48 0.45

2010

1.00 0.78 0.71 0.67 0.64 0.60 0.57 0.54 0.50 0.48

2009

1.00 0.78 0.71 0.67 0.63 0.59 0.56 0.53 0.51

2008

All correlation coefficients are significant at the level of 1%

2011 2010 2009 2008 2007 2006 2005 2004 2003 2002 2001 2000 1.00 0.77 0.71 0.66 0.62 0.59 0.55 0.53

2007

1.00 0.78 0.71 0.66 0.62 0.57 0.55

2006

Table 13.1 Correlation analysis of export product relatedness during 2000–2011

1.00 0.77 0.70 0.65 0.60 0.58

2005

1.00 0.75 0.68 0.62 0.60

2004

1.00 0.75 0.66 0.63

2003

1.00 0.71 0.65

2002

1.00 0.70

2001

1.00

2000

13.3 Evolution of Export Product Space in China 289

290

13 What Drives the Evolution of Export Product Space in China?

Fig. 13.2 (a) The export product space in China, 2000. (b) The export product space in China, 2011

Figure 13.2b shows that the export product space has changed substantially during 2001–2011. The original main core develops into a larger one, including more products such as plastics and rubber, pulp and paper, and furniture products. However, the sub-core of electric and electronic products becomes less clear and disappears in 2011. Chemical and metal products that used to be in the main core in 2000 remain in the main core and develop new linkages with products in the

13.4

Econometric Analysis

291

periphery. The periphery of the product space becomes increasingly denser, implying that the export product space in China has evolved and stabilized considerably during 2000–2011. This echoes with the findings of Guo and He (2017) based on China’s industry data. To examine the effect of technological relatedness, this study analyzes the relationship between export/import density and the evolution of export products. Figure 13.3a, b shows the relationship between the probability of developing a comparative advantage in a new product 5 years later and the average export/import density of products without a comparative advantage at the beginning. Higher levels of density correspond to higher probabilities of developing a comparative advantage in new products, implying that technological relatedness serves as a driver for the evolution of export product space. Moreover, the correlation in Fig. 12.3a is much stronger than that in Fig. 13.3b, indicating a larger impact of export density over regional industrial diversification than that of import density.

13.4

Econometric Analysis

13.4.1 Variables and Model Specification Following Hidalgo et al. (2007) and Boschma and Capone (2016), we estimate the following econometric equation: xi, c, tþk ¼ α þ β1 xi, c, t þ β2 di, c, t þ β3 idi, c, t þ β4 ndi, c, t þ β5 NGi, t þ β6 LGc, t þ β7 NGi, t  di, c, t þ β8 LGc, t  di, c, t þ β9 NGi, t  LGc, t  di, c, t þ REGION þ PRODUCT þ YEAR þ εi, c, t ð13:2Þ The dependent variable takes the value of 1 if province c has a comparative advantage in product i in year t + k (k ¼ 3) and 0 otherwise. di,c,t is the density of product i in province c in year t, and idi,c,t is the import density of product i in province c in year t. ndi,c,t is the average density of product i in neighboring provinces of province c in year t. PRODUCT and YEAR are vectors of product and year dummy variables, which are included to control for any product characteristics and time-varying effects. The dummy variable for product (PRODUCT) is measured at the 2-digit level. We divide China into four big areas: East China, Central China, West China, and Northeast China.3 The dummy variable, AREA, is included to control for any area characteristics, given the enormous regional disparity in China.

3 East China, Hebei, Beijing, Tianjin, Shandong, Jiangsu, Shanghai, Zhejiang, Fujian, and Guangdong; Central China, Shanxi, Henan, Anhui, Hubei, Hunan, and Jiangxi; West China, Xinjiang, Inner Mongolia, Gansu, Ningxia, Shaanxi, Qinghai, Tibet, Sichuan, Yunnan, Guizhou, and Guangxi; Northeast China, Heilongjiang, Jilin, and Liaoning.

292

13 What Drives the Evolution of Export Product Space in China?

Fig. 13.3 Relationship between the probability of developing a comparative advantage in a new product (5 years later) and average export/import density of products without a comparative advantage at the beginning

In this chapter, we differentiate two types of government intervention: national and local governance. In the econometric equation, NGi,c,t refers to national governance variables. First, export tax rebate policy is one form of national governance. The export tax rebate policy is a financial tool employed by the central government in order to encourage the exports of certain types of products. Since the tax reform in

13.4

Econometric Analysis

293

Fig. 13.4 Export tax rebate rates in different types of products (2002–2011)

1994, China’s central government has adjusted the rate of export tax rebate several times (Fig. 13.4). In the late 1990s, China increased export tax rebate rate particularly after the financial crisis in 1997 to sustain its exports. However, since 2004, to upgrade its productive structure, the central government has increased the tax rebate rate of high value-added and high-tech products and decreased the tax rebate rate of “two highs, one resource” (lianggao yizi) products4 (Fig. 13.4) (see also Wu 2011). Such strategies would definitely have an impact over regional industrial development. We apply export tax rebate data at 4-digit product level in 2002–2011 and include the ratio of export tax rebate to export price in our model as a proxy of this type of government intervention at the national level (Rebate). FDI-related policies are another form of national governance. FDIs can raise host country’s efficiency and promote the competitiveness of export products, especially in developing countries (Kojima 1978). In order to entice FDIs in certain industries while prohibiting and limiting the development of FDIs in some other industries, China’s central government has issued the Catalogue for the Guidance of Foreign Investment Industries in 1995 and revised it in 1997, 2002, 2004, 2007, 2011, and 2015, according to changing national strategic and developmental needs. These policies would also strongly affect regional industrial evolution in China. In this

4 “Two highs, one resource” refers to high energy consuming, high pollution, resource-intensive production.

294

13 What Drives the Evolution of Export Product Space in China?

Fig. 13.5 Share of 4-digit products for which FDIs are encouraged or limited/prohibited in 2-digit products (2002)

chapter, we match the items included in the Catalogue for the Guidance of Foreign Investment Industries during 2000–2011 to 4-digit (HS code) export products. For instance, one item included in the 2002 Catalogue is “Production of biological fertilizer, high concentration of fertilizer (potash, phosphate), compound fertilizer,” classified as “encouraged.” We go through all 4-digit (HS code) export products and pinpoint products that fall into this category. In this case, five 4-digit products are identified, for which FDIs are encouraged: animal or plant fertilizer (3101); mineral nitrogen fertilizer and chemical nitrogen fertilizer (3102); mineral phosphate fertilizer and chemical phosphate fertilizer (3103); mineral potash and chemical potash (3104); mineral or chemical fertilizer containing two or three elements of nitrogen, phosphate, and potash; and other fertilizers (3105). In 2002, for 407 products, FDIs are welcomed, while for 87 products FDIs are limited or prohibited. As is shown in Fig. 13.5, for capital- and technology-intensive products (e.g., general machinery, optical instruments and apparatus), FDIs are by and large welcomed, whereas an opposite attitude has been taken by the central government regarding pollutionintensive products (e.g., textile and chemical articles). We thus define two dummy variables: FDI_E taking the value of 1 if FDIs are encouraged for a product and FDI_L&P capturing if FDIs are limited or prohibited for a product. We expect FDI_E has a positive coefficient, whereas FDI_L&P has a negative effect. In the econometric equation, LGi,c,t represents local governance variables. The areas of development zones at the provincial and national level are used as a proxy of

13.4

Econometric Analysis

295

local governance. There are four types of development zones in China: economic and technological development zone, high and new technological development zone, export-processing zone, and tax-protected zone. The economic and technological development zone and high and new technological development zone are both designed to promote the development of knowledge- and tech-intensive industries, while the export processing zone and tax-protected zone are constructed specifically for the development of export-oriented industries. Besides, development zones also provide a variety of tax credits and facilitate the formation of industrial agglomeration (Wu 2014). Thus, we calculate the areas of four kinds of development zones at the provincial and national level (DZ_P and DZ_N) in 2000–2006 (subject to data availability constraints) to test the effect of local governance on the evolution of regional productive structure. The data is derived from the China’s Development Zones Audit Bulletin Catalog (2006 edition), which provides information on the location, area, approval time, approval bodies, and leading industry of 222 national development zones and 1346 provincial development zones. The coefficients β7 and β8 in the equation capture the ways in which institutional arrangements may affect the relationship between regional industrial diversification and regional preexisting productive structure. The coefficients β9 shows whether local and national governance co-shape regional industrial dynamics in a conflicting or complementary way. All density variables and institution-specific variables are normalized by subtracting the mean and divided by the standard deviation.

13.4.2 Econometric Results Since the dependent variable is a dummy variable, we estimate the equation with the probit model. Correlation analysis indicates that the correlations between most independent variables are moderate or low, suggesting that there are no serious problems of multicollinearity (Table 13.2). The econometric results are reported in Table 13.3. Model 1 shows that both d (export density) and id (import density) have a positive effect, which is consistent with theoretical proposition (Hypothesis 13.1). The coefficient of d (export density) is larger than that of id (import density), implying that knowledge base within a region plays a much more important role in triggering the emergence of new products than learning from imports (see also Boschma et al. (2013) for a similar argument). nd (the average export density of a product in neighboring provinces) also has a positive and significant effect, indicating that knowledge spillovers can take place between neighboring provinces (Hypothesis 13.2). The findings support the first and second hypotheses. Besides, the coefficient of xi,c,t is positive and significant, suggesting that products having a comparative advantage at the beginning of the period are more likely to have a comparative advantage at the end of the period, which echoes with the findings of Boschma et al. (2012) and Zhu et al. (2017). Overall, the significance of d, id, and xi,c,t indicates that the evolution of

296

13 What Drives the Evolution of Export Product Space in China?

Table 13.2 Pearson correlation matrix d id nd Rebate FDI_E FDI_L&P DZ_N DZ_P

d

id

nd

Rebate

FDI_E

FDI_L&P

DZ_N

DZ_P

1.000 0.575 0.494 0.009 0.025 0.014 0.469 0.588

1.000 0.281 0.009 0.037 0.005 0.773 0.587

1.000 0.003 0.024 0.011 0.201 0.226

1.000 0.020 0.039 0.003 0.044

1.000 0.192 0.000 0.001

1.000 0.000 0.001

1.000 0.263

1.000

Table 13.3 Empirical results d id nd Rebate FDI_E FDI_L&P DZ_N DZ_P xi,c,t AREA dummies HS dummies YEAR dummies No. of obs. Pseudo R2 Log likelihood

Model 1 0.54*** 0.02*** 0.08***

Model 2

Model 3

0.05*** 0.04*** 0.01**

1.49*** Yes Yes Yes 354,609 0.35 64,183

1.55*** Yes Yes Yes 275,807 0.347 90,153

0.04*** 0.09*** 1.81*** Yes Yes Yes 277,109 0.295 96,947

Note: Dependent variable: xi,c,t + 3 * p < 0.1, **p < 0.05, ***p < 0.01

export product spaces in China is path-dependent. Results on id highlight the probability of new path creation through learning from imports. The coefficient of Rebate is positive and significant in Model 2, suggesting that a high level of export tax rebate rate raises the probability of developing a comparative advantage in a new industry. China’s export tax rebate policy indeed has affected the transformation of export product structure. FDI_E has a positive and significant coefficient, while FDI_L&P has a negative impact. In industries where FDIs are welcomed, it is much easier to develop a comparative advantage in new export products, while in industries where FDIs are limited or prohibited, FDIs play a minor role in the emergence of new industries. The findings confirm the effect of exportand FDI-related policies on the evolution of export productive structure. Statistical results in Model 3 show that government policies with respect to development zones at the national and provincial level have also shaped the evolution of export

