G L O B A L VA L U E C H A I N D E V E L O P M E N T R E P O R T 2 0 17 MEASURING AND ANALYZING THE IMPACT OF GVCs ON ECONOMIC DEVELOPMENT © 2017 International Bank for Reconstruction and Development/The World Bank 1818 H Street NW Washington DC 20433 Telephone: 202–473–1000 Internet: www.worldbank.org This work is a product of the World Bank Group, the Institute of Developing Economies, the Organisation for Economic Co-operation and Development, the Research Center of Global Value Chains within the University of International Business and Economics, and the World Trade Organization and is based on joint research efforts to better understand the ongoing development and evolution of global value chains and their implications for economic development. The findings, interpretations, and conclusions expressed in this work are those of the authors and do not necessarily reflect the views of the co-publishing partners, their Boards of Executive Directors, or the governments they represent. The co-publishing partners do not guarantee the accuracy of the data included in this work. The boundaries, colors, denominations, and other information shown on any map in this work do not imply any judgment on the part of the co-publishing partners concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Rights and Permissions The material in this work is subject to copyright. Because the co-publishing partners encourage dissemination of their knowledge, this work may be reproduced, in whole or in part, for noncommercial purposes as long as full attribution to this work is given. Any queries on rights and licenses, including subsidiary rights, should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; fax: 202–522–2625; e-mail: pubrights@worldbank.org. ISBN 978-92-870-4125-8 This publication uses U.S. spelling. All mentions of dollars refer to U.S. dollars, unless otherwise indicated. The term “billion” refers to a thou- sand million. The Research Center of Global Value Chains acknowledges the financial support from the Bill & Melinda Gates Foundation. iii  Foreword by Michael Spence T he global economy is evolving rapidly. It presents a with multiple participants and numerous cross-border flows, are complex and ever-changing picture. And it is impor- literally invisible when the focus is on gross flows. tant. The trends, opportunities, and challenges affect Bilateral trade balances shift in a major way when viewed in the lives of every person on the planet. But the forces value-added terms. While economists may deem these to be not at work and the results for trade, the structure of economies and so important, they are politically salient and strongly influence employment, incomes, and human capital values have been and public sentiment and hence attitudes toward trade, trade agree- still are at best incompletely understood. ments, and indeed fairness. There is a growing body of research on the impacts of glo- Services are extremely important and represent a growing balization and digital technology on individual economies. And fraction of trade. But to see that in detail, one needs to break there has been a huge and productive effort to reconfigure and down manufacturing value chains to expose the very large refine trade data so as to expose the complex value-added struc- services components that are embedded in them. This point ture of trade in goods and services, led by the Organisation for deserves emphasis. A careful study of GVCs in sectors classified Economic Co-operation and Development, the World Bank, the as manufacturing (presumably because a physical product is World Trade Organization, and a number of other institutions. delivered to the final consumer­ —­ which may be a firm that is pro - What has been largely missing is a comprehensive and ducing something else) reveals that a large fraction of the value detailed picture of the dynamic network structure of the global added is in services, broadly defined, with the fraction depend - economy. How economies are linked, specialize, and grow (or ing on the industry. These services are both upstream and down- not) is captured in the way global value chains (GVCs) are put stream from the physical production of components and from together. That is what this report is about. It is a huge contribu- assembly. Advanced economies, where the higher valued-added tion to our deepening understanding of what the global econ- components tend to be located, thus show up in GVCs in the omy really means and how it is changing. One can think of it as a upstream and downstream components. This gives rise to a different viewpoint, complementary to the single-economy stud- picture of GVCs composed of the participants along the value ies that focus on structural shifts and economic performance at chain correlated with their stage of development, referred to as the national level and that try to capture the impacts of trade and a “smile curve” because of its shape. This report does a superb technology on growth, incomes, jobs, and more. job of moving the quantitative analysis forward on this front. The insights in this volume are far too numerous to document Barriers to trade in services are declining, slowly, but are in a foreword. I encourage everyone, scholars, policymakers, and much higher than those in the movement of goods, narrowly leaders in business and civil society organizations as well as curi- construed. These barriers have multiple sources, including regu- ous and at times concerned citizens, to read the studies. If we lation, legal institutions, infrastructure, and simple capacity. do this, it will expand our shared understanding of the forces at The patterns of specialization across countries are much more work and facilitate productive discussion of how to adapt and visible and clearly defined when viewed through the lens of com- benefit from the global economy and how to deal with some of plex value-added chains. Through this lens you can detect, with the distributional challenges that come with it. It is natural to much greater precision, where employment is created, what drives see globalization through the lens of its effects on the domestic productivity growth, and what factors are affecting income distri- economy. But it is eye opening to see it also as a complex evolv- bution in a wide range of developed and developing countries. ing network. As China’s incomes rise and the tradables side of its econ- Here is a small sample of the insights that emerge from this omy shifts away from labor-intensive process manufacturing and valuable collection of studies. assembly, one would think that these components of complex Global trade looks very different when detailed quantitatively GVCs would shift to lower income countries, creating growth and in value-added terms rather than as gross flows of exports and development opportunities and momentum. To some extent, imports. Complex value chains (a growing fraction of global this is happening and will expand. But there are impediments. trade, especially trade in high-value manufacturing and services), This volume makes it clear that low wages are not enough. iv • Measuring and Analyzing the Impact of GVCs on Economic Development Connectivity and, with it, reasonably efficient processes for The report argues that connectivity in the networks that logistics and for meeting standards and regulatory requirements define the evolving architecture of GVCs is important. This is are critical. And lots of countries currently lose out on this front. another aspect of neighborhood mattering. GVCs properly doc- In the report, these issues are brought into sharp focus by distin- umented in value-added terms provide a detailed picture of the guishing between wages and unit labor costs and by highlight- network connectivity of an individual economy and hence of ing the factors that can differentially drive a wedge between the which parts of the global economy will strongly influence that two. Clearly, for competitiveness, unit labor costs are the critical economy and its various sectors. factor. Under the heading of neighborhood effects, the authors There are many more insights in this volume. I was espe- find that even well-structured and connected economies with cially interested in the way the evolving pattern of specialization relatively low unit labor costs and high connectivity will suffer if in production and services within global supply chains helps their immediate neighbors fall short on the same metrics. “Bad explain the divergent distributional impacts of globalization neighbors” have a depressing effect on trade and presumably across developed and developing countries. on growth. This may result from depressing effects on local trade The report helpfully distinguishes elements of an economy or other factors. One suspects this negative spillover is more sig- that are tradable and the large set that are nontradable. Clearly nificant in the case of landlocked countries, but that is specula- the tradables set is expanding with the support of enabling tion on my part. technology. For example, small and medium-size businesses The Global Value Chain Development Report is the result of can access global markets in a way that was simply impossible intensive and detailed work in assembling and analyzing data on before because the transaction costs of doing so were prohib- the structure of economies and on how they are linked. It creates itively high. But the nontradables part of any economy remains a much clearer picture of evolving patterns of independence. It very large. The linkages between the tradables and nontradables also presents a much clearer picture of comparative advantage. parts of an economy on both the supply and demand sides are No country has a comparative advantage in making iPhones. crucial in understanding the growth patterns. This volume makes What they do have is comparative advantage in generating ele- a good start at exposing the linkages between the tradables and ments of the iPhone GVC, in key services, components, and nontradables sides of an economy. These linkages are complex. assembly. The multinationals are the architects of the supply On the supply side they come through labor market shifts, and chains. And there are firms that are specialists in structuring on the demand side through spillover effects of rapid income GVCs in various groups of industries­ —­firms that are not asso - growth arising from specialization and growth on the tradables ciated with the production of a particular set of products. One side. There is more to do here, but this volume is a very good might think of them as pure GVC architects. start. Trade is a microeconomic phenomenon. Countries don’t In trade and in economies generally, your neighborhood mat- trade GDP. They specialize in trade in goods and services. And ters. Michael Porter and others in the classic book, The Compet- now we see that specialization for much of trade needs to be itive Advantage of Nations, documented that in virtually every thought of in terms of components of GVCs. With value-added industry there are a relatively small number of concentrated trade and GVCs you can see in detail, really for the first time, how centers of excellence where efficiency and innovation are high. trade catalyzes growth, especially in developing countries, and Proximity and agglomeration benefits matter. We continue to how it relates to employment, productivity, and income growth. see these patterns today, whether they be in autos, electronics, I learned a tremendous amount from the research reported financial centers, or in innovation hubs like that in Shenzhen in in this volume, and I highly recommend it to all who are inter- southern China. This has obvious implications for trade, partic- ested in growth and development, in trade and international ularly in services, including those contributing to manufacturing investment, and in policies that promote or retard growth and GVCs. development. Michael Spence Nobel Laureate in Economics v  Foreword by Pol Antràs O n 19 April 2017, international trade economists cel- manipulate, and interpret international trade statistics. In partic- ebrated the 200th birthday of their field. The date ular, a remarkable body of work has devised ingenious empirical marks the publication of David Ricardo’s Principles of methods to disentangle the value-added and intermediate input Political Economy and Taxation. Its landmark chapter contents of gross bilateral international trade flows. It has also 7, “On Foreign Trade,” introduced the concept of comparative developed theoretical models for interpreting the new data and advantage to economics. In his famous example, David Ricardo offering insights into the likely consequences of future trade lib- demonstrated that it was in England’s interest to exchange cloth eralization episodes. for wine with Portugal even if English workers could produce both This report offers a superb overview of some of the key devel- goods more efficiently than workers in Portugal could. opments in this body of work and breaks new ground on the Although the concept of comparative advantage is as rele- study of the rise of GVCs. I have spent the better part of my aca- vant today as it was 200 years ago, the nature of international demic career thinking about the growing disintegration of pro- trade flows has changed dramatically in recent decades. The duction processes across countries, and yet the various chapters information and communication revolution (and the attendant here have revealed a great number of new insights. radical decline in the cost of processing and transmitting infor- From the initial chapters that masterfully overview and doc- mation at long distances), a wave of regional trade agreements ument several empirical facts related to the participation of in various corners of the world, and the fall of communism (which various countries in GVCs and their relative positioning within brought into the capitalist system hundreds of millions of work- those chains, to the subsequent five chapters focused on spe - ers) led to the disintegration of production processes across bor- cific topics, the report offers an enormous amount of food for ders, as firms found it more profitable to organize production on thought. What are the consequences of the cascading effects of a global scale. World production is now structured into global trade costs along GVCs? What is the role of GVC participation in value chains (GVCs) in which firms source parts, components, and escaping the middle-income trap? How does services trade com- services from producers in several countries and in turn sell their plement and support merchandise trade in GVCs? What are the output to firms and consumers worldwide. The typical “Made in” consequences of variation in institutional quality for the geog- labels in manufactured goods have become archaic symbols of raphy of GVCs? And how should one design trade policy in the an old era. Today, most goods are “Made in the World.” new age of GVCs? The rise of GVCs has naturally captured the attention of inter- These are questions I feel much better equipped to answer national trade economists eager to bridge the apparent gap after working through this fascinating report. I am sure it will be between the new characteristics of the international organiza- a very useful reference for academics and practitioners for years tion of production and the standard methods used to collect, to come. Pol Antràs Robert G. Ory Professor of Economics Harvard University vi • Measuring and Analyzing the Impact of GVCs on Economic Development Contents Foreword by Michael Spence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . iii Foreword by Pol Antràs. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . v Partners . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xi Contributors. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xii Acknowledgments . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiii Abbreviations and acronyms . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . xiv Executive summary . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 1 Chapter 1 Analytical frameworks for global value chains: An overview . . . . . . . . . . . . . . . . . . 15 The global value chain paradigm: New-New-New Trade Theory?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 15 Concept development. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 17 Economic modeling. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 20 Empirical challenges . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 22 So, what’s next?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 27 Meta-methodological considerations . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 28 Chapter 2 Recent trends in global trade and global value chains. . . . . . . . . . . . . . . . . . . . . 37 Intermediate trade in manufactured goods and global business cycles . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 38 Decomposing domestic value added and final goods production into global value chain and other activities . . . . . . . . . 41 Global value chain production activities in the global business cycle. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 43 Measuring smile curves in global value chains: Creation and distribution of value added and job opportunities. . . . . . . . 52 Conclusion . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 60 Chapter 3 From domestic to regional to global: Factory Africa and Factory Latin America?. . . . . . . 69 What is upgrading?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 69 Integration for growth: Imports for exports . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 72 Regional value chains as enablers of integration into global supply chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . 74 National experience with value chain upgrading and integration: Automotive sector . . . . . . . . . . . . . . . . . . . . . . 81 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 89 Chapter 4 Accumulated trade costs and their impact on domestic and international value chains . . . . 97 Tariffs, cascading transaction costs, and competitiveness. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 98 Cascading transaction costs in the world trade network . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 106 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 110 Contents • vii Chapter 5 The middle-income trap and upgrading along global value chains . . . . . . . . . . . . . . 119 Definitions and implications of the middle-income trap. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 121 The relevance of global value chain trade for understanding the middle-income trap. . . . . . . . . . . . . . . . . . . . . 126 Conclusion and policy implications. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 134 Chapter 6 Services trade and global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 141 The role of services in global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 142 How services participation in global value chains is measured. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 143 The emerging patterns of services in global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 The implications of services in global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 148 Policy implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 155 Chapter 7 Institutional quality and participation in global value chains . . . . . . . . . . . . . . . . . 161 Institutions and participation in global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 161 African involvement in global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 166 Policy implications . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 Chapter 8 Preferential trade agreements and global value chains: Theory, evidence, and open questions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 175 Evolution of preferential trade agreements . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 Global value chains and the rationale for trade agreements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 178 Do deep agreements promote global value chains?. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 179 Global value chains and the content of preferential trade agreements. . . . . . . . . . . . . . . . . . . . . . . . . . . . . 180 Global value chains and the choice of preferential trade agreement partners. . . . . . . . . . . . . . . . . . . . . . . . . . 181 The future of the relationship between deep agreements and global value chains . . . . . . . . . . . . . . . . . . . . . . . 182 Conclusions. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 Appendix 1 Authors’ conferences. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 First Authors’ Conference: Background Papers . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 187 Second Authors’ Conference: Chapters. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 189 Annexes 1.1 Typology of global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .29 1.2 Governance arrangements along a production sequence. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 30 2.1 Shifting roles in global value chains for intermediate and final goods. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 61 2.2 Key to country abbreviations and sector codes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.1 Result, variables, and data sources for the study by Kowalski and Lopez-Gonzales (2016) . . . . . . . . . . . . . . . . . . . . . . . .91 4.1 Extended effective protection rates and the relative price of value added . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 113 4.2 Measuring the length of global value chains and the number of border crossings . . . . . . . . . . . . . . . . . . . . . . . . . . . 114 5.1 Attempts to quantify the middle-income trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 7.1 Results for Dollar, Ge, and Yu (2016) and Miranda and Wagner (2015). . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 viii • Measuring and Analyzing the Impact of GVCs on Economic Development Boxes 2.1 Identifying global value chain activities with new indicators. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .38 2.2 Identifying which types of production are global value chain activities and which are not. . . . . . . . . . . . . . . . . . . . . . . .41 2.3 The evolution of cross-border production sharing in complex global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . 51 4.1 Tariff and nontariff measures . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 103 4.2 Transaction costs, trade, and foreign direct investment . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 108 5.1 Lessons from China for apparel upgrading . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 133 Figures 1 Global value chains were expanding, until the financial crisis . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 2 2 Nominal growth rates of different value-added creation activities, 1996–2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 3 3 The estimated smile curve for China’s exports of electrical and optical equipment deepened between 1995 and 2009. . . . . . . . 4 4 Efficiency and factor income distribution in the information and communication industry in the United States, 1995–2009 . . . . . . 5 5 Efficiency and factor income distribution in the information and communication industry in China, 1995–2009 . . . . . . . . . . . . . 5 6 Trade in components shows three interrelated production hubs . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 7 7 Developing countries deeply involved in global value chains have low unit labor costs but not low wages, 2000 and 2010 . . . . . . 8 8 Relationship between the Logistics Performance Index and a centrality measure of country involvement in global value chains . . . . 9 9 The share of services is higher and has increased more sharply in trade in value added than in trade in gross terms, 1980, 1995, and 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 10 The share of services in exports is higher for developed countries, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .10 11 Developing countries maintain high restrictions on services trade . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 11 1.1 Genealogical map of analytical frameworks for global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 16 1.2 Three cascading constraints of globalization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .18 1.3 Modes of organizing value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .19 1.4 Typology of global value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .20 1.5 Optimal form of outsourcing options . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .21 1.6 Sequential choices for organizing value chains. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .23 1.7 Bilateral trade and value-added balances for the United States, by partner, 2004 . . . . . . . . . . . . . . . . . . . . . . . . . . . .24 1.8 China’s real effective exchange rates . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25 1.9 Gross trade accounting framework . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .25 1.10 Relative line position of countries in the regional production networks of East Asia, 1985, 2005 . . . . . . . . . . . . . . . . . . . . 27 2.1 Trends in global GDP and manufacturing trade before and after recent economic downturns, 1995–2015 . . . . . . . . . . . . . . .39 2.2 Evolution of intraregional trade in intermediate and final manufactured goods, 1995–2015. . . . . . . . . . . . . . . . . . . . . . 40 2.3 Trends in production activities as a share of global GDP, by type of value-added creation activity, 1995–2014. . . . . . . . . . . . 43 2.4 Simple global value chain production activities as a share of total global value chain production activities, 1995–2014 . . . . . . . 43 2.5 Nominal growth rates of value-added creation activities during the global business cycle at the global level, 1996–2014. . . . . . 44 2.6 Nominal growth rates of value-added creation activities during the global business cycle at the manufacturing and services sector level, by forward and backward linkages, 1996–2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .45 2.7 Structural changes in different types of value-added creation activities at the global level between 2011 and 2015 . . . . . . . . . 46 2.8 Structural changes in different types of value-added creation activities at the sectoral level between 2011 and 2015 . . . . . . . . 46 2.9 Structural changes in different type of value-added creation activities between 2011 and 2015 at the country level. . . . . . . . . .47 2.10 Structural changes in different type of value-added creation activities between 2011 and 2015, at the country and sector levels. . 48 2.11 Change in average annual growth rate by type of value-added creation activity between 2003–08 and 2011–15 . . . . . . . . . . . 49 2.12 Changes in growth of different types of value-added creation activities between country groups with positive and those with negative GDP growth between 2011 and 2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .50 2.13 Trend in production length by different types of value-added creation activities, world average, 2000–14 . . . . . . . . . . . . . . . 52 2.14 Change in production length for different types of value-added creation activities at the global level between 2011 and 2015. . . .52 2.15 Change in production length for different types of value-added creation activities at the sector and economy levels between 2011 and 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53 2.16 The decline in the number of border crossings drives the declining length of global value chain production at the country level between 2011 and 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .53 2.17 Estimated smile curve for China’s exports of electrical and optical equipment, 1995 . . . . . . . . . . . . . . . . . . . . . . . . . . 55 Contents • ix 2.18 Estimated smile curve for China’s exports of electrical and optical equipment, 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . 55 2.19 Estimated smile curve for Germany’s for automobile exports production, 1995 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 57 2.20 Estimated smile curve for Germany’s for automobile exports production, 2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . .57 2.21 Labor productivity and income distribution for the United States, 1995–2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .58 2.22 Labor productivity and income distribution for China, 1995–2009 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 59 A2.1.1 Eastern European economies’ trade of intermediate manufactured goods with Europe, 1995–2015. . . . . . . . . . . . . . . . . . . 61 A2.1.2 Poland’s exports of final and intermediate goods (motor vehicles) to Germany, 2000–15 . . . . . . . . . . . . . . . . . . . . . . . . 61 A2.1.3 Bilateral exports of Poland and Czech Republic, by manufacturing technology, 2000 and 2015. . . . . . . . . . . . . . . . . . . . .62 A2.1.4 Intra-NAFTA trade in intermediate manufactured goods, by main destination and origin, 1995, 2005, and 2015. . . . . . . . . . . 63 A2.1.5 Mexico’s trade with the United States in final and intermediate goods related to motor vehicles, 1995–2015. . . . . . . . . . . . . 63 A2.1.6 Evolution of net trade in intermediate and final manufactured goods of Cambodia and Thailand with China, the European Union, and the United States, 2000–15 and 1991–2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 64 A2.1.7 Evolution of net trade in intermediate and final textile products of Cambodia and Viet Nam with China and the world, 2000–15. . 64 3.1 The smile curve of the global value chain, 1970s and 2000s. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .70 3.2 Labor productivity: Manufacturing relative to business services in selected Organisation for Economic Co-operation and Development countries, 2010. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .71 3.3 China’s exports of textiles, by origin of value added, 1991 and 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 71 3.4 Significant determinants of a change in domestic value added in exports for developed and emerging economies. . . . . . . . . .73 3.5 Evolution of unit labor costs and average wages, 2000 to 2010 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .75 3.6 Determinants of change in domestic value added in exports, by sector . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 76 3.7 Changes in measures of integration into global value chains between 1995 and 2011 for the 63 economies in the Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database . . . . . . . . . . . . . . 76 3.8 Intraregional and extraregional value chains, by region, for the 63 economies in the OECD–WTO Trade in Value-Added database, 1995 and 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 77 3.9 Extraregional and intraregional trade in intermediates, 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 78 3.10 Composition of trade in low-income countries by intermediate and final goods, 2000–13. . . . . . . . . . . . . . . . . . . . . . . .78 3.11 Convergence in income per capita and exports in Africa relative to the United States between 1995 and 2014 . . . . . . . . . . . . 79 3.12 Economic complexity rankings in Africa, 1995 and 2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .79 3.13 Economic complexity rankings in Latin America and the Caribbean, 1995 and 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . .79 3.14 Economic complexity rankings in Asia and Central Europe, 1995 and 2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 80 3.15 Correlation of change in economic complexity rankings and change in foreign value-added content of exports between 1995 and 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .81 3.16 Regional exports by share of technological intensity, 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 82 3.17 Number of exporters in Latin America and the Caribbean by main export destination, 2011 . . . . . . . . . . . . . . . . . . . . . . 83 3.18 Selected regional and megaregional agreements, 2016. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 84 3.19 Gross exports of motor vehicles and parts by region and origin of value added, 2011. . . . . . . . . . . . . . . . . . . . . . . . . 84 4.1 Ten countries with the highest and lowest trade cost in all sectors, 2011 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 105 4.2 The global value chain zoo: spiders, snakes, and hybrid “snikers”. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 107 4.3 Ad valorem and accumulated trade costs in snakes and spiders. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 109 4.4 Graphical representation of trade in intermediate goods, 2011. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 111 4.5 PageRank scores and Logistics Performance Index values, 2011. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 112 5.1 GDP per capita growth, by income group, 2006–15 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.2 Ohno’s stages of catch-up industrialization. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 122 5.3 Product space maps of Peru and the Republic of Korea in 2009. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 125 5.4 Achieving functional, product, and intersectoral upgrading through skills, capital, and process upgrading . . . . . . . . . . . . . 127 5.5 Transmission channels from global value chain participation to the domestic economy . . . . . . . . . . . . . . . . . . . . . . . . 128 5.6 Growth of global value chain integration and GDP per capita by income category. . . . . . . . . . . . . . . . . . . . . . . . . . . 129 5.7 How global value chain determinants evolve as the engagement in global value chain changes . . . . . . . . . . . . . . . . . . . 130 6.1 Deconstructing services value added in gross exports. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 144 6.2 Gross exports of goods and services as a percentage of total world gross exports, 1980, 1995, and 2009 . . . . . . . . . . . . . . 145 6.3 Value-added exports of goods and services as a percentage of total world value-added exports, 1980, 1995, and 2009 . . . . . . 145 6.4 Share of direct, indirect, and foreign services value added in world gross exports, 1995–2011 . . . . . . . . . . . . . . . . . . . . 145 6.5 Direct and indirect domestic services value added and foreign services value added in gross exports, by country, 2011. . . . . . 146 6.6 Decomposition of services value added in world gross exports, by manufacturing industry, 2011. . . . . . . . . . . . . . . . . . .147 6.7 Decomposition of jobs embodied in gross manufacturing exports, by business function by industry, 2011. . . . . . . . . . . . . . 148 x • Measuring and Analyzing the Impact of GVCs on Economic Development 6.8 Financial development and revealed comparative advantage, by degree of embodied services, 2005 . . . . . . . . . . . . . . . . 151 6.9 Direct and indirect value added exports of financial services as shares of GDP, 1995–2009 . . . . . . . . . . . . . . . . . . . . . . 152 6.10 Direct and indirect value-added exports of business services as shares of GDP, 1995–2009. . . . . . . . . . . . . . . . . . . . . . 152 6.11 Services trade restrictiveness by services sector and region. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 154 7.1 Value added in business services as a percentage of GDP in China and the United States, 1985–2005 . . . . . . . . . . . . . . . . 162 7.2 Relationship of own and neighboring countries’ judicial quality (rule of law) . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 165 7.3 Average revealed comparative advantage across industries and countries, controlling for local institutional effects. . . . . . . . . 166 7.4 Depth of integration in global value chains of Sub-­ Saharan African and comparator countries, 2008–12 . . . . . . . . . . . . . . . 166 7.5 Institutional quality relative to development level in African countries: Association of Rule of Law Index and GDP per capita, 2010 . 167 7.6 Potential increase in trade in Sub-­ Saharan Africa from improvements in investment climate. . . . . . . . . . . . . . . . . . . . . . 168 8.1 The number and content of preferential trade agreements, 1951–2015. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 176 8.2 “WTO-plus” policy areas in preferential trade agreements, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 8.3 “WTO-extra” policy areas in preferential trade agreements, 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 177 8.4 Relationship between depth of preferential trade agreements and global value chain trade . . . . . . . . . . . . . . . . . . . . . 