13.4

Econometric Analysis

297

Table 13.4 Empirical results d Rebate Rebate*d FDI_E FDI_E*d FDI_L&P FDI_L&P*d DZ_N DZ_N*d DZ_P DZ_P*d xi,c,t AREA dummies HS dummies YEAR dummies No. of obs. Pseudo R2 Log likelihood

Model 1 0.35*** 0.05*** 0.01*** 0.03*** 0.03*** 0.02** 0.01

1.54*** Yes Yes Yes 275,807 0.337 90,139

Model 2 0.64***

0.09*** 0.04*** 0.22*** 0.04*** 1.44*** Yes Yes Yes 277,109 0.347 89,831

Note: Dependent variable: xi,c,t + 3 *p < 0.1, **p < 0.05, ***p < 0.01

productive structure and boosted the emergence of new industries, which is consistent with our theoretical predication. Table 13.4 reports statistical results on the interaction effects of government intervention and technological relatedness over the evolution of regional export productive structure. Model 1 presents the effects of the interaction terms between national governance and technological relatedness on the evolution of export products. The coefficient of the interaction term between technological relatedness and export tax rebate (Rebate) is positive and significant, implying that the implementation of China’s export tax rebate policy is reinforcing the effect of export density. Similarly, the interaction term between technological relatedness and FDI_E has a positive and significant effect, while the interaction term between technological relatedness and FDI_L&P is insignificant. This suggests that FDI restrictions do not increase or decrease the effect of export density, whereas policies attracting FDIs increase the effect of export density. These findings support the third hypothesis (Hypothesis 13.3). Finally, the impact of the interaction terms between local governance and technological relatedness over regional industrial evolution is reported in Model 2. The coefficient of both interaction terms is significant and positive, indicating development zones at the national and provincial level both increase the effect of export density (Hypothesis 13.4). In short, statistical results imply that Chinese governments should take advantage of technological relatedness while pushing forward regional industrial restructuring and support the development of

298

13 What Drives the Evolution of Export Product Space in China?

industries that are closely related to regional existing productive structure. Otherwise, government intervention could be less efficient. Table 13.4 shows China’s central and local governments have both taken advantage of technological relatedness while pushing forward new path creation. However, it is often argued that local governments do not always share the concerns that motivate the national government and at times evade or subvert central policies and adopt different development trajectories (Zhu and Pickles 2014). Table 13.5 reports the empirical results showing whether national and local governance have co-shaped the evolution of China’s export productive structure in conflicting or complementary ways. The key findings are as follows. The parameters of the interaction terms between national governance variable (Rebate or FDI_E), local governance variable (DZ_N or DZ_P), and d (export density) in all models are negative and significant. This indicates the presence of diminishing returns, suggesting that the partial effect of local/national governance is reduced when the other force is present. In other words, when one type of policies (policies issued by national or local governments) is sufficiently active, the marginal contribution of the other one is weakened. This implies that China’s national and local governments are in conflict in terms of their plans on industrial restructuring and regional economic development. In this case, their policies are inconsistent with one another, resulting in negative effects on economic efficiency (Hypothesis 13.5). As a robustness check, we re-estimate all models by using the logit model. All models are also estimated by using different threshold values (0.8 and 1.2) to determine a comparative advantage. We also conduct these analyses with 4-year and 5-year intervals for robustness check. Compared with the results presented above, these changes produce only minor effects. Hence, estimation results for these robustness checks are not reported here.

13.5

Conclusion

Regional development is closely related to its preexisting productive structure. EEG proposes that region tends to branch into related industries, and regional development is often path-dependent. Technological relatedness therefore serves as a key driver for regional industrial evolution. However, literature on relatedness has paid insufficient attention to the effect of institutional context on regional industrial diversification. To fill this gap, this study has explored the evolution of export product space in China and tested the impact of government intervention over regional industrial evolution based on China custom data during 2000–2011. Empirical analyses suggest that China has experienced substantial structural transformation in terms of its export productive structure during 2000–2011. China’s export product space presents a clear core-periphery structure. Its export products are increasingly related with each other, and the export product space has become denser during 2000–2011. The evolution of export products in China is significantly underpinned by technological relatedness in terms of both exports and

Model 2 0.180*** 0.503***

0.147**

0.250***

1.558*** Yes Yes Yes 236,406 0.3293 78091.40

0.319***

0.493*** 0.727***

0.297***

1.538*** Yes Yes Yes 236,406 0.3314 77847.05

Model 1

Note: Dependent variable: xi,c,t+3 *p < 0.1, **p < 0.05, ***p < 0.01

Rebate Rebate*d FDI_P FDI_P *d DZ_N DZ_N*d DZ_P DZ_P*d Rebate*d*DZ_N Rebate*d*DZ_P FDI_P*d* DZ_N FDI_P*d* DZ_P xi,c,t AREA dummies HS dummies YEAR dummies No. of obs. Pseudo R2 Log likelihood

Table 13.5 Empirical results

1.540*** Yes Yes Yes 236,406 0.3314 77849.22

0.331***

0.458*** 0.699***

0.371***

Model 3

1.541*** Yes Yes Yes 236,406 0.3307 77933.24

0.271***

0.120***

Model 4 0.218*** 0.570***

1.693*** Yes Yes Yes 236,406 0.3033 81126.68

0.053***

0.136*** 0.079*** 0.131***

Model 5

1.694*** Yes Yes Yes 236,406 0.3093 80424.71

0.167***

0.146***

0.686*** 0.483***

Model 6

0.110*** 1.633*** Yes Yes Yes 236,406 0.3167 79559.19

0.601*** 0.745***

0.151***

Model 7

0.220*** 1.682*** Yes Yes Yes 236,406 0.3113 80194.30

0.161***

0.780*** 0.560***

Model 8

13.5 Conclusion 299

300

13 What Drives the Evolution of Export Product Space in China?

imports within a region. In other words, Chinese regions tend to branch into industries that are technologically related to existing productive structure, showing that the evolution of export product space is path-dependent. In addition, knowledge spillovers also take place among neighboring regions. Two types of government intervention affect the emergence of new industries and industrial diversification in Chinese regions: national and local governance, implying that Chinese governments can restructure export product space by using export- and FDI-related policies and setting up development zones to catalyze the formation of new industries. However, the combination of national and local governance exhibits diminishing returns. That is, when one type of governance is sufficiently active, the other one does not contribute as much. Local and national governance have played somehow conflicting roles in China’s regional industrial dynamics. This study contributes to the EEG literature by exploring the evolution of export produce space in a transitional economy where market force and government intervention work hand in hand in shaping regional economic development. Empirically, this study may help policy-makers to better understand how regions develop and what national and local governments can do to promote economic growth and push forward industrial restructuring. The state should focus on the development of industries that are closely related to regional existing productive structure; otherwise, government intervention could be unproductive. Even more attention should be directed toward the gap between national and local interests, as well as the constraint this gap places on the state’s ability to govern effectively.

References Alcacer, J., & Oxley, J. (2014). Learning by supplying. Strategic Management Journa, 35(2), 204–223. Bahar, D., Hausmann, R., & Hidalgo, C. A. (2014). Neighbors and the evolution of the comparative advantage of nations: Evidence of international knowledge difussion? Journal of International Economics, 92(1), 111–123. Barbieri, E., Tommaso, D., Marco, R., & Bonnini, S. (2012). Industrial development policies and performance in southern China: Beyond the specialized industrial cluster program. China Economic Review, 23(3), 613–625. Bathelt, H., Malmberg, A., & Maskell, P. (2004). Clusters and knowledge: Local buzz, global pipelines and the process of knowledge creation. Progress in Human Geography, 28(1), 31–56. Beaudry, C., & Schiffauerova, A. (2009). Who’s right, Marshall or Jacobs? The localization vs urbanization debate. Research Policy, 38(2), 318–337. Bishop, P., & Gripaios, P. (2010). Spatial externalities, relatedness and sector employment growth in great Britain. Regional Studies, 44(4), 443–454. Boschma, R. (2005). Proximity and Innovation: A critical assessment. Regional Studies, 39(1), 61–74. Boschma, R., & Capone, G. (2015). Institutions and diversification: Related versus unrelated diversification in a varieties of capitalism framework. Research Policy, 44(10), 1902–1914. Boschma, R., & Capone, G. (2016). Relatedness and diversification in the European union (EU-27) and European neighbourhood policy countries. Environment and Planning C: Government and Policy, 34(4), 617–637.

References

301

Boschma, R., & Frenken, K. (2011). The emerging empirics of evolutionary economic geography. Journal of Economic Geography, 11(2), 295–307. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R. A., & Wenting, R. (2007). The spatial evolution of the British automobile industry: Does location matter? Industrial and Corporate Change, 16(2), 213–238. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new industries at the regional level in Spain: A proximity approach based on product relatedness. Economic Geography, 89 (1), 29–51. Boschma, R., Martin, V., & Minondo, A. (2017). Neighbour regions as the source of new industries. Papers in Regional Science, 96 (2), n/a–n/a. Castellani, D., Serti, F., & Tomasi, C. (2010). Firms in international trade: Importers’ and exporters’ heterogeneity in Italian manufacturing industry. The World Economy, 33(3), 424–457. Cheung, K. Y., & Lin, P. (2004). Spillover effects of FDI on innovation in China: Evidence from the provincial data. China Economic Review, 15(1), 25–44. Cohen, W. M., & Levinthal, D. A. (1990). Absorptive capacity: A new perspective on learning and innovation. Administrative Science Quarterly, 35(1), 128–152. Donaldson, J. (2016). Assessing the balance of power in central-local relations in China. London: Routledge. Essletzbichler, J. (2007). Diversity, stability and regional growth in the United States, 1975–2002. In K. Frenken (Ed.), Applied evolutionary economics and economic geography (pp. 203–229). Cheltenham: Edward Elgar. Essletzbichler, J. (2015). Relatedness, industrial branching and technological cohesion in US Metropolitan areas. Regional Studies, 49(5), 752–766. Filippetti, A., & Cerulli, G. (2015). Are decentralized regions ruled better? Evidence from European regions using a dose-response approach. Social Science Electronic Publishing. Filippetti, A., & Sacchi, A. (2016). Decentralization and economic growth reconsidered: The role of regional authority. Environment and Planning C: Government and Policy, 34(8), 1793–1824. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Fujita, M., & Hu, D. (2001). Regional disparity in China 1985–1994: The effects of globalization and economic liberalization. The Annals of Regional Science, 35(1), 3–37. Glaeser, E. L., Kallal, H. D., Scheinkman, J. A., & Shleifer, A. (1992). Growth in cities. Journal of Political Economy, 100(6), 1126–1152. Guo, Q., & He, C. (2017). Production space and regional industrial evolution in China. GeoJournal, 82(2), 379–396. Hartog, M., Boschma, R., & Sotaruta, M. (2012). The impact of related variety on regional employment growth in Finland 1993–2006: High-tech versus medium/low-tech. Industry and Innovation, 19(6), 459–476. He, C., Wei, Y. H. D., & Xie, X. (2008). Globalization, institutional change, and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. He, C., Yan, Y., & Rigby, D. (2016a). Regional industrial evolution in China. Papers in Regional Science. https://doi.org/10.1111/pirs.12246. He, C., Zhu, S., & Yang, X. (2016b). What matters for regional industrial dynamics in a transitional economy? Area Development and Policy. https://doi.org/10.1080/23792949.2016.1264867. Heebels, B., & Boschma, R. (2011). Performing in dutch book publishing 1880–2008: The importance of entrepreneurial experience and the Amsterdam cluster. Journal of Economic Geography, 11(6), 1007–1029. Henderson, V. (1997). Externalities and industrial development. Journal of Urban Economics, 42 (3), 449–470.

302

13 What Drives the Evolution of Export Product Space in China?

Hidalgo, C. A., & Hausmann, R. (2009). The building blocks of economic complexity. Proceedings of the National Academy of Sciences of the United States of America, 106(26), 10570–10575. Hidalgo, C. A., Klinger, B., Barabási, A.-L., & Hausmann, R. (2007). The product space conditions the development of nations. Science, 317(5837), 482–487. Isaksen, A. (2014). Industrial development in thin regions: Trapped in path extension? Journal of Economic Geography, 15(3), 1–16. Jacobs, J. (1969). The economy of cities. New York: Vintage Books. Jaffe, A. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. The Quarterly Journal of Economics, 108(3), 577–598. Klepper, S., & Simons, K. L. (2000). The making of an oligopoly: Firm survival and technological change in the evolution of the US. Tire Industry. Journal of Political Economy, 108(4), 728–760. Kojima, K. (1978). Direct foreign investment: A Japanese model of multinational business operations. London: Croom Helm. Lileeva, A., & Trefler, D. (2010). Improved access to foreign markets raises plant-level productivity. . .for some plants. Quarterly Journal of Economics, 125(3), 1051–1099. Marshall, A. (1920 [1890]). Principles of economics, 8th Edition. London: Macmillan. Martin, R., & Sunley, P. (2006). Path dependence and regional economic evolution. Journal of Economic Geography, 6(4), 395–437. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Nooteboom, B. (2000). Learning and innovation in organizations and economies. Oxford: Oxford University Press. Pan, F., Zhang, F., Zhu, S., & Wójcik, D. (2016). Developing by borrowing? Inter-jurisdictional competition, land finance and local debt accumulation in China. Urban Studies, 54(4), 897–916. Poncet, S., & Waldemar, F. S. (2013). Export upgrading and growth: The prerequisite of domestic embeddedness. World Development, 51(0), 104–118. Redding, S. J. (2011). Theories of heterogeneous firms and trade. Annual Review of Economics, 3 (1), 1–54. Wagner, J. (2012). International trade and firm performance: A survey of empirical studies since 2006. Review of World Economics, 148(2), 235–267. Wu, R. (2011). Study on the effect of export tax rebate on China’s export trade. Ph.D. thesis, Inner Mongolia Agricultural University. Wu, Y. (2014). The influence of industrial aggregation on exportation——based on the study of China’s high-tech development zones. Ph.D. thesis, Zhejiang University. Xiang, A., & Zhang, W. (2013). Characteristics, evaluation of Chinese industrial policy and adjustment recommendations. China Development Observation, 12, 19–21. Xu, C. (2011). The fundamental institutions of China’s reforms and development. Journal of Economic Literature, 49(4), 1076–1151. Yu, J., Zhou, L. A., & Zhu, G. (2016). Strategic interaction in political competition: Evidence from spatial effects across Chinese cities. Regional Science and Urban Economics, 57, 23–37. Zhao, X., & Zhang, L. (1999). Decentralization reforms and regionalism in China: A review. International Regional Science Review, 22(3), 251–281. Zhu, S., & He, C. (2016). Global and local governance, industrial and geographical dynamics: A tale of two clusters. Environment and Planning. C, Government & Policy, 34(8), 1453–1473. Zhu, S., & Pickles, J. (2014). Bring in, go up, go west, go out: Upgrading, regionalisation and delocalisation in China’s apparel production networks. Journal of Contemporary Asia, 44(1), 36–63. Zhu, S., He, C., & Zhou, Y. (2017). How to jump further and catch up? Path-breaking in an uneven industry space. Journal of Economic Geography, 17(3).

Chapter 14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

14.1

Introduction

The level, causes, and development of regional income inequalities in both developing and developed countries have received considerable and interdisciplinary attention, resulting in an extensive literature emerging in recent years that have sought to measure inequality and examine its evolution over time (Lewis and Williams 1981; Yuen Tsui 1991; Martin and Sunley 1998; Hudson 2007; Pike and Tomaney 2009; Wei and Liefner 2012). The spatial dimension of inequality has also attracted interests from policy-makers and governments, since regional disparities in economic activities boost overall income inequality. Besides being a determinant of interpersonal inequality, regional inequality matters since it is often associated with political and ethnic conflicts, causing social dissatisfaction and political instability (Kanbur and Zhang 2005). Existing literature tends to focus on a region’s different factor conditions, the quantity and quality of input factors (e.g., capital, labor, technology, and infrastructure), and their contributions to regional inequality (Florida and Kenney 1988; Wei 2001; Wan et al. 2007; Wei et al. 2011), while few studies have pointed out the importance of factor productivity as a driving force of regional inequality. This chapter contributes to the literature on the causes of regional inequality by paying special attention to factor productivity—the effectiveness with which accumulated factors of production, or capital, are used to produce output—rather than factor accumulation. The impressive economic growth of China since the late 1970s is associated with uneven regional development (Sutherland and Yao 2011). The wealthiest regions are mostly located in the coastal provinces, while inland regions have lagged behind in terms of economic development. China’s most developed regions (e.g., Beijing and

Modified article originally published in [He, C., Zhou, Y. and Zhu, S. (2017), Firm Dynamics, Institutional Context, And Regional Inequality Of Productivity In China. Geogr Rev, 107: 296–316.]. Published with kind permission of © [Wiley, 2018]. All Rights Reserved. © Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_14

303

304

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Shanghai) have a human development index higher than 0.8, similar currently to Hungary, Portugal, and Argentina, whereas the least developed provinces such as Xinjiang and Ningxia stand at around 0.6, similar to Cambodia, Vietnam, and Laos (UNDP 2013). In short, significant regional differences and imbalanced spatial development in terms of income and human development can be identified across China’s regions, making China an ideal case to examine regional inequality and its complicated relationship with region’s factor productivity. This chapter moves away from focusing on the relationship between factor accumulation and regional inequality and pays more attention to the ways in which regional inequality has articulated with factor productivity. It does so by using a large firm-level dataset on China’s manufacturing industry to measure and examine regional inequality of factor productivity in China at various geographical scales. In addition, we decompose the dynamics of regional inequality of productivity and investigate its relationship with institutional context and firm dynamics. The following section presents a conceptual account on the interconnectedness between regional inequality of productivity, firm dynamics, and institutional context. The third section introduces the data and our methodology for empirical analysis. Section 14.4 discusses regional inequality of productivity in China, while the fifth one decomposes productivity and analyzes regional inequality at the regional level. After a regression analysis in the sixth section, the last one concludes the chapter.

14.2

Regional Inequality of Productivity, Firm Dynamics, and Institutional Context

There has long been concern about inequalities between regions in both developing and developed countries, whether they are increasing or decreasing and the extent to which they can be reduced and through what measures (Dunford 1994; Rey and Janikas 2005; Shorrocks and Wan 2005; Kuijs and Wang 2006; Sutherland and Yao 2011). This has resulted in various strands of literatures, including the convergence school that proposes the level of regional inequality will decline over time and the divergence school that carries a relatively pessimistic perspective and believes regional inequality persists or even deteriorates. Neoclassical theory of convergence tends to highlight the role of markets in resource allocation and argues that regional inequality is temporary (Solow 1956). Kuznets (1955) and Williamson (1965) have further suggested an inverted-U trajectory of regional inequality and argued that as an economy transitions from one based mainly on agriculture toward an industrydominated one, regional inequality increases at first and then peaks and decreases in the long term. In contrast, divergence thoughts such as the new economic geography have emphasized the persistence of a core-periphery structure due largely to increasing returns and agglomeration externalities (Krugman 1991; Krugman and Venables 1995). More recently, some empirical studies increasingly adopt a more complicated

14.2

Regional Inequality of Productivity, Firm Dynamics, and Institutional Context

305

view as they envisage regional inequality as contingent not only on the developmental stages of the economy but also on a variety of factors, such as geographical scale, spatial division of labor, institutional context, and region’s integration into the global economy (Wei 2002; Bradshaw and Vartapetov 2003; Barnes et al. 2004; Lessmann 2014). While explaining regional inequality, existing empirical studies tend to focus on the quantity and quality of region’s input factors (e.g., capital, labor, technology, and infrastructure), whereas insufficient attention has been paid to how efficiently factor accumulation (input) has been translated into economic development and income increase (output)—i.e., factor productivity (Aiello and Scoppa 2000). Low-productivity and slow-productivity growth, as opposed to impediments to factor accumulation, may be the key to understanding why a certain regions face low income and stagnant economic development relative to others and why a certain less developed regions fail or manage to catch up (Young 1995; Kim and Lau 1994; Hsieh and Klenow 2007; Bosworth and Collins 2008). By productivity, we are referring to factor productivity that is often defined as the effects in total output not caused by inputs and factor accumulation but by an economy’s long-term technological change or technological dynamism (Nishimizu and Page 1982; Aiello and Scoppa 2000; Kumbhakar et al. 2000; Kim and Han 2001). Existing literature has pointed out that technological dynamism and productivity change are further affected by firm dynamics in two ways: on the one hand, the introduction of new technologies and organizational changes in incumbent firms may boost productivity growth (Disney et al. 2003a; Dosi and Nelson 2010; Baldwin and Gu 2011); on the other hand, productivity also increases as a result of a resource reallocation process where low-productivity firms exit and high-productivity firms enter (Disney et al. 2003a; Aghion et al. 2004; Lee and Mukoyama 2008). This idea that productivity growth involves restructuring and resource reallocation across firms also echoes with Schumpeter’s (1939, 1942) “creative destruction” argument. In short, firm dynamics play a crucial role in region’s productivity change through a process of reallocation of input factors from low-productivity firms to high-productivity firms and further contribute to regional inequalities. Another issue that has received much attention in the public debate as well as in the academic literature is the role of institutional context in regional inequalities, which should not be underestimated particularly in transitional economies like China (Wei 2002; Wei and Ye 2009; Liao and Wei 2012). Since the launch of China’s reform and opening-up policies, China has undergone dramatic economic growth and has experienced three fundamental transformations: (1) from a state-owned, collective economy dominated by state-owned enterprises (SOEs) to one with growing level of private ownership and market orientation (marketization); (2) from a centrally planned to an increasingly decentralized economy (decentralization); and (3) from a partially closed economy to one oriented toward export markets (globalization) (Wei 2001; He et al. 2008; Zhu and He 2013). These institutional transformations have profound influences on resource mobility and (re)allocation and, thus, on firm dynamics, productivity, and regional inequalities in China. Coastal regions that have been one step ahead in reforms have developed at

306

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

a faster pace based on the growth of non-SOEs and the agglomeration of foreign direct investments (FDIs), while inland China where marketization and globalization have been implemented half-heartedly is still reliant on SOEs and lags behind in terms of economic development. Decentralization has further created a geographically uneven institutional and economic landscape, as it granted local governments more autonomy and allowed them to take different routes. These result in a gap between coastal and inland regions in terms of not only local institutional context (e.g., the extent to which local economy is globalized/marketized) but also economic growth and regional development, providing an ideal case to study regional inequality and how it has been shaped by institutional context as well as firm dynamics.