178 8.5 The impact of deep preferential trade agreements on two types of global value chain trade . . . . . . . . . . . . . . . . . . . . . 179 8.6 The depth of preferential trade agreements by the type or provisions and the development level of members . . . . . . . . . . . 181 Maps 6.1 Restrictiveness of services trade policy, 2008–10. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 153 7.1 Rule of law across Africa. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 168 Tables 2.1 Contribution to the change in global manufacturing trade by trade type, 1995–2015 . . . . . . . . . . . . . . . . . . . . . . . . . .39 A2.2.1 Country abbreviations. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 A2.2.2 Sector codes. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 66 3.1 Average wages and unit labor costs in manufacturing in selected developing and emerging economies, 2000 and 2010. . . . . . .74 3.2 Number of exported products by destination from countries in Latin America and the Caribbean, 2014 . . . . . . . . . . . . . . . 83 3.3 Main characteristics of targeted programs to promote the automotive industry in selected countries, 2014 . . . . . . . . . . . . . .85 3.4 Multi-institution and multidimensional policy mix targeted to the automotive industry, 2014 . . . . . . . . . . . . . . . . . . . . . .86 3.5 Policy mix of the National Incentive Scheme, Czech Republic, 2014. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .87 3.6 Horizontal programs that can support the automotive sector in Mexico, 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 3.7 Turkey’s investment incentives system, 2014 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 88 A3.1.1 Determinants of changes in the domestic value added in exports (standardized coefficients). . . . . . . . . . . . . . . . . . . . . .91 A3.1.2 Description of variables . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92 A3.1.3 Descriptive statistics. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .92 4.1 Amplification effect of trade costs on value added and profit margin. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 99 4.2 Magnification effect on tariff costs under global value chain trade, 2004. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 100 4.3 Influence of trade cost on value added and profit margin, domestic versus export prices . . . . . . . . . . . . . . . . . . . . . . . 101 4.4 Incidence of trade costs on output and input prices, 2006–11. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 104 5.1 Share of all countries in a given income group in 2000 and 2015 . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 120 5.2 Countries that have escaped the middle-income trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 124 5.3 Trajectories in global value chain engagement. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 131 5.4 The institutional dimension of global value chain engagement . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 132 A5.1.1 Country status relative to indicators associated with middle-income trap . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 135 7.1 Ten most and ten least contract-intensive manufacturing industries . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 163 A7.1.1 Summary of Dollar, Ge, and Yu (2016) industry results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 170 A7.1.2 Summary of Dollar, Ge, and Yu (2016) firm results . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 171 A7.1.3 Summary of Miranda and Wagner (2015) main . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 172 8.1 The trust dilemma of deep integration . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . 183 xi  Partners T he World Bank is an international development institu- whose mission is to promote policies that will improve the eco- tion established by the Articles of Agreement adopted nomic and social well-being of people around the world: Better by its member countries. The World Bank’s overarch- Policies for Better Lives. The OECD does this by providing a ing mission is to reduce poverty, improve living con- forum for governments to share experiences and seek solutions ditions, and promote sustainable and comprehensive develop- to common problems. ment in its developing member countries. It has two ambitious The Research Center of Global Value Chains is a global aca- goals to anchor its mission: end extreme poverty within a gener- demic think tank headquartered at the University of International ation and boost shared prosperity. It will achieve these goals by Business and Economics, focusing on basic and interdisciplinary providing loans, concessional financing, technical assistance, and research on the development of global value chains and their knowledge-sharing services to its developing member countries implications for global economies. and through partnerships with other organizations. The World Trade Organization (WTO) is an international organi- The Institute of Developing Economies (IDE–JETRO) is a zation that deals with the global rules of trade between countries. government-affiliated research institute that conducts basic It administers agreements, negotiated and signed by its members, and comprehensive research on economics, politics, and social which provide the legal ground rules for international commerce. issues in developing countries. Through its research, IDE–JETRO The purpose is to help trade flow as freely as possible for the eco- contributes to knowledge on developing economies and better nomic development and the welfare of its members’ citizens. The understanding of the regions to the government and public. WTO is serviced by a secretariat that provides expert, impartial, The Organisation for Economic Co-operation and Develop- and independent support to member governments, including ment (OECD) is an international and intergovernmental organiza- research, analysis, and statistical information related to the role tion comprising the world’s main industrialized market economies and developments of trade in the global economy. xii • Measuring and Analyzing the Impact of GVCs on Economic Development Contributors David Dollar Hubert Escaith Senior Fellow, John L. Thornton China Center, Brookings Institution Former Chief Statistician, World Trade Organization Satoshi Inomata Jakob Engel Chief Senior Researcher, Institute of Developing Economies, Economist, Trade and Competitiveness, World Bank Japan External Trade Organization Daria Taglioni Christophe Degain Lead Economist, Trade and Competitiveness, World Bank Senior Statistician, World Trade Organization Cecilia Heuser Bo Meng Research Analyst, Development Economics Research Group, Senior Researcher, Institute of Developing Economies, Japan World Bank External Trade Organization Aaditya Mattoo Zhi Wang Research Manager, Trade and International Integration, World Professor and Director, Research Center of Global Value Chains, Bank University of International Business and Economics Matthew Kidder Nadim Ahmad Assistant Professor, University of International Business and Head of Trade and Competitiveness Statistics Division, Statistics Economics Directorate, Organisation for Economic Co-operation and Development Michele Ruta Lead Economist, Trade and Competitiveness, World Bank Annalisa Primi Head of Structural Policies and Innovation Unit, Development Jose Guilherme Reis Centre, Organisation for Economic Co-operation and Development Manager, Trade and Competitiveness, World Bank xiii  Acknowledgments T he Global Value Chains Development Report is a joint the University of Hong Kong), Maggie Chen (professor at the publication of the World Bank Group, the Institute of George Washington University), Marcel Timmer (professor at Developing Economies (IDE–JETRO), the Organisation the University of Groningen), Gary Hufbauer (senior fellow at the for Economic Co-operation and Development (OECD), Peterson Institute for International Economics), and Nicholas C. the Research Center of Global Value Chains (RCGVC-UIBE), and Hope (director of the China program at the Stanford Center for the World Trade Organization (WTO), based on joint research International Development). The editors are grateful to Michael efforts to better understand the ongoing development and evo- Spence, Pol Antràs, and Shang-Jin Wei for their invaluable exper- lution of global value chains and their implications for economic tise and advice on the overall narrative of the report. development. The editors also thank Robert Koopman, chief economist This first report draws contributions from 16 background of the World Trade Organization, and Anabel Gonzalez, senior papers presented and discussed at the conference “Making director of the World Bank Group’s Trade and Competitiveness Global Value Chains Work for Economic Development and Global Practice, for their guidance and support during the joint Shared Prosperity” in Beijing during March 17–18, 2016, orga- research process as well as research contributions from the Asian nized by the RCGVC and the China Development Research Foun- Development Bank, Inter-American Development Bank, United dation. Drafts of the eight chapters of the report were presented Nations Industrial Development Organization, and International and discussed at the second Authors’ Conference in Washing- Center for Trade and Sustainable Development. ton, DC, during November 28–29, 2016, organized by the World The editors are grateful to a team at Communications Develop- Bank. The editors thank the authors of background papers and ment Incorporated, led by Bruce Ross-Larson and including Jon- individual chapters and the discussants and participants in the athan Aspin, Joe Caponio, Mike Crumplar, Meta de Coquereau- two conferences for insightful comments and suggestions that mont, Debra Naylor, Christopher Trott, John Wagley, and Elaine helped draft and improve the chapters (see appendix 1 for the Wilson, who edited, designed, and produced the report. programs) and participants at the two conferences. The RCGVC would like to acknowledge the financial support Special thanks go to our external reviewers: Laura Alfaro (pro - from the Bill & Melinda Gates Foundation. fessor at Harvard Business School), Zhi Gang Tao (professor at Editors: David Dollar, Jose Guilherme Reis, and Zhi Wang. xiv • Measuring and Analyzing the Impact of GVCs on Economic Development Abbreviations and acronyms ADB Asian Development Bank ASEAN Association of Southeast Asian Nations CEPII Institute for Research on the International Economy EU European Union FDI Foreign direct investment G7 Group of Seven G20 Group of Twenty GDP Gross domestic product GVC Global value chain ICT Information and communication technology IDE–JETRO Institute of Developing Economies–Japan External Trade Organization NAFTA North America Free Trade Agreement OECD Organisation for Economic Co-operation and Development UN United Nations WTO World Trade Organization Executive summary DAVID DOLLAR G lobal value chains (GVCs) break up the production production process. Data are now available on the value added process so at different steps can be carried out in traded among major economies during 1995–2014. This first different countries. Many smart phones and televi- Global Value Chain Development Report draws on the expand- sions, for example, are designed in the United States ing research that uses data on the value added in trade. Its main or Japan. They have sophisticated inputs, such as semiconduc- objective is to reveal the changing nature of international trade tors and processors, which are produced in the Republic of that can be seen only by analyzing it in terms of value added and Korea or Chinese Taipei. And they are assembled in China. They value chains. are then marketed and receive after-sale servicing in Europe and A natural place to start is with the theoretical foundation of the United States. These complex global production arrange - GVCs (chapter 1). Why do we care about analyzing GVCs? For ments have transformed the nature of trade. But their complexity two main reasons. First, GVCs provide new opportunities for has also created difficulties in understanding trade and in formu- developing countries to increase their participation in global lating policies that allow firms and governments to capitalize on trade and to diversify their exports. Without GVCs, a develop - GVCs and to mitigate negative side effects. ing country would have to be able to produce a complete prod- Today’s official statistical information systems, designed to uct in order to expand into a new line of business. Historically, measure economic activity in a pre-GVC world, have struggled developing countries have tended to export unprocessed raw to keep pace with these changes. Conventional measures of materials, suggesting that the jump to producing finished goods trade, important though they remain, measure the gross value was difficult. Today, because of the opportunities for integrating of transactions between partners and so are not able to reveal in specific parts of the value chain, many developing countries how foreign producers, upstream in the value chain, are con- are exporting primarily manufactured goods. (In the spirit of this nected to final consumers at the end of the value chain. For report, it would be more accurate to say that they export primar- example, conventional statistics suggest that the Republic of ily manufacturing value added.) The development of GVCs has Korea exports a lot to China. In fact, much of this trade con - no doubt contributed to this diversification of exports. Still, only sists of components that are ultimately destined for the Euro- a small number of developing economies are deeply involved in pean and U.S. markets. So it would be more accurate to say for GVCs, China being the best example. So how can developing these products that Korea exports a lot to advanced consumer countries deepen their involvement in GVCs? And how can they markets. move up the value chain? The importance of the GVC phenomenon has stimulated A second reason to analyze valued added in trade and GVCs researchers to develop statistics and analysis based on the is that data on the gross value of trade can be misleading. This value added in trade. The GVC phenomenon also demands report highlights how shifting the analysis to value added radi- that researchers analyze the discrete tasks or phases in the cally changes the picture. 1 2 • Measuring and Analyzing the Impact of GVCs on Economic Development Characterizing global value chains FIGURE 1 Global value chains were expanding, until the financial crisis To capture the variation in the extent of offshoring and production Percent sharing by sector and country, the report develops a GVC index 86 Asian Dot-com Global system that includes three indexes to characterize the nature of financial crisis bust financial crisis GVCs: a production length index for the average number of pro - 84 duction stages and complexity of the value chain, a participation index for the intensity of a country-­sector’s engagement in GVCs, 82 and a position index for the location of a country-­ sector pair on Pure domestic a GVC­ —­that is, the relative distance of a particular production production 80 stage to both ends of a GVC (chapter 2). All these indexes are built through a system of global input-output tables that under- 10 pin all trade in value added data and that provide the basis for Simple GVC decomposing gross domestic product (GDP) into broad catego- ries of activity based on forward industrial linkages. Traditional trade production Pure domestic value-added production activities are those 5 Complex GVC that are completely produced and consumed within one country, such as a haircut. When these goods or services are exported to another country, that transaction conforms to the classical idea of trade, with production occurring completely in 0 one country and consumption in another. República Bolivariana 1995 2000 2005 2010 2014 de Venezuela exporting oil to the United States is an example. Value added created by production across national borders Source: University of International Business and Economics global value chain (embodied in intermediate trade flows) are GVC activities, which indexes derived from the 2016 World Input-Output Database. can be further decomposed into simple and complex cross- border production-sharing activities based on the number of border crossings. In simple GVCs value added crosses national (simple and complex combined) accounted for 60–67% of global borders only once during the production process, with no indi- trade in value-added terms, reflecting the importance of the rect exports via third countries or re-exports or re-imports. In GVC phenomenon. complex GVCs value added crosses national borders at least Further insight into the changing pattern of value-added twice (Wang, Wei, and Zhu 2017a). Using the GVC index system, creation can be gained by looking at the nominal growth of the report characterizes cross-border production-sharing pat- value added separately for purely domestic production, tradi- terns and GVC activities for 35 sectors and more than 40 coun - tional trade, and GVCs between 1995 and 2014 (figure 2). From tries over 20 years based on the World Input-Output Database 1996 through 2007 value added in complex GVCs grew faster (Timmer and others 2015). than other components of GDP (so its share was rising). This was especially pronounced in 2002–08, the heyday of GVC expan - sion. The acceleration of GVC expansion occurred shortly after Global value chains were expanding until the China joined the World Trade Organization (WTO), and China’s global financial crisis growing participation in GVCs is probably one factor at work here. During 2002–08 not only was the share of GVCs rising, It will come as no surprise that, in general, GVC production has but the rate of nominal value-added growth was also very high been increasing during the modern era of globalization. Most in all parts of value added because of rapid real growth, mod- value added is still domestically produced and consumed, but erate inflation, and appreciation of most currencies against the the share of this part of GDP declined markedly until the global U.S. dollar. The period 2009–11 then represents the crisis and financial crisis, shrinking from 85% of global value added in 1995 initial rebound. What is striking since 2011, however, is how to less than 80% in 2008 (figure 1). Different types of trade all slowing rates of GDP growth appear to have had a dispropor- expanded their shares during this period, but the most rapid tionate impact on GVC channels, particularly for more complex increase was for complex GVCs. The 2008–09 global finan- GVCs, which were the key driver of growth in preceding eco- cial crisis was naturally a disruption, but trade rebounded fairly nomic cycles. quickly. What is surprising is the lack of further expansion in the The decomposition also allows for the characterization of dif- shares of either traditional trade or GVC trade since 2011. The ferent stages along GVCs: at each stage value added is counted share of purely domestic value added has increased slightly since as the gross output of some industry. This report also draws new 2008. It is too soon to know for sure, but it may be that the pro- insights on the changing pattern of GVCs through a new type cess of deeper integration associated with GVCs has stalled or of “smile curve” (chapter 2). The smile curve is best explained even started to reverse. Still, throughout this period, GVC trade through an example, as in figure 3. For China’s exports of Executive summary • 3 FIGURE 2 Nominal growth rates of different value-added creation activities, 1996–2014 Percent 30 Asian Dot-com Global Slow financial bust financial economic crisis crisis recovery 20 10 0 –10 Pure domestic production Traditional trade production –20 Simple GVC Complex GVC –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Source: University of International Business and Economics global value chain indexes derived from the 2016 World Input-Output Database. electrical and optical equipment in 1995 and again in 2009, the remained very low on the smile curve. But the bubble that data points are represented by circles indicating country-sector shows the total value added produced by CHN14 expanded pairs that contribute in production, with the letters denoting the about 10-fold. China may have held a low position in the value country and the number the industry. The size of the circles rep- chain throughout this period, but it brought a huge number resents the absolute value added gained by joining the value of workers from its impoverished countryside to work in the chain (in millions of constant U.S. dollars). An estimated curve is related factories. fitted through these points, and the shape of the curve is that Figure 3 captures anxieties felt by both rich and poor coun- of a “U” or a “smile.” The vertical axis plots labor compensa- tries in contemplating contemporary trade. Rich-country elec- tion per hour in the country-sector, indicating high- versus low-­ torates worry that manufacturing is being hollowed out­ —­that value-­added activities. The horizontal axis plots the total forward is, that semiskilled production jobs have moved to developing linkage–based production length between global consumers countries or, to the extent that such jobs still remain in advanced of electrical and optical equipment and a specific participating economies, have suffered downward pressure on wages. Poor industry in the corresponding GVC. countries worry that they are trapped in low-value-added activ- The logic of the smile shape is as follows. Research and ities and are locked out of the higher value-added activities in design activities for critical components of the electrical and design, key technological inputs, and marketing. optical equipment occur early in the production process (left side of the figure). These knowledge activities tend to be high-­ value-added activities in GVCs and tend to be carried out in Within-country distributional impacts more advanced economies. For example, in the 1995 curve Japan and the United States (JPN28 and USA28) are in the The changes in technology and global trade highlighted by upper left corner, reflecting the high-value-added contributions the smile curves can also be seen in statistics on factor use and from these two countries’ financial services sector. The Chinese income distribution in developed and developing countries. industry that manufactures the good, Chinese electrical and Here, this is shown using the information and communication optical (CHN14), is at the bottom point of the curve, reflecting technology industry in the United States and China as examples assembly activity at low wages. The activities closest to the (figures 4 and 5). consumer are marketing, logistics, and after-product servicing. For the United States the left panel in figure 4 tracks the evo - These market knowledge industries are also high value added, lution of factor return shares (left scale) and labor productivity as shown by the upward-sloping part of the smile curve on the (right scale). Labor’s share in returns rose from 60% to more than right. And they tend to be carried out in advanced economies, 70%, highlighting the important role of human capital in this high- where the mass consumption products are eventually purchased tech industry. Over 15 years the share of medium- and low-skilled by households. workers in the total number of hours worked declined (middle The comparison of the same country-sector export in 1995 panel), while the share of high-skilled workers (college educated and 2009 reveals that the smile curve for this product has and above) increased sharply, from about a third to a half of total deepened. Compensation in the USA28 industry rose from hours worked. The distribution of compensation across skill levels about $25 an hour to $60 an hour, whereas Chinese wages (right panel) reveals that proportionally more of the benefit went 4 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3 The estimated smile curve for China’s exports of electrical and optical equipment deepened between 1995 and 2009 Compensation per hour ($) Source: Meng, Ye, and Wei 2017. Note: See annex 2.2 in chapter 2 for a key to country abbreviations and sector codes. to high-skilled workers; compensation was flat for low-skilled Now consider the analogous analysis for China’s information workers and increased only slightly for medium-skilled workers. and communication technology industry (see figure 5). The first These shifts are consistent with the overall transformation of thing to notice is that labor productivity growth was phenomenal, the information and communication technology industry in the increasing some six times over 15 years (right scale, left panel). United States over the period, which went from producing goods During the period, labor’s share dropped from more than 40% to to primarily designing and providing support services. about 30%, while capital’s share rose from less than 60% to nearly Executive summary • 5 FIGURE 4 Efficiency and factor income distribution in the information and communication industry in the United States, 1995–2009 2009 $ Percent of GDP (thousands) Percent of total hours worked Compensation ($ per hour) 100 300 100 80 Capital High-skilled labor High-skilled labor compensation 75 250 75 60 Labor compensation Medium-skilled labor Medium-skilled labor 50 200 50 40 Labor productivity 25 150 25 20 Low-skilled labor Low-skilled labor 0 100 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Source: Meng, Ye, and Wei 2017. FIGURE 5 Efficiency and factor income distribution in the information and communication industry in China, 1995–2009 2009 renminbi Percent of GDP (thousands) Percent of total hours worked Compensation (renminbi per hour) 100 700 100 High-skilled labor 20 Medium-skilled labor High-skilled labor 600 Medium-skilled labor 75 75 15 500 400 Capital 50 compensation 50 10 300 Low-skilled labor Low-skilled labor 200 25 25 5 Labor productivity 100 Labor compensation 0 0 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Source: Meng, Ye, and Wei 2017. 70% (left scale). Clearly, capital was able to reap much of the ben- workers (with high school degrees) went up about 80%. Even low- efit of the productivity gain. It should be emphasized that the skilled workers saw their pay rise more than 50%. The distribu- gain accrued to the capital deployed in China, and that included tion of hours worked by different skill classes in China is basically multinational corporations engaged in GVCs. Other research has a mirror image of that for the United States. The overwhelming shown that most of the value added in China’s exports has come share of labor input in China’s information and communication from the domestic private sector, and multinational corporations technology industry over the period was low- and medium-­ produce a substantial amount as well. Thus, much of the benefit skilled, though their shares did decline somewhat, from more from the expansion of Chinese GVCs has gone to private owners than 95% of hours to 90% (middle panel of figure 5). High-skilled of capital. But there have also been significant wage increases for input was very small, about 5% of hours by the end of the period. all workers­—­albeit starting from a very low base (right panel). The These distributional findings shed some light on the grow- big proportional gain went to skilled labor, whose compensation ing protectionist sentiment in some advanced economies­ —­ and nearly doubled (right panel). Compensation for medium-skilled on the fact that globalization remains popular in developing 6 • Measuring and Analyzing the Impact of GVCs on Economic Development countries that are deeply involved in GVCs, such as China, Countries with high labor productivity will have higher wages and Mexico, and Viet  Nam. These findings do not permit drawing still be low-cost producers. The countries more deeply involved strong causal conclusions, but the analysis is consistent with a in GVCs (identified in orange in figure 7) all stand out as having story in which the benefits from GVC-related trade have been low unit labor costs, but not necessarily low wages. In contrast, in distributed highly unevenly. For the United States the big win- each time period there is a circle of countries that have very low ners appear to be high-skilled workers and multinational cor- wages but high unit labor costs. These are mostly African econo- porations. GVCs have enabled them to benefit from enormous mies. Other costs in the production process offset any potential productivity gains in developing countries such as China. Ordi- advantage from low wages. nary workers in the United States have not seen much (if any) One of the most important impediments for developing benefit. In China ordinary workers have benefited. Even at the countries is trade costs, examined in chapter 4. Today, nontariff beginning of the period factory wages in China were far ahead trade costs (freight, insurance, and other cross-border-related of rural incomes. And those wages doubled over 15 years. The fees) tend to be much larger than any remaining import tariffs as wage gains are a driving factor behind the impressive decline products travel through production stages. Those trade costs, of absolute poverty in China. Relatively speaking, however, the which vary by country and sector, have a monetary dimension big benefits in China accrued to the small number of high-skilled (for example, transportation, insurance, and other fees) but also a workers and to the owners of capital, including foreign investors. more intangible dimension that encompasses information costs, nonmonetary barriers (regulation, licensing, and so on), insecure contracts, and weak trade governance leading to uncertainty. Developing country participation in global These impediments to trade can be expressed as ad valorem value chains tariff equivalents and are generally much higher than tariffs. In sectors with complex value chains, such as motor vehicles, com- Witnessing this rise of GVCs, stakeholders in developing coun- puters, and machinery, trade costs are more than four times tries typically want to see their country more involved in value higher than tariffs. In traditional traded goods, such as agricul- chains and moving to higher value-added activities within the tural products, minerals, and wood, these trade costs tend to be chains over time. GVC research can help identify factors asso- less of an impediment. ciated with integration into GVCs, such as the related issues of So while weak transportation links, inefficient customs clear- developing country involvement in GVCs, trade costs, and the ance, bureaucracy, and red tape all tend to impede trade, their middle-income trap (chapters 3, 4, and 5). effects are most pernicious in sectors requiring that parts move For the involvement of developing countries in GVCs, geog- back and forth across borders. The costs of impediments cas- raphy clearly matters. The world seems to have three intercon- cade. Countries with very high trade costs will not be able to nected production hubs for the extensive trade in parts and participate in GVCs, and any exports are likely to be traditional components (figure 6): one centered on the United States, one goods, often primary products. Developing countries try to on Asia (China, Japan, Republic of Korea), and one on Europe address this problem by establishing special export process- (especially Germany). Figure 6 shows the important bilateral ing zones, which have superior logistics and expedited customs flows of parts and components, with the countries that are most clearance (as well as through duty drawbacks on any remaining deeply involved highlighted in red. China aside, developing import tariffs). The problem with this second-best approach is countries are generally on the periphery and tend to trade with that it limits participation in GVCs to the small number of firms in the hub that is geographically closest. Many developing regions the export processing zones, while other domestic firms, espe - are barely involved at all. Most African countries are far from cially small ones that might become parts suppliers, are left to existing hubs. And within developing countries, it is large firms stumble in a world with high transaction costs. A better approach that tend to be involved in global production networks. In Latin is to improve trade facilitation for all firms in the economy. America, for instance, small firms rarely trade outside the region. China provides some interesting lessons. China is known for having started its economic reform with four special economic zones that fit the model of export processing zones, with favored Unit labor costs and trade costs infrastructure and customs clearance. What is less known is that within a short time China had expanded these benefits to more How to explain the differential participation of developing coun- than 30 cities nationwide. Competition among the cities has tries in GVCs? Low wages are often thought to be an important enabled quite a few of them to emerge as locations with low factor. But low wages exist across developing countries, yet trade costs and deep participation in GVCs. Research into the only a few locations are involved in GVCs. Low unit labor costs value added of trade has shown that the majority of the domes- (the ratio of average wages to per capita GDP) turn out to be tic value added in China’s exports comes from private domes- much more important than low wages. Figure 7, which plots unit tic firms. Foreign firms are often the processing exporters from labor costs against wages in 2000 and 2010 for a large number China, but the successful expansion of value chains to domes- of developing countries, show no positive relationship between tic firms within China has resulted in most of the value added them because labor productivity varies so much across countries. coming from the domestic private sector. Executive summary • 7 FIGURE 6 Trade in components shows three interrelated production hubs Czech Republic Austria Hungary Sweden Poland France Ireland Netherlands Italy Russian Federation United Kingdom Germany Canada Belgium Turkey Switzerland Mexico United States Spain Korea, Rep. China Brazil Chinese Taipei Portugal Japan Singapore India Australia Hong Kong, China Israel South Africa Thailand Malaysia Indonesia Argentina Chile Viet Nam Source: Diakantoni and others 2017, based on the UN Comtrade database (https://comtrade.un.org). Note: Includes the 61 economies in the Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database and their most important bilateral gross trade flows. Further evidence on the importance of reducing transactions on to consumers. A clear relationship emerges between better costs comes from the World Bank’s Logistics Performance Index, logistics performance and deeper involvement in GVCs when which captures how well infrastructure and bureaucracy work the Logistics Performance Index is plotted against a centrality together to move goods through the production process and indicator of each country’s role in GVCs (an indicator that ranks 8 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 7 Developing countries deeply involved in global value chains have low unit labor costs but not low wages, 2000 and 2010 Unit labor costs (ratio of average wages to GDP per capita) 2000 8 6 4 Morocco South Africa Tunisia 2 Latvia Colombia Brazil Malaysia Poland Mexico Egypt China Czech Republic 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 2010 8 6 4 Morocco 2 South Africa Egypt China Tunisia Brazil Latvia Mexico Czech Republic Colombia Malaysia Poland 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 Average wages ($) Source: Ceglowski and others 2015. Note: Unit labor costs are the ratio of average wages to per capita GDP. a country or industry’s centrality to GVCs taking into account Global value chains and the middle-income trap direct and indirect trade flows to and from trading partners in the global production network; figure 8). The link is not that tight One of the most hotly debated issues in development is the (R 2 = 0.29), however, indicating that other factors are at work as “middle-income trap” (chapter 5). This is the idea that it is rel- well. But it is interesting that there are no countries in the lower- atively easy to grow from low income to middle income, by imi- right quadrant: no countries