14.3

Research Design

This chapter uses the ASIF dataset. We adopt Brandt et al.’s (2012) method and calculate the total factor productivity (TFP) index to measure firm’s productivity.1 TFP is the difference of output growth rate and the weighted average of growth rate of input factors, capturing mainly the contribution of technological dynamism and institutional changes, rather than the effects in total output caused by inputs and factor accumulation. We then calculate the annual average TFP of a region as the sum of TFP of all firms located in the region weighted by the share of firm’s industrial output: TFPi, t ¼

X e

se, t TFPe, t  P e s e, t

ð14:1Þ

where i, e, and t denote region, firm, and year, respectively. TFPi,t is the average weighted TFP of region i in year t, while TFPe,t is firm e’s TFP in year t. se,t is the industrial output of firm e in year t. To better understand the dynamics of regional inequality of productivity, we use Foster et al.’s (2001) method to decompose a region’s TFP growth into five parts as follows: se, t we, t ¼ P e s e, t

1

ð14:2Þ

This method has been widely used in recent studies (Brandt et al. 2012; Hsieh and Klenow 2009; Yang and He 2014). However, it is quite complicated; details thus have not been included here due to space limitation. Please see Yang and He (2014) for a detailed description of TFP calculation.

14.3

Research Design

ΔTFPi, t ¼

307

X

X  w ΔTFP þ ðTFPe,t1  TFPi,t1 Þ  Δwe, t e , t1 e , t e2C e2C X X  þ e2C Δwe, t ΔTFPe, t þ ðTFPe, t  TFPi,t1 Þ  we, t e2N h X i þ  e2X ðTFPe,t1  TFPi,t1 Þ  we,t1 ð14:3Þ

where C, N, and X denote surviving, entering, and exiting firms, respectively. In Eq. (14.3), the first term in this decomposition represents a within-firm component based on firm-level TFP changes, weighted by firm’s initial shares in the region (within share). This term captures TFP growth and decline of surviving firms (i.e., firms that have survived throughout this time period) in the region. The second term represents a between-firm component that reflects changing shares, weighted by the deviation of initial firm productivity from the initial regional index (between share). For a surviving firm, an increase in the share of its industrial output contributes positively (negatively) to the between-firm component if the firm has higher (lower) productivity than initial average productivity of the region. In other words, if the second term is positive (negative), it means that in the region, high-productivity firms have largely upsized (downsized), while low-productivity firms have mainly downsized (upsized). The third term represents a cross (i.e., covariance-type) term (cross share). The last two terms represent the contribution of entering and exiting firms, respectively (entry and exit share). An exiting firm contributes positively (negatively) if the firm exhibits productivity lower (higher) than the initial average productivity of the region, and an entering firm contributes positively (negatively) if the firm has higher (lower) productivity than the initial average productivity of the region. Empirical studies suggest that low-productivity firms are more likely to exit, while new entrants tend to be more productive through a redeployment of resources released by firm exit (Olley and Pakes 1996; Baily et al. 1992). If the fourth and fifth terms are positive, it means high-productivity firms enter and low-productivity firms exit. The first two terms thus indicate the extent to which the introduction of new technologies and organizational changes in incumbent firms may boost productivity growth, while the last two terms should fluctuate as a result of a resource reallocation process through firm entry and exit. Following Brandt et al. (2012), we consider a firm as an entrant in year t, if it is reported in the ASIF in year t but not in year t-1. Likewise, if firm i is reported in the ASIF in year t-1 but not in year t, it is assumed that firm i exits in year t. Since ASIF dataset only includes non-state-owned enterprises with annual sales of 5 million RMB or more besides state-owned enterprises, firm exit is likely to be slightly overestimated due to the fact that non-state-owned enterprise that passes the threshold (annual sales of 5 million RMB or more) in year t but fails to do so in year t+1 will be treated as an exiting firm. Nonetheless, this flaw only slightly affects research results2; ASIF has been widely used to study firm exit and entry (Brandt et al. 2012; Yang and He 2014). 2 We acknowledge that Mergers & Acquisitions may compromise the effectiveness of our measure. However, based on China’s Mergers & Acquisitions Yearbook, we found that this only has minor impact. For example, in 2006, there were 1,784 M&A cases (the largest number during 1998–2008) in China, while the number of firm exit and entry in 2006 were 20,400 and 47,616, respectively.

308

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

This decomposition method and its variations have been often used to examine the extent to which each component contributes to TFP growth (or decline) (Rigby and Essletzbichler 2000; Foster et al. 2001; Disney et al. 2003b). Most empirical studies based on data collected in developed countries have suggested that productivity growth is largely accounted for by high-productivity firms replacing low-productivity firms. In this research, we build on these studies and testify the contribution of each component to region’s TFP change in a transitional economy—China, as TFP change is also a key source of regional inequality. In addition to firm dynamics, we also seek to emphasize the role of institutional context as it potentially has fundamental effects on resource mobility and (re)allocation, and therefore on TFP growth and regional inequalities.

14.4

Regional Inequality of Productivity in China

Table 14.1 shows the temporal change of average regional TFP in China during 1998–2007. We also employ several indices to measure regional inequality of productivity—Gini index, Theil index, and the coefficient of variation (CV). The mean value of regional TFP has increased steadily from 2.40 to 3.56 during the study period, indicating a continuous productivity improvement throughout the country. More importantly, CV, Gini, and Theil indices have changed from 0.30, 0.16, and 0.04 in 1998 to 0.18, 0.10, and 0.02 in 2007, respectively, suggesting that regional inequality of productivity has been gradually declining in China. In other words, regions have been converging in terms of factor productivity. These findings also relate to recent debates on China’s development model. On the one hand, some empirical studies start to question the sustainability of China’s export-oriented and investment-driven industrialization model. A popular argument is that there has been only modest growth in terms of efficiency and factor productivity in China during the last few decades, with economic growth mainly driven by increasing use of input factors, suggesting such a growth model relies heavily on debt and state-led investment and is therefore unsustainable (Young 2003). On the other hand, Borensztein and Ostry (1996) and Bosworth and Collins (2008) have adopted a relatively more optimistic attitude toward China’s economic development and shown that in addition to relying on input accumulation, China has achieved an average annual TFP growth rate of about 3% during the last few decades. Our findings thus support the second view and highlight that China’s economic development is much more complicated than predicted by the first view. To better understand regional inequality of productivity in China, we show the spatial variation of regional TFP at the prefecture level in 1999 and 2007 (Fig. 14.1). Driven by the export-oriented industrialization model, the coastal region expanded its production capacity and factor productivity more rapidly than central and western regions (Fleisher and Chen 1997; Zhu and He 2013). In 1999, high TFP production was mainly located in the coastal region, with primary concentrations in Shandong, Fujian, Guangdong, Shanghai, Beijing, Jiangsu, and Zhejiang and some outliers in

14.4

Regional Inequality of Productivity in China

309

Table 14.1 Temporal change of regional TFP in China (1998–2007) Year 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007

Number of regions 308 338 319 340 340 340 337 339 339 338

Mean 2.40 2.43 2.58 2.69 2.83 2.96 3.03 3.15 3.39 3.56

CV 0.30 0.29 0.29 0.26 0.26 0.25 0.24 0.25 0.21 0.18

Gini index 0.16 0.15 0.16 0.14 0.14 0.14 0.13 0.13 0.11 0.10

Theil index 0.04 0.04 0.04 0.03 0.03 0.03 0.03 0.03 0.02 0.02

Fig. 14.1 Spatial variation of regional TFP in China in 1999 (left) and 2007 (right)

regional centers, such as those in Central and Western China along the Yangtze River. The late 2000s, however, started to witness a diffusive trend. While the coastal region has kept its leading position and became more productive, some regions in Central and Western China have also managed to upgrade their production and improve productivity dramatically (e.g., Chongqing, Chengdu, and Changchun). Finally, some regions in Western China (e.g., Yuxi in Yunnan province) had abnormally high TFP value in both 1999 and 2007, because local economy in these regions was mainly dominated by one or several large, highly productive, and often state-owned enterprises in resource-based and/or strategically important industries (e.g., tobacco processing industry and non-ferrous metal smelting and pressing industry). Figure 14.2 shows the geographical variation of regional TFP change from 1999 to 2007 in China. Most coastal regions, particularly those in the wealthiest and most developed areas such as Pearl River Delta, Yangtze River Delta, and Shandong Peninsula, have undergone a modest TFP growth (around 0.5–1.5). In contrast, TFP has increased by more than 1.5 in some inland regions in Central China and

310

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Fig. 14.2 Spatial distribution of regional TFP change from 1999 to 2007

relatively less developed regions in coastal China, suggesting a convergence tendency between these regions and the most developed regions in coastal China. Finally, TFP dynamics in Western China has been much more complicated, as some regions have achieved high TFP growth (more than 1.5), and some have failed to increase their productivity. Fleisher and Chen (1997) have shown that TFP in inland China was less than half of that in coastal provinces during the 1980s and early 1990s, and inferior factor productivity in noncoastal regions was a principal reason for their low economic growth, further resulting in a widening gap between coastal and noncoastal provinces in terms of economic development and productivity, while our research suggests that the gap has been shrinking during the late 1990s and 2000s (Figs. 14.1 and 14.2).

14.5

Decomposition of Regional TFP Growth

As China’s economic growth slows down, there is a growing voice suggesting that it may fall into the middle-income trap, which is characterized by a sharp deceleration in growth and more importantly in the pace of productivity increase (World Bank 2012). Eichengreen et al. (2012) argue that economic growth slowdown is essentially productivity growth slowdown and specifically they point out that 85% of slowdowns in the rate of output growth are due more to a slowdown in the rate of productivity growth than to any slowdown in input factor accumulation. In this section, we thus examine regional TFP growth in China and explore the source of regional productivity growth by using the abovementioned decomposition method. Table 14.2 reports productivity growth at the national level and in Western, Central,

14.5

Decomposition of Regional TFP Growth

311

Table 14.2 Decomposition of regional TFP growth (1999–2007) National Eastern China Central China Western China

Within share 0.343 0.324 0.400 0.352

Between share 0.056 0.056 0.017 0.161

Cross share 0.098 0.080 0.160 0.146

Entry share 0.776 0.757 0.838 0.640

Exit share 0.031 0.036 0.061 0.151

and Eastern China separately. Entry share has played a dominant role both at the national level and in all three regions, followed by the within share, indicating China’s regional TFP growth has been largely driven by productivity improvement of existing firms and entrance of firms with high productivity during 1999–2007. The exit share in Western China was the highest, suggesting a process of reallocation of input factors from low-productivity firms to high-productivity firms in this region as high-productivity firms entered and low-productivity firms exited. Finally, the fact that the between share has been mostly negative (except in Central China) means high-productivity firms have largely downsized, while low-productivity firms have mainly upsized. This process of inefficient resource reallocation may be due to government intervention and policies in favor of certain types of firms (e.g., SOEs) through subsidies and political supports. Below we will investigate TFP growth decomposition at the regional level and focus on the within, between, entry, and exit share.3

14.5.1 Within Share As shown in Table 14.2, the within share has played an important role in regional TFP growth during 1999–2007, indicating existing firms have become more productive in China. Table 14.3 shows the within share of regional TFP growth generated by SOEs, Hong Kong, Macau, and Taiwan’s firms (HMT firms), foreign firms (excluding HMT firms), and private domestic firms.4 Private firms and SOEs have accounted for the majority of the within share of TFP growth during 1999–2007, probably due to China’s reform of SOEs and the process of marketization. On the one hand, the reform of SOEs has sought to (1) increase operational autonomy of SOE managers; (2) introduce a market orientation into SOEs; (3) emphasize the corporation, modernization, and rationalization of SOEs and strengthen the regulation of state assets management (e.g., SOEs supervision); (4) sell off unproductive SOEs and lay off SOE workers to improve efficiency; (5) and invigorate and upgrade key SOEs, particularly those in industrial sectors with high strategic value (e.g., petroleum and automobile industry) (Ho and Young 2013). 3

We do not pay much attention to cross share, because it is just a covariance-type term. A firm is considered as a state-owned/HMT/foreign/private firm, if the share of state-owned/HMT/ foreign/private capital is the largest in the firm.

4

312

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Table 14.3 Within share generated by various types of firms Period 1999–2007 1999–2003 2003–2007

SOEs 0.073 0.068 0.029

Foreign firms 0.057 0.042 0.035

HMT firms 0.026 0.013 0.019

Private firms 0.185 0.077 0.128

On the other hand, marketization transformed China’s economic system from a planning one that was dominated by SOEs to one with growing level of private ownership and thus allowed non-SOEs, particularly private, domestic firms and foreign firms to better capitalize on region’s comparative advantage and take advantage of resources released in the process of the reform of SOEs. Finally, the reform of SOEs has slowed down since the early 2000s. As a result, within share of TFP growth generated by SOEs accounted for a smaller portion of the total in 2003–2007 than in 1999–2003, while private firms’ impact increased from the first half to the second half of the study period. As shown in Fig. 14.3, regions with a high level of the within share were mainly China’s traditional manufacturing bases, concentrating in the Yangtze River Delta, Pearl River Delta, and Shandong Peninsula, as well as regional centers in inland China (e.g., Chengdu-Chongqing area). Since the within share indicates productivity improvement of existing firms, these preferential locations of manufacturing industry with a large number of existing firms are more likely to outperform the rest. In contrast, it is extremely difficult for unfavorable locations, particularly those in Western China, with a limited number of firms to stand out in terms of the within share of regional TFP growth. Since SOEs and private firms have made the largest contribution to the within share (see Table 14.3), we here examine the spatial distribution of the within share generated by SOEs and private firms (Fig. 14.4). Regions with a high level of the within share generated by SOEs were largely concentrated in Central and Western China (e.g., Chengdu-Chongqing area and Northeast China). Private domestic firms played a much more important role in the coastal region (e.g., Zhejiang, Fujian, and Shandong provinces), even though some regions in Chengdu-Chongqing area and Northeast China also had high levels of the within shares derived from private firms. This is consistent with China’s uneven institutional landscape in terms of marketization, privatization, and the reform of SOEs. While the reform of SOEs has boosted SOEs’ efficiency in Central and Western China, privatization and marketization have been implemented wholeheartedly in the coastal region where non-state capital and privately owned firms have had a much bigger impact on economic development.

14.5

Decomposition of Regional TFP Growth

313

Fig. 14.3 Spatial distribution of the within share of regional TFP growth from 1999 to 2007

Fig. 14.4 Spatial distribution of the within share generated by SOEs (left) and private firms (right)

14.5.2 Between Share The between share was negative and relatively small during 1999–2007 (Table 14.2), suggesting that the effect of low-productivity firms downsizing and high-productivity firms upsizing was slightly smaller than that of high-productivity firms downsizing and low-productivity firms upsizing. This is also evident at the regional level. Figure 14.5 shows that the vast majority of regions in China have witnessed a negative and small between share (0.1 to 0) from 1999 to 2007, indicating that the between share has had a minor, negative effect over the dynamics of regional productivity. This implies that the process of resource reallocation in China was not efficient, due possibly to the process of decentralization, which has

314

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Fig. 14.5 Spatial distribution of the between share of regional TFP change from 1999 to 2007

granted local governments more power to intervene in local economic development, by providing favorable policies and political, financial, and technological supports (e.g., subsidies and tax credits) to certain types of firms (e.g., SOEs or foreign firms) and firms in certain industries that local authorities deemed as strategically important (e.g., automobile and electronics).

14.5.3 Entry and Exit Share Entry share is the dominant driving force of TFP growth during 1999–2007 in China (Table 14.2). Its role in China is much more remarkable than suggested by recent empirical studies based on TFP growth in developed countries (e.g., US) (Rigby and Essletzbichler 2000; Foster et al. 2001; Disney et al. 2003a). On the one hand, in developed, market-oriented economies, productivity improvement of mature, incumbent firms (within share) accounts for a larger share than it does in China. On the other hand, in a transitional economy like China, economic reform and industrial restructuring imply that it is more common to see high firm entry and exit rate, and therefore, the replacement of low-productivity firms by highproductivity ones plays a more critical role in China. Table 14.4 shows that foreign firms and private, domestic firms have made the largest contribution to the entry share of regional TFP growth. China’s reform policies and the subsequent process of marketization have gradually lifted restrictions on business licensing and encouraged private entrepreneurship (He et al. 2008). The opening-up policies and the process of globalization have attracted a considerable number of FDIs. As a result, private and foreign firms have started to exert an

14.5

Decomposition of Regional TFP Growth

315

Table 14.4 Entry and exit share generated by various types of firms Entry share

Exit share

Period 1999–2007 1999–2003 2003–2007 1999–2007 1999–2003 2003–2007

SOEs 0.038 0.009 0.017 0.082 0.067 0.034

Foreign firms 0.212 0.070 0.096 0.031 0.013 0.046

HMT firms 0.074 0.017 0.035 0.019 0.008 0.001

Private firms 0.453 0.111 0.111 0.001 0.007 0.039

Fig. 14.6 Spatial distribution of the entry (left) and the exit (right) share of regional TFP change from 1999 to 2007

increasingly important effect in economic development and productivity dynamics. Finally, SOEs have a leading effect over the exit share of TFP growth, indicating that marketization and the reform of SOEs have forced plenty of unproductive SOEs to exit. The exit share generated by SOEs was lower during 2003–2007 since the reform of SOEs slowed down in the early 2000s. Figure 14.6 shows that regions with a high level of entry share during 1999–2007 were mainly located in Central and Western China (e.g., Hunan, Sichuan, Jiangxi, Anhui, and Inner Mongolia) and relatively less developed areas in coastal provinces, such as Northern Guangdong, Western Shandong, and Northern Liaoning. Meanwhile, regions with a high level of exit share also concentrated in Central and Western China (e.g., Hunan, Hubei, and Guizhou), while the coastal region had observed a lower level of exit share. One possible explanation is the emerging process of industrial relocation within China particularly in the 2000s. As marketization and globalization proceed in China since the late 1970s, the growth of exportoriented production has driven the geographical distribution of the manufacturing industry from a broadly-based industry to one concentrated in coastal regions (He et al. 2008). However, since the early 2000s, this model of industrialization has encountered a series of constraints, particularly in the coastal region, including rising labor cost and labor shortages fueled by low wages and poor working conditions. As competitive pressures intensified in the coastal region, recent years

316

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Fig. 14.7 The most important component of TFP growth in China’s regions during 1999–2007 (left), 1999–2003 (middle) and 2003–2007 (right)

have seen rapid industrial restructuring and spatial shifts in production pattern, in particular industrial relocation from the coastal region to inland China, resulting in higher entry rate in Central and Western China as well as relatively less developed areas in coastal provinces (Zhu and He 2013). Such industrial relocation has further crowded out unproductive firms in those regions and led to a high level of exit share. Finally, we examine the leading component of TFP growth in China’s regions (Fig. 14.7). From 1999 to 2007, the entry share was the dominant component driving TFP growth in most regions. However, over a shorter period of time (e.g., 1999–2003 and 2003–2007), the within share became more important. In other words, TFP growth is driven mainly by a resource reallocation process, where low-productivity firms exit and high-productivity firms enter in the long run, but changes within existing firms, such as the introduction of new technology and organizational changes, also matter in the short run (see also Foster et al.’s 2001 work on TFP growth in the USA). Furthermore, TFP growth in the coastal region relied heavily on the entry share during 1999–2003; however, the role of the within share became more significant during 2003–2007. As marketization was progressively introduced into the economy, a large number of private firms emerged in the coastal region since the late 1970s. Globalization further attracted a remarkable inflow of FDIs to locate in China’s coastal region in order to take advantage of cheap labor and other low-cost input factors. However, as labor cost rose and competitive pressures increased in the coastal region in the early 2000s, these processes have slowed down (Zhu and He 2013; Zhu and Pickles 2014). The entry share therefore played a less important role in the second stage, as the impact of the within share grew.