with poor logistics performance are tating successful countries and expanding factors of production central to GVCs. For countries that want to get more involved in (labor force growth and investment). But it is harder to move GVCs, trade facilitation and infrastructure are obvious places to from middle income to high income, which in general is based start. more on innovation and creativity than on extensive growth. Executive summary • 9 FIGURE 8 Relationship between the Logistics Performance Index and a centrality measure of country involvement in global value chains Logistics Performance Index 4.3 4.1 3.9 R2 = 0.29 3.7 3.5 3.3 3.1 2.9 2.7 2.5 0.0 0.5 1.0 1.5 2.0 2.5 3.0 PageRank centrality indicator Source: Diakantoni and others 2017. Note: The centrality indicator ranks a country’s centrality to global value chains, taking into account direct and indirect trade flows to and from trading partners in the global production network. It turns out that there is mixed empirical evidence for a that ratio had barely changed (left panel of figure 9). Most of the middle-income trap. Chapter 5 finds substantial upward mobility goods trade was manufactures, with the remainder being agri- between 2000 and 2015, particularly for middle-income coun- cultural and mining products. Economists refer to many services tries, with 79 of 133 countries that were low or middle income in as “nontradables,” meaning that they cannot be directly traded 2000 improving their income status and none declining. internationally. Haircuts and dry cleaning are common examples. While there is only weak evidence for a generalized growth Higher end services such as health care and legal advice are also slowdown in middle-income countries, there is still the concern hard to directly trade internationally. That is starting to change that in any period some countries are moving ahead rapidly with some remote services trade, but statistically the share is while others are stagnating or moving ahead less rapidly. Fur- very small. thermore, problems of the structural transformation of industries However, analysis of value added shows that the share of are quite specific to middle-income countries, and this more lim- services in trade nearly doubled between 1980 and 2008 (right ited understanding of a middle-income trap is usefully explored panel of figure 9). Another way of looking at this statistic is that in the chapter. One clear empirical regularity is that upwardly much of the value in manufactured goods comes from inputs mobile countries have considerably more involvement in GVCs of services industries. The reasons for these developments are than do languishing countries. Care is required in interpreting variants of the older arguments for why the share of services in this kind of association, but it is consistent with the notion that GDP tends to grow: the splintering or outsourcing of services GVCs have given developing countries new opportunities to par- activities from manufacturing firms; the growing importance in ticipate in a global division of labor. For the countries that have a GVC world of connecting services like telecommunications been able to respond effectively to the opportunities, that has and transport; the growing services component in sophisticated in turn led to faster productivity growth and economic advance. manufacturing goods, such as software in cars; and the increase in relative prices of services tasks because manufacturing tasks are easier to offshore to lower cost locations. Services and trade restrictiveness This tendency for value-added exports of services to be greater than the direct export of services is true in all major econ- A key perception of international trade that changes when value omies, though the share varies considerably. Figure 10 ranks added replaces gross value in the analysis concerns the relative countries in the services share of value added exported and role of goods and services (chapter 6). In 1980 the split between in the services share of gross exports, which is smaller in every trade in goods and direct trade in services was 80:20. By 2008 case. In general, developed countries have services shares in 10 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 9 The share of services is higher and has increased more sharply in trade in value added than in trade in gross terms, 1980, 1995, and 2009 Gross exports of goods and services Value-added exports of goods and services Share of total world gross exports (%) Share of total world value-added exports (%) 100 Goods 100 Goods Services Services 75 75 50 50 25 25 0 0 1980 1995 2009 1980 1995 2009 Source: Authors’ calculations based on Johnson and Noguera 2016. FIGURE 10 The share of services in exports is higher for developed countries, 2011 Percent 100 Domestic value added Foreign value added OECD average 75 Total services export 50 25 0 Mexico Norway Korea, Rep. Romania China Russian Federation Canada Slovak Rep. Czech Rep. Bulgaria Japan Germany Hungary Poland Slovenia Turkey Finland Italy United States Austria Portugal Iceland Spain Sweden Estonia Lithuania Israel Latvia Switzerland France United Kingdom Greece Denmark Belgium Croatia Ireland Netherlands Malta Cyprus Luxembourg Source: Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database 2015. Executive summary • 11 value-added exports above 50%. About 55% of the value added would feed into more-­ competitive and more-productive manu- exported from the United States comes from services sectors. facturing sectors. Figure 11 shows measure of import protection The shares are even higher for European economies. For the in key services sectors for different regions. As the benchmark, Netherlands, well known as an exporter of agricultural products Organisation for Economic Co-operation and Development and manufactures, services account for nearly 70% of the value (OECD) countries are very open to imports of financial, tele - of its gross exports. communications, and retailing services and moderately open to Emerging market economies that are major exporters of man- trade in transportation services. Professional services, such as ufactured products have somewhat lower but still surprisingly law, medicine, and architecture, on the other hand, remain rel- high services shares. For example, China, Mexico, and Viet Nam atively protected. For many services it is difficult to trade inter- have very little direct export of services, but in value added nationally without investment in establishing a local presence. terms about 40% of their exports come from services. They can OECD economies are also very open to direct investment in expect that share to rise as they develop further and move up services sectors, contributing to their competitive character and the value chain. high-productivity outcomes. While the links between manufacturing and services are Developing countries have embraced import openness for deepening, many developing countries continue to carry out manufactured products, especially machinery and parts that dualistic policies between manufacturing and services. Pro- enable them to participate in the international division of labor. tection tends to be stronger against imports of services, even But they continue to protect imports of services (see figure 11). though more-open policies would help countries develop more-­ Countries in East Asia and Pacific have much higher levels of pro - competitive and more-productive services sectors, which in turn tection than OECD countries. Countries in Latin America and FIGURE 11 Developing countries maintain high restrictions on services trade Professional services Transportation Europe and Central Asia Retail Telecommunications Financial OECD Latin America and Caribbean Sub-Saharan Africa East Asia and Pacific Middle East and North Africa South Asia Gulf Cooperation Council 0 20 40 60 80 Services Trade Restrictions Index Source: Borchert, Gootiiz, and Mattoo 2014. Note: This figure compares the restrictiveness of services trade policy across countries based on the World Bank Services Trade Restrictions Index, which ranges from 0 (completely open) to 100 (completely closed). 12 • Measuring and Analyzing the Impact of GVCs on Economic Development Central Asia are modestly more open but still less open than policy, participation in deep preferential trade agreements turns OECD countries. Countries in Africa and South Asia, home to out to be an effective way to expand involvement in GVCs. The most of the world’s remaining extreme poor, are generally the new areas covered in these agreements facilitate the operations most closed. For developing countries wishing to participate of complex production structures that span multiple borders. more in GVCs and to move up the value chain, one obvious mea- Participating in deep preferential trade agreements increases a sure is to open services to import competition and direct foreign country’s trade in parts and components, an important measure investment. Improved access to finance, communications, trans- of GVC activity. port, and other services, through reform in general foreign direct While strengthening institutions and reducing trade costs, investment in particular, enhances manufacturing firms’ produc- perhaps through deep preferential trade agreements, are effect­ tivity and other aspects of the performance of downstream firms. ive routes for developing countries to become more involved in GVCs, some sobering research shows that in addition to one’s own institutions, the quality of neighboring countries’ institu- Institutions and deep trade agreements tions matters as well. In contract-intensive sectors (such as those with complex value chains), countries with “bad” neighbors have Another way to think about products that have complex value fewer exports, even after controlling for the country’s own insti- chains is that they are contract-intensive goods. That is, they tutions. This result implies that deep agreements would be more often involve many exchanges among different firms, each facing effective if a group of neighboring economies all signed up for some risk of contract nonperformance by others in the chain. the same agreement. In the case of the Trans-Pacific Partnership, GVC research shows that, other things equal, countries with for example, several countries in the Association of Southeast better institutions such as stronger property rights and rule of Asian Nations (ASEAN), such as Singapore and Viet Nam, are law participate more in GVCs (chapter 7). Research for this report partners to the agreement, as are several Latin American coun- found a similar result within China across a large number of cities. tries (such as Chile, Mexico, and Peru). The benefits would be Cities with better measures of contract enforcement, faster cus- greater if all of ASEAN countries and the Pacific countries in Latin toms clearance, and deeper financial systems participated more America signed on. In the wake of the 2016 U.S. presidential in GVCs. election, U.S. President Donald Trump pulled the United States The idea of improving institutions and lowering trade costs out of the agreement, but the remaining 11 countries are dis- across the board through better infrastructure, control of corrup- cussing whether to proceed without the United States. tion, reduction of red tape, and zero tariffs on imported inputs For developing countries the agenda of reform needed to (including services) is clear. But developing country leaders nat- participate more deeply in GVCs is challenging. Moreover, access urally wonder how to pursue this agenda. It turns out that one to finance remains an issue in less advanced economies that are effective route is through “deep” trade agreements, agreements prone to market and public governance failures. While joining that go beyond simple tariff cutting and involve legal commit- GVCs improves the prospects of attracting private foreign direct ments on laws and regulations (chapter 8). The different rounds investment, the poorest countries may still require substantial of agreements within the framework of the WTO have involved additional financing, if only to improve the public transport and primarily reducing import tariffs, and these have had the most telecommunication infrastructure as well as trade facilitation. effect on trade in manufactures. It has proved more difficult to go In this respect, the 2015 Addis Ababa Action Agenda provides beyond tariff cutting in the WTO. Although significant progress a new global financing framework to mobilize and deliver the has been made in recent years with the WTO Trade Facilitation resources, technology, and partnerships needed to improve Agreement, the abolition of agricultural export subsidies, and many of the structural and institutional conditions required for several other agreements, progress has stalled within the WTO fostering export-oriented industrial activities (UN 2016). on new global agreements. Preferential trade agreements­ —­ in which a group of like-minded countries negotiate agreements on policy areas that build on WTO commitments­ —­have prolif- Toward more inclusive globalization erated. In practice, the most important areas concern services trade, investment, competition policy, and intellectual property This report provides some insight into how GVCs are advancing rights protection. the development process and how they are creating distribu- Between 1958 and 2014, 279 preferential trade agreements tional conflict, especially in advanced countries. The rapid pro - were notified to the WTO. This report rates the “depth” of each ductivity growth within GVCs shows that they are an efficient agreement based on the number and share of legally enforceable form of production. They have enabled developing countries provisions. The North American Free-Trade Agreement among in particular to move into new activities and rapidly raise their Canada, Mexico, and the United States is a deep agreement, as productivity. To be sustainable, however, globalization needs to is the Trans-Pacific Partnership, which has been negotiated but become more inclusive in at least three dimensions. not yet ratified or implemented among 12 Asia–Pacific econo - First, in developing countries deeply involved in GVCs, virtu- mies. Because deep integration often involves leveling the play- ally the entire population benefits from the expanded trade and ing field for investment, intellectual property, and competition faster growth, though not all to the same extent. In developed Executive summary • 13 countries, by contrast, the benefits of expanded international East Asia, in particular, has taken advantage of the opportuni- trade and investment are highly concentrated among the very ties provided by globalization. But increasingly, the remaining skilled in the workforce and the owners of capital. Both groups extreme poor are concentrated in South Asia and Africa. Coun- are already high up in the income distribution, and globalization tries in these regions can help themselves through further trade increases their share of the pie. and investment liberalization, especially trade facilitation that There is no simple agenda to spread the benefits more improves infrastructure and import or export processes so that widely. A protectionist sentiment is arising in developed coun- goods can move easily around the world. One of the interesting tries. Historical evidence suggests that cutting themselves off trends identified in GVC research is that more and more of the from the global market through import restrictions will almost value added traded in the world comes from services sectors. certainly backfire. That is likely to lead to slower global growth Opening services sectors to foreign trade and investment is a and poor results all around. Evidence has shown that effective smart strategy for deepening integration. Participating in deep responses may include active labor market policies to provide trade and investment agreements can advance that agenda, and training and retraining so that workers have the skills demanded such agreements will be most powerful if they involve several in the market, a stronger safety net of minimum income support, neighboring countries. and support to communities hit hard by changes in production A third dimension of inclusion concerns small firms and the arising from trade or technological change. Also important is informal sector. Most job creation in the world is in small and developing more detailed official national data that can inform medium-size firms, so GVC involvement by these firms is crucial policymakers. Considerable improvements have been made on for maximizing the positive impact from trade. Poor infrastruc- the data front in recent years, notably through trade in value ture, corruption, and red tape tends to hamstring smaller com- added–type measures. But with few exceptions these provide a panies more than larger ones since large firms can often finance wide-angled view, whereas what is increasingly needed is a more their own infrastructure and finds ways to operate in corrupt and granular view, at least a view that zooms in on workers, occupa- bureaucratic environments. Special export zones can be a way tions, and skills. for a developing country to begin to participate in GVCs, but for Second, while GVCs have enabled many developing coun- the benefits to spread throughout the economy, it is important tries to increase their participation in global trade and raise their that the zones are seen as stepping-stones to economywide productivity, too many countries and regions are still left out. improvements. 14 • Measuring and Analyzing the Impact of GVCs on Economic Development References Borchert, I., B. Gootiiz, and A. Mattoo. 2014. “Policy Barriers to Interna- Meng, B., M. Ye, and S.-J. Wei. 2017. “Value-added Gains and Job Oppor- tional Trade in Services: Evidence from a New Database.” The World tunities in Global Value Chains.” IDE Discussion Paper No.  668, IDE– Bank Economic Review 28 (1): 162–88. JETRO, Chiba City, Japan. Ceglowski, J., S. Golub, A. Mbaye, and V. Prasad. 2015. “Can Africa Com- OECD (Organisation for Economic Co-operation and Development) and pete with China in Manufacturing? The Role of Relative Unit Labor Added (TiVA) WTO (World Trade Organization). 2015. Trade in Value-­ Costs.” Working Paper 201504, Development Policy Research Unit, database. Paris: OECD. Available at: http://www.oecd.org/sti/ind University of Cape Town, Cape Town, South Africa. /measuringtradeinvalue-addedanoecd-wtojointinitiative.htm. Diakantoni, A., H. Escaith, M. Roberts, and T. Verbeet. 2017. “Accumulat- Timmer, M., R. Stehrer, and G. de Vries. 2015. “Occupations in Global ing Trade Costs and Competitiveness in Global Value Chains.” World Value Chains: Patterns of International Specialisation.” OECD Trade Trade Organization (WTO) Working Paper Economic Research and Sta- Policy Paper, OECD, Paris. tistics Division (ERSD) 2017–2, WTO, Geneva. UN (United Nations). 2016. Addis Ababa Action Agenda: Monitoring Com- EC (European Commission). 2013. World Input-Output Database. Brussels: mitments and Actions. New York: UN. EC. Available at: http://www.wiod.org/release13. ———. UN Commodity Trade Statistics Database (Comtrade). New York: ———. 2016. World Input-Output Database. Brussels: EC. Available at: UN. Available at: https://comtrade.un.org/. http://www.wiod.org/database/wiots16. Wang, Z., S.-J. Wei, X. Yu, and K. Zhu. 2017. “Measures of Participation Johnson, R., and G. Noguera. 2016. “A Portrait of Trade in Value Added in Global Value Chains and Global Business Cycles.” Working Paper over Four Decades.” NBER Working Paper No. 22974, NBER, Cam- 23222, National Bureau of Economic Research, Cambridge, MA. bridge, MA. CHAPTER 1 Analytical frameworks for global value chains: An overview SATOSHI INOMATA, IDE–JETRO I n a keynote speech at a seminar on global value chains of international trade theories. The second traces the develop- (GVCs), Richard Baldwin delivered wittily, with his mischie- ment of the GVC concept, with some reference to the evolution vous smile, a rather provocative statement: “The term ‘global of global production networks. The third introduces the main value chains’ doesn’t describe what we see today in the world theoretical achievement in GVC studies. The fourth summarizes economy”1 because: the challenges for a quantitative description of GVCs, particularly • The world economy is not global; it remains regionally seg- for the innovative use of multicountry input-­ output tables. The regated, such as Factory Asia, Factory Europe, and Factory fifth addresses pressing issues for advancing GVC research. The North America. last section presents some meta-­ methodological considerations • What matters is not value (added) but jobs, especially good on the development of GVC analyses.3 jobs. • Production systems are not configured as a linear sequence of production stages like chains but consist of complex networks The global value chain paradigm: New-New- of hubs and spokes. New Trade Theory? This is alarming. However, it is also true that many people now use the term “GVCs”­ — often inconsistently across contexts. Since David Ricardo established the foundation of international With that as the backdrop, this chapter cultivates some trade theory two centuries ago, mainstream thought, from common ground for approaching this new area of academic inter- Heckscher-Ohlin to Samuelson, has hinged on three classic est by tracing the development of relevant studies. This is not an premises (figure 1.1): encyclopedic literature survey; it focuses only on the strands of • Markets are perfectly competitive, and producers operate at research that explicitly consider vertical (supply–use) relations of constant returns to scale. cross-border production sharing and their impact on distributing • An industry consists of homogeneous producers. value among the parties­ —­which is at the heart of GVC studies.2 • Countries trade only final products­—­ traditionally phrased as The first section of the chapter considers why GVC studies are —­ Portuguese wine for English cloth­ and each product is made important from the viewpoint of their contribution to the history using the production factors of only the exporting country. The author would like to express his sincere gratitude to Laura Alfaro, Pol Antràs, Richard Baldwin, Rudolfs Bems, Juan Blyde, Gaaitzen de Vries, Gary Ger- effi, Robert Johnson, Robert Koopman, Manfred Lenzen, Kiyoyasu Tanaka, Zhigang Tao, Marcel Timmer, Zhi Wang, and Deborah Winkler for their highly valuable comments and suggestions. He is also grateful to the Economic Research Center of Graduate School of Economics, Nagoya University, for its support by offering him a position as a domestic visiting scholar from April 1 to September 30, 2015. 15 16 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 1.1 Genealogical map of analytical frameworks for global value chains (Neo) classical theory of international trade Ricardo, Heckscher-Ohlin - (Vanek), Samuelson Paradigm shift Constant returns Homogeneous Portuguese wine Strong association to scale producers for English cloth Weak association 1970 Intra-industry trade Gruben & Lloyd 1980 New Trade Theory Helpman & Krugman Fragmentation 1990 Jones & Kierzkowski Deardorff Global Increasing returns Impact of offshoring Firm-level microdata commodity chains to scale Trade in intermediates on domestic incomes Gereffi & Korzeniewicz Bernard & Jensen Feenstra & Hanson Feenstra & Hanson Campa & Goldberg 2000 Yeats Vertical specialization Global value chains Firm’s choice of Supply chain length Firm heterogeneity Unbundling Hummels, Ishii, & Yi Gereffi, Humphrey, & GVC governance Dietzenbacher, Sturgeon Antràs & Helpman Melitz Baldwin Romero, & Bosma Fally Value-added export Impact of task trade on Chen, Cheng, Fung, & Lau factor productivity Koopman, Wang, & Wei Contract theory Heterogeneous Grossman & Theory of firms producers Heterogeneous producers Rossi-Hansberg Heterogeneous production factors Trade in value added Timmer, Erumban, Johnson & Noguera Product-level empirics Firm-level microdata Los, Stehrer & deVries Dedrick, Kraemer, & Linden Tomiura Trade in tasks Xing & Detert Bernard, Jensen, Input-output table by Gross export Sturgeon, Nielsen, Linden, GVC sequentiality Redding, & Schott firm characteristics decomposition Gereffi, & Brown Antràs & Chor Koopman, Wang, & Wei Ma, Wang, & Zhu Source: Author’s drawing. The first premise was shaken in the 1970s and 1980s when an explanation for these observations, advancing in the quest a new school of thought, New Trade Theory, emerged. Its key for what was later called New-New Trade Theory. By assuming a feature, pioneered by Krugman (1979, 1980) and generalized by fixed cost of entering export activities, the model considers the Helpman and Krugman (1985), was the theoretical scope for con- mechanism of a firm’s endogenous selection on market entry or sidering production technology with increasing returns to scale exit and thereby provides a powerful explanation for the coexis- (paired with the love of variety), which underpins the analytical tence of heterogeneous firms within an industry.5 frameworks of international trade under imperfect competition. A third wave of reconstructing classical theory is now under way, The models provided a plausible explanation for the prevalence and the literature on GVCs is generally linked to this development of intra-industrial trade between countries with similar technol- strand. With the dramatic advance of transportation modes and ogy and resource endowments­ —­a phenomenon that cannot be information and communication technology, production processes explained by the orthodox notion of comparative advantage.4 can now be “sliced” into several production segments, each corre- The evolution of theoretical frameworks is generally driven sponding to a particular task­ —­such as design, parts procurement, by the need to fill a gap between a newly discovered stylized assembly, and distribution. These segments are relocated, often fact and the predictions of prevailing models. Just as the empir- across national borders, to the places where the tasks can be per- ical findings on intra-industry trade, notably those of Grubel and formed most efficiently. Thus the core subject of the literature today Lloyd (1975), were followed by New Trade Theory, so too was the is not only the movement of final products, as classical theories have second classic premise of homogeneous producers reconsidered focused on (under the third premise), but also the cross-national following evidence in the late 1990s. Bernard and Jensen’s (1995, transfer of tasks, or the value added generated by these tasks. 1999) detailed examination of firm-level microdata revealed sub - The main characteristic of the GVC paradigm is the vari- stantial heterogeneity in firm productivity between exporters ety of its intellectual origins. The initial theory of production and nonexporters in a given industry. Melitz (2003) pioneered fragmentation (Jones and Kierzkowski 1990) was followed by Analytical frameworks for global value chains: An overview • 17 increasing observations of trade in intermediate goods (Feens- various scientific subfields in different ways at different times. tra and Hanson 1996b; Campa and Goldberg 1997; Yeats 1998), The ideas only recently started to cross over academic borders, which brought about further elaboration of key concepts such and they continue to evolve along dynamic interactions of theo- as unbundling (Baldwin 2006) and trade in tasks (Grossman and ries and empirics. Rossi-Hansberg 2008a). In parallel, methodological frameworks also advanced in Unbundling economies: Baldwin’s historical perspective sociology. Drawing on analytical scopes of academic fields, from When the movement of goods, people, and ideas was not as business management to industrial organization theory, a com- frictionless as it is today, economic activities were organized prehensive study on the structure and mechanism of value dis- mostly within the boundaries of a small-scale community (figure tribution among countries led to the term “global value chains” 1.2).7 Farmers harvested wheat and milled flour for a bakery a (Gereffi, Humphrey, and Sturgeon 2005). few blocks away, and the baker baked loaves of bread for the The empirical aspect of GVC studies is newer. Earlier value- neighbors who walked into the shop every morning. Economic added analyses based on firms’ business records (Dedrick, Krae- self-sufficiency was achieved with the points of production and mer, and Linden 2008; Xing and Detert 2010) are now comple- consumption in close proximity. Extraterritorial business was mented by input-­ output analysis, in which various GVC metrics rare, except perhaps for the merchant voyages of a sailing ship or were devised using multicountry input-­ output databases, such the Silk Road caravans. And those cross-border trades dealt only as trade in value added (Johnson and Noguera 2012) and supply with a handful of luxury items such as spices and silk products, chain length (Dietzenbacher, Romero, and Bosma 2005; Fally 2011). sold at high prices to compensate for the risk incurred and the One of the key integrating forces was Antràs and Helpman time spent during the journey. (2004), who featured the legacies of both the New Trade Theory International trade began to develop at the beginning of the (increasing returns to scale) and the New-New Trade Theory (firm 19th century when steam engines rapidly improved land trans- heterogeneity) in a study based on the frameworks of contract port (by locomotives) and water transport (by steamships), trig- theory, while contract theory can be associated with sociolo- gering unprecedented expansion of trade activities beyond local gists’ approaches to GVCs. The properties of the model were communities. The economies of scale from mass logistics further carried over to Antràs and Chor (2013), who further incorporated lowered transportation costs. The point of consumption was the methodological progress in input-­ output economics. unbundled from the point of production, and goods travelled all The interdisciplinary characteristic of the GVC paradigm over the world in search of the most profitable markets. allows for large-scale research collaboration across the social sci- Paradoxically, the geographical unbundling of economies ences, as demonstrated in this report. Topics in the GVC litera- between production and consumption coincided with the ture, some of which are highly politically relevant, include:6 agglomeration of production activities in large-scale factories • Industrialization strategy (full-set versus GVC-driven industri- in industrial zones. Because of the increase in potential custom- alization). ers created by international trade, the mass production system • Labor issues (impact of globalization on employment and became an appropriate manufacturing mode at the time. The income distribution). key to high productivity in manufacturing is the division of labor, • Regional development (trickle-down effect through domestic as seen in Adam Smith’s classic example of pin-making,8 where production linkages). workers specialize in a particular task to raise their competencies • Innovation and technological spillovers (learning through through intensive learning of a specific routine. However, division GVC participation). of labor entails delicate coordination among the different stages • Economic crisis (propagation of external shocks on produc- because the variety of tasks must collectively produce a homo- tion and trade). geneous product. Accordingly, the different productive func- • Supply chain resilience (impact of natural or human-caused tions were brought together under the same roof (a factory) to disasters on supply chains). facilitate communication and create harmony among the various • Environmental protection (carbon footprints and global tasks. governance). The information technology revolution in the 1980s completely • Consumer protection (food safety and certification). changed this picture. With telexes, facsimiles, and the Internet­ —­ • Poverty alleviation (fair trade and corporate social responsibility). along with high-speed international communication networks­ —­it • Trade regimes (World Trade Organization and regional trade became cheaper and easier to coordinate production units in agreements). different locations. Sales forecasts and procurement schedules • National accounts (statistical bias of gross trade data). could be instantly delivered to production lines, and the elec- tronic profiles of minute product designs and specifications could be shared with and adjusted by every production site. Productive Concept development functions no longer had to be confined within proximate spaces. The technological unbundling of production activities has accel- The concept of GVCs did not follow a linear development path. erated, with some segments relocated across borders to exploit The basic images of the term were conceived and fostered in the cost differentials of production factors in various countries. 18 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 1.2 Three cascading constraints of globalization n io e at ce ad ic fa s un tr - st -to st ds st m co om co ace co oo s s G C F Pre-globalized High High High world Steam revolution Stage A Stage B First High High unbundling Stage C Low Information and communication technology revolution Second High unbundling Stage A Stage B Customs Stage C Low Low Source: Adapted from Baldwin 2013. Vertical integration activities involves nontrivial administrative and bureaucratic Richard Baldwin’s unbundling concept captures one important costs. Accordingly, the governance schemes are chosen to aspect of the dynamics of the world economy. But there is minimize the production inefficiencies attributed to a trading another critical dimension of the analytical perspective for the relationship by weighing the transaction costs of spot-market development of GVCs. dealings against the bureaucratic costs of unified hierarchical In the beginning of the 20th century Henry Ford devised and organizations (firms).10 implemented a business model that aimed to integrate various From the viewpoint of transaction cost economics the costs of segments (functions) of a production process under a single concern include not only the direct costs of writing, monitoring, capital and management umbrella through the acquisition of a and enforcing contracts, but also the ex post performance inef- variety of companies. The model, later known as a vertical inte- ficiencies caused by contractual hazards within the relationship. gration strategy, became a modus operandi in the era of mass One of the basic tenets of transaction cost economics is that con- production.9 tracts are incomplete­ in that the terms of exchange between —­ Early studies of vertical integration focused on market imper- the parties cannot be disciplined ex ante because of information fections. A firm integrates other entities to redress pre-­existing asymmetry.11 When the parties are locked in to the transaction, market power distortions, such as double marginalization, the incompleteness of contracts evokes contractual hazards of free-riding, or entry foreclosure (Tirole 1989). various types, yet vertical integration pre-empts these hazards Another strand of thought considers the preclusion of trans- by internalizing ex post quasi-rents into the unified objective action costs as a main motive for vertical integration, where inter- function of the integrated firm. So vertical integration becomes nalizing production activities is a measure to avoid the potential a preferred mode of organizing value chains when the benefit of costs of establishing formal business relations at arm’s length. attenuating the opportunistic behavior of parties within the rela- Given these benefits of integration, why then do some firms tionship outweighs the cost of inefficiently allocating resources not choose to integrate? Because the internal arrangement of associated with bureaucratic arrangements (Joskow 2003). Analytical frameworks for global value chains: An overview • 19 And today vertical integration in the multicountry dimen- FIGURE 1.3 Modes of organizing value chains sion refers to the emergence of business entities called multi- national corporations. Foreign direct investment by multinational National corporations is the main driver of global production networks, boundaries decisively influencing the distribution of value added across countries.12 Domestic Foreign direct VERTICAL INTEGRATION Accordingly, there are four modes of organizing value chains, in-house investment along the axes of whether the task is done in-house or out- procurement sourced and of whether it is carried out domestically or across national borders (figure 1.3). Firm boundaries Value chains and global value chains The term “value chains” was conceived in business management Domestic Arm’s length studies. Porter (1985) tailored the concept as a basic framework outsourcing offshoring for developing a corporate strategy to promote firm competi- tiveness by directing attention to the entire system of activities involved in producing and consuming a product. A corporate UNBUNDLING entity is first decomposed into a set of business activities with individual functions that constitute analytical units for diagnos- ing the firm’s competitive advantage. When a firm has a relatively Source: Author’s drawing, based on Kimura and Ando 2005. atomized organizational structure, the task of each unit (business —­ activity)­ such as product design, materials procurement, mar- keting, and distribution­ —­ tends to be defined in a way to pursue answering these questions, GVC studies pay attention to the forms the individual objective of that particular unit, which may or of transactions, codified or otherwise, between stakeholders. This may not conflict with the objective of other units. However, in is because the way transactions are made reflects the structure of the value chain perspective all activities should be collectively power relations between the parties, which ultimately determines organized to ensure the optimal functioning of the corporate the scope and magnitude of value distributions within the game. entity as a whole. To this end, the nature of linkages between The vertical integration type of GVC is based on the hierarchi- activities (value chains) is carefully examined­ —­just as if drawing cal structure that assumes an absolute and unidirectional control an anatomical chart of a firm­ —­ to internalize potential externali- of the parent company over its subsidiaries. The activities and ties through cross-functional coordination, which is an important performance of subsidiaries are strictly monitored and assessed source of the firm’s competitive advantage.13 in line with their headquarter management strategies. In con- In contrast, GVC studies originated in sociology. Unlike Por- trast, outsourcing options tend to generate leveled relationships ter’s value chain concept, which is concerned primarily with how between clients (buyers) and subcontractors (service suppliers), firm strategies can be renovated by shifting the focus to the con- and the power exercise is more or less mutual, unlike the vertical figuration of business activities, GVC studies consider the gener- integration type. ation and transfer of value within the system as a consequence Within this dichotomy, Gereffi, Humphrey, and Sturgeon of firm efforts to optimize production networks and, conversely, (2005) set out a GVC typology in a higher resolution spectrum the mechanism of how the value distribution structure affects the in accord with power relations between the contracting parties. firm’s choice of the organizational form of international produc- Figure 1.4 illustrates five variants of GVC governance. The rect- tion networks. GVC analysis is not a global extension of Porter’s angles represent the firm’s boundary, and their size indicates the value chain approach because the scope and motivation differ, strength of bargaining power in relation to the other party. The as described below.14 arrows show the direction and extent of business intervention in the partners’ activities, which can be supportive, such as to draw Typology of global value chains “win-win” scenarios in the long-term perspective, or predatory, The main objective of GVC studies is to explore the interplay by focusing on uptakes of quick profits in the short run. Toward between value distribution mechanisms and organization of the the right of the diagram, the clients (the headquarters in the case cross-border production–consumption nexus. The concept was of the “hierarchy” type) possess greater bargaining powers and first collectively framed in the discussions of the Global Value so are considered to exert a strong influence over the distribu- Chains Initiative (2000–05), sponsored by the Rockefeller Founda- tion of value added. (See annex 1.1 for a detailed description.) tion,15 and further crystallized by Gereffi, Humphrey, and Sturgeon Gereffi, Humphrey, and Sturgeon (2005) also considered the (2005), whose analytical focus rests on the governance structure of dynamics of the GVC configuration by factoring out three param- organizing international production networks. Who are the players eters: complexity of transactions, ability to codify transactions, in the game? What kinds of rules exist? Is it a competitive or a and capabilities in the supply base (known as the “3 C’s model”– cooperative play? What generates the winning opportunities? In Complexity, Codifiability, and Capabilities). For example, the 20 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 1.4 Typology of global value chains Degree of power asymmetry and extent of explicit coordination with partner’s production activities Low High Market type Modular type Relational type Captive type Hierarchy type Customers Client Client Parent company Client (headquarters) MARKETS Supplier Supplier Subsidiary Suppliers Supplier Complexity of transactions Low High High High High Ability to codify transactions High High Low High Low Capabilities in the supply base High High High Low Low Source: Author’s drawing, based on Gereffi, Humphrey, and Sturgeon 2005. shift in the type of value chains from market to relational is asso- a firm whose production technology contains elements of increas- ciated with an increase in the complexity of transactions. The ing returns to scale. The line Fd1 represents the cost schedule of shift from relational to modular assumes an increase in the abil- the traditional method, with all production stages concentrated in ity to codify transactions. And the improving capabilities in the one location. When a part of the production process is outsourced supply base, other things equal, drive value chains from the cap- to a domestic partner, two things occur, as shown in the move- tive type toward the market type. And so on.16 ment of the cost curve from Fd1 to Fd2. First, the curve becomes By probing the mechanism of GVC configurations, the model flatter, indicating an improvement in productivity caused by the helps identify the policy instruments to facilitate the transfor- division of labor. Second, the curve shifts upward, indicating an mation of value chains from one type to another, especially in increase in fixed costs (from c1 to c 2) because of the need for coor- the light of industrial upgrading and the GVC-driven growth of dination between the production units in different locations.19 developing countries.17 Here, the least costly form of production will switch from the tradi- tional method to outsourcing at the output level q1. When outsourcing options are enlarged to include the inter- Economic modeling national context, two other aspects are also taken into account. • Production factor costs are considered to be more diverse In principle, economists’ analytical focus on GVCs has been on between countries than within a country, so productivity will three issues: the mechanism of the fragmentation of production rise more when outsourcing takes place across borders in processes,18 the impacts of offshoring on domestic factor incomes accord with comparative advantage. and welfare, and the firm’s choice of an organizational form of GVCs. • Connecting production units in different countries is more costly than connecting production units within the same Mechanism of the fragmentation of production processes country. International logistics is generally more expensive, Jones and Kierzkowski (1990) provide a model of outsourcing and marked up by import duties and costs for clearing customs set out the factors that affect the degree and form of the fragmen- and the like. There also are nontrivial communication costs for tation of production activities. Figure 1.5a illustrates the relation coordinating production units in countries with different lan- between output level (market size) and total cost of production for guages, legal systems, and business ethics. Analytical frameworks for global value chains: An overview • 21 FIGURE 1.5 Optimal form of outsourcing options (a) (b) Total cost Total cost Fd1 Fd2 Fd1 Fd2 Fw1 Fw1 Fw2 Fw3 … c3 c2 c1 0 q1 q2 Output level 0 Output level Source: Author’s drawing, based on Jones and Kierzkowski 1990. These features are represented by line Fw1, which has a flat- Traditionally, the effect of international trade on the labor ter slope for increased productivity and a higher intercept for market has been considered in regard to a resource shift an extra top-up of the fixed cost (from c 2 to c 3). Then, the opti- between industrial sectors caused by import competition, with- mal form of production will switch from domestic outsourcing to out much attention to the change in the within-sector compo- cross-border outsourcing (offshoring) at the output level q 2. sition of different types of labor. Newer globalization literature In this light, it is possible to consider where multiple countries seizes on this point, recognizing that offshoring is a cross-border are involved in the production process (Fw2, Fw3, …). Different movement of a production activity corresponding to a task for a schedules can be drawn for various outsourcing options, as in particular type and skill of labor.21 figure 1.5b, and the shaded boundary defines the optimal form Feenstra and Hanson (1996a, 1996b) considered the impact of production arrangement at each level of output. of offshoring that follows the liberalization of foreign ownership The model’s implications for a global production arrange- in developing countries. Substantial movements of capital from ment are threefold. Other things being equal, the production developed countries to developing countries are accompanied process will be more prone to international fragmentation when: by transfers of some segments of production processes that are • The targeted market is larger, so that it has more room to considered more skill-intensive by the standard of developing absorb the increased supply of goods from the organization countries but less skill-intensive by the standard for developed of more efficient divisions of labor across borders. countries. Accordingly, the demand for labor becomes skewed • The costs of connecting the production activities in different toward higher skilled labor in the light of the respective skill stan- countries are less inhibitive. dard of each economy, so the relative wages of low-skilled labor • The countries in the production networks are more diverse in fall in both developed and developing countries.22 their factor costs, so there is a better chance for offshoring Grossman and Rossi-Hansberg (2008a) then introduced a firms to exploit comparative advantage. “trade in tasks” concept to explain how an increase in offshoring feasibility affects the productivity and factor incomes of the off- Impacts of offshoring on domestic factor incomes and shoring country. They emphasized the need to shift the analytical welfare focus from goods, as in the conventional trade theory (Portu- The offshoring model was further developed to address income guese wine for English cloth), to tasks that line up in a production —­ distribution and welfare­ a natural response to mounting politi- process, in order to capture the rising prevalence of offshoring cal concerns about the potentially detrimental effect of offshor- activities in a firm’s business strategies. ing on the domestic labor market (the industrial hollowing-­ out In the model the offshoring feasibility is parameterized as an problem).20 improvement in the coordination capability between the firm’s 22 • Measuring and Analyzing the Impact of GVCs on Economic Development headquarters and its foreign suppliers through transportation imperfect competition (Helpman and Krugman 1985) to explain and communication technologies. The sensitivity to the change the asymmetric prevalence of intrafirm trade in capital-inten- in offshoring feasibility is assumed to vary across different types sive industries and between capital-abundant countries. The of tasks. Some tasks (such as those akin to codified description) firm’s dual motives for minimizing transaction costs (by assign- are easy to offshore, while others (such as those relying on per- ing property rights) and factor costs (by exploiting comparative sonal tacit knowledge) are not.23 advantages) are analyzed in the unified theoretical framework. The impact of the improved prospect for offshoring is consid- The model expands the margins of analytical scope in figure 1.3 ered through three channels: to cover the range of value chain variations for both spatial and • A labor-supply effect. Moving some tasks to foreign coun- organizational dimensions. tries frees up the domestic labor that would otherwise carry Antràs and Helpman (2004) introduced another dimension to out these tasks, so it has an effect analogous to increasing the analysis: firm heterogeneity. Drawing on Melitz (2003), Antràs the supply of labor in the market. Such an implication, widely and Helpman investigated the impact of within-sector hetero- discussed in the mass media and political circles, generally geneity in firm productivity on the firm’s globalization decision. evokes opinions against a firm’s offshoring activities for fear of The model predicts that different degrees of entry cost to global lowering the real wages of offshored labor or losing domestic activities bring about the productivity ranking among firms on jobs when wages are sticky. the choice of globalization modes. The most productive firms • A relative-price effect. A country offshores low-skilled labor would choose to undertake foreign direct investment, the next when its cross-country comparative advantage is weaker in most productive firms would choose to engage in arm’s length that type of task than in the tasks of high-skilled labor. The offshoring, and so on down to the least productive firms, which country would then specialize in exporting goods that are would choose to engage only in domestic procurement. intensive in high-skilled labor, as conventional trade theory Further to these approaches, Antràs and Chor (2013) shed new predicts. Accordingly, if an increase in exports leads to a light on the line of analyses by considering a technological order- deterioration in the country’s terms of trade, it would create ing of production stages­ —­ —­ a crucial attribute of value chains­ to a negative impact on the welfare of its high-skilled labor address the traditional make-or-buy question for each segment through the Stolper–Samuelson mechanism. (However, this of a production process along a value chain. Incompleteness of effect comes into play only when the country is large enough contract, as previously defined, entails strategic consideration by to affect the international relative prices of goods.) a lead firm (final good producer) in choosing the form of value • A productivity effect. This effect is a unique feature of the chain governance. And the key prediction of the model is that model that is not fully considered in other studies on the the lead firm should differentiate the governance forms between topic. When the prospect for offshoring improves­ —­say, by an upstream and downstream suppliers for optimizing the gains increase in communication capabilities­ —­an offshoring firm’s from the set of transactions. profitability will rise in proportion to the extent that the firm The model identifies two types of value chains, determined relies on the offshoring business. Such a productivity effect is by the nature of the final product: sequential complements and equivalent to the consequence of factor-augmenting techno- sequential substitutes. The type of sequentiality that character- logical progress, so it is able to bring a positive impact on the izes the production process affects the lead firm’s decision on employment of domestic workers (across all industries) whose the governance arrangements along that value chain (figure 1.6). task levels are similar to those of offshored labor. For sequential complements the lead firm chooses to integrate The net impact of offshoring on factor incomes is the sum of downstream suppliers while outsourcing its upstream produc- these three effects. And in most cases the empirical consider- tion stages. For sequential substitutes upstream suppliers are ation is reduced to whether the productivity effect will dominate vertically integrated, while the transactions with downstream the other two effects­ —­if so, the argument turns in favor of off- suppliers are carried out at arm’s length. (See annex 1.2 for a shoring activities.24 brief description of the argument.)26 The property-rights theory on the firm’s choice of an organi- Firm’s choice of an organizational form of global value zational form is highly resonant with the sociologists’ analytical chains insights about value chain governance because, broadly speak- The factors that determine whether a transaction is mediated ing, both approaches engage the contractibility of transactions through markets or within firm boundaries have long been a sub - as a core parameter of the models. The topic is thus one of the ject of inquiry in industrial organizational theory. The question most promising areas for extensive interdisciplinary dialogue on has been addressed in many ways since Ronald Coase docu- synergetic development of the GVC analysis. mented his insights on the nature of the firm, 25 and it has been brought into the international context in studies on intrafirm trade and multinational corporations. Empirical challenges Antràs (2003), one of the earliest efforts in pursuing this direction, synthesized firm theory under incomplete contracts The rapid progress of empirical analysis on GVCs has been backed (Grossman and Hart 1986) and international trade theory under up by two substantial changes in the research environment. One Analytical frameworks for global value chains: An overview • 23 FIGURE 1.6 Sequential choices for organizing value chains equivalent to the U.S. trade deficit of the product in relation to China. The study shows that the U.S. deficit of $1.9  billion for iPhone trades is reduced to $73 million if viewed in value-­ added UPSTREAM DOWNSTREAM SUPPLIERS SUPPLIERS terms and broken down to include the deficits with other countries such as Japan and Germany, which are the core parts suppliers. These product-level approaches are useful in drawing the Pr Re uct M arc esig Pa ria nd As M mb D eti C ibu us ti ist ng at h ar ly od rts l p d se se d actual structure of production chains because they directly use to on e a n k r pr roc eve m data provided by individual firms rather than resorting to statisti- oc ur lo er ur em pm se cal inference. But the weakness is apparent in the flipside.28 em e e rv ic en nt nt First, these approaches have limited applicability when con- es t sidering macroeconomic issues such as trade policies, because the analytical focus is cast only on a particular product or on the Sequential activity of a few firms. This is far from sufficient to capture the Outsourcing Integration complements entire value flows in the national context. Second, as Dedrick, Kraemer, and Linden (2008) pointed Sequential Integration Outsourcing out, most firm data do not explicitly present compensation of substitutes employees, an important component of value-added items in Source: Author’s drawing. the national accounting framework, but merge it with other types of production costs. Third, because values are generated at every point of the is the increasing availability of relevant data and statistics, espe- production process, the value-added analysis should be able cially multicountry input-output tables and firm-­ level micro­ data. to trace all the production stages along the entire supply chain. The other is the advance in data-­ processing capacity of per- However, the product-level approach considers only the value- sonal computers for handling these massive datasets as well as added structure of direct input suppliers (the first tier), leaving the information and communications infrastructure that allows the rest of the value-added stream untracked. For example, a for efficient shared use of the databases. What was impossible hard-disk drive in an iPhone contains subparts produced in dif- 20 years ago is common practice today, and the empirical chal- ferent countries and thereby requires further decomposition of lenges of GVC analysis are entering a new phase of development. the value-added sources. Mapping global value chains by firm business records output tables Mapping global value chains by input-­ The initial efforts to quantitatively describe GVCs can be found Given the limitations of the conventional approach, multicountry in studies that use firm-specific business records. These studies input-­output tables have received increased attention. A multi- typically aim to identify the composition of inputs procurement country input-­ output table provides a comprehensive map of or the sales networks of a product on the basis of data provided international transactions of goods and services in a massive by the manufacturers themselves or from the teardown reports dataset that combines the national input-­ output tables of vari- of private consulting companies­ —­ or, for the average breakdown ous countries at a given point of time. Because the tables con- of an industry’s generic product type, the information from the tain information on supply–use relations between industries relevant industry associations (Sturgeon and others 2013). and across countries­ which are totally absent from foreign —­ Earlier studies of this kind include Dedrick, Kraemer, and —­ trade statistics­ it is possible to identify the vertical structure of Linden (2008), who analyzed the value-added structure of four international production sharing. And unlike the product-level representative products­ —­Apple’s iPod and video iPod and Hew- approach, input-­ output analysis covers an entire set of industries lett Packard’s and Lenovo’s laptop personal computers­ —­using that make up an economic system, thus enabling the measure- information from business reports.27 They found that a video ment of cross-border value flows for a country or region. Theo - iPod with a retail price of $299 in 2005 was associated with a retically, such analysis has the capacity to track the value-added breakdown of $144 for the product’s factory cost, $75 for dis- generation process of every product in every country at every tribution margins and $80 for the profit of the lead firm (Apple), production stage. while within the factory cost only $3.86 was estimated for the The input-­output approach has weaknesses as well. Sturgeon assembly services in China. The original motivation of the study and others (2013) pointed out the limitations of (multicountry) was to investigate how firms benefit from technological innova- input-­output analyses arising from the statistical characteris- tion through production sharing, but it came to elucidate a sepa- tics of input-­output tables. First, the table’s sectoral classifica- rate and even more alarming question about the validity of con- tion is based on industrial categories so that the value-added ventional trade statistics based on gross values. of a specific task such as product design or assembly cannot In this context, Xing and Detert (2010) addressed U.S.–China be identified. Second, transactions are recorded on a domestic trade imbalances. iPhones were not sold in China in 2009, which basis, so production activities are circumscribed by territorial implies that China’s exports of iPhones to the United States were borders rather than by the nationality that the produced goods 24 • Measuring and Analyzing the Impact of GVCs on Economic Development are associated with, which may cause (analytically) inappropriate about 50% in 2002, more than double what would have been attribution of value added among countries.29 Third, information obtained by a straightforward application of the vertical special- on the nature of specific transactions is totally absent from input-­ ization metric. It quantitatively demonstrates the importance of output statistics, making qualitative analyses of value chains dif- measuring trade in value added terms, as well as the significant ficult, if not impossible. analytical impact of overlooking processing trade. In a nutshell the product-level approach is relevant for ana- While these empirical exercises rely on the national input-­ lyzing qualitative aspects of individual value chains, such as the output tables of individual countries, Daudin, Rifflart, and Sch- form of governance arrangement or the mode of technological weisguth (2006) used the database of the Global Trade Analy- transfer between parties, while the multicountry input-­ output sis Project to construct a multicountry input-­ output table of 70 approach captures a general picture of value chain configuration countries and their composite regions in order to calculate the in the larger context from a systematic point of view. They are domestic value-added content of exports, alongside indices of not exclusive substitutes but must be employed in a complemen- vertical specialization and regionalization. Johnson and Nogu- tary manner, depending on the type of research questions. era (2012) calculated the ratio of value-added exports to gross GVC studies using input-­ output tables have become increas- exports as a metric of international production sharing, again ingly common in the last decade. Their origin can be traced back using the Global Trade Analysis Project database.31 They exten- to Hummels, Ishii, and Yi (2001), who introduced the concept of sively discussed the impact of production sharing on the scale vertical specialization­—­ defined as the amount of imported inter- of bilateral trade balances with respect to multiple countries, mediate inputs used to produce an exported good or, put dif- not to mention the U.S. trade deficit with China, which shows a ferently, the import content of exports, which is presented as a 30–40% drop in value added terms from the traditional calcula- measure of international production sharing. tion (figure 1.7).32 Chen and others (2004) first brought the idea into the value- Bems and Johnson (2012) present an interesting extension added context in relation to the statistical distortion caused by of the trade in value added approach to international macro- ignoring the presence of processing trade and by measuring economics by proposing the concept of the value-added real international trade in terms of gross exports. Here the long-de - effective exchange rate. Real effective exchange rates are com- bated issue of U.S.–China trade imbalances was fully consid- monly used to measure country export competitiveness by eval- ered in the value-added perspective. Koopman, Wang, and Wei uating the magnitude of price adjustments necessary to clear (2012) further developed and methodologically formalized the the external imbalances or, put differently, the extent of nominal approach for separating China’s national input-­ output matrices exchange rate misalignments. into two components, one for the export processing sectors and Conventional real effective exchange rates are often calcu - one for the rest of the economy.30 They showed that the foreign lated from a weighted basket of consumer price indices, where content of value added in China’s manufacturing exports was weights are based on bilateral gross trade flows. However, with FIGURE 1.7 Bilateral trade and value-added balances for the United States, by partner, 2004 $ (billions) n tio m ra ei do de ip ng p. Ta Fe e Re Ki y or lia sia m se an a n o nd ap ce d a, ad ra iu ia n ay ic ne na m n il ite pa re lg ss st an ng la ai ex az ly an er al hi hi 25 Un Au Ko Be Ire Sp Ru Ita Ja Br Fr M M Si G C C C 0 –25 –50 –75 –100 Trade deficit –125 Value-added deficit Value-added deficit (adjusted) –150 Source: Author’s drawing, based on Johnson and Noguera 2012. Analytical frameworks for global value chains: An overview • 25 rapid globalization, conventional rates became an inappro- FIGURE 1.8 China’s real effective exchange rates priate measure in two respects. First, because real effective Change from 1995 value (%) exchange rates are used to assess country export competitive - ness in the world market, approximating price developments 0.4 with consumer price indices is not ideal because consumer Value-added real effective exchange rate price indices summarize the prices of products whose value- added origins could be fragmented across different countries. 0.3 Second, using the same line of logic, the values of gross trade flows cannot serve as unbiased weights because they do not represent today’s economic reality of increasing production sharing among countries. 0.2 The value-added real effective exchange rate overcomes these problems by using gross domestic product (value- added) deflators, instead of consumer price indices, to mea - sure price changes, and bases its weights on value-added 0.1 bilateral trade flows, instead of gross trade flows. Figure 1.8 shows that the gap between China’s conventional and value- Conventional real effective exchange rate added real effective exchange rates increased substantially from 2000 onward. 33 0.0 1995 2000 2005 2010 One of the most recent achievements in this strand of analy- ses is from Koopman, Wang, and Wei (2014), who devised a full Source: Author’s drawing, based on Bems and Johnson 2012. decomposition method of gross exports into various sources of value added. Gross exports are first decomposed into four cate - gories: domestic value added absorbed abroad, domestic value various preceding formulas for measuring value-added trade are added first exported then returned home, foreign value added, systematically integrated into a single accounting framework. In and pure double-counted terms; each category is then further particular, the method enables the isolation of double-counting decomposed by trading mode (figure 1.9). The result is a com- elements in gross exports, which have long haunted trade econ- plete picture of the value-added generation process, in which omists conducting empirical analyses. FIGURE 1.9 Gross trade accounting framework Gross exports Domestic value added Pure double-counted Domestic value added first exported then Foreign value added absorbed abroad terms returned home Intermediates Foreign Intermediate Pure double Foreign sent to first value added Pure double Final goods and exports counting from value added importer and contained in counting from services exports absorbed by domestic contained then re-exported intermediate foreign sources direct importer sources in final exports to third country exports Domestic value added Vertical specialization Source: Author’s drawing, based on Koopman and others 2016. Note: This figure is a revised version from the one presented in Koopman, Wang, and Wei 2014 in response to the comment by Los, Timmer, and de Vries 2016. 