14.6

Institutional Context and Regional Inequality of Productivity

14.6

317

Institutional Context and Regional Inequality of Productivity

To better understand the extent to which regional inequality of productivity has been shaped by institutional context, we formulate the following model: yi, t ¼ Globalizationi,t1 þ Marketizationi,t1 þ Decentralizationi,t1 þ Labor i,t1 þ Capitali,t1 þ Agglomerationi,t1 þ yt1 þ εi

ð14:4Þ

where i and t denote region and year, respectively. Three key independent variables are included to examine the impact of institutional context. Globalizationi,t-1 is the export intensity—the ratio of export to total output—of region i in year t-1, indicating the degree of globalization and the extent to which a region is export-oriented and linked to foreign direct investments. The presence of a large number of SOEs often reflects a low level of market orientation (He et al. 2008). We use the proportion of non-SOEs’ output in region i in year t-1 (Marketizationi,t-1) as a proxy of marketization. The ratio of the number of firms that have received government subsidies to the total in region i in year t-1(Decentralizationi,t-1) captures the magnitude of local government intervention under China’s decentralization system. We also add some control variables. The level of labor cost (Labori,t-1) is included, measured as the total expense on wages and welfare benefits for workers of all firms in region i in year t-1 divided by the total employment. Capitali,t-1 is the capital intensity of production, measured as fixed asset per employee in region i in year t-1. Agglomerationi,t-1 is the total industrial output of region i in year t-1, capturing agglomeration economies. We also include TFPi,t-1 as a control variable. Correlation analysis indicates that correlations of independent variables are moderate or low, suggesting no serious problem of multicollinearity. Table 14.5 reports the results of the estimations of Eq. (14.4) based on fixed effects panel regression. In model (1)–(6), dependent variables (yi,t) are TFP, TFP change, within share, between share, entry share, and exit share in region i in year t, respectively. In all models, labor, capital, and agglomeration show a relationship with various dependent variables that are consistent with theoretical prediction: input factors and agglomeration economies have had a positive impact over productivity and productivity growth. The coefficient of TFPt-1 in model (1) is positive and statistically significant, indicating that regions that have already accomplished a high level of productivity are more likely to have a high-level productivity in the future. However, the coefficient of TFPt-1 in model (2)–(6) is negative and statistically significant. This means that as regions move upward in terms of productivity, it becomes increasingly difficult to keep raising productivity level. Moving on to the results connected more closely with the central argument, note that in model (1)–(2), globalization has a positive and statistically significant effect on TFP and TFP growth, suggesting that integration into international markets has boosted regional TFP growth. Firms have benefitted from globalization as they learned from foreign firms that started to swarm into China since the 1980s on the

Note: *p < 0.05, **p < 0.01, ***p < 0.0

Globalization Marketization Decentralization Labor Capital Agglomeration TFPt-1 _cons N F R2

(1) TFP 0.242** 0.565*** 0.441* 5.726 11.36* 4.365* 0.705*** 0.460*** 1888 339.4 0.689

(2) ΔTFP 0.242** 0.565*** 0.441* 5.726 11.36* 4.365* 0.295*** 0.460*** 1888 14.78 0.121

(3) Within share 0.156* 0.351*** 0.048 4.115 2.012 1.670* 0.088*** 0.035 1888 9.255 0.063

(4) Between share 0.041 0.050 0.298* 0.936 8.555*** 1.017 0.084*** 0.127*** 1888 6.411 0.105

(5) Entry share 0.069* 0.037 0.087 1.709 3.465* 1.524** 0.094*** 0.186*** 1888 18.08 0.063

(6) Exit share 0.023 0.102*** 0.085 0.016 1.053 0.480 0.045*** 0.091*** 1888 4.029 0.024

14

Table 14.5 Regression results

318 How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

14.7

Conclusion

319

one hand and were increasingly encouraged to established extra-regional linkages with global lead firms in the North by participating in global production networks, on the other hand (Zhu and He 2013). The impact of the remarkable inflow of FDIs into Chinese regions is also reflected in the entry share model, as the coefficient of globalization is positive and statistically significant. This resonates with our findings above (see also Fig. 14.7). Marketization presents a positive and significant relationship not only with TFP and TFP growth but also with within share and exit share. On the one hand, the reform of SOEs has shut down inefficient and unproductive SOEs while simultaneously consolidated and rationalized the rest. On the other hand, increasing level of market orientation in the economy has provided firms with a stable and transparent business environment, characterized by low institutional barriers and bureaucratic costs as well as efficient market competition. These resulted in a large number of exits of inefficient and unproductive firms (particularly SOEs) (see model (6)) and productivity increase of surviving firms (see model (3)) (see also Tables 14.3 and 14.4). Finally, decentralization and government subsidies inhibit TFP growth, indicating that government intervention tends to be a disturbance to the market economy, and may compromise the efficiency of resource mobility and (re)allocation between firms. This effect is particularly evident in the between share model, echoing with our analysis in Sect. 14.5.2 on Fig. 14.5.

14.7

Conclusion

This chapter has examined the ways in which regional inequality of productivity has unfolded in China’s manufacturing industry and how it has interacted with firm dynamics and China’s peculiar institutional context. At the present time, a great deal of scholarly attention is directed toward the quantity and quality of region’s input factors, whereas less effort has been made to understand the effectiveness with which accumulated factors of production are used to produce output and the importance of factor productivity as a driving force of regional inequality. This chapter thus builds on existing research and focuses on factor productivity rather than factor accumulation, by disclosing how factor productivity is not only interconnected with firm dynamics (technological and organizational change of existing firms, firm entry and exit) but also shaped by institutional context, particularly in the context of China where a triple process of decentralization, globalization, and marketization has given rise to vast spatial variation of economic and institutional landscape. Using firm-level data of China’s industries during 1998–2007, we show that the degree of regional inequality of productivity in China has been declining as relatively less developed Central and Western China have been catching up with China’s wealthy coastal regions. Furthermore, our decomposition method shows that the entry share has played a dominant role, followed by the within share, indicating China’s regional productivity growth has been largely driven by productivity

320

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

improvement of existing firms and entrance of firms with high productivity during 1999–2007. Finally, marketization, decentralization, and globalization all have had some fundamental influence over productivity growth and regional inequality of productivity. Marketization has reformed and upgraded SOEs on the one hand and encouraged private and foreign entrepreneurship on the other. Globalization has allowed Chinese enterprises not only to access financial capital, more advanced technologies, knowledge and management skills, and high-quality input factors on the international market but also to learn from foreign firms that started to swarm into China since the 1980s. Decentralization and government intervention in the form of subsidies have, however, compromised the efficiency of resource mobility and (re) allocation and curbed regional productivity growth. Empirically, this research seeks to find brighter future for countries like China with a high level of regional inequality that may trigger political, social, and even security problems, by pointing out that marketization and globalization may enable regions to push forward economic growth, while government intervention could be counterproductive. This policy implication is also of central importance for developing countries that are facing the so-called middle-income trap as productivity growth is one potential way to transcend this trap. Theoretically, while interpreting regional inequality, we bring institutional context and firm dynamics to the forefront. We show that institutional factors have long-lasting consequences on regional inequality of productivity by using both a decomposition method and regression analysis. In drawing from analysis of firm-level data from 1998 to 2007, we realize the potential limitations of our study. First, our firm-level dataset, which covers all Chinese industrial state-owned enterprises and non-state-owned enterprises with annual sales of 5 million RMB or more, does not enable us to examine the geographical dynamics of small firms and may cause overestimation/underestimation of regional TFP growth and its components. Nonetheless, this flaw only slightly affects research results since the firms included in the dataset contribute to more than 90% of the total industry output. Second, our calculation on TFP and TFP change of HMT, private, and foreign firms may be affected by the frequent use of tax havens and “round tripping” of capital (Sutherland and Anderson 2015). Finally, even though we acknowledge that regional productivity has been converging, it is also pointed out that income inequality in China has been increasing (Sutherland and Yao 2011). Benjamin et al. (2008) have argued that regional inequality in terms of economic development may play a less important role in income inequality than one might anticipate, as at least half of estimated income inequality is driven by income differences between “neighbors” as opposed to income differences at the regional level. Their study shows that regional inequality only contributed a portion of China’s total income inequality. More research is therefore needed to disclose the relationship between regional inequality of economic development and income inequality in China as well as other developing countries and to understand income inequality at the micro level (e.g., household level), which we leave for future research.

References

321

References Aghion, P., Blundell, R., Griffith, R., Howitt, P., & Prantl, S. (2004). Entry and productivity growth: Evidence from microlevel panel data. Journal of the European Economic Association, 2(2–3), 265–276. Aiello, F., & Scoppa V. (2000). Uneven regional development in Italy: Explaining differences in productivity levels, Giornale Degli Economisti E Annali Di Economia, Nuova Serie, 60 (Anno 113) (2), 270–98. Baily, M. N., Hulten, C., Campbell, D., Bresnahan, T., & Caves, R. E. (1992). Productivity dynamics in manufacturing plants, Brookings Papers on Economic Activity. Microeconomics, 1992(January), 187–267. Baldwin, J. R., & Gu, W. (2011). Firm dynamics and productivity growth: A comparison of the retail trade and manufacturing sectors. Industrial and Corporate Change, 20(2), 367–395. Barnes, T., Peck, J., Sheppard, E., & Tickell, A. (2004). Reading economic geography. New York: Wiley. Benjamin, D., Brandt, L., Giles, J., & Wang, S. (2008). Income inequality during China’s economic transition. In China’s great economic transformation (pp. 729–75), Cambridge, UK: Cambridge University Press. Borensztein, E., & Ostry, J. D. (1996). Accounting for China’s growth performance. The American Economic Review, 86(2), 224–228. Bosworth, B., & Collins, S. M. (2008). Accounting for growth: Comparing China and India. Journal of Economic Perspectives, 22 (1): 45–66. Bradshaw, M. J., & Vartapetov, K. (2003). A new perspective on regional inequalities in Russia. Eurasian Geography and Economics, 44(6), 403–429. Brandt, L., Van Biesebroeck, J., & Zhang, Y. (2012). Creative accounting or creative destruction? Firm-level productivity growth in Chinese manufacturing. Journal of Development Economics, 97(2), 339–351. Disney, R., Haskel, J., & Heden, Y. (2003a). Entry, exit and establishment survival in UK manufacturing. The Journal of Industrial Economics, 51(1), 91–112. Disney, R., Haskel, J., & Heden, Y. (2003b). Restructuring and productivity growth in UK manufacturing. The Economic Journal, 113(489), 666–694. Dosi, G., & Nelson, R. R. (2010). Chapter 3 – Technical change and industrial dynamics as evolutionary processes. In H. H Bronwyn, & N. Rosenberg (Eds.), Handbook of the economics of innovation (Vol. 1, pp. 51–127). Dunford, M. (1994). Winners and losers the new map of economic inequality in the European Union. European Urban and Regional Studies, 1(2), 95–114. Eichengreen, B., Park, D., & Shin, K. (2012). When fast-growing economies slow down: International evidence and implications for China. Asian Economic Papers, 11(1), 42–87. Fleisher, B. M., & Chen, J. (1997). The coast–noncoast income gap, productivity, and regional economic policy in China. Journal of Comparative Economics, 25(2), 220–236. Florida, R. L., & Kenney, M. (1988). Venture capital, high technology and regional development. Regional Studies, 22(1), 33–48. Foster, L., Haltiwanger, J. C., & Krizan, C. J. (2001, January). Aggregate productivity growth. Lessons from microeconomic evidence. NBER, 303–372. He, C., Wei, Y., & Xie, X. (2008). Globalization, institutional change and industrial location: Economic transition and industrial concentration in China. Regional Studies, 42(7), 923–945. Ho, D., & Young, A. (2013). China’s experience in reforming its state-owned enterprises: Something new, something old and something Chinese. International Journal of Economy, Management and Social Sciences, 2(4), 84–90. Hsieh, C. T., & Klenow, P. J. (2007). Misallocation and manufacturing TFP in China and India. Working Paper 13290. National Bureau of Economic Research.

322

14

How Do Firm Dynamics Affect Regional Inequality of Productivity in China?