26 • Measuring and Analyzing the Impact of GVCs on Economic Development For trade policies the channels of domestic value added first input-­ output tables of respective countries into a single matrix exported then returned home have important implications. For to account for regional heterogeneity within a country in a multi- example, the antidumping measure that the European Commis- country input-­ output framework. The table allows for economic sion imposed on the import of footwear from China and Viet Nam linkages across borders to be studied on a region-to-region in 2006 is known to have had a detrimental impact on service basis­ —­say, between Huanan in China and Kyushu in Japan.36 industries in the European Union because these imported items Domestic linkages between regions are particularly relevant contained considerable value added originating in the European when considering regional (within-­ country) development. For design and distribution sectors. Such consequences could have example, China built strong economic linkages with neighboring been avoided by due reference to a detailed presentation of the countries after the launch of the Reform and Open-Door Policy in value-added sources of traded products.34 1978, but the benefit of economic globalization was not equally shared within the country. Income disparities immediately wid- Heterogeneity considered ened between coastal and inland regions, and it took time for Another important development in the quantitative analyses of the positive impact from abroad to trickle down to inner China GVCs, with a theoretical foundation in Melitz (2003), is account- through domestic linkage effects. In this sense, regional aspects ing for within-sector heterogeneity in firm characteristics when are crucial in accounting for the process of economic develop- constructing input-­ output tables. Conventional input-­ output ment, especially for spacious and less integrated economies.37 tables do not differentiate the input structure of different types Finally, consider heterogeneity in labor markets. The impact of producers in the same industry. However, export-­ oriented of GVCs on employment has been the subject of heated discus- firms, especially those in the processing trade, generally have sion, especially around the industrial hollowing out problem. Ear- higher import intensity in sourcing intermediate inputs than do lier globalization debates addressed the issue primarily in terms domestic-­ oriented producers. This implies that conventional of the industrial structural change brought about by opening input-­output tables, which provide information only on the aver- the domestic economy to global competition (leading to iden- age input structure across all types of producers, may bias ana- tification of declining, stagnant, and expanding industries). The lytical results for countries where processing trade is prevalent current arguments from the GVC perspective engage in more (notably China and Mexico). microscopic analysis by looking into the wealth distribution at As stated earlier, Koopman, Wang, and Wei (2012) were first to the task level within production chains, often epitomized by the formally address this problem, by presenting a method to split the so-called “smiley curve.” Chinese input-­ output tables into subaccounts that align export Along these lines, Timmer and others (2014) conducted processing activities with the rest of the sector. Tang, Wang, and empirical research on value-added distribution among heteroge- Wang (2014) further elaborated the approach, by considering neous labor markets with different types of skill (upon recogniz- variation in such firm characteristics as size (large scale or small ing that each task in the production processes can be associated to medium scale) and ownership structure (domestic or foreign, with a particular level of labor skill). They employed the European private or state-owned). They also used the Chinese input-­ output Commission–­ funded World Input-­ Output Database augmented tables but combined them with data from China’s industrial census by the EU KLEMS database for information on factor inputs, in and trade statistics by firm type. Importantly, the information on which three types of labor (low skilled, medium skilled, and high ownership structure allows the impact of China’s privatization pro- skilled) were identified on the basis of educational attainment. gram on domestic value-chain upgrading to be assessed. For most of the countries in the database the value-added share Ma, Wang, and Zhu (2015) integrated these approaches by of high-skilled labor increased substantially from 1995 to 2008, considering firm heterogeneity in dual dimensions­ —­trading while that of less-skilled labor declined. The results agree with mode (processing exporters or normal exporters plus nonexport- the findings of Feenstra and Hanson (1996a, 1996b) and have ers) and firm characteristics (domestic-owned or foreign-owned). important implications for recent political events in Europe and Using the information of ownership structure, they worked out the United States.38 the distribution of domestic value added according to factor ownership, which contributes to the conversion of measurement Distance matters: “length” analyses of value chains from gross domestic product to gross national income by taking The theory of fragmentation predicts that if the production pro- into account firm heterogeneity.35 cess of a good has the potential for further segmentation by the Heterogeneity can also be considered from a geographic per- change in production technologies or consumption markets, spective. The current setup of multicountry input-­ output tables then there is an opportunity for a finer division of labor that will regards a country as a point of transaction in global production lead to better allocation of resources and lower marginal cost of networks. However, a national economy has a spatial dimension. production. This is especially true with access to international Brazil and China cannot be treated the same way in the input-­ markets, because the differences in factor endowments (and thus output matrices that Costa Rica and Singapore are. Inomata and comparative advantage) are even more salient across borders.39 Meng (2013) introduced the Transnational Inter­ regional Input-­ Accordingly, the study on fragmentation concerns the number Output Table for China, Japan, and Korea, constructed by the of production stages in a production process­ —­comparing alter- Institute of Developing Economies, which links the interregional native technologies that produce the same good, one with few Analytical frameworks for global value chains: An overview • 27 production stages and another with many. Empirical research FIGURE 1.10 Relative line position of countries in the requires an overall perspective for the entire structure of the pro- regional production networks of East Asia, 1985, 2005 duction sequence. What matters is not only the strength (magni- Forward average propagation length tude) of production linkages, but also the length of the linkages, 3.9 Upstreamness Longer supply chains determined by the number of production stages. The traditional input-­output approach to analyzing produc- tion networks is generally concerned with the interconnected- 3.7 1985 ness or strength of linkages between industries. The “length” 2005 dimension of production linkages was first addressed by the 3.5 Japan input-­output model of average propagation length developed Rep. of by Dietzenbacher, Romero, and Bosma (2005). The average United States Korea propagation length model represents the average number of 3.3 China production stages lining up in every branch of production net- Chinese Taipei works, so it effectively measures an industry’s fragmentation.40 3.1 Philippines Dietzenbacher and Romero (2007) further applied the model to Thailand Malaysia the international context by analyzing the cross-national linkages Indonesia 2.9 of major European economies using the 1985 European multi- Singapore country input-­ output table. Shorter supply chains Downstreamness Fally (2011) developed a model for measuring fragmentation 2.7 that was based on a philosophy similar to that of the average 2.7 2.9 3.1 3.3 3.5 3.7 3.9 propagation length model. The major difference is that Fally’s Backward average propagation length model, as well as Antràs and others’ (2012) variation, captures the average number of production stages by pegging the endpoint Source: Author’s drawing, based on Escaith and Inomata 2013. of the sequence at final consumption, which enables measuring the distance to final demand of a product along the production chains. Those studies rely on national input-­ output tables of the country within the regional production networks (as determined United States and other selected countries, but De Backer and by the ratio of forward and backward average propagation Miroudot (2012) later applied Fally’s (2011) model to the inter- lengths). For example, China moved along the path that is far- country input-­ output tables of the Organisation for Economic thest from the bottom-left to top-right diagonal, indicating that Co-operation and Development covering 56 countries for 1995, it stayed in the most downstream segment of the regional supply 2000, and 2005.41 chains throughout the period, which reflects the country’s domi- One application of the “length” model in the GVC context is nant role as a final assembler of regional products.42 to identify countries’ (or industries’) relative position within the The line position of industries and countries within a produc- global production system. If a country’s representative produc- tion system is particularly important for considering the varia- tion chains toward final products are longer than those toward —­ tions in sectoral characteristics along value chains­ for example, primary products, the country is considered to operate in a rela- value-added ratios as signified by the “smiley curve” (Baldwin, tively upstream position (and conversely if a country’s represen- Forslid, and Ito 2016; Ye, Meng, and Wei 2015) or the mode of tative production chains toward final products are shorter than value chain governance (Antràs and Chor 2013). those toward primary products, the country operates in a rela- tively downstream position). Because the average propagation length can be measured both in forward (cost-push) and back- So, what’s next? ward (demand-pull) directions along production lines, it is possi- ble to identify the relative position of a country within the global Perhaps the most pressing issue for the GVC research community production networks by comparing the pairs of forward-­ length is to accelerate the development of relevant data. Until now, a and backward-­ length values. large share of empirical work for testing GVC governance models Inomata (2008) and Escaith and Inomata (2013) are among of firm theory has relied on data from official merchandise trade the earliest efforts to develop the idea of measuring the rela- statistics.43 Some country databases (such as the Related Party tive production positions of countries. They elucidated the struc- Trade Database from the U.S. Census Bureau) contain informa- tural change of the regional production system in two dimen- tion on whether shipping involves transactions between related sions, using data for East Asia (figure 1.10). With the horizontal or nonrelated parties, which can be used to sketch out the pres- axis for backward average propagation length and the vertical ence of multinational firms in international trade.44 axis for forward average propagation length, the bottom-left to Despite the observable advantages of the data (notably top-right direction presents the changes in the entire length of accessibility and availability), researchers face several challenges the supply chains that countries participate in, and the top-left to using it appropriately. Antràs (2011) set out four of them. First, to bottom-right direction draws the relative line position of each the product-level information aggregates the sourcing decisions 28 • Measuring and Analyzing the Impact of GVCs on Economic Development of multiple firms, so some approximation is imposed for testing researchers can choose any combination of regions or sectors the model of firm-level sourcing behavior. Second, the data do to assemble the multicountry input-­ output tables most suited to not provide information about the users of the products being their research interests. By developing a Wikipedia­-like common shipped, so it is impossible to identify which sector of the econ- e‑infrastructure, the lab’s setup optimizes the use of available omy has absorbed the imported product (or even whether it is information, enhances flexibility in data construction, and saves for intermediate use or final consumption). Third, as for the ship - resources by avoiding duplication of work among different insti- ping between related parties, the data tell neither which party is tutions (Lenzen and others 2017). owned by whom, nor the degree of control or ownership share of the parent company. The second and third points pose a practical problem when relating observations in intrafirm trade Meta-methodological considerations with the characteristics of importers (headquarters, in the case of backward integration), as modeled in Antràs (2003). Fourth, GVC studies have evolved along three distinctive modes of the data report only the information on incoming and outgo- analyses: spot analysis, sequence analysis, and network analysis. ing shipments from the viewpoint of a home country. But multi- Gary Gereffi’s earlier model, global commodity chains, consid - national firms often engage in global sourcing, involving ship - ered the power relation between a lead firm and a set of multiple ments between third countries (for example, Apple headquarters subcontractors that operate at different tiers along production in the United States may source Korean Samsung’s inputs being chains (Gereffi and Korzeniewicz 1994). “One versus many” was shipped to Foxconn factories in China for assembly). thus the basic setup for analyzing the nature of governance. In Firm-level microdata, which have become increasingly avail- contrast, Gereffi, Humphrey, and Sturgeon (2005) and later able in recent years, may provide the information needed to studies moved the analytical target to one-to-one transactions develop empirical tools that overcome these problems.45 The within a particular pair of a lead firm and a supplier (Bair 2008). benefit of the datasets rests on their representativeness of var- So the modal shift in GVC studies among sociologists was from ious aspects of firm operations. For example, the Basic Survey sequence (that is, one versus many) to spot (that is, one versus of Japanese Business Structure and Activities (Kigyo-katsudou one) analysis­ —­ or, in the Euclidean sense of the word, from one-­ kihon chosa toukei) by Japan’s Ministry of Economy, Trade, and dimensional to zero-dimensional spatiality. Industry, has annual survey data (mandatory under the Statistics In international trade theories the analytical focus of GVC Act of Japan) that cover multiple types of information on firms, studies has been primarily on a particular supply–use relation such as sales, costs, employment, capital expenditures, exports, between trading partners, especially for a firm’s “make-or-buy” imports, and foreign direct investment.46 choice of intermediate inputs. The dominant mode of analysis has Even so, unlike those Japanese data, many firm-level micro - thus been spot analysis, yet Antràs and Chor (2013) have opened data come from one-shot industrial surveys and thus are avail- a new path toward sequence analysis by considering a techno- able only for particular countries in particular years. The datasets logical ordering of production stages (from zero-­ dimensional to also differ in the dimensions of representativeness. Accordingly, one-dimensional spatiality). in order to apply these datasets to a general equilibrium setup output economics has by its nature always been con- Input-­ like the input-­output system, they should be used, for example, cerned with a sequence, whether in the traditional Leontief to provide combined structural information for estimating the impact models or in the latest supply chain length models. How- relevant coefficients along with appropriate constraints and a ever, recent work engages network theory by applying the con- balancing algorithm. cept of network centralities to input-­ output matrices (Carvalho Another aspect to consider is the integration of databases, 2012; Escaith 2014) and thereby shows some movement from especially of multicountry input-output tables. Currently, various sequence to network analysis (from one-dimensional to two-­ institutions construct competing tables, each designed for a spe- dimensional spatiality). cific analytical objective, so their presentation format, sectoral These observations suggest that the analytical frameworks of classification, and types of ancillary information (such as environ- GVC studies are diverging rather than converging over time­ —­ mental accounts) differ.47 and that the prospect for overall consolidation of methodologies A team at the University of Sydney recently launched the is limited in the near future. However, this is not necessarily bad Global Multi-Region Input-­ Output Lab, which aims to build a news. The diversity and multiplicity of methodological frame- cloud-computing platform that allows participants to use each works imply that a wider scope of analysis is available. It is only other’s individually developed statistical resources. The infor- a matter of how best to combine the relevant frameworks in an mation from the aforementioned multicountry input-­ output appropriate way for each research question, just as with inte- databases, together with national accounts and foreign trade grating various tasks into an optimal configuration of production statistics, are expected to be input in the platform. Then, a chains. Keeping and facilitating interdisciplinary dialogues are highly detailed regional-sectoral taxonomy (the root classifica- essential, and the Global Value Chain Development Report will tion) linked to the data pool will serve as a feedstock from which serve as a core platform for this end. Analytical frameworks for global value chains: An overview • 29 ANNEX 1.1 Typology of global value chains Gereffi, Humphrey, and Sturgeon (2005) set out a typology of Relational-type global value chain five global value chains (GVCs) on the basis of the structure of When the manufacturing process involves specialized equip- power relations between the contracting parties. ment (for example, the mold for a product of a particular shape), transactions become asset-specific, and the contracting par- Market-type global value chain ties become mutually dependent. The equipment for a specific Producing a commodity of a generic nature does not require any purpose has limited scope for alternative uses, so its productiv- specific investment in production facilities for a particular trans- ity will drop considerably when it is applied in other contexts. action, so both customers and suppliers have countless choices Accordingly, the service suppliers (the holders of the specialized for alternative partners. They are connected mainly through open equipment) are not motivated to look for other potential clients. spot-market transactions in a shoulder-to-shoulder relationship. But it is also difficult, or at least costly, for the client to expect Also, the procurement of a generic commodity will not neces- the same level of performance from other third suppliers with- sitate an exchange of detailed product specification between out these specialized facilities. As a result, both parties have little contractors because the key information is mostly reduced to incentive to search for alternative business relations. Further, the preset price of the product that can be found in a book of reinvestment in the specialized equipment for raising productiv- catalogs. The transaction cost for changing business partners is ity deepens the asset-specificity of the transaction, thus trapping almost negligible, leaving the value chains in a constant state of the parties in even more mutually dependent relationships. flux because of their high price elasticity. Captive-type global value chain Modular-type global value chain This type of transaction assumes an overwhelming disparity in In business management or industrial engineering the word power exercise among the parties, as seen in the business rela- “module” generally refers to a composite of subcomponents tions between a lead firm of global brands and its subcontracting grouped by the types of functions that are assumed in making local small companies. Service suppliers are expected to follow up the final product.48 The possibility of different combinations the client’s instructions word for word and are subject to strict of differentiated modules enables producers to design multiple surveillance on product quality and delivery times. Unlike suppli- variants of a product. By the same token, if a complex transaction ers in the market-type GVC, captive service suppliers have nei- can be accommodated in the supply base by adjusting the com- ther sufficient productive capacity to enjoy the scale of mass pro - bination of multipurpose equipment, the supplier will not have to duction, nor the specialized production facilities needed to claim incur transaction-specific investment (no hold-up problem) and its uniqueness, as attributed to the suppliers in the relational-­ is thus able to spread the equipment’s use across a wide range type GVC. The availability of only mediocre production capabil- of potential clients. Even though the information to be delivered ity greatly narrows their opportunities to look for alternative busi- between the contractors may be considerable (say, for produc- ness relations, imposing a captive position toward their clients. ing a complex product), the relative codifiability of transactions, as presumed in this type of GVC governance, compresses the Hierarchy-type global value chain volume of interventions, and the supplier is able to take overall As stated earlier, this type of GVC generally refers to the rela- control of its own production process. This implies that the trans- tions within a vertically integrated firm, as with multinational action cost for changing business partners remains relatively low. corporations. 30 • Measuring and Analyzing the Impact of GVCs on Economic Development ANNEX 1.2 Governance arrangements along a production sequence In the setup of Antràs and Chor (2013), in which a contract is In contrast, if the lead firm has substantial market power and incomplete, a lead firm (final good producer) and a supplier thus operates along an inelastic downward-sloping demand (intermediate input producer) need to bargain ex post over their curve, the firm’s revenue function becomes highly concave to respective share of an incremental surplus (quasi-rent) generated (quality-adjusted) output, and marginal revenues fall at a rela- at the corresponding stage of the production sequence. Follow- tively fast rate along the production sequence. As a result, the ing Grossman and Hart (1986), the lead firm acquires a better large upstream investment dampens the revenue prospect of bargaining position and thus gains a higher share of the surplus downstream suppliers by reducing the value of undertaking when its supplier is integrated than when its supplier remains future investment. The former investment options of suppliers independent.49 are called sequential complements, and the latter sequential Since the supplier’s investment is assumed to be relation-­ substitutes.50 And the type of sequentiality that characterizes specific to the lead firm’s final product (for example, invest- the production process affects the lead firm’s decision about the ing in the mold for a distinctive shape), the investment has no organizational form of value chains. value outside this production sequence, which causes a familiar Recall the lead firm’s tradeoffs: the rent-extraction opportu- hold-up problem, such that the vertically integrated supplier nities by integration, on the one hand, and the investment ineffi- tends to underinvest in its production capacity in anticipation of ciencies caused by such integration, on the other. On this basis, exploitation by the lead firm. the lead firm should weigh the costs and benefits of integrating So the lead firm faces tradeoffs. If it integrates the supplier, the suppliers. it can extract a higher share of the surplus from that particular For sequential complements the investment-curbing effect of production stage, but doing so may induce underinvestment by integration is more costly in upstream production stages because the supplier, which would constrain the output or quality of the it dampens the positive spillover of investment incentives to the final product. downstream suppliers. So the lead firm should seek better bar- Here, the lead firm’s strategic space depends critically on the gaining positions by integrating downstream segments of the nature of the final product that it produces. Suppose that the production process, rather than upstream ones. For sequential product has a quite elastic market demand, so that the lead firm substitutes, the potential underinvestment by the upstream sup- is able to generate larger revenues by producing more. Since the pliers can be compensated for by the downstream suppliers. investment decision of each intermediate input supplier depends The lead firm is then able to place a relatively high weight on the on the prospect of final product turnover, which further depends rent-extraction motive in the upstream stages without worrying on how much the upstream suppliers prior to the current produc- too much about the overall underinvestment. tion stage have already invested in their production capacities, The corollary of the argument is summarized in figure 1.6 by it follows that higher investment by upstream suppliers induces a couple of the lead firm’s decisions about the organization of more investment by downstream suppliers. value chains. Analytical frameworks for global value chains: An overview • 31 Notes in contrast, focuses on the dynamic interplay among parties, where lead firms actively engage in strategic maneuvers for turning the cost 1. Global Value-Chain Training and Research Workshop, June 30–July structures to their own favor, such as spurring competition among 11, 2014, University of International Business and Economics, Beijing, suppliers or promoting supply-base capabilities. It is also important to China. consider the role of government in affecting the choice of value chain 2. A particular concern is the difficulty of delineating a boundary arrangement where markets may fail. The provision of public goods between GVC studies and international trade literature. Apparently, such as transport infrastructure gives a straightforward example. The these two areas overlap in many respects, and the relevant work is fre- underinvestment caused by the hold-up problem presents another quently cross-referenced. However, the characterization of GVC stud- case of market failure due to information asymmetry which calls for ies stated here aims to limit the number of references relating to the government intervention to, say, tighten up contract enforcement vast range of important literature in international economics. schemes. These issues are discussed in chapter 1 of Blyde (2014), with 3. Throughout this chapter the following terms are considered to carry respect to Latin American and Caribbean economies. more or less the same meaning: international (cross-border) produc- 11. For example, even in the case of a dispute, the arbitrator cannot tion sharing, international (cross-border) fragmentation of production, judge whether the delivered good may accord with the product the second unbundling, trade in tasks, and vertical specialization, specification or whether the supplier has put sufficient effort into its each referring to the process and consequence of offshoring activities. productive activities. Contracts cannot be written on sales revenues, 4. This theoretical breakthrough paved several development pathways either. in the days that followed. Aided by the analytical model of oligopoly seeking pur- 12. Firms may carry out foreign direct investment for market-­ formalized in the theory of industrial organization, it factored in the poses (horizontal foreign direct investment) rather than for exploit- strategic aspects of trade policies using the language of game theory. ing factor cost differences (vertical foreign direct investment). In the Also, the element of increasing returns was further embodied and former case, foreign direct investment may not be associated with advanced in other subfields of economics, such as the endogenous vertical integration. growth model and the new economic geography (spatial economics). 13. For example, the Toyota Production System, well known for its just- 5. As a result, industry became an inappropriate analytical unit for the in-time delivery, can be considered as an ultimate form of value chain study of international trade. See the later discussion on firm heteroge- management, where information sharing and task coordination across neity for the empirical challenges to tackle this problem. different divisions are implemented and achieved at the highest level 6. A more extensive discussion of these topics can be found in many of synchronization. other GVC-related materials. See especially the comprehensive 14. There are other terminologies of a similar kind in the field. Global review in OECD (2013). output sequence of supply chain is a generic label for a physical input-­ 7. See Baldwin (2006) for the comprehensive argument of his view intro- value-adding activities across borders, used mainly in business stud- duced in this section. ies that focus on logistics management or trade facilitation (how to 8. Smith (1776, p. 15): “One man draws out the wire, another straights reduce costs and lead times for delivery). Global commodity chain, as it, a third cuts it, a fourth points it, a fifth grinds it at the top for receiv- developed in Gereffi and Korzeniewicz (1994, p. 2), addresses wealth ing, the head; … and the important business of making a pin is, in distribution by showing “how production, distribution, and consump- this manner, divided into about eighteen distinct operations, which, in tion are shaped by social relations (including organizations)….” In some manufactories, are all performed by distinct hands….” this sense, global commodity chains can be considered a predeces- 9. Ronald Coase is said to have opened the horizon for theorizing sor to the GVC concept in spirit, though their analytical frameworks about the mechanism of vertical integration. Until then, a firm was are somewhat different (producer-driven and buyer-driven chains of conceptualized as a production set that defines and implements global commodity chains, compared with the five types of GVC gov- the most efficient arrangement for transforming inputs into outputs ernance in figure 1.4). through multiple interactions with markets. That is, markets and firms 15. See Gereffi and Kaplinsky (2001). were considered to be complementary in their respective functions. 16. Sectoral examples include bicycles (from hierarchy to market), apparel Coase’s insight about the nature of the firm has altered this view. Mar- (from captive to relational), fresh vegetables (from market to rela- kets and firms are more like substitutes, in the sense that they are tional), and electronics (from hierarchy to modular). just different types of coordination arrangements for resource alloca- 17. The governance structure of value chains is particularly important for tion; one through the price mechanism and the other through entre- generating and diffusing the knowledge-based capital that leads to preneurship. So, for the issue of vertical integration, “What has to innovation and industrial upgrading. See the case studies in Kawakami be explained is why one integrating force (the entrepreneur) should and Sturgeon (2011) for East Asian economies and Blyde (2014) for be substituted for another integrating force (the price mechanism)” Latin American and Caribbean economies about the industries that (Coase 1937, p. 398). are learning and upgrading through participation in GVCs. 10. Milberg and Winkler (2013) point out that transaction cost economics 18. Deardorff (2001, p. 122) defines fragmentation as “the splitting of a essentially operates within the static framework of constrained optimi- production process into two or more steps that can be undertaken in zation such that firms would choose the most efficient form of value different locations but that lead to the same final product.” chain governance (make-or-buy) in the face of a given set of transac- 19. The original setup in the study postulates that the firm invests in a new tional and bureaucratic cost structures. The resource-based approach, production facility for the fragmented tasks rather than outsourcing 32 • Measuring and Analyzing the Impact of GVCs on Economic Development them, so there is an extra span in the upward shift of the cost curve in database for trade in terms of value added today is the Organisation the diagram. World Trade Organi- for Economic Co-operation and Development–­ 20. The Economist 2004. zation Trade in Value-Added database. The latest release (reference 21. In the United States the issue has evolved in a wider context: whether year 2015) covers 34 industries for 64 countries (including rest of the jobs are destroyed by foreign competition or by technological prog- world). For a general description of the data, see www.oecd.org/sti ress. U.S. workers are competing with cheap labor abroad and with /ind/tiva/tivasourcesandmethods.htm. For a quick guide to the con- robots at home, and which of those is a worse enemy has been a ­nomata (2014) or WTO and cept of the trade in value added, see I topic of heated debate. See, for example, Spence (2011) for a dis- IDE–JETRO (2011). cussion of the impact of globalization on U.S. job markets along the 33. The increasing gap between the values of two indicators is accounted dimensions of tradeable versus nontradeable sectors and high-skilled for mainly by the shift of the base from consumer price index to gross versus low- and medium-skilled labor. domestic product deflators, rather than the change in weights from 22. However, the declining relative wage does not necessarily make gross to value added terms. unskilled workers worse off because, from a general equilibrium per- 34. One of the key properties of the accounting framework for trade in spective, the increased supply of goods to the market brought about value added is the mathematical identity between a country’s total by finer division of labor may lower the goods prices of both countries trade balance measured in gross terms and that in value added terms. through trade, perhaps offsetting the nominal wage reduction. Kuboniwa (2014a, 2014b) provide rigorous proofs of the relevant 23. See, for example, Blinder (2009). In the base model of Grossman and propositions. Rossi-Hansberg (2008a) only low-skilled labor is assumed to be feasi- 35. Similar efforts have been made by Ahmad and others (2013) for ble for offshoring. Turkey, by Fetzer and Strassner (2015) for the United States, and by 24. The implication of offshoring between similar countries is discussed in Piacentini and Fortanier (2015) for member countries of the Organi- Grossman and Rossi-Hansberg (2008b). sation for Economic Co-operation and Development. Liu and others 25. See endnote 9. Legros and Newman (2014) give an overview for the (2016) extend the method to the application in environmental analy- recent arguments on firm boundaries. ses. If carbon emissions from production activities are regarded 26. For the empirical specification the study refers to the latest develop- as negative value added, the carbon footprint analysis using multi- ment in quantifying an industry’s upstreamness and downstreamness output tables can also be considered one form of GVC country input-­ output model of Antràs and others (2012). by employing the input-­ studies (especially the topic on the political interplay among countries Also, Alfaro and others (2015) develops the benchmark model of based accounts and consumption-­ over production-­ based accounts of Antràs and Chor (2013) with three extensions. First, it considers asym- carbon emissions). metric differences in input contractibility; second, it incorporates the 36. Other efforts with a similar motivation include Cherubini and Los productivity heterogeneity of final good producers (as in Antràs and (2013) for Italy, Dietzenbacher, Guilhoto, and Imori (2013) for Brazil, Helpman 2004); and third, it accommodates the case in which integra- and Meng, Wang, and Koopman (2013) for China. These studies tion is not feasible for certain segments of the production processes output table in embed the respective country’s interregional input-­ because of external factors. funded World Input-­ the European Commission–­ Output Database. 27. If nonacademic literature is included, Tempest’s (1996) account of the 37. See further discussion in Meng, Wang, and Koopman (2013). Barbie Doll is one of the earliest. class workers in nonurban 38. The mounting discontent of middle-­ 28. The product-level approaches introduced here should be strictly dis- regions, who were left disconnected from the benefit of economic tinguished (in terms of the scope of analyses) from the strand of stud- globalization, was a driving force of Brexit and the latest U.S. presi- ies using industrywide microdata of firms, such as those available from dential election. However, offshoring activities alone cannot explain industrial censuses. whether globalization will create or destroy domestic jobs because 29. The efforts to alleviate these potential drawbacks are introduced the structural changes in labor markets are also triggered by techno- below in the section on firm heterogeneity. logical innovations and switches in consumer demand. 30. The same exercise is carried out in De La Cruz and others (2011) for 39. See the model of Jones and Kierzkowski (1990). Mexico, where processing trade is also prevalent. 40. However, Dietzenbacher, Romero, and Bosma (2005) do not explicitly 31. Los, Timmer, and de Vries (2015) implement a similar exercise but use the word “fragmentation.” with a different motivation. They conduct a longitudinal analysis of 41. Recent studies aim to decompose the length model into domestic the tension between a force toward regionalization and one toward and international segments, which enables one to depict the “gen- globalization in the organization of international production net- uine” international fragmentation of the production process. These works. They conclude that increasing globalization (less segmentation efforts include Hagiwara (2016) on the average propagation length into regional blocs) has been a dominant trend during the period of model and Wang and others (2016) on the Antràs and others (2012) analysis. model. 32. To be precise, the Institute of Developing Economies was the first 42. The more formal documentation of the idea is in Miller and Temur- to develop and publish such measurements in the 1980s for seven shoev (2015) and Wang and others (2016), although their models Asian countries and the United States using the reference year of have different specifications and are more rigorously articulated than 1975. However, the measurements were called the impact of final those in Inomata (2008) and Escaith and Inomata (2013). demand on value added rather than trade in value added. The major 43. See, for example, Antràs (2003) and Bernard and others (2010). Analytical frameworks for global value chains: An overview • 33 44. In the U.S. data, partners are related if either party owns at least 10% References of the other party. 45. Tomiura (2007) is one of the earliest studies using firm-level micro- Ahmad, N., S. Araujo, A. Lo Turco, and D. Maggioni. 2013. “Using Trade data. It applies the data to an investigation of the relation between Microdata to Improve Trade in Value-Added Measures: Proof of Con- firm productivity and globalization decisions and derives results that cept Using Turkish Data.” In Trade in Value Added: Developing New are consistent with the predictions of Antràs and Helpman (2004) Measures of Cross-Border Trade, edited by A. Mattoo, Z. Wang, and about the productivity ranking of different globalization modes. S-J. Wei, 187–219. Washington, DC: World Bank. 46. The data cover only medium and large firms with 50 or more employ- Alfaro, L., P. Antras, D. Chor, and P. Conconi. 2015. “Internalizing Global ees and whose paid-up capital is more than 30 million yen. However, Value Chains: A Firm-Level Analysis.” National Bureau of Economic given that global sourcing matters for large enterprises, the threshold Research (NBER) Working Paper No. 21582, NBER, Cambridge, MA. is unlikely to limit the analyses. Antràs, P. 2003. “Firms, Contracts, and Trade Structure.” Quarterly Journal funded World Input-­ 47. The European Commission–­ Output Database of Economics 118 (4): 1375–1418. operation and and EXIOBASE, the Organisation for Economic Co-­ ———. 2011. “Grossman-Hart (1986) Goes Global: Incomplete Contracts, Output Tables, Purdue University’s Development’s Inter-Country Input-­ Property Rights, and the International Organization of Production.” Regional Input-­ Global Trade Analysis Project Multi-­ Output Database, NBER Working Paper No. 17470, NBER, Cambridge, MA. and the University of Sydney’s Eora Database, among others. Diet- Antràs, P., and D. Chor. 2013. “Organizing the Global Value Chain.” Econo- zenbacher and Tukker (2013) introduce the major multicountry input-­ metrica 81 (6): 2127–204. output table projects, and Inomata and Owen (2014) discuss the ana- Antràs, P., D. Chor, T. Fally, and R. Hillberry. 2012. “Measuring the lytical implication of using different databases. Upstreamness of Production and Trade Flows.” American Economic management 48. For example, a modular car may consist of a power-­ Review: Papers & Proceedings 102 (3): 412–16. module (a composite of compressors and charge controls), a drive-as- Antràs, P., and E. Helpman. 2004. “Global Sourcing.” Journal of Political sisting module (a composite of sensors, cameras, light emitting Economy 112 (3): 552–80. diodes), and so on. Bair, J. 2008. “Analysing Global Economic Organization: Embedded Net- 49. Grossman and Hart (1986) define integration as the purchase by one works and Global Chains Compared.” Economy and Society 37 (3): firm of the residual rights of control over another firm’s assets. While 339–64. transaction cost economics is concerned with inefficiencies arising Baldwin, R. 2006. Globalisation: The Great Unbundling(s). Helsinki: Eco- from both the ex post haggling by the parties over quasi-rents and nomic Council of Finland. the consequent ex ante underinvestment (and its negative impact on ———. 2013. “Global Supply Chains: Why They Emerged, Why They ex post performance), the property-rights literature focuses on the Matter, and Where They Are Going.” In Global Value Chains in a impact of property-rights assignment (the choice of organizational Changing World, edited by D. Elms and P. Low, 13–59. Geneva: World form) on ex post bargaining, which is assumed to be efficiently con- Trade Organization. ducted, and that, in turn, affects the party’s decision about ex ante Baldwin, R., R. Forslid, and T. Ito. 2016. “Unveiling the Evolving Sources investment. of Value Added in Exports.” Joint Research Program Series No. 161, 50. More specifically, sequential complementarity and substitutability are Institute of Developing Economies–Japan External Trade Organization determined by the relative magnitudes of (1) the market demand elas- (IDE–JETRO), Chiba City, Japan. ticity for the final product and (2) the elasticity of substitution among Bems, R., and R. Johnson. 2012. “Value-Added Exchange Rates.” NBER intermediate inputs. If (1) is larger than (2), the investment options are Working Paper No. 18498, NBER, Cambridge, MA. sequential complements; otherwise, they are sequential substitutes. Bernard, A. B., and J. B. Jensen. 1995. “Exporters, Jobs, and Wages in U.S. In the usual sense of the word the suppliers’ investments are always Manufacturing, 1976–1987.” Brookings Papers on Economic Activity, complementary. Only when the standard complementarity of interme- Microeconomics Vol. 1995 (pp. 67–119), Washington, DC. diate inputs is dominated by the effect of a quick erosion of revenue Bernard, A. B., J. B. Jensen, S. J. Redding, and P. K. Schott. 2010. “Intrafirm prospect due to the low demand elasticity of the final product does Trade and Product Contractibility.” American Economic Review: Papers the relation turn from complements to substitutes. & Proceedings 100 (2): 444–48. Blinder, A. 2009. “How Many U.S. Jobs Might Be Offshorable?” World Eco- nomics 10 (2): 41–78. Blyde, J. S., ed. 2014. Synchronized Factories: Latin America and the Carib- bean in the Era of Global Value Chains. New York: Springer. Campa, J., and L. Goldberg. 1997. “The Evolving External Orientation of Manufacturing Industries: Evidence from Four Countries.” NBER Work- ing Paper No. 5919, NBER, Cambridge, MA. Carvalho, V. M. 2012. Input-Output Networks: A Survey. A report for the European Commission (EC) under the Complexity Research Initiative for Systemic Instabilities consortium agreement. Brussels: EC. Chen, X., L. Cheng, K. C. Fung, and L. Lau. 2004. “The Estimation of Domestic Value-added and Employment Induced by Exports: An 34 • Measuring and Analyzing the Impact of GVCs on Economic Development Application to Chinese Exports to the United States.” In China and Fetzer, J. J., and E. H. Strassner. 2015. “Identifying Heterogeneity in the Asia: Economic and Financial Interactions, edited by Y. W. Cheung and Production Components of Globally Engaged Business Enterprises in K. Y. Wong. New York: Routledge. the United States.” Working Paper WP2015–13, Bureau of Economic Cherubini, L., and B. Los. 2013. “Regional Employment Patterns in a Glo- Analysis, Washington, DC. balizing World: A Tale of Four Italies.” Paper presented at the Bank of Gereffi, G., J. Humphrey, and T. Sturgeon. 2005. “The Governance of Italy workshop, “Global Value Chains: New Evidence and Implications,” Global Value Chains.” Review of International Political Economy 12 (1): Rome, June 22. 78–104. Coase, R. 1937. “The Nature of the Firm.” Economica 4 (16): 386–405. Gereffi, G., and M. Korzeniewicz, eds. 1994. Commodity Chains and Global Daudin, G., C. Rifflart, and D. Schweisguth. 2006. “Who Produces for Capitalism. Westport, CT: Praeger. Whom in the World Economy?” Document de travail No. 2009–18, Gereffi, G., and R. Kaplinsky, eds. 2001. The Value of Value Chains: Spread- Observatoire Francais des Conjonctures Economiques, Paris. ing the Gains from Globalisation. Special issue of the Institute of Devel- Deardorff, A. V. 2001. “Fragmentation in Simple Trade Models.” North opment Studies Bulletin (IDS) Bulletin. Brighton, U.K.: IDS. American Journal of Economics and Finance 12 (2): 121–37. Grossman, G. M., and E. Rossi-Hansberg. 2008a. “Trading Tasks: A Simple De Backer, K., and S. Miroudot. 2012. “Mapping Global Value Chains.” Theory of Offshoring.” American Economic Review 98 (5): 1978–97. Working Paper Series No. 1677, European Central Bank, Frankfurt. ———. 2008b. “Task Trade between Similar Countries.” NBER Working Dedrick, J., K. Kraemer, and G. Linden. 2008. Who Profits from Innovation Paper No. 14554, NBER, Cambridge, MA. in Global Value Chains? A Study of the iPod and notebook PCs. Report Grossman, S., and O. Hart. 1986. “The Costs and Benefits of Ownership: A prepared for the Sloan Industry Studies Annual Conference, Boston, Theory of Vertical and Lateral Integration.” Journal of Political Economy May 1–2. 94 (4): 691–719. De La Cruz, J., R. B. Koopman, Z. Wang, and S. J. Wei. 2011. “Estimat- Hagiwara, T. 2016. “Decomposition of Average Propagation Length.” ing Foreign Value-Added in Mexico’s Manufacturing Exports.” Office of Paper presented at the 24th International Input-Output Conference, Economics Working Paper, 2011–04A, U.S. International Trade Commis- Seoul, July 4–8. sion, Washington, DC. Helpman, E., and P. Krugman. 1985. Market Structure and Foreign Trade. Dietzenbacher, E., and A. Tukker. 2013. “Global Multiregional Input-Out- Cambridge, MA: MIT Press. put Frameworks: An Introduction and Outlook.” Economic Systems Hummels, D., J. Ishii, and K-M. Yi. 2001. “The Nature and Growth of Ver- Research 25 (1): 1–19. tical Specialization in World Trade.” Journal of International Economics Dietzenbacher, E., and I. Romero. 2007. “Production Chains in an Inter- 54 (1): 75–96. regional Framework: Identification by Means of Average Propagation Inomata, S. 2008. “Average Propagation Lengths: A New Concept of the Lengths.” International Regional Science Review 30 (4): 362–83. ‘Distance’ between Industries, with an Application to the Asia-Pacific Dietzenbacher, E., I. Romero, and N. S. Bosma. 2005. “Using Average Region.” Sangyo-Renkan 16 (1): 46–55. Propagation Lengths to Identify Production Chains in the Andalusian ———. 2014. “Trade in Value Added: Concept, Development, and an East Economy.” Estudios de Economia Aplicada 23 (2): 405–22. Asian perspective.” In A World Trade Organization for the 21st Century, Dietzenbacher, E., J. Guilhoto, and D. Imori. 2013. “The Role of Brazilian edited by R. Baldwin, M. Kawai, and G. Wignaraja, 48–70. Cheltenham, Regions in the Global Value Chain.” School of Economics, Business U.K.: Edward Elgar. and Accounting Working Paper 2013–15, University of São Paulo, São Inomata, S., and A. Owen. 2014. “Comparative Evaluation of MRIO Frame- Paulo. works.” Economic Systems Research 26 (3): 239–44. The Economist. 2004. “The Great Hollowing-Out Myth.” February 19. Inomata, S., and B. Meng. 2013. “Transnational Interregional Input-Out- Escaith, H. 2014. “Mapping Global Value Chains and Measuring Trade in put Tables: An Alternative Approach to MRIO?” In Sustainability Prac- Tasks.” In Asia and Global Production Networks: Implications for Trade, titioner’s Guide to Multi-Regional Input-Output Analysis, edited by Incomes and Economic Vulnerability, edited by B. Ferrarini and D. Hum- J. Murray and M. Lenzen, 33–42. Champaign, IL: Common Ground mels, 287–337. Mandaluyong, Philippines and Cheltenham, U.K.: Asian Publishing. Development Bank and Edwards Elgar Publishing. Johnson, R. C., and G. Noguera. 2012. “Accounting for Intermediate Pro- Escaith, H., and S. Inomata. 2013. “Geometry of Global Value Chains in duction Sharing and Trade in Value Added.” Journal of International East Asia: The Role of Industrial Networks and Trade Policies.” In Global Economics 86 (2): 224–36. Value Chains in a Changing World, edited by D. Elms and P. Low, 135– Jones, R., and H. Kierzkowski. 1990. “The Role of Services in Production 57. Geneva: World Trade Organization. and International Trade: A Theoretical Framework.” In The Political Fally, T. 2011. On the Fragmentation of Production in the US. Boulder, CO: Economy of International Trade, edited by R. Jones and A. Krueger, University of Colorado-Boulder. 31–48. Oxford, U.K.: Basil Blackwell. Feenstra, R. C., and G. H. Hanson. 1996a. “Foreign Investment, Outsourc- Joskow, P. L. 2003. “Vertical Integration.” In Handbook of New Institutional ing, and Relative Wages.” In The Political Economy of Trade Policy: Economics, edited by C. Menard and M. Shirley, 319–48. New York: Papers in Honor of Jagdish Bhagwati, edited by R. C. Feenstra, G. M. Springer. Grossman, and D. A. Irwin, 89–127. Cambridge, MA: Massachusetts Kawakami, M., and T. Sturgeon, eds. 2011. The Dynamics of Local Learning Institute of Technology (MIT) Press. in Global Value Chains: Experiences from East Asia. New York and Bas- ———. 1996b. “Globalization, Outsourcing, and Wage Inequality.” Ameri- ingstoke, U.K.: Palgrave Macmillan. can Economic Review 86 (2): 240–45. Analytical frameworks for global value chains: An overview • 35 Kimura, F., and M. Ando. 2005. “Two-dimensional Fragmentation in East Ng, F., and A. Yeats. 1999. “Production Sharing in East Asia: Who Does Asia: Conceptual Framework and Empirics.” International Review of What for Whom, and Why?” Policy Research Working Paper 2197, Economics and Finance 14 (3): 317–48. World Bank, Washington, DC. Koopman, R., Z. Wang, and S. Wei. 2012. “Estimating Domestic Content in OECD (Organisation for Economic Co-operation and Development). 2013. Exports when Processing Trade is Pervasive.” Journal of Development Interconnected Economies: Benefitting from Global Value Chains. Paris: Economics 99 (1): 178–89. OECD. ———. 2014. “Tracing Value-Added and Double Counting in Gross Piacentini, M., and F. Fortanier. 2015. “Firm Heterogeneity and Trade in Exports.” American Economic Review 104 (2): 259–94. Value Added.” OECD Working Paper No. 2015/23, OECD, Paris. Koopman, R., Z. Wang, S. Wei, and K. Zhu. 2016. “Tracing Value-Added Porter, M. E. 1985. Competitive Advantage, Creating and Sustaining Supe- and Double Counting in Gross Export: Reply to the Comment by Bart rior Performance. New York: Free Press. Los, Marcel Timmer, and Caaitzen J. de Vries.” Draft Manuscript. Smith, Adam. 1776. An Inquiry into the Nature and Causes of the Wealth of Kuboniwa, M. 2014a. “Trade in Value Added Revisited: A Comment on R. Nations. Reprinted in 1976. Oxford, U.K.: Oxford University Press. Johnson and G. Noguera, Accounting for Intermediates: Production Spence, M. 2011. “The Impact of Globalization on Income and Employ- Sharing and Trade in Value Added.” Discussion Paper Series A, No. ment: The Downside of Integrating Markets.” Foreign Affairs 90 (4): 599, Institute of Economic Research, Hitotsubashi University, Tokyo. 28–41. ———. 2014b. “Fundamental Theorem on the Relationship between Trade Sturgeon, T., P. Nielsen, G. Linden, G. Gereffi, and C. Brown. 2013. “Direct Balances in Value Added and Gross Terms: Amendment.” Discussion Measurement of Global Value Chains: Collecting Product- and Firm- Paper Series A, No.600, Institute of Economic Research, Hitotsubashi Level Statistics on Value Added and Business Function Outsourcing University, Tokyo. and Offshoring.” In Trade in Value Added: Developing New Measures Legros, P., and A. Newman. 2014. “Contracts, Ownership, and Industrial of Cross-Border Trade, edited by A. Mattoo, Z. Wang, and S-J. Wei, Organization: Past and Future.” Journal of Law Economics and Organi- 289–320. Washington, DC: World Bank. zation 30 (S1): i82-i117. Tang, H., F. Wang, and Z. Wang. 2014. “The Domestic Segment of Global Lenzen, M., A. Geschke, M. Abd Rahman, Y. Xiao, E. Dietzenbacher, S. Supply Chains in China under State Capitalism.” Working Paper No. Inomata, K. Kanemoto, B. Los, D. Moran, R. Reyes, H. den Bäumen, 186, Federal Reserve Bank of Dallas Globalization and Monetary Policy A. Tukker, T. Walmsley, T. Wiedmann, R. Wood, and N. Yamano. 2017. Institute, Dallas. —­ “The Global MRIO Lab­ Charting the World Economy.” Economic Sys- Tempest, R. 1996. “Barbie and the World Economy.” Los Angeles Times, tems Research 29 (2): 158–86. September 22. Liu, Y., B. Meng, K. Hubacek, J. Xue, K. Feng, and Y. Gao. 2016. “‘Made in Timmer, M., A. Erumban, B. Los, R. Stehrer, and G. de Vries. 2014. “Slic- China’: A Reevaluation of Embodied CO2 Emissions in Chinese Exports ing Up Global Value Chains.” Journal of Economic Perspectives 28 (2): Using Firm Heterogeneity Information.” Applied Energy 184: 1106–13. 99–118. Los, B., M. Timmer, and G. de Vries. 2015. “How Global Are Global Value Tirole, J. 1989. The Theory of Industrial Organization. Cambridge, MA: MIT Chains? A New Approach to Measure International Fragmentation.” Press. Journal of Regional Science 55 (1): 66–92. Tomiura, E. 2007. “Foreign Outsourcing, Exporting, and FDI: A Productivity ———. 2016. “Tracing Value-Added and Double Counting in Gross Comparison at the Firm Level.” Journal of International Economics 72 Exports: Comments.” American Economic Review 106 (7): 1958–66. (1): 113–27. Ma, H., Z. Wang, and K. Zhu. 2015. “Domestic Content in China’s Exports Wang, Z., S. Wei, X. Yu, and K. Zhu. 2016. “Characterizing Global Value and Its Distribution by Firm Ownership.” Journal of Comparative Eco- Chains.” Working Paper No. 578, Stanford Center for International nomics 43 (1): 3–18. Development, Stanford, CA. Melitz, M. J. 2003. “The Impact of Trade on Intra-Industry Reallocations WTO (World Trade Organization) and IDE–JETRO (Institute of Developing and Aggregate Industry Productivity.” Econometrica 71 (6): 1695–725. Economies–Japan External Trade Organization). 2011. Trade Patterns Meng, B., Z. Wang, and R. Koopman. 2013. “How are Global Value Chains and Global Value Chains in East Asia: From Trade in Goods to Trade in Fragmented and Extended in China’s Domestic Production Networks?” Tasks. Geneva: WTO. IDE Discussion Paper Series No. 424, IDE–JETRO, Chiba, Japan. Xing, Y., and N. Detert. 2010. “How the iPhone Widens the United States Milberg, W., and D. Winkler. 2013. Outsourcing Economics: Global Value Trade Deficit with the People’s Republic of China.” Asian Development Chains in Capitalist Development. Cambridge, U.K.: Cambridge Univer- Bank Institute (ADBI) Working Paper Series No 257, ADBI, Tokyo. sity Press. Yeats, A. J. 1998. “Just How Big is Global Production Sharing?” Policy Miller, R., and U. Temurshoev. 2015. “Output Upstreamness and Input Research Working Paper 1871, World Bank, Washington, DC. Downstreamness of Industries/Countries in World Production.” Interna- Ye, M., B. Meng, and S-J. Wei. 2015. “Measuring Smile Curves in Global tional Regional Science Review. Value Chains.” IDE Discussion Paper No. 530, IDE–JETRO, Chiba, Japan. CHAPTER 2 Recent trends in global trade and global value chains CHRISTOPHE DEGAIN, BO MENG, AND ZHI WANG D uring a long period after World War II, global trade three previous growth periods of the last 20 years. First, there grew several times faster than global GDP. Since 2012, has been a reduction in cross-country production sharing in com- however, the world may have entered a period of plex GVCs during the current economic recovery, contrary to the trade growth that is almost in line with GDP growth. rapid production globalization driven by the growth of complex Is this pattern cyclical or structural? Can value-added trade data GVC activities in previous periods. Second, again unlike the pro- and information on global value chains (GVCs) help explain these duction structure of the previous economic growth periods, the developments? Are GVCs, which involve intermediate products recent economic recovery has been driven mainly by traditional crossing national borders, unwinding? What does this trend trade to satisfy foreign demand and pure domestic production mean for developing countries? This chapter addresses these activities in the United States and several major emerging econo- questions through in-depth analysis of available trade and global mies, such as China. Third, participation in simple GVCs has been input-­output statistics. mixed, rising in some developed economies but falling in most The chapter looks first at the changing patterns of trade in emerging Asian economies. global intermediate goods during the last two decades and ana- GVC production length (the average number of production lyzes the major factors driving these changes. Then it describes stages between primary inputs and final products) has short- the structural change in global production and analyzes its rel- ened, reflecting mainly the declining number of national border evance for the recent global trade slowdown by distinguishing crossings. The production length before and after national GVC and non-GVC activities in GDP and final goods production. border crossings has actually increased, indicating the poten- Last, it discusses the income distribution issues resulting from tial deepening division of labor within national borders despite the development of GVCs and potential contributions to recent the decline in cross-border production-sharing activities. The trade slowdowns and the growing antiglobalization sentiment. It reduced number of national border crossings for production can does this by numerically estimating the “smile curve,” a graphical be observed in all countries, regardless of whether their GDP outline of the value-added potential of each production stage in grew or shrank during this period. a value chain for various industries, based on recently developed Changes in the global production structure are consistent GVC length and participation indexes (box 2.1). with three factors. First is the rising tide of protection around The value-added creation structure that has emerged during the globe after the global financial crisis. Second is the substitu- the slow economic recovery since 2012 is quite different from the tion of domestically produced intermediate inputs for imported Other contributors to this chapter include Xin Li from Beijing Normal University on intermediate goods trade, Xinding Yu from the University of International Busi- ness and Economics on the global macroeconomic circle, and Ming Ye from the Organisation for Economic Co-operation and Development on the smile curves. 37 38 • Measuring and Analyzing the Impact of GVCs on Economic Development BOX 2.1 Identifying global value chain activities with new indicators The rise of global value chains (GVCs) in the past two Applying this new GVC accounting system to the most decades has dramatically altered the world economy. But up-to-date intercountry input-­ output databases (World with the increasing complexity and sophistication of cross- Output Database 2013, 2016; Asian Development Input-­ border production-­ sharing activities, the use of only official Bank Multi-Region Input-Output Database 20161) makes it trade data (such as gross exports and imports) and GDP possible to identify the production length (more or fewer statistics has not revealed the significance and nature of production stages between primary inputs and final goods) changes in the global business cycle. An important reason and degree of participation (simple or complex) in GVCs at is that indicators based on official trade and production country and sector levels. data cannot identify and distinguish which types of trade are GVC activities and which are not, thus making it diffi- Note cult to evaluate the relation between changes in global 1. The Asian Development Bank Multi-Region Input-Output Data- trade and changes in GDP growth. This chapter introduces base data cover a time-series intercountry input-output table, recently developed GVC indicators, which make it possible compiled by the Asian Development Bank in 2016 using the World to decompose a country or sector’s GDP and final goods Input-Output Database and other Asian countries’ input-output production into GVC and non-GVC activities (see box 2.2). tables. intermediate inputs in major emerging developing economies, financial crisis and the dot-com bust, suggesting that about 4% such as China. And third is the technology innovation and reshor- is the steady state for the world economy. GDP growth initially ing that deepened the domestic division of labor for major devel- recovered to about 4% after the global financial crisis but then oped economies, such as Japan and the United States. Whether fell back and stabilized at roughly 2.5%, hinting that structural such changes are temporary or permanent can be determined factors in addition to cyclical factors may be affecting global only as more data become available. economic growth (see figure 2.1). Complex GVCs were the most important driving force for The 2008–09 global financial crisis may have also changed globalization and the growth of global GDP during 1995–2000 the pattern of global trade. Unlike the 1997 Asian financial crisis, and 2000–08. But during 2012–15, complex GVC–related cross- the global financial crisis had large negative impacts on both the border production-sharing activities declined. Industry upgrad- level and the growth of trade. The rapid trade growth from 2001 ing occurred within emerging economies, especially in China, to 2008 contrasts sharply with the much slower growth starting accompanied by a decline in processing trade. Trade protection- in 2009. The decline in intermediate goods trade in 2015 pushes ism has increased because of the slow pace of economic recov- the world economy closer to precrisis levels, thus challenging the ery after the financial crisis. recovery six years after the crisis. There seems to be a clear link Smile curves show that countries and sectors, depending on between the patterns of trade and the global business cycle. What their position and degree of participation, can show very dif- roles have cross-country production sharing and GVCs played in ferent value added and job gains along GVCs. Joining a GVC such a global business cycle? As GVCs involve intermediate goods increases economic efficiency, but this can have a distributional crossing national borders, trade in such goods provides the first impact. piece of information to help understand what is going on. The evolution of global manufacturing trade from 1995 to 2015 exhibits six phases (see figure 2.1). The Asian financial crisis Intermediate trade in manufactured goods and severely damaged domestic demand in several Asian economies global business cycles over 1995–2000, but total manufacturing trade still grew, albeit slowly, and reached a low peak in 2000. Due to the dot-com bust The global economy recently went through three short down- in 2000–01, manufacturing trade declined slightly. In 2001–08, turns centered on the 1997–98 Asian financial crisis, the 2000–01 and accompanying China’s accession to the World Trade Orga- dot-com bust, and the 2008–09 global financial crisis (figure 2.1). nization (WTO) at the end of 2001, total manufacturing trade The global financial crisis precipitated the only global reces- increased substantially. With the 2008–09 global financial crisis, sion, defined by negative GDP growth for a period of at least however, total manufacturing trade dropped sharply. But then in two consecutive quarters. And it seems to have had a struc- 2010–14, it showed a rapid V-shaped recovery, before dropping tural impact on the global economy, both on economic growth again slightly in 2014–15. and on patterns of trade. Global GDP grew at about 4% a year There is no clear indication of which product type contrib- during the precrisis and postrecovery periods of both the Asian utes more to growth in total manfacturing trade, intermediate Recent trends in global trade and global value chains • 39 FIGURE 2.1 Trends in global GDP and manufacturing trade before and after recent economic downturns, 1995–2015 $ (trillions) Percent Rapid growth After-crisis recovery 16 Total manufacturing 5 trade 14 4 12 GDP growth rate 3 Asian 10 financial crisis 2 Dot-com bust Trade in 8 intermediate goods 1 6 0 4 –1 2 Global –2 financial crisis 0 –3 1995 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 Source: For real GDP growth rate, World Development Indicators database; for trade in goods, data on total imports from the Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 4 (2016 edition). or final goods. Trade in intermediate goods contributed more 2015­ —­ showing that European industrial inputs originate essen- than trade in final goods did to the growth of total manufactur- tially from European supply chains. ing trade in 2001–08 and 2009–14 and to its decline in 2000–01 The share of intra-Americas exports in intermediate goods and 2008–09 (table 2.1). Trade in final goods contributed more trade also gradually increased (from 51% in 1995 to 58% in 2015), to the growth of manufacturing trade during 1995–2000 and to while the share of intra-Americas imports in intermediate goods its recent decline in 2014–15. trade drifted downward and reached its lowest point in 2015 (41%, The weight of intraregional exports in trade in intermediate down from 48% in 1995). The shares of manufacturing inputs in and final manufactured goods over 1995–2015 for Europe, the trade within both North and South America are relatively low, but Americas, Asia, and the rest of the world highlights the large those between North America and South America are higher. shares of intraregional linkages among them (figure 2.2). It North American exports of intermediate goods to South America confirms that GVCs are organized mainly at the regional level, accounted for 14% of its total exports of intermediate goods in similar to findings by Baldwin and Lopez (2013) using data from 1995 and 25% in 2015. The share of South American exports to 2009. North America rose from 40% to 50% in the same period. Despite a 6% decrease in the share of intra-Europe trade in The two way intra-Asia trade in intermediate goods fluctuated total European intermediate goods trade during 1995–2015 (due while increasing overall between 1995 and 2015 and reached largely to the emergence of China), intra-Europe trade remained more than two-thirds of total manufacturing trade during substantial in both exports and imports­ —­at around 70% in the period. Similar to Europe, this highlights the sustainable TABLE 2.1 Contribution to the change in global manufacturing trade by trade type, 1995–2015 Percent Contribution to growth of total manufacturing trade Contribution to decline in total manufacturing trade Trade type 1995–2000 2001–08 2009–14 2000–01 2008–09 2014–15 Trade in intermediate goods 45.3 52.0 50.2 79.0 55.4 47.0 Trade in final goods 54.7 48.0 49.8 21.0 44.6 53.0 Source: Authors’ calculations based on data from the Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End- use database, International Standard Industrial Classification, Revision 3. 40 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.2 Evolution of intraregional trade in intermediate and final manufactured goods, 1995–2015 Percent of regional total Intermediate goods 80 Exports 80 Imports Europe Europe Asia Asia 60 60 The Americas The Americas 40 40 Rest of the world 20 20 Rest of the world 0 0 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 Final goods 80 Exports 80 Imports Europe Europe The Americas 60 60 Asia Asia 40 40 The Americas Rest of the world Rest of the world 20 20 0 0 1995 2000 2005 2010 2015 1995 2000 2005 2010 2015 Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 4 (2016 edition). industrial linkages arrangement of “Factory Asia.” About 60% of GVCs are still largely regional, despite the trend of increas- Asia’s exports of final manufactured goods over the period went ing globalization before the recent global financial crisis (see also to extraregional markets, but only about 40% of the Americas’ annex 2.1). Developing economies are increasingly participating exports did, an imbalance that began to change after the global in GVCs through exports and imports of intermediate manufac- financial crisis. Compared with Asia and the Americas, Europe’s tured goods. And some emerging economies are upgrading final goods trade has been more balanced during the last two along GVCs­ —­for example, China tends to export more interme - decades, with a slight decline in intraregional trade from more diate goods to other low-income downstream countries to sup- than 70% in 1995 to about 66% in 2015. port their final goods exports to the global market. Recent trends in global trade and global value chains • 41 Decomposing domestic value added and final 2. Production of value added embodied in the export of goods production into global value chain and final goods and services, or traditional trade. This involves other activities domestic value added produced to satisfy foreign final demand. Domestic factor content is embodied in final goods A country’s GDP by industry can be decomposed into four types that cross national borders for consumption only; therefore, based on whether there are cross-border production-sharing it is very similar to traditional trade, such as “French wine for activities (box 2.2; Wang and others 2017a). English cloth.”2 This is labeled V_RT. The first two production processes described here are pure In the next two production processes, domestic value added domestic production activities. No domestic factor content is used in production activities outside the source country and crosses national borders for production purposes, so there is no is contributed by the source country’s production factors to cross-country production-sharing:1 cross-country production-sharing GVC activities: 1. Production of domestically produced and consumed value 3a. Simple cross-border production-sharing activities, or added, or pure domestic production. This involves domes- simple GVCs. This involves domestic value added crossing tic value added produced to satisfy domestic final demand, national borders for production only once. Value added with no participation in international trade; an example is a is embodied in intermediate exports and used by trading haircut. This is labeled V_D in the figure in box 2.2. partners to produce domestic goods consumed in the direct BOX 2.2 Identifying which types of production are global value chain activities and which are not Global value chains (GVCs) depend on products and serv- The production decomposition method used in this ices that are used as inputs in production processes that report, based on System of National Accounts standards cross national borders, so the first major issue in measur- and adopted from Wang and others (2017a), classifies ing GVCs is separating final and intermediate use in cus- embedded factor content as GVC or non-GVC activities toms trade statistics. But thousands of products are clas- according to whether they cross national borders for pro- sified by customs product codes (such as the U.S. 10-digit duction. Value-added creation is classified as a GVC activ- Harmonized Tariff Schedule), and even within the 10-digit ity only when embedded factor content crosses a national product groups, the heterogeneity is tremendous. So prop - border for production purposes. Domestic and foreign erly identifying final use is not easy. Furthermore, measures factor content in various production activities are distin- of supply chain trade or cross-border production-sharing guished using domestic input-output coefficient matrixes appearing in the literature­ —­such as vertical specialization and import input-output coefficient matrixes in an inter- (Hummels, Ishii, and Yi 2001) and import to produce and country input-­output table, including their local and global import to export (Baldwin and Lopez 2013)­ —­are recursive Leontief inverse matrixes. concepts with pervasive double counting. From a factor content perspective a complete decomposi- To overcome these difficulties, factor content or value- sector’s value added or final goods produc- tion of a country-­ added trade is emerging as the mainstream measure of tion needs to consider both forward and backward industrial cross-border production-sharing activities. Since produc- linkages (Wang and others 2017a). The forward linkage–­ tion factors such as land, labor, and capital are relatively based decomposition views a country-­ sector’s engagement easy to classify, production activities based on factor con- in GVC activities from a producer perspective. It classifies as tent can be classified according to a uniform standard, GVC production activities the portion of GDP created (in a which makes analytical work tractable. When traditional country-­sector) by domestic production factor content that trade dominated international commerce, factors were crosses borders for production at least once. It classifies as less mobile across countries, and factor content embodied domestic production the portion of GDP created by domes- in final goods crossed national borders only for consump - tic factor content that stays within national borders over the tion. In today’s world economy dominated by regional and entire production process. It decomposes values but not global value chains, some production factors directly cross goods. The backward industrial linkage–based decomposi- a national border, such as foreign direct investment, while tion views a country or sector’s engagement in GVC activi- many others still do not but are instead embedded in final ties from a user perspective. It traces all primary factor inputs and intermediate trade flows across national borders. embodied in the final goods produced by the country-­ sector (continued) 42 • Measuring and Analyzing the Impact of GVCs on Economic Development BOX 2.2 (continued) Identifying which types of production are global value chain activities and which are not sector sources and consistently clas- to the original country-­ type 3 is embodied in trade in intermediate goods and can sifies embodied domestic or foreign factor content into GVC be decomposed further into two types. Type 3a is value and non-GVC production activities based on whether they added embedded in intermediate goods absorbed by the have crossed a national border for production. direct importer and in which cross-border production activ- Both ways to decompose production activities in a ities are conducted, but only within the direct importing country-sector pair include the four types described in country (without further border crossing)­ —­thus, these are the text. Factor content or value added in types 1 and 2 simple GVCs. Type 3b is value added that crosses borders involves no cross-border production activities and satisfies at least twice to satisfy domestic and foreign final demand, domestic (type 1) and foreign (type 2) demand. Factor con - respectively­—­thus, these are complex GVCs. These last two tent or value added in type 2 crosses borders once but only types measure cross-country production-sharing activities. for consumption activities since all value-added embodied They exclude domestic value added measured by the first in the good’s intermediate inputs are derived from domes- two types because those production activities are accom- tic sources; therefore, it is traditional trade in value added plished completely within national borders and so can be terms (French wine for English cloth). Factor content in treated as pure domestic production activities. Decomposing GDP and final goods production by country or sector Forward linkage-based: Producer perspective Backward linkage-based: User perspective Which types of GDP production activities belong to GVCs? Which types of final goods production belong to GVCs? A country-sector’s Production of final total value added (V) goods and services GDP by industry by country/sector (Y) 0 1 0 1 In production of final In production In production of Domestic value added Domestic Domestic and foreign goods and services to of final exports intermediate exports in domestically used value added in value added in domestic market directly (V_RT) (V_GVC) final goods (Y_D) final exports (Y_RT) intermediate imports directly (V_D) (Y_GVC) 1 ≥2 1 ≥2 Absorbed by Partner value added Re-export/re-import In production of direct importer in production of Complex GVCs exported goods Simple GVCs domestically used (V_GVC_C) (Y_GVC_C) (V_GVC_S) goods (Y_GVC_S) Source: Adapted from Wang and others 2017a. Note: Numbers in circles are number of border crossings Blue circles represent border crossings for consumption. Orange circles represent border crossings for production. importing country. No indirect exports via third countries countries. Domestic factor content crosses the border at or re-exports or re-imports of the source countries’ factor least twice and is used by the partner country to produce content occur. For example, Chinese value added is embod- intermediate or final product exports either for re-export ied in its steel exports to the United States and used in U.S. to the home country (such as an Apple engineer’s salary house construction. This is labeled V_GVC_S. embodied in an iPhone bought by an American consumer) 3b. Complex cross-border production-sharing activities, or or for re-export to other countries (such as Japanese value complex GVCs. This involves domestic value added that is added embodied in electronic chips installed in Chi- embodied in intermediate exports and used by a partner na-made toy exports to the United States). This is labeled country to produce exports (intermediate or final) for other V_GVC_C. Recent trends in global trade and global value chains • 43 Global value chain production activities in the Pure domestic production activities increased, and cross- global business cycle border production-sharing activities (both simple and com- plex GVCs) decreased in 2001. The four types of value-added creation activities were decom- • Third, as the global economy recovered in 2001 and China posed following the GDP decomposition method proposed by joined the WTO at the end of that year, production globaliza- Wang and others (2017a) and using the recently released World tion resumed in 2002 and accelerated from 2003 until 2008. Output Database (Timmer et. al. 2016). The global produc- Input-­ Up dramatically were GVC production activities as a share tion structures in different types of value-added creation activi- of total global production, as were complex cross-border ties were then plotted for the past two decades (figure 2.3). production-­ sharing activities as a share of total GVC produc- tion activities (figure 2.4). The changing relative importance of different types of • Fourth, the 2008–09 global financial crisis caused a signifi- value-added creation activities in the global business cycle cant setback in production globalization. The share of pure Before the 2008–09 global financial crisis, the dominant trend in domestic production activities rose and the share of all production activities was the decline of pure domestic produc- trade-related production activities fell, especially the cross- tion activities. Although all trade-related production activities border production-sharing activities of complex GVCs (see were increasing, cross-border GVC production-sharing activi- figures 2.3 and 2.4). But unlike the recoveries after the 1997– ties were growing faster than traditional trade production activ- 98 Asian financial crisis and the 2000–01 dot-com bust, the ities. Then four important events affected the global production recovery after the 2008–09 global financial crisis was short. pattern. The production globalization trend not only slowed, but there • First, the financial crisis struck several Asian developing were signs of reversal (see below). countries in 1997–98. GDP growth declined more than 1 per- centage point, but trade in manufactured products was less The changing growth rate of different value-added affected (see figure 2.3; as shown later, the impact was mainly creation activities in the global business cycle on pure domestic production). Some stylized facts emerge from closer analysis of the rate of • Second, the 2000–01 dot-com bust resulted in a minor set- change for the different types of value-added creation activities back for globalization that was similar to the effect of the 2008–09 global financial crisis but on a much smaller scale. FIGURE 2.4 Simple global value chain production activities as a share of total global value chain production FIGURE 2.3 Trends in production activities as a share of activities, 1995–2014 global GDP, by type of value-added creation activity, Percent 1995–2014 70 Asian Dot-com Global Percent financial crisis bust financial crisis 86 Asian Dot-com Global financial crisis bust financial crisis 68 84 82 66 Pure domestic production 80 64 10 Simple GVC Traditional trade production 62 5 Complex GVC 60 1995 2000 2005 2010 2014 0 1995 2000 2005 2010 2014 Source: University of International Business and Economics global value chain indexes derived from the 2016 World Input-Output Database. Source: University of International Business and Economics global value chain Note: At the global level the forward and backward industrial linkage–based indexes derived from the 2016 World Input-Output Database. decomposition methods give the same results. 44 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.5 Nominal growth rates of value-added creation activities during the global business cycle at the global level, 1996–2014 Percent 30 Asian Dot-com Global Slow financial bust financial economic crisis crisis recovery 20 10 0 –10 Pure domestic production Traditional trade production –20 Simple GVC Complex GVC –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Source: University of International Business and Economics global value chain indexes derived from the 2016 World Input-Output Database. year by year for the three growth periods and the three eco- in a reversed pattern. The growth of pure domestic production nomic downturns. activities was slow but steady, faster than that of complex cross- Before 2000–01, growth was slow for all types of value-­ border production sharing activities, which had negative or near added production activities, but GVCs, especially cross-border zero growth. And the growth of simple cross-border activities production-sharing activities of complex GVCs, increased every (those with only one border crossing) increased much faster than year, even during the 1997–98 Asian financial crisis, and began to that of complex GVC activities. Both patterns were completely accelerate toward the end of the period (figure 2.5). Global econ- different from those during the earlier economic recoveries. omies took off in 2003–08 after the 2000–01 dot-com bust, and To minimize the impact of price fluctuations in crude oil and there was a dramatic expansion of GVCs, especially those with bulk commodities (the “commodity super-cycle”) on the nominal complex production-sharing activities. Economic recovery was GDP growth rate in figure 2.5, growth rates were examined at rapid for two years following the 2008–09 global financial crisis. the sector level (figure 2.6). The growth patterns just discussed But the growth rate fell sharply for all types of GDP production in still hold for both forward and backward linkage–based decom- 2012–14, with an obvious slowdown in cross-border production- position of production activities, and there is no significant dif- sharing GVC activities. ference between manufacturing and services. Before the 2000–01 dot-com bust and the 2008–09 global financial crisis, trade-related production activities, especially The new pattern of global production during the complex GVC production-sharing activities grew much faster economic recovery after the global financial crisis than pure domestic production activities. During the crises, pure Signs of a different pattern of global production emerged during domestic production activities were least affected (0.5% in 2001 the slow economic recovery following the quick rebound in 2010 and 1.7% in 2009). While the production of traditional trade was and 2011. At the global level the share of both types of cross- the second-least affected type of value-added creation activity, border production-sharing GVC activities declined, whereas cross-border GVC production activities, especially for complex the shares of pure domestic and traditional trade value-added GVCs, were the most affected, falling 4% in 2001 and 17% in creation activities increased, implying a nearly 3 percentage 2009 for simple GVCs and 6% and 29% for complex GVCs. But point decline in the aggregate GVC participation rate globally the two types of GVC production activities also had the fastest between 2011 and 2015 (figure 2.7). postcrisis recovery. So, despite the difference in magnitude, the To exclude the effects of fluctuations in commodity and crude impact of the two economic crises on types of value-added cre- oil prices, the decomposition is further broken down into four ation activities was similar. broad economic sectors (agriculture, manufacturing, mining, and The impacts of the 2000–01 dot-com bust and the 2008– services) and into both forward and backward industrial linkages. 09 global financial crisis on the global production pattern had The results confirm the relative decline of GVC production activi- many similarities, but the recoveries from the two shocks were ties (figure 2.8). The general pattern of change in the global pro - very different. Although the recovery of production globaliza- duction structure among the four types of value-added creation tion was quick in 2010 and 2011, the growth rate slowed signifi- activities holds for most sectors, except for the forward linkage– cantly after that. Total global GDP still grew during 2012–14, but based decomposition of pure domestic production in services Recent trends in global trade and global value chains • 45 FIGURE 2.6 Nominal growth rates of value-added creation activities during the global business cycle at the manufacturing and services sector level, by forward and backward linkages, 1996–2014 Percent Pure domestic production Traditional trade production Simple GVC Complex GVC Manufacturing 30 Forward linkage Asian Dot-com Global Slow financial bust financial economic 20 crisis crisis recovery 10 0 –10 –20 –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 30 Backward linkage Asian Dot-com Global Slow financial bust financial economic 20 crisis crisis recovery 10 0 –10 –20 –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Services 30 Forward linkage Asian Dot-com Global Slow financial bust financial economic 20 crisis crisis recovery 10 0 –10 –20 –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 30 Backward linkage Asian Dot-com Global Slow financial bust financial economic 20 crisis crisis recovery 10 0 –10 –20 –30 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 Source: University of International Business and Economics global value chain indexes derived from the 2016 World Input-Output Database. 46 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.7 Structural changes in different types of value- and the backward linkage–based decomposition of traditional added creation activities at the global level between trade in agriculture. 2011 and 2015 After 2011, complex GVC activities declined in all G7 coun- Share in 2015 minus share in 2011 (percentage points) tries and in major Asian emerging economies except Viet  Nam (figure 2.9). In backward linkage–based decomposition the 2 changes in simple GVC activities were mixed across countries. At the same time, implying weak domestic demand for major world economies, pure domestic production declined in almost all G7 1 countries except the United States and in China and a few other emerging economies. The share of production for traditional trade, which satisfies foreign demand, increased for all G7 and most Asian emerging economies. 0 To ensure the robustness of these results, the changes in major portions of production activities based on the decomposition of both forward and backward linkages were compared for the four largest economies ranked by GDP­ —­ the United States, China, –1 Japan, and Germany (figure 2.10). This analysis confirms the pro - duction structure changes identified at the aggregate level. The structure of value-added creation during the slow eco- nomic recovery since 2011 is quite different from that during the –2 three previous economic growth periods in the last 20 years. Pure Traditional Simple Complex First, unlike the rapid production globalization driven by the domestic trade GVC GVC growth of complex GVC activities in previous periods, during the production production current economic recovery the pattern was reversed, with less Source: University of International Business and Economics global value cross-border production-sharing activities in complex GVCs. chain indexes derived from the 2016 Asian Development Bank Multi-­ Region Again contrary to the production structure of the previous eco- ­Input-Output Database. nomic growth periods, the current economic recovery has been driven mainly by traditional trade production to satisfy foreign FIGURE 2.8 Structural changes in different types of value-added creation activities at the sectoral level between 2011 and 2015 Share in 2015 minus share in 2011 (percentage points) Forward linkage Backward linkage 20 5.0 Agriculture Agriculture 15 Manufacturing Manufacturing Mining Mining Services Services 2.5 10 5 0.0 0 –5 –2.5 –10 –15 –5.0 Pure Traditional Simple Complex Pure Traditional Simple Complex domestic trade GVC GVC domestic trade GVC GVC production production production production Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. Recent trends in global trade and global value chains • 47 FIGURE 2.9 Structural changes in different type of value-added creation activities between 2011 and 2015 at the country level Share in 2015 minus share in 2011 (percentage points) Forward linkage 6 Emerging Asia G7 Emerging Asia G7 Emerging Asia G7 Emerging Asia G7 4 2 0 –2 –4 –6 Pure domestic production Traditional trade production Simple GVC Complex GVC Backward linkage 6 Emerging Asia G7 Emerging Asia G7 Emerging Asia G7 Emerging Asia G7 4 2 0 –2 –4 –6 Pure domestic production Traditional trade production Simple GVC Complex GVC Emerging Asia China Viet Nam Philippines Rep. of Korea Malaysia Indonesia G7 United Kingdom United States Germany Canada France Italy Japan Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. demand and the growth of pure domestic production in the declined, with the average growth rate of complex GVC activ- United States and several other major emerging economies, such ities declining more. as China. Finally, in the current growth period, participation in The impact of these two factors is even clearer when the 48 simple GVCs has been mixed, increasing for some developed economies in the ADB database are divided into two groups economies but decreasing for most emerging Asian economies. based on positive and negative GDP growth during 2011–15 (figure 2.12). Compared with GDP in 2011, GDP in 2015 increased Factors behind the slow growth of GDP during the recent in 24 economies and decreased in 24 economies. Decomposing economic recovery the total GDP of each group into GVC and non-GVC production GDP growth has been slower during the recent economic recov- activities shows the following: ery than during the previous growth period (figure 2.11) for two • The change in pure domestic production to meet domes- key reasons: tic demand explained the largest portion of the change • Weak domestic demand. The average annual growth rate of in GDP for both groups; all economies with negative GDP pure domestic production (orange points in figure 2.11) declined growth experienced a significant decrease in pure domestic significantly, reflecting weak domestic demand for most econo- production. mies. The growth rate of traditional trade production activities • Traditional trade production increased, while cross-border (blue points) for foreign demand actually grew more rapidly in production-sharing GVC activities decreased. In con- the second period for most of the 48 economies in the Asian trast, during the precrisis period (2003–08) cross-border Development Bank (ADB) data sample than in the first period. production-sharing GVC activities grew much faster than tra- • The slowdown of production globalization. The average ditional trade-related domestic production activities. growth rate of both complex GVC value-added creation activ- • The decline of cross-border production-sharing GVC activities ities (black points) and simple GVC activities (gray points) was driven by complex GVC activities. Simple GVC activities 48 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.10 Structural changes in different type of value-added creation activities between 2011 and 2015, at the country and sector levels Share in 2015 minus share in 2011 (percentage points) Agriculture Manufacturing Mining Services Forward linkage 15 China 15 United States 10 10 5 5 0 0 –5 –5 –10 –10 –15 –15 Pure Traditional Simple Complex Pure Traditional Simple Complex domestic trade GVC GVC domestic trade GVC GVC production production production production 15 Germany 15 Japan 10 10 5 5 0 0 –5 –5 –10 –10 –15 –15 Pure Traditional Simple Complex Pure Traditional Simple Complex domestic trade GVC GVC domestic trade GVC GVC production production production production Backward linkage 15 China 15 United States 10 10 5 5 0 0 –5 –5 –10 –10 –15 –15 Pure Traditional Simple Complex Pure Traditional Simple Complex domestic trade GVC GVC domestic trade GVC GVC production production production production 15 Germany 40 Japan 10 20 5 0 0 –5 –20 –10 –15 –40 Pure Traditional Simple Complex Pure Traditional Simple Complex domestic trade GVC GVC domestic trade GVC GVC production production production production Output Tables. Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-­ Recent trends in global trade and global value chains • 49 FIGURE 2.11 Change in average annual growth rate by type of value-added creation activity between 2003–08 and 2011–15 Change in GDP growth rate (percentage points) 20 Pure domestic production Traditional trade production Simple GVC 10 Complex GVC 0 –10 –20 –30 –30 –25 –20 –15 –10 –5 0 5 10 15 Change in growth rate by type of GDP creation activity (percentage points) Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. declined in countries with negative GDP growth during 2012– based on the decomposition of domestic value added into GVC 15 but kept growing in countries with positive GDP growth. and non-GVC activities (figure 2.13). The units here are the aver- Even in the country group with positive growth in total cross- age number of stages in the production process: that is, at each border production-sharing GVC activities, the production stage, value added is counted as the gross output of some activities of complex GVCs declined. In contrast, complex GVC industry. activities were the fastest growing portion of GDP production This decomposition of production length reveals several in most countries during the precrisis period (2003–08). patterns. First, the breakdown of the production process into Network analysis based on decomposing bilateral gross trade more stages is not a general phenomenon, either within or flows, proposed by Koopman, Wang, and Wei (2014), confirms among countries. The length of domestic production chains the decline of complex GVC activities during 2011–15 (box 2.3). is quite stable, though production chains for traditional trade increased very slightly. The main reason that production chains Why complex cross-border production-sharing activities have lengthened, on average, is that the length of value-added declined during the recovery following the financial crisis production activities that cross national borders increased sig- What drives the recent pattern of global production? The mea- nificantly during 2002–12 for both simple and complex GVCs, sure of total and GVC production length proposed by Wang and but especially for complex GVCs. This pattern changed during others (2017b) can shed light on this question. the recovery period, however. At the global level, production Average production length is a measure of the average time length increased during 2011–15 for all value-added produc- that value added created by production factors employed in a tion activities except complex GVC production, which declined country or sector is counted as gross output in the economy. (figure 2.14), running counter to its pattern in the precrisis When value added is used as an input in a production stage, period. either as a primary or intermediate input, it is counted as gross Second, the decline in production length of complex GVC output where it is used. Therefore, the length of a production activities can also be observed clearly at the sector level (figure chain is the number of times value added is counted as an output 2.15). For almost all country-sector pairs except agriculture and in the production chain from the first time it is used as a primary mining in emerging economies, the production length of com- input until it is absorbed by a final product. It reflects the com- plex GVC activities declined. The decline in manufacturing was plexity of production processes. So the finer the division of labor, more severe in emerging economies than in advanced econo- the longer the production length, which can be computed as the mies, and the opposite was true for the decline in services. The ratio of value added to its induced gross output. production length of simple GVC activities in manufacturing Newly released data from the World Input-Output Database also increased in emerging economies but not as much as in (Timmer and others 2016) can be used to decompose produc- advanced economies. The direction of change is again opposite tion length for the four types of value-added creation activities for services in advanced economies. 50 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.12 Changes in growth of different types of value-added creation activities between country groups with positive and those with negative GDP growth between 2011 and 2015 $ (trillions) Forward linkage Countries with negative GDP growth Decomposition changes of GVC activities 1 0.0 0 –0.2 –1 –2 –0.4 –3 –0.6 –4 –5 –0.8 GDP = Pure + Traditional + Composite Composite GVC = Simple GVC + Complex GVC domestic trade GVC production production 20 Countries with positive GDP growth 0.6 Decomposition changes of GVC activities 15 0.3 10 0.0 5 0 –0.3 GDP = Pure + Traditional + Composite Composite GVC = Simple GVC + Complex GVC domestic trade GVC production production Backward linkage Countries with negative final output growth Decomposition changes of GVC activities 1 0.0 0 –0.2 –1 –2 –0.4 –3 –0.6 –4 –5 –0.8 Final Pure Traditional Composite Composite GVC = Simple GVC + Complex GVC output = domestic + trade + GVC production production 20 Countries with positive final output growth 0.6 Decomposition changes of GVC activities 15 0.3 10 0.0 5 0 –0.3 Final Pure Traditional Composite Composite GVC = Simple GVC + Complex GVC output = domestic + trade + GVC production production Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. Recent trends in global trade and global value chains • 51 BOX 2.3 The evolution of cross-border production sharing in complex global value chains Given the complexity and sophistication of cross-border In 2005 the Asia Pacific community separated into two production-sharing, network analysis can illuminate the evo- groups: the United States maintained connections only with lution of global value chains (GVCs) (box figure). For simplic- Canada and Mexico, while China became the new core of ity, the analysis considers vertical specialization (Hummels, the East Asia + Association of Southeast Asian Nations Ishii, and Yi 2001) as an example and uses network tools community, with strong connections to Japan, the Republic (Zhong and others 2014) to show the topology of foreign of Korea, and Chinese Taipei. value added embodied in manufactured exports (one part In 2011 dramatic changes were evident across the entire of complex GVCs) at the bilateral level. network, and the magnitude of connections strengthened. In 2000 the entire network was dispersed, and the Euro - China became the core of the Asia–Pacific community by pean community (with Germany as the core) had no con- transferring a large portion of foreign value added to other nection with the Asia–Pacific community. The United States countries. The relative distance between the European and was the core of the Asia–Pacific community, with strong Asia–Pacific communities shortened, reflecting that com - connections to Canada, Mexico, Brazil, and Australia. The plex GVCs had developed globally, and more countries United States also had a “chain” connection with Japan joined GVCs through some of the main hubs (the United through the Republic of Korea and had connections with States, China, Germany, and the Republic of Korea). China through Korea and Chinese Taipei. Korea and Chinese In 2015 a recession likely occurred in the complex GVCs Taipei, a sub-hub in the Asia Pacific community, were linked networks; in particular, the North American Free Trade with most Association of Southeast Asian Nations (ASEAN) Agreement, East Asia + ASEAN, and Europe were again economies. isolated. This phenomenon is consistent with the decline of complex GVCs. The typology of foreign value added embedded in bilateral manufactured exports, 2000–15 2000 2005 Denmark Ireland Australia Bangladesh Estonia United Mongolia Japan Finland Austria Kingdom Hungary Latvia India Sri Lanka Estonia Luxembourg Luxembourg Sweden Ireland Chinese China Malaysia France Poland Lithuania Greece Taipei Italy Finland Belgium Bangladesh India Belgium Netherlands Germany Philippines United Kingdom Russian Cyprus Netherlands Thailand Czech Thailand Austria Federation Rep. Korea, Rep. Hungary Romania Malaysia Mongolia Slovak Rep. Turkey France Viet Nam Portugal Philippines Bulgaria Czech Rep. Slovenia Germany Spain China Japan Spain United Mexico States Indonesia Sweden Denmark Chinese Taipei Mexico Malta Portugal Malta Italy Bulgaria Canada Turkey Sri Lanka Korea, United Slovak Lithuania Rep. States Rep. Viet Nam Greece Romania Russian Indonesia Brazil Cyprus Slovenia Poland Brazil Federation Canada Latvia Australia 2011 2015 Chinese Taipei Thailand Mongolia Malaysia Australia Malaysia Bangladesh Philippines Thailand Japan Japan Australia Sri Lanka Philippines Viet Nam Slovak Rep. Ireland China Korea, Brazil Korea, Rep. Malta United IND Rep. Indonesia China United Cyprus States Viet Kingdom Czech Rep. Romania Denmark Malta Cyprus Sri Lanka Romania Nam Indonesia Chinese Austria Italy Bulgaria Austria Brazil Turkey Taipei Russian Hungary Mexico Greece India Federation Canada Slovak Rep. Italy France Sweden Slovenia Mongolia Sweden Portugal Bangladesh Finland Estonia Germany Finland Netherlands Spain Germany United Ireland United Denmark Bulgaria Belgium Kingdom States Greece Turkey Lithuania Poland Estonia Lithuania Netherlands Luxembourg Spain Latvia Poland Hungary Canada Slovenia Luxembourg Portugal France Russian Federation Latvia Czech Rep. Mexico Belgium Asia–Pacific Europe Source: Author’s calculation based on Xiao and others’ 2017 method and data from Asian Development Bank Inter-Country Input-Output Tables. 52 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.13 Trend in production length by different types Caution is required in interpreting these conclusions because of value-added creation activities, world average, 2000–14 official statistics always lag behind the real world economy. For Average number of stages in the production process example, many aspects of new economies, such as cross-border 6 business-­ to-­business e-commerce, are not easy to measure under the current national account system, so the analysis may Complex GVC underestimate cross-border production-sharing activities. How- 5 ever, stylized facts on changes in the global production structure Composite GVC as summarized from the data are consistent with the following Simple GVC factors. 4 • The rising tide of protection around the globe after the global financial crisis. 3 • The substitution of domestically produced intermediate inputs for imported intermediate inputs in major emerging Total economies, such as China. When the domestic division of 2 labor deepens in emerging economies, more intermediate Traditional trade production Pure domestic production inputs are produced domestically, so the domestic value 1 chain lengthens, and cross-border production-sharing activ- 2000 2002 2004 2006 2008 2010 2012 2014 ities may decline as major emerging economies upgrade Source: University of International Business and Economics global value chain along GVCs. indexes derived from the 2016 World Input-Output Database. • Technological innovation and reshoring also deepen the domestic division of labor for major developed economies, such as the United States and Japan. Finally, the main reason for the decline in complex GVC pro - It remains to be seen whether such changes are temporary or duction length is the declining number of national border cross- permanent. ings for production. The production length before and after national border crossing actually increased, indicating the poten- tial deepening division of labor within national borders despite Measuring smile curves in global value chains: the decline in cross-border production-sharing activities. The Creation and distribution of value added and reduced number of national border crossings for production job opportunities can be observed in every country in the ADB database (figure 2.16), regardless of whether its GDP grew or declined during this From a development perspective, GVCs have at least three period. positive aspects. First, by linking into GVCs, firms, especially in FIGURE 2.14 Change in production length for different types of value-added creation activities at the global level between 2011 and 2015 Number of stages 0.15 0.10 0.05 0.00 –0.05 Total Pure domestic Traditional trade Composite GVC Simple GVC Complex GVC production production Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. Recent trends in global trade and global value chains • 53 FIGURE 2.15 Change in production length for different types of value-added creation activities at the sector and economy levels between 2011 and 2015 Number of stages Emerging economies Advanced economies 0.2 Manufacturing 0.2 Manufacturing Services Services 0.1 0.1 0.0 0.0 -0.1 –0.1 –0.2 –0.2 al n n VC VC VC al n n VC VC VC tio tio tio tio t t To To G G G G G G uc uc uc uc e ex e ex od od od od pl pl pl pl m m pr pr pr pr m m Si Si Co Co tic de tic de es es ra ra lt lt m m na na do do tio tio re re di di Pu Pu a a Tr Tr Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. FIGURE 2.16 The decline in the number of border crossings drives the declining length of global value chain production at the country level between 2011 and 2015 Difference in number of border crossings between 2011 and 2015 0.5 Before first border crossing Crossing 0.4 After first border crossing 0.3 0.2 0.1 0 –0.1 –0.2 –0.3 –40 –20 0 20 40 60 80 100 Change in nominal GDP, 2011–15 (%) Source: University of International Business and Economics global value chain indexes derived from 2016 Asian Development Bank Inter-Country Input-Output Tables. 54 • Measuring and Analyzing the Impact of GVCs on Economic Development developing economies, do not need to build the entire course developing countries integrate into GVCs successfully and then of production capacity for a product. Instead, they can use their move up from the low end to the high end of the smile curve? comparative advantage to concentrate on a specific production Answers to these questions are crucial for designing development process or task, which enables them to integrate into the global strategies, industrial policies, and international governance. This economy more rapidly than was possible in the previous industri- section considers several highly fragmented exporting industries alization period (Kowalski and others 2015). Second, becoming a in some countries to show how value added and job opportunities part of GVCs can create more job opportunities (UNCTAD 2013). are created and distributed in GVCs along various smile curves. For example, jobs are created in developing countries through iPhone assembly in China, call center operations in the Philip- China’s information and communication technology pines and India, Nike shoe production in Viet Nam, and automo- industry export-related smile curves: Distribution of value bile and auto parts production in Mexico and Thailand. Third, added and job opportunities GVCs also provide the opportunity for technology transfers or Ye, Meng, and Wei (2015) and Meng, Ye, and Wei (2017) consis- spillovers from developed countries to developing countries tently measure both the value-added gain from GVC participa- through local learning (Pietrobelli and Rabellotti 2010; Kawakami tion and the distance (total production length) between produc- and Sturgeon 2012). ers and consumers. Following their approach, smile curves can However, as mentioned in OECD, WTO, and World Bank be drawn for various GVCs. A good starting point would be the Group (2014, p. 4), “Gains from GVC participation are not auto- iPhone, labeled “Designed by Apple in California; assembled in matic. Benefits of GVCs can also vary considerably depending China.” But it is difficult to isolate the iPhone industry in exist- on whether a country operates at the high or at the low end of ing intercountry input-­ output databases. Here, the first step is the value chain.” Thus, developed and developing countries may to examine the information and communication technology (ICT) face quite different costs and risks in joining GVCs (Baldwin, Ito, industry (industry 14, Electrical and Optical Equipment, in the and Sato 2014). Because of differences in comparative advan- World Input-Output Database) as a proxy to show how, and to tage across countries in GVC participation, developed countries what extent, countries and industries are involved in the GVCs tend to engage in high-end and intangible production activities, for China’s ICT product exports. such as research and development, design, and brand building In figures 2.17 and 2.18 the y-axis shows labor compensation in the prefabrication stages and in after-sales services and mar- per hour (a proxy for technology level or a first-order approxima- keting in the postfabrication stages. Thus, these countries may tion of labor productivity in current U.S. dollars), and the x-axis worry about the hollowing out of their economies as manufac- denotes distance, measured by the total forward linkage–based turing jobs are offshored to low-technology, low-wage countries. production length between global consumers of ICT products Developing countries, in contrast, tend to focus on low-end and and a specific participating industry in the corresponding GVC. tangible production activities such as manufacturing and assem- The 2013 version of World Input-Output Database data are used bly. So, they may worry that they are getting the wrong types of here, covering 41 economies and 35 industries, with the total jobs and that their economies could be locked in to the bottom number of GVC participants (41 × 35 = 1,435) represented as a of the GVC “smile curve,” which presents an outline of the value- circle in the figure. The size of the circle represents the absolute added potential of each production stage in a value chain for value added gained by joining the corresponding GVC (thresh- various industries. old equals 0.1% of the total value added gain). The smooth line The concept of the smile curve was first proposed around is fitted by local polynomial regression–smoothing weighted by 1992 by Stan Shih, the founder of Acer, a technology company its value-added gain, and the shadowed area represents the con- headquartered in Chinese Taipei (Shih 1996). In the personal fidence interval around the smooth line. Using the smile curve computer industry, Shih observed that both ends of the value can lead to an understanding of the participants (countries and chain bring higher value added to the product than the middle industries) of a specific GVC as well as their positions and gains part. The logic of the smile curve has been widely used and in the chain. discussed in the context of GVCs. However, most research has The plotted GVC for China’s ICT exports to the world market focused on product-level case studies rather than the economy- clearly appears as a smile curve; to save space, only values for wide implications. 1995 and 2009 are shown (see figures 2.17 and 2.18). Several styl- Smile curves can help answer numerous questions at the econ- ized facts emerge from these curves: omy level. What is the relationship between developed and devel- • China had the largest value-added gain in this GVC. China’s oping countries in the creation and distribution of value added ICT industry (CHN14) was the most affected industry based and job opportunities in GVCs? Do smile curves exist for country on China’s production of ICT exports through domestic back- or industry GVCs? If yes, have smile curves deepened or flattened ward and intra-industrial linkages. over the years? Have developing countries been locked into the • Several other Chinese domestic industries whose lowest low end of GVCs? Which policies can help countries maintain or labor compensation placed them at the low end of the smile improve their competitiveness on the smile curve? And how can curve also benefited by participating in prefabrication stages Recent trends in global trade and global value chains • 55 FIGURE 2.17 Estimated smile curve for China’s exports of electrical and optical equipment, 1995 Compensation per hour ($) Source: Meng, Ye, and Wei 2017. Note: See annex 2.2 for a key to country abbreviations and sector codes. FIGURE 2.18 Estimated smile curve for China’s exports of electrical and optical equipment, 2009 Compensation per hour ($) Source: Meng, Ye, and Wei 2017. Note: See annex 2.2 for a key to country abbreviations and sector codes. 56 • Measuring and Analyzing the Impact of GVCs on Economic Development of this GVC (CHN12 basic metals, CHN28 financial interme - might otherwise be the case. Thus, China’s ICT value-added diation, CHN20 wholesale, CHN9 chemicals, CHN30 renting activities are naturally located in the middle of this smile curve. of machinery and equipment and other business activities, Another reason is that labor compensation per hour in current CHN2 mining, CHN10 rubber and plastics). This was due to U.S. dollars increased slightly during the target years but not to the fact that most intermediate inputs needed directly and a very high level because of the abundant labor supply in China. indirectly to produce China’s ICT exports were presumed to This explains the almost unchanged position of China’s ICT on come from the Chinese domestic market. the y-axis. • ICT industries in other economies (DEU14, USA14, JPN14, The confidence interval of the smile curve widened consider- KOR14, TWN14), located in the upstream portion of this GVC, ably between 1995 and 2009. This widening was driven mainly also obtained a relatively large part of the value-added gain. by the expanding differentials for labor compensation per hour The main reason is that a majority of transactions involved among GVC participants. The labor compensation of U.S. ICT cross-border, intra-industry trade, given the broad indus- workers (USA14) soared from $18.10 an hour in 1995 to $52.20 try classification in the World Input-Output Database. This in 2009, while for China (CHN14) labor compensation went up result is also consistent with the finding of a case study of the only slightly, from $0.60 an hour in 1995 to $1.60 in 2009. In other iPhone’s supply chain that Japan, the Republic of Korea, Chi- words, the U.S. ICT industry concentrated increasingly on high- nese Taipei, and the United States were the main suppliers tech production of more complex intermediate goods (such as of parts and components for iPhone assembly in China (Xing computer processors), as China took on more tasks using its low- and Detert 2010). skilled labor (such as assembling final products). Also changing • Renting machinery and equipment and other business activi- the confidence interval is the deep involvement of more foreign ties (30) and financial intermediation (28) provided by foreign and Chinese domestic service industries in this GVC, who wanted countries (USA30, USA28, JPN30, JPN28, KOR30, and KOR28) a large share of the value-added gain. are located at the high end of the prefabrication stages of China’s other domestic participating industries are at the low this GVC, with higher labor compensation. ICT products pro- end of the smile curve, but their value-added gain has risen in duced in China, dominated by foreign-invested enterprises, absolute terms (note the change in circle size between 1995 and may need inputs of intermediate services directly imported 2009 in figures 2.17 and 2.18). In other words, China’s domestic from the United States, Japan, and the Republic of Korea. industries, without directly exporting goods to the world market, But this kind of service may also be embodied in the inter- also participate in GVCs by providing intermediate goods and mediate goods produced in the United States, Japan, and services to its exporting industries, like ICT. the Republic of Korea and exported to China to support the production of China’s ICT goods indirectly through various Global value chains can also frown GVC routes. For an inverted smile curve, consider value-added activities in • Postfabrication service industries with higher labor compen- the German auto industry. Given the higher labor compensation sation per hour­ —­such as wholesale (20) and inland transpor- in Germany’s auto industry and lower labor compensation in tation (23) in the United States, Japan, Germany, and France­ both upstream and downstream industries, the entire GVC looks —­ were the main beneficiaries in the postfabrication stage of like an inverted smile curve­ —­a frown (figures 2.19 and 2.20). To this GVC. China’s ICT goods exported to the United States, some extent, this may reflect the successful transition of the Japan, and Germany had to be delivered to their domestic German auto industry from traditional mass producer to mass consumers mainly through those countries’ domestic whole- customizer and to individual design based on digital technology sale and transportation service industries. and artificial intelligence. The mass customized and individual design manufacturing stage accounts for a relatively large por- Changes in China’s information and communication tion of the total value gain, while the traditional high-end design technology industry export-related smile curves over time and sales functions account for only a small portion of total value China’s ICT industry was located at the low end of the GVC in gain and mostly in foreign countries. This is contrary to intuitions 1995 and that position did not change much between 1995 and from the smile curve, in which traditional manufacturing stands 2009­ —­for two likely reasons. One is the relatively high share of at the low end of the GVC, such as China’s ICT exports. But it processing trade in this industry, which can explain China’s posi- could also reflect the ongoing structural change in global GVCs, tion on the x-axis of figures 2.17 and 2.18 measuring production such as the emergence of the customer to manufacturing busi- distance. China’s participation in the GVC at the early stage ness model in several industries. reflects its acceptance of outsourcing tasks such as assembling The most important changes between 1995 and 2009 were iPhones. Compared with the traditional production of ICT prod- the increasing number and variation of foreign participants and ucts, assembly is much more labor-intensive and depends on the increasing length of the curve. In 1995 developed Euro - more foreign parts and components. In addition, export prod - pean countries, the United States, and Japan dominated foreign ucts processed in China are intended for export only (no domes- participants, while in 2009 more countries and industries were tic consumption), so more foreign after-service industries have involved, especially in Eastern Europe, China, and the Repub- been involved in the postfabrication stages in this GVC than lic of Korea. This clearly reflects the increasing diversity and Recent trends in global trade and global value chains • 57 FIGURE 2.19 Estimated smile curve for Germany’s for automobile exports production, 1995 Compensation per hour ($) Source: Meng, Ye, and Wei 2017. Note: See annex 2.2 for a key to country abbreviations and sector codes. FIGURE 2.20 Estimated smile curve for Germany’s for automobile exports production, 2009 Compensation per hour ($) Source: Meng, Ye, and Wei 2017. Note: See annex 2.2 for a key to country abbreviations and sector codes. 58 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 2.21 Labor productivity and income distribution for the United States, 1995–2009 Industrial total 2009 $ Percent of GDP (thousands) Percent of total hours worked Compensation ($ per hour) 100 300 100 80 High-skilled labor 75 250 75 60 Capital compensation High-skilled labor Medium-skilled labor Labor compensation 50 200 50 40 Medium-skilled labor 25 Labor productivity 150 25 20 Low-skilled labor Low-skilled labor 0 100 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Information and communication technology 100 300 100 80 Capital High-skilled labor High-skilled labor compensation 75 250 75 60 Labor compensation Medium-skilled labor Medium-skilled labor 50 200 50 40 Labor productivity 25 150 25 20 Low-skilled labor Low-skilled labor 0 100 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Construction 100 300 100 80 High-skilled labor Capital compensation Medium-skilled labor Labor compensation 75 250 75 60 High-skilled labor 50 200 50 40 Medium-skilled labor Labor productivity Low-skilled labor 25 150 25 20 Low-skilled labor 0 100 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Source: Meng, Ye, and Wei 2017. Recent trends in global trade and global value chains • 59 FIGURE 2.22 Labor productivity and income distribution for China, 1995–2009 Industrial total 2009 renminbi Percent of GDP (thousands) Percent of total hours worked Compensation (renminbi per hour) 100 700 100 20 High-skilled labor High-skilled labor 600 75 75 Medium-skilled labor 15 500 Capital compensation Medium-skilled 400 labor 50 Labor 50 10 compensation 300 25 200 25 5 Labor productivity Low-skilled labor 100 Low-skilled labor 0 0 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Information and communication technology 100 700 100 High-skilled labor 20 Medium-skilled labor High-skilled labor 600 Medium-skilled labor 75 75 15 500 400 Capital 50 compensation 50 10 300 Low-skilled labor Low-skilled labor 200 25 25 5 Labor productivity 100 Labor compensation 0 0 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Construction 100 700 100 20 High-skilled labor 600 75 Capital compensation 75 15 500 Medium-skilled labor High-skilled labor Labor compensation 400 50 50 10 Medium-skilled labor 300 25 200 25 5 Low-skilled labor 100 Labor productivity Low-skilled labor 0 0 0 0 1995 2000 2005 2009 1995 2000 2005 2009 1995 2000 2005 2009 Source: Meng, Ye, and Wei 2017. 60 • Measuring and Analyzing the Impact of GVCs on Economic Development complexity of international fragmentation in Germany’s auto but highly correlated income distribution changes. The United exports. In addition, given the increase in labor compensation States is facing the challenge of job offshoring for medium- and and absolute value-added gain in Germany’s auto industry and low-skilled workers and downward pressure on their wages. the relatively low labor compensation of upstream participants Until 2009, China faced the challenge of unequal income distri- from China, the slope of the entire curve became much steeper. bution between capital and labor, with very low compensation for low-skilled labor. In the United States the big winners appear Labor productivity and income distribution in global to be high-skilled workers and multinational corporations. GVCs value chains enabled them to benefit from the enormous productivity gains in Smile curve mapping can be a touchstone for better understand- developing countries such as China. In China, by contrast, ordi- ing various country and industry positions and value-added gains nary workers benefited. Even at the beginning of the process, from participating in GVCs. The empirical results presented so far factory wages in China were far ahead of rural incomes. And raise an important issue about the relation between economic effi- those wages doubled over 15 years. This is one of the driving fac- ciency and income (or job opportunity) distribution along GVCs tors behind the impressive decline of absolute poverty in China. from the perspective of economic development. Following Meng, But the really big benefits in China accrued to the small number Ye, and Wei (2017), this section considers the ICT industry, an of high-skilled workers and to the owners of capital, including industry dominated by international production sharing, and the foreign investors. construction industry, one of the most domestic-­ oriented indus- In summary, while developed and developing countries may tries (relatively less influenced by international trade). It also con- face quite different costs and risks in joining GVCs, doing so may siders the United States and China as country comparisons since lead to efficiency improvements. But without proper domestic both are active in GVCs. The United States joins GVCs mainly from labor market adjustment policies and universal-coverage safety upstream­ —­ such as exporting complex intermediate goods, parts, nets, as well as better international governance, medium-skilled and components­ —­ or through foreign direct investment outflows and especially low-skilled workers can become the most easily to developing countries. China joins GVCs mainly from downstream­ injured groups in both developed and developing economies. —­ such as exporting assembled final goods­ —­ or through foreign direct investment inflows (before the global financial crisis). U.S. labor productivity (measured as output per person eco- Conclusion nomically engaged, in 2009 national currency) increased rapidly from 1995 to 2009 as the U.S. economy became more efficient, The rise of GVCs has dramatically changed the world economy. with income distribution between capital and labor a relatively After explaining the changing patterns of global GDP and trade stable in their shares in total value added (figure 2.21). But high- growth and the limitation of traditional trade indicators, this skilled workers received more job opportunities, with increased chapter showed how to use the most recent GVC indicators to compensation per hour, while medium- and low-skilled workers decompose country and sector GDP and final goods production lost jobs gradually, with only a small increase in pay for medium- into GVC and non-GVC activities. These new indicators were skilled workers and almost no change in compensation for low- used to identify the production length and degree of partic- skilled workers over the 15 years for which data are available. ipation (simple or complex) in GVCs at the country and sector This phenomenon was more pronounced in ICT industries, while levels. This analysis found that complex GVC-related cross- no significant change was observed in the income distribution border production-sharing activities were the most important between skilled and unskilled workers in construction. In other force driving globalization and the growth of global GDP during words, the rise of GVCs (and technological innovation) may lead 1995–2000 and 2000–08 before declining during 2012–15. to greater efficiency in the U.S. economy but may also leave low- Why did complex GVC activities decline? As industrial upgrad - skilled workers worse off, especially in industries with more out- ing occurred in emerging economies, especially in China, pro- sourcing of production tasks to low-wage developing countries cessing trade declined. Trade protectionism may have increased such as China. due to the slow pace of economic recovery after the 2008–09 For the same industries in China, the evolution is very differ- global financial crisis. And some types of manufacturing jobs ent, but it may be highly correlated with the U.S. phenomenon. may have returned to source countries (reshoring) in response to China’s labor productivity also increased, but more value added technology innovation. was distributed to capital than to labor (figure 2.22). The gain Smile curve analysis was used to show how these new phe- accrued to the capital deployed in China, and that would include nomena affect the distribution of value added and job oppor- multinational corporations involved in GVCs. Given this, and the tunities in GVCs across countries. This analysis shows that coun- very large portion of low-skilled workers in China’s domestic tries and sectors can achieve very different value added and job labor market, the slow growth in compensation per hour for low- gains along GVCs depending on their position and degree of skilled workers should come as no surprise. participation. Joining a GVC increases economic efficiency, but China, with the world’s largest pool of low-skilled labor, meets this can have a distributional impact. The remaining chapters in the United States, the world’s largest capital-abundant coun- this report discuss how to resolve the distribution issue and help try, through GVCs. This intersection generates very different participants from developing countries move up the smile curve. Recent trends in global trade and global value chains • 61 ANNEX 2.1 Shifting roles in global value chains for intermediate and final goods Rising impact of Eastern European economies in intra- machinery and equipment, and motor vehicles. Furthermore, the Europe exchanges of intermediate products comparable size of exports and imports of manufacturing inputs During the last two decades, especially since joining the Euro- between the two economies and Germany ($55  billion and pean Union in 2004, Eastern European countries have developed $48 billion in 2015) suggests the two-way trade usually found in intensive bilateral trade linkages in industrial inputs with other supply chains, with Germany acting both upstream and down- European countries (figure A2.1.1). Joining the European Union stream for its Eastern European partners. and adopting EU regulations have been conducive to the devel- Figure A2.1.2 highlights the divergence of final and interme - opment of these ties within European GVCs. The Czech Repub- diate exports from the Polish motor vehicles industry after 2009 lic, Hungary, and Poland, the largest players in intraregional and reveals a change in the position and role of Poland in Euro- trade in manufacturing inputs among the European economies, pean car production chains. Poland reduced its exports of final accounted for more than 11% of intra-Europe exports in inter- cars while developing a specialization in the upstream produc- mediate goods in 2015, a share that more than quadrupled since tion of medium- and high-technology car parts. The production 1995. and export of final cars are facilitated by other European part- Germany is by far the main trading partner for Eastern Euro- ners, such as the Czech Republic, that increasingly export auto - pean economies in both regional and global value chains, mobiles for various foreign car makers to the European market with most of the trade involving intermediate rather than final ($4.5 billion in 2015, with a 2000–15 average growth rate of 8%). goods. The share of intermediate goods in total Eastern Euro- Poland’s upgrading along the production chain is also con- pean exports to Germany fluctuated around 60% between 1995 firmed through its bilateral trade with Czech industries. Bilateral and 2015. In the same period, Germany accounted for 30% of exports of manufacturing inputs between Poland and the Czech Poland’s exports and 27% of its imports of manufacturing inputs. Republic increased rapidly between 2000 and 2015, at an aver- The shares for the Czech Republic stood at 36% and 32%. The two age annual rate of 13.7%. Poland is a net exporter to the Czech economies’ main trade with Germany is in medium-technology Republic ($7.1  billion exports of manufacturing inputs versus intermediates related to chemicals (excluding pharmaceuticals), $3.5 billion imports), and the Czech Republic’s share in Poland’s FIGURE A2.1.1 Eastern European economies’ trade of FIGURE A2.1.2 Poland’s exports of final and intermediate intermediate manufactured goods with Europe, 1995–2015 goods (motor vehicles) to Germany, 2000–15 $ (billions) $ (billions) 100 6 Intermediate goods Poland 75 5 Exports Czech Republic 50 Hungary 4 25 0 3 25 Hungary 2 Final goods Imports 50 Czech Republic 1 75 Poland 100 0 1995 2000 2005 2010 2015 2000 2005 2010 2015 Source: Organisation for Economic Co‑operation and Development Bilateral Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard In- Trade in Goods by Industry and End-use database, International Standard In- dustrial Classification, Revision 4 (2016 edition). dustrial Classification, Revision 4 (2016 edition). 62 • Measuring and Analyzing the Impact of GVCs on Economic Development total exports of manufacturing inputs doubled during that The Czech Republic’s exports to Poland are quite diversified, period, reaching 8% in 2015. mainly machinery equipment, chemicals, and motor vehicles. In 2015 approximately 80% of the intermediate goods The share of motor parts exports to Poland decreased drastically exchanged between the two countries relied on medium tech- in 2000–15 as Poland took the lead and upgraded in that indus- nologies. Between 2000 and 2015 a significant shift occurred try. In contrast, exports to Poland from the machinery sector between the medium–low and the medium–high technology (medium–low technology) increased 10-fold. inputs exchanged by the two economies. The share of medium– high goods Poland exported to the Czech Republic rose 13 per- Rising role of Mexico in intra-NAFTA trade in centage points while that of medium–low intermediate goods intermediate manufactured goods fell almost proportionally, reflecting a larger upgrade of Polish North America Free Trade Agreement (NAFTA) exports of manu- manufacturing industries in the European production chains than facturing inputs for Canada, Mexico, and the United States were of its neighbor’s manufacturing industries (figure A2.1.3). This is 52.6% in 2015, up from 45.4% in 1995, indicating that supply largely a result of the substantial increase in Poland’s exports to chains have been developed and consolidated. the Czech Republic in motor vehicles and transport equipment. One major feature of the evolution of intra-NAFTA trade is Exports from the two sectors rose by around 27% a year on aver- the growing role of Mexico in the exchange of manufacturing age between 2000 and 2015, when they made up nearly 50% inputs among NAFTA countries (figure A2.1.4). Mexico’s share in ($3.4 billion) of Poland’s exports of manufacturing inputs to the intra-NAFTA trade of intermediate goods increased continuously Czech Republic. between 1995 and 2015, while Canada’s share declined progres- The inverse evolution is observed for medium technology sively and the U.S. share varied within a large range. Although exports from the Czech Republic to Poland. Between 2000 and the United States was the main destination of intra-NAFTA 2015 Czech industries gradually began to specialize in medium– exports of industrial inputs over the period, with a 24.5% share in low technology intermediate goods for export to Poland, and 2015, Mexico rose and surpassed Canada as the second export bilateral exports of medium–high technology goods fell. This destination within NAFTA, receiving 15.3% of NAFTA exports is the typical situation in GVCs: partner countries specialize in of industrial inputs in 2015. Mexico’s GVC-related trade in inter- industrial technologies and tasks that complement each other. mediate goods is essentially with the United States, with 83% of Poland took the lead for the production and export of medium– Mexico’s exports in manufacturing intermediates destined for high technology inputs, while the Czech Republic specialized in the United States in 2015. medium–low technology. Mexico developed its exports of inputs to the United States The share of high-technology intermediates in Czech exports mainly in machinery and transport equipment, which accounted to Poland increased between 2000 and 2015, reaching 4.2% of for 70.4% of Mexico’s exports of intermediate manufactured total inputs sent to Poland, thus reflecting a high level of special- goods to the United States in 2015 ($51  billion for machinery ization for some Czech companies and raising the potential of and $32 billion for transport equipment). Within machinery, the developing foreign market share. electrical machinery and apparatus sector amounted to half the FIGURE A2.1.3 Bilateral exports of Poland and Czech Republic, by manufacturing technology, 2000 and 2015 Share in total exports of intermediate manufactured goods (%) Poland to Czech Republic, 2000 60 Poland to Czech Republic, 2015 Czech Republic to Poland, 2000 50 Czech Republic to Poland, 2015 40 30 20 10 0 Low tech Medium–low tech Medium–high tech High tech Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 3. Recent trends in global trade and global value chains • 63 FIGURE A2.1.4 Intra-NAFTA trade in intermediate manufactured goods, by main destination and origin, 1995, 2005, and 2015 Share in total NAFTA trade (%) 30 1995 2005 Exports 20 2015 10 0 10 Imports 20 1995 2005 30 2015 United States Mexico Canada Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 4 (2016 edition). exported intermediate goods in 2015. Exports of auto parts to intermediate goods with other industrial countries. But it has the United States, $29  billion in 2015, had the fastest growth already become an important supplier of manufactured interme- of all exporting industries to the United States in 2009–15, an diate goods for many lower-wage countries in its neighborhood, annual average of 18.4%. In recent years Mexico supplanted such as Cambodia, India, Malaysia, Thailand, and Viet Nam. Sim- Canada as the main provider of automotive components to the ilar to China, these emerging Asian economies all run surpluses U.S. market. And Mexico’s imports of auto parts from the United States grew at a similar pace (16.8% on average between 2009 and 2015), but at a slightly lower magnitude ($22 billion). FIGURE A2.1.5 Mexico’s trade with the United States in Mexico not only trades car components with the United final and intermediate goods related to motor vehicles, States but also exports final vehicles. In 2015 Mexico became 1995–2015 the world’s seventh largest car producer and the largest in Latin $ (billions) America, with 3.4 million vehicles. A comparison of the growth of Mexico’s car exports to the United States with its trade in car 60 parts (import and exports) finds that the three trade flows fol- lowed similar upward trends, with average increases of 17–19% 50 Exports of final cars between 2009 and 2015 (figure A2.1.5). This highlights the inter- dependency between the export of final cars and the import of parts and accessories when growth in vehicle production inevita- 40 bly leads to an increased demand for imports of car parts and a Exports of wider range of components for assembly companies. car parts 30 Shifts in the division of roles in Southeast Asian global value chains 20 Over the past two decades production networks in Asia have Imports of car parts developed tremendously and have become increasingly fragmented, providing incentives and opportunities to less-­ 10 industrialized economies to join the manufacturing process. For instance, labor-intensive assembling of final goods used to be the major comparative advantage of China, but such assembly is 0 1995 2000 2005 2010 2015 now being transferred out of China as emerging economies from the Association of Southeast Asian Nations have been increas- Source: Organisation for Economic Co‑operation and Development Bilateral ingly integrated into Factory Asia. Trade in Goods by Industry and End-use database, International Standard In- China still runs large trade surpluses in final goods with EU dustrial Classification, Revision 4 (2016 edition). countries and the United States, along with a trade deficit in 64 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE A2.1.6 Evolution of net trade in intermediate and final manufactured goods of Cambodia and Thailand with China, the European Union, and the United States, 2000–15 and 1991–2015 $ (billions) Cambodia Thailand 6 40 Final goods with the European Union and the United States Final goods with the European Union and the United States 30 4 20 2 10 0 0 –10 –2 Intermediate goods with China –20 Intermediate goods with China –4 –30 2000 2002 2004 2006 2008 2010 2012 2015 1991 1995 2000 2005 2010 2015 Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 4 (2016 edition). FIGURE A2.1.7 Evolution of net trade in intermediate and final textile products of Cambodia and Viet Nam with China and the world, 2000–15 $ (billions) Cambodia Viet Nam 1.0 4 Exported final goods to the rest of the world 0.5 Exported final goods to the rest of the world 2 0.0 0 –0.5 –1.0 –2 –1.5 –4 –2.0 Imported intermediate goods with China Imported intermediate goods with China –2.5 –6 2000 2002 2004 2006 2008 2010 2012 2015 2000 2002 2004 2006 2008 2010 2012 2014 Source: Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End-use database, International Standard Industrial Classification, Revision 4 (2016 edition). Recent trends in global trade and global value chains • 65 on manufactured final goods with the United States and EU coun- The level and type of industrial partnership between South- tries (figure A2.1.6). Despite the fact that China is still a global east Asian economies and China depend on their endow- center for the final assembly of numerous manufactured prod - ment and stage of development. As illustrated in figure A2.1.7, ucts, some labor-intensive final assembly activities have shifted low- and middle-income countries, such as Cambodia and to other low-cost economies. Viet Nam, absorb labor-intensive manufacturing inputs from Breaking down Asian trade by sector and end-use accord- China for their production and export. Upper-middle-income ing to Organisation for economic Co-operation and Develop- countries, such as Malaysia and Thailand, and larger economies, ment technology intensity also reveals triangular trade relations such as India, tend to import medium–low or medium–high tech- (figure A2.1.7). Less developed economies, such as Cambodia, nology inputs from China since they have already upgraded in partnered with China in the textile industry, mainly by importing the chain and have the industrial capacity to produce and export low-technology fabrics for manufacturing final goods for EU and high-technology products. U.S. consumer markets. 66 • Measuring and Analyzing the Impact of GVCs on Economic Development ANNEX 2.2 Key to country abbreviations and sector codes TABLE A2.2.1 Country abbreviations AUS Australia DNK Denmark IRL Ireland POL Poland AUT Austria ESP Spain ITA Italy PRT Portugal BEL Belgium EST Estonia JPN Japan ROM Romania BGR Bulgaria FIN Finland KOR Korea, Rep. RUS Russian Federation BRA Brazil FRA France LTU Lithuania SVK Slovak Republic CAN Canada GBR United Kingdom LUX Luxembourg SVN Slovenia CHN China GRC Greece LVA Latvia SWE Sweden CYP Cyprus HUN Hungary MEX Mexico TUR Turkey CZE Czech Republic IDN Indonesia MLT Malta TWN Chinese Taipei DEU Germany IND India NLD Netherlands USA United States RoW Rest of the world Source: Authors’ calculations based on data from the Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End- use database, International Standard Industrial Classification, Revision 3. TABLE A2.2.2 Sector codes 1 Agriculture, hunting, forestry, and fishing 20 Wholesale trade and commission trade, except of motor vehicles and motorcycles 2 Mining and quarrying 21 Retail Trade, except of motor vehicles and motorcycles; repair of 3 food, beverages, and tobacco household goods 4 Textiles and textile products 22 Hotels and restaurants 5 Leather, leather and footwear 23 Inland transport 6 Wood and products of wood and cork 24 Water transport 7 Pulp, paper, paper, printing, and publishing 25 Air transport 8 Coke, refined petroleum, and nuclear fuel 26 Other supporting and auxiliary transport activities; activities of travel agencies 9 Chemicals and chemical products 27 Post and telecommunications 10 Rubber and plastics 28 Financial intermediation 11 Other nonmetallic mineral 29 Real estate activities 12 Basic metals and fabricated metal 30 Renting of machinery and equipment and other business 13 Machinery, not elsewhere classified activities 14 Electrical and optical equipment 31 Public administration and defense; compulsory social security 15 Transport equipment 32 Education 16 Manufacturing, not elsewhere classified; recycling 33 Health and social work 17 Electricity, gas, and water supply 34 Other community, social, and personal services 18 Construction 35 Private households with employed persons 19 Sale, maintenance, and repair of motor vehicles and motorcycles; retail sale of fuel Source: Authors’ calculations based on data from the Organisation for Economic Co‑operation and Development Bilateral Trade in Goods by Industry and End- use database, International Standard Industrial Classification, Revision 3. Recent trends in global trade and global value chains • 67 Notes References 1. This means that the production of final goods and services can be ADB (Asian Development Bank). Inter-Country Input-Output Tables (data- classified as GVC production only when it is combined with foreign base). Mandaluyong, Philippines: ADB factor content (value added) or returned domestic value added. See ———. Multi-Regional Input-Output (MRIO) Tables (database). Mandalu­ backward industrial linkages based on the decomposition in Wang yong, Philippines: ADB. Available at: http://www.wiod.org/otherdb#ADB. and others (2017a) for details. The production of foreign affiliates Baldwin, R., and J. Lopez-Gonzalez. 2013. “Supply-Chain Trade: A Portrait may also be considered a type of GVC activity since current resi- of Global Patterns and Several Testable Hypotheses.” National Bureau dence-based national account rules treat all firms within national bor- of Economic Research (NBER) Working Paper No. 18957, NBER, Cam- ders as domestic firms; therefore, they treat foreign affiliates’ value- bridge, MA. added creation as part of domestic GDP production. No intercountry Baldwin, R., T. Ito, and H. Sato. 2014. The Smile Curve: Evolving Sources of input-output table currently exists that can be used to separate pro- Value Added in Manufacturing. Joint Research Program Series. Chiba duction activities between domestic firms and foreign affiliates. So City, Japan: Institute of Developing Economies–Japan External Trade the GDP decomposition method here may underestimate GVC pro- Organization. duction activities. EC (European Commission). 2013. World Input-Output Database. Brussels: 2. In David Ricardo’s time, exports were 100% domestically produced EC. Available at: http://www.wiod.org/release13. value added, while today, foreign value added is always embodied ———. 2016. World Input-Output Database. Brussels: EC. Available at: even in final product exports from a country; therefore, domestically http://www.wiod.org/database/wiots16. produced value added becomes only a part of exports. However, Hummels, D., J. Ishii, and K. Yi. 2001. “The Nature and Growth of Vertical using the decomposition method applied here, we are still able to Specialization in World Trade.” Journal of International Economics 54 compute the portion of “classical trade” analytically. (1): 75–96. Kawakami, M., and T. J. Sturgeon, eds. 2012. The Dynamics of Local Learn- ing in Global Value Chains: Experiences from East Asia. London: Pal- grave Macmillan. Koopman, R., Z. Wang, and S-J. Wei. 2014. “Tracing Value-added and Double Counting in Gross Exports.” American Economic Review 104 (2): 459–94. Kowalski, P., J. Lopez-Gonzalez, A. Ragoussis, and C. Ugarte. 2015. “Par- ticipation of Developing Countries in Global Value Chains: Implica- tions for Trade and Trade-Related Policies.” Organisation for Economic Co-operation and Development (OECD) Trade Policy Papers No. 179, OECD, Paris. Meng, B., M