Hsieh, C. T., & Klenow, P. J. (2009). Misallocation and manufacturing TFP in China and India. Quarterly Journal of Economics, 24(4), 1403–1448. Hudson, R. (2007). Regions and regional uneven development forever? Some reflective comments upon theory and practice. Regional Studies, 41(9), 1149–1160. Kanbur, R., & Zhang, X. (2005). Fifty years of regional inequality in China: A journey through central planning, reform, and openness. Review of Development Economics, 9(1), 87–106. Kim, S., & Han, G. (2001). A decomposition of total factor productivity growth in Korean manufacturing industries: A stochastic frontier approach. Journal of Productivity Analysis, 16 (3), 269–281. Kim, J. I., & Lau, L. J. (1994). The sources of economic growth of the East Asian newly industrialized countries. Journal of the Japanese and International Economies, 8(3), 235–271. Krugman, P. (1991). Increasing returns and economic geography. Journal of Political Economy, 99 (3), 483–499. Krugman, P., & Venables, A. J. (1995). Globalization and the inequality of nations. Quarterly Journal of Economics, 110(4), 857–880. Kuijs, L., & Wang, T. (2006). China’s pattern of growth: Moving to sustainability and reducing inequality. China & World Economy, 14(1), 1–14. Kumbhakar, S. C., Denny, M., & Fuss, M. (2000). Estimation and decomposition of productivity change when production is not efficient: A paneldata approach. Econometric Reviews, 19(4), 312–320. Kuznets, S. (1955). Economic growth and income inequality. American Economic Review, 45(1), 1–8. Lee, Y., & Mukoyama, T. (2008). Entry, exit and plant-level dynamics over the business cycle (SSRN Scholarly Paper ID 1171122). Rochester: Social Science Research Network. Lessmann, C. (2014). Spatial inequality and development – is there an inverted-U relationship? Journal of Development Economics, 106(January), 35–51. Lewis, J. R., & Williams, A. M. (1981). Regional uneven development on the European periphery: The case of Portugal, 1950–1978. Tijdschrift voor Economische en Sociale Geografie, 72(2), 81–98. Liao, F. H. F., & Wei, Y. D. (2012). Dynamics, space, and regional inequality in Provincial China: A case study of Guangdong Province. Applied Geography, 35(1–2), 71–83. Martin, R., & Sunley, P. (1998). Slow convergence? The new endogenous growth theory and regional development. Economic Geography, 74(3), 201–227. Nishimizu, M., & J.M.P, Jr. (1982). Total factor productivity growth, technological progress and technical efficiency change: Dimensions of productivity change in Yugoslavia, 1965–78. The Economic Journal, 92(368), 920–936. Olley, G. S., & Pakes, A. (1996). The dynamics of productivity in the telecommunications equipment industry. Econometrica, 64(6), 1263–1297. Pike, A., & Tomaney, J. (2009). The state and uneven development: The governance of economic development in England in the post-devolution UK. Cambridge Journal of Regions, Economy and Society, 2, 13–34. January, rsn025. Rey, S. J., & Janikas, M. V. (2005). Regional convergence, inequality, and space. Journal of Economic Geography, 5(2), 155–176. Rigby, D. L., & Essletzbichler, J. (2000). Impacts of industry mix, technological change, selection and plant entry/exit on regional productivity growth. Regional Studies, 34(4), 333–342. Schumpeter, J. A. (1939). Business cycles: A theoretical, historical, and statistical analysis of the capitalist process. New York: McGraw-Hill. Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York: Harper & Row. Shorrocks, A., & Wan, G. (2005). Spatial decomposition of inequality. Journal of Economic Geography, 5(1), 59–81. Solow, R. M. (1956). A contribution to the theory of economic growth. The Quarterly Journal of Economics, 70(1), 65–94.

References

323

Sutherland, D., & Anderson, J. (2015). The pitfalls of using foreign direct investment data to measure Chinese multinational enterprise activity. The China Quarterly, 221(March), 21–48. Sutherland, D., & Yao, S. (2011). Income inequality in China over 30 years of reforms. Cambridge Journal of Regions, Economy and Society, 4(1), 91–105. UNDP. (2013). China human development report 2013: Sustainable and liveable cities: Toward ecological urbanisation. Beijing: China Translation and Publishing Corporation. Wan, G., Lu, M., & Chen, Z. (2007). Globalization and regional income inequality: Empirical evidence from within China. Review of Income and Wealth, 53(1), 35–59. Wei, Y. D. (2001). Decentralization, marketization, and globalization: The triple processes underlying regional development in China. Asian Geographer, 20(1–2), 7–23. Wei, Y. D. (2002). Multiscale and multimechanisms of regional inequality in China: Implications for regional policy. Journal of Contemporary China, 11(30), 109–124. Wei, Y. H. D., & Liefner, I. (2012). Globalization, industrial restructuring, and regional development in China. Applied Geography, 32(1), 102–105. Wei, Y. H. D., & Ye, X. (2009). Beyond convergence: Space, scale, and regional inequality in China. Tijdschrift voor Economische en Sociale Geografie, 100(1), 59–80. Wei, Y. D., Yu, D., & Chen, X. (2011). Scale, agglomeration, and regional inequality in provincial China. Tijdschrift voor Economische en Sociale Geografie, 102(4), 406–425. Williamson, J. G. (1965). Regional inequality and the process of national development: A description of the patterns. Economic Development and Cultural Change, 13(4), 1–84. World Bank. (2012). China 2030: Building a modern, harmonious, and creative high-income society. Yang, R., & He, C. (2014). The productivity puzzle of Chinese exporters: Perspectives of local protection and spillover effects. Papers in Regional Science, 93(2), 367–384. Young, A. (1995). The tyranny of numbers: Confronting the statistical realities of the East Asian growth experience. The Quarterly Journal of Economics, 110(3), 641–680. Young, A. (2003). Gold into base metals: Productivity growth in the People’s Republic of China during the reform period. Journal of Political Economy, 111(6), 1220–1261. Yuen, T. K. (1991). China’s regional inequality, 1952–1985. Journal of Comparative Economics, 15(1), 1–21. Zhu, S., & He, C. (2013). Geographical dynamics and industrial relocation: Spatial strategies of apparel firms in Ningbo, China. Eurasian Geography and Economics, 54(3), 342–362. Zhu, S., & Pickles, J. (2014). Bring in, go up, go west, go out: Upgrading, regionalisation and delocalisation in China’s apparel production networks. Journal of Contemporary Asia, 44(1), 36–63.

Chapter 15

Summary and Conclusion

Based on recent insights in evolutionary economic geography on regional development and its association with local knowledge spillovers, the cultivation of a local pool of specialized labor, local interfirm synergies and the division of labor, and institutional contexts, this book strives to re-examine regional industrial dynamics by employing “evolutionary” metaphors, concepts, and terminology. Specifically, the central thesis of the research demonstrates the ways in which regional industrial dynamics, firm dynamics, and product dynamics have been co-shaped by technological relatedness and, more importantly, some other peculiar factors in the context of China. It is divided into four sections: (1) industrial dynamics, path dependence and path creation; (2) firm dynamics, such as firm survival, entry and exit; (3) product dynamics and the evolution of China’s product space; and (4) regional industrial dynamics and the consequence. There has been a long-standing debate in economic geography, regional science, and urban economics, on the effects of agglomeration externalities on firm performance and regional economic growth. This strand of literature tends to distinguish between external economies that are restricted within particular sectors of the economy (i.e., localization externalities) and those that flow across sectors of the economy, urbanization externalities. For Marshall (1920 [1890]), economies from specialization in industry towns derived from local input-output networks, dense local labor pools, and knowledge spillovers. Jacobs (1969) was more generally concerned with diversification externalities resulting from interaction, generation, replication, recombination, and modification of ideas and applications across different sectors. The gist of their arguments is that knowledge tends to spillover in spatially proximate locations (Jaffe et al. 1993). Geographical proximity matters since actors are likely to enhance ties with partners that they have the most interactions with and are geographically close to, especially if those ties involve complicated interactions and trust (Barnes and Gertler 1999; Storper 1997). Other localized mechanisms, such as labor mobility, firm diversification, social networking, and

© Springer Nature Singapore Pte Ltd. 2019 C. He, S. Zhu, Evolutionary Economic Geography in China, Economic Geography, https://doi.org/10.1007/978-981-13-3447-4_15

325

326

15

Summary and Conclusion

entrepreneurial spinoffs, are also important to induce local knowledge spillovers, resulting in localized capabilities (Boschma and Frenken 2011; Maskell and Malmberg 1999; Tanner 2014). However, geographical proximity is neither a necessary nor a sufficient condition for spurring knowledge diffusion between firms (Boschma 2005). Empirical work in economic geography and regional science has confirmed that knowledge spillovers can spur firm innovation and performance and regional economic development, as long as the right extent of cognitive distance exists among economic agents (Boschma 2005; Nooteboom 2000). The notion that an optimal level of cognitive distance may exist in knowledge spillovers and sharing predicates on Nooteboom’s (2000) work, who claims that “information is useless if it is not new, but it is also useless if it is so new that it cannot be understood” (p. 153). Hence, the role of geographical proximity in determining economic phenomena is complemented and enriched by the notion of cognitive proximity (Lo Turco and Maggioni 2015). This new perspective based on technological relatedness, defined as interactive learning among economic actors stemming from the existence of a certain degree of cognitive proximity among them (Boschma and Iammarino 2009; Boschma et al. 2012, 2013; Frenken et al. 2007; Neffke et al. 2011), has given new momentum to debates in economic geography on the relative importance of specialization versus diversity economies in the mechanics of agglomeration externalities for firm innovation and regional development (Lo Turco and Maggioni 2015).

15.1

Industrial Dynamics

Based on a firm-level dataset of China’s manufacturing industries during 1998–2008, this part first shows that technological relatedness does underpin regional industrial dynamics in China. By dividing the country into several big regions, we first point out that the process of path-dependent regional economic development varies across regions. Specifically, while technological relatedness has played a fundamental role in regional industrial dynamics, Northwest and Southwest China, somehow, have managed to break the development routes set up by technological relatedness. This provides some preliminary evidences that some other factors that could potentially inflect the impact of technological relatedness in the process of industrial evolution may have been overlooked in canonical EEG studies, an argument that is further confirmed in the next two chapters of this part. In addition to path-dependent regional economic development underpinned by technological relatedness, regional industrial dynamics may also unfold in ways that are much more path-breaking and enable regions to create new, relatively unrelated paths of industrial development. It suggests that regional industrial dynamics are not only conditioned by preexisting regional capabilities and technological relatedness but also by the ways in which technological relatedness is interconnected with industry attributes and more importantly region’s institutional context. Here, in terms of industry attributes, we differentiate four types of industries: supplierdominated industries, science-based industries, SOE-dominated industries, and

15.2

Firm Dynamics

327

export-oriented industries. By regional institutional context, we are referring to a region’s global linkages, economic liberalization, and state intervention. We find that regional industrial development is a path-dependent process in China, one that is also inflected by industry characteristics. In addition, institutional context plays an even more complicated role in this process, as it reinforces the role of technological relatedness on the one hand and generates some opportunities for Chinese regions to transcend the confinement of technological relatedness on the other hand. In this part, we argue that EEG would profit from incorporating insights generated from the Chinese context. This point is also confirmed and explicated in other parts.

15.2

Firm Dynamics

This part starts with an analysis of the impact of localization and urbanization economies on entrepreneurship in China. Empirical results show that localization economies, especially supplier-customer linkages, have played a positive role in China’s new firm formation, while the effect of urbanization economies on entrepreneurship in China is quite unclear. We follow Frenken, Van Oort, and Verburg (2007) and Boschma and Iammarino (2009) and divide Jacobs’s externalities into related and unrelated variety. It turns out the former significantly promotes entrepreneurship, while the latter in most cases discourages entrepreneurship. This provides some preliminary evidence that Jacobs’s externalities do not necessarily result in knowledge spillover, instead, it only takes place effectively when complementarities and technological relatedness exist among industrial sectors in terms of shared competences. New firm formation is only one side of a coin. The other, equally important side of the coin is firm exit and firm survival, which are often interconnected with firm entry. While Schumpeter (1939, 1942) has focused on radical technological shifts and on how new entrants bring in new products and more advanced technologies, making existing technology regime and products of incumbents obsolete and forcing them to exit or catch up, we have examined industrial renewal and restructuring through a different perspective that emphasizes a local creative destruction process through which technologies change incrementally. Unlike Schumpeter who examines the impact of the introduction of radical innovation and entry of new firms over exit of incumbents, we show that firm exit actually creates a stimulus for entry of new firms, resulting in incremental innovation and productivity increase. In other words, we agree that firm exit and entry interact with one another but disagree on the direction of causality. In addition, we rely on econometric analyses to show that the relationship between firm exit and entry has been constantly shaped by an assemblage of various factors, including not only firm characteristics but also industrial linkages, and, most importantly, national and regional institutional contexts, particularly in the context of China where a triple process of decentralization, globalization, and privatization has resulted in enormous spatial and temporal variation of economic and institutional landscape.

328

15

Summary and Conclusion

Firm exit and firm survival have thus been heavily shaped by China’s peculiar institutional context that is characterized by a triple process of marketization, globalization, and decentralization. Marketization and globalization have clearly introduced harsh market competition, but decentralization has granted local governments strong incentives to protect and support local businesses. This provides institutional advantage for firm survival. We argue that competition effect, learning effect, and governmental support underpin firm survival and failure in China. Competition effects crowd out less productive firms. Local protectionism and supportive policies can certainly reduce the chance of firm failure. However, governmental intervention would generate negative externality, reducing the survival chances of productive firms without supports. This is the social cost of governmental intervention in industrial development. Furthermore, in addition to the interaction between the state and market forces and their influence on firm survival, we argue that firm heterogeneity should also be taken into account as the exit and survival of foreign-/state-owned/private enterprises, large/small firms, and old/young firms have been affected by global, regional, and local factors to different extents. After analyzing firm dynamics in China’s manufacturing industry, we scale down and focus on one industry, the apparel industry, in one chapter and one type of institutional arrangements—environmental regulations—in another chapter. The former chapter participates in the recent literature on “flying geese” theory by paying attention to the “domestic flying geese” model in which the labor-intensive industries relocate within China because of the large gap in economic development among different parts of the country. This chapter examines the multi-scalar process of spatial restructuring of firm demography and points out that gross job flow analysis demonstrated that the “domestic flying geese” model in China’s apparel industry was primarily driven by differences in the rate of job creation, especially through the entry of start-up firms. Once again, we stress that the case of China’s apparel industry illustrates the interaction of state involvement and market forces, which in turn influences firm dynamics and their spatial distribution. The blueprint for central government to relocate the apparel industry across the country, based upon the competitive advantage of individual regions, is constantly modified by industrial policies of subnational governments with a myopic focus on the potential benefits for their jurisdictions. The regionally decentralized authoritarian regime mediates the effect of market forces on firms, and the process of spatial restructuring in China’s apparel industry. The chapter on environmental regulations picks another angle to investigate the role of institutional context in firm dynamics and focuses particularly on the interaction between environmental regulations and firm dynamics. Recent studies on environmental regulation and industrial dynamics tend to predicate either on the pollution haven hypothesis (PHH) or on the Porter hypothesis (PH) and argue that environmental regulations may either render firms less competitive due to the additional costs required to comply with regulations and force them to exit or relocate or encourage firms to upgrade their production through industrial innovations. Existing studies thus fall short in uncovering the whole picture where various interconnected factors all have potentials to inflect the relationship between

15.3

Product Dynamics

329

environmental regulations and industrial dynamics. In addition, most researches have either adopted quantitative methods analyzing industrial dynamics at the industry and regional level or employed qualitative analyses such as case studies. Findings from the former approach have limitations due to the high level of aggregation, while the latter researches could be biased due to the limited sample size. This chapter contributes to recent studies by examining the relationship between environmental regulation and industrial dynamics at the firm level based on both the PHH and the PH. By combining a firm-level industrial dataset of China’s manufacturing industries and a dataset on polluting firms in China, we show the coexistence of the PHH and the PH in China, linking this story to two key factors that have been largely overlooked in existing studies—firm heterogeneity and government intervention.

15.3

Product Dynamics

By using the proximity product index, recent studies have argued that regional diversification emerged as a path-dependent process, as regions often branch into industries that are related to preexisting industrial structure. It is thereafter claimed that developed countries that start from the core, dense areas in the uneven industry space have more opportunities to jump to new related industries and therefore have more opportunities to sustain economic growth than do developing countries that jump from peripheral, deserted areas. In Chap. 11, we differentiate two types of regional diversification—path-dependent and path-breaking—and ask questions from a different angle: Can developing countries/regions jump further in the industry space to break path-dependent development trajectories and more importantly catch up with developed ones? Based on China’s export data, this chapter shows that regions can jump further by investing in extra-regional linkages and internal innovation. Not only do these two sets of factors promote regions jumping capability, but they also contribute to regions’ capability of maintaining a comparative advantage in technologically distant and less related industries. In addition, different extraregional linkage and internal innovation factors have affected path-breaking regional diversification to different extents, and these effects also vary across regions and industries. Empirically, this chapter seeks to find a brighter future for developing countries/regions. Theoretically, our research testifies some key findings of theoretical works in evolutionary economic geography by using a quantitative framework. In addition, this chapter includes some economic and institutional factors that have been left out in previous studies. What is more interesting is that Chinese various levels of governments may play different roles in product dynamics. Fiscal decentralization after the reform has allowed regional administrations to take different routes, resulting in a geographically uneven economic and institutional context. In other words, government intervention at the national and local level should be both taken into account separately to better understand China’s regional industrial dynamics. Those two types of

330

15

Summary and Conclusion

government intervention—national and local governance—both affect the emergence of new industries and industrial diversification in Chinese regions, implying that Chinese governments can restructure export product space by using export- and FDI-related policies and setting up development zones to catalyze the formation of new industries. However, the combination of national and local governance exhibit diminishing returns. That is, when one type of governance is sufficiently active, the other one does not contribute as much. Local and national governance have played somehow conflicting roles in China’s product dynamics.

15.4

Impact of Industrial Dynamics on Regional Inequality

One consequence of regional industrial dynamics is regional disparity. We have examined the ways in which regional inequality of productivity has unfolded in China’s manufacturing industry and how it has interacted with firm dynamics and China’s peculiar institutional context. At the present time, a great deal of scholarly attention is directed toward the quantity and quality of region’s input factors, whereas less effort has been made to understand the effectiveness with which accumulated factors of production are used to produce output and the importance of factor productivity as a driving force of regional inequality. We thus build on existing research and focus on factor productivity rather than factor accumulation, by disclosing how factor productivity is not only interconnected with firm dynamics (technological and organizational change of existing firms, firm entry and exit) but also shaped by institutional context, particularly in the context of China where a triple process of decentralization, globalization, and marketization has given rise to vast spatial variation of economic and institutional landscape. We seek to find brighter future for countries like China with a high level of regional inequality that may trigger political, social, and even security problems, by pointing out that marketization and globalization may enable regions to push forward economic growth, while government intervention could be counterproductive. This policy implication is also of central importance for developing countries that are facing the so-called middle-income trap as productivity growth is one potential way to transcend this trap. To sum up, our book provides the first book length, systematic treatment of the articulations between path dependence, path-breaking, technological relatedness, and institutional contexts in China. It is the first book examining the role-played technological relatedness and more importantly China’s peculiar institutional arrangements in regional industrial diversification. This book provides perhaps the first detailed account of the complex geographical dynamics restructuring China’s manufacturing industries from the evolutionary economic geography perspective. As such, the book builds on but is distinct from recent literature. Furthermore, by examining the ongoing debates based on China, we hope to point out how Western economic geography can benefit from critical engagements with China as well as other important and yet overlooked regions, such as Africa, East

References

331

Europe, South Asia, and South America. We see great value in testing Western theories and concepts in the non-Western context where nontraditional factors may come into focus. We also believe this will enable us to challenge and/or broaden the explanatory power of existing mainstream, Western-centric paradigms.

References Barnes, T., & Gertler, M. (1999). The new industrial geography: Regions, regulation and institutions. London: Routledge. Boschma, R. (2005). Proximity and innovation: A critical assessment. Regional Studies, 39(1), 61–74. Boschma, R., & Frenken, K. (2011). Technological relatedness and regional branching. In H. Bathelt, M. Feldman, & D. F. Kogler (Eds.), Beyond territory: Dynamic geographies of knowledge creation, diffusion and innovation (pp. 64–81). London: Routlege. Boschma, R., & Iammarino, S. (2009). Related variety, trade linkages, and regional growth in Italy. Economic Geography, 85(3), 289–311. Boschma, R., Minondo, A., & Navarro, M. (2012). Related variety and regional growth in Spain. Papers in Regional Science, 91(2), 241–256. Boschma, R., Minondo, A., & Navarro, M. (2013). The emergence of new industries at the regional level in Spain: A proximity approach based on product relatedness. Economic Geography, 89 (1), 29–51. Frenken, K., Van Oort, F., & Verburg, T. (2007). Related variety, unrelated variety and regional economic growth. Regional Studies, 41(5), 685–697. Jacobs, J. (1969). The economy of cities. New York: Vintage. Jaffe, A. B., Trajtenberg, M., & Henderson, R. (1993). Geographic localization of knowledge spillovers as evidenced by patent citations. The Quarterly Journal of Economics, 108(3), 577–598. Lo Turco, A., & Maggioni, D. (2015). On firms’ product space evolution: The role of firm and local product relatedness. Journal of Economic Geography, 16, 975–1006. Marshall, A. (1920 [1890]). Principles of economics (8th ed.). London: Macmillan. Maskell, P., & Malmberg, A. (1999). The competitiveness of firms and regions: ‘Ubiquitification’ and the importance of localized learning. European Urban and Regional Studies, 6(1), 9–25. Neffke, F., Henning, M., & Boschma, R. (2011). How do regions diversify over time? Industry relatedness and the development of new growth paths in regions. Economic Geography, 87(3), 237–265. Nooteboom, B. (2000). Learning and innovation in organizations and economies. Oxford: Oxford University Press. Schumpeter, J. A. (1939). Business cycles: A theoretical, historical, and statistical analysis of the capitalist process. New York: McGraw-Hill. Schumpeter, J. A. (1942). Capitalism, socialism and democracy. New York: Harper & Row. Storper, M. (1997). The regional world: Territorial development in a global economy. New York: Guilford Press. Tanner, A. N. (2014). Regional branching reconsidered: Emergence of the fuel cell industry in European regions. Economic Geography, 90(4), 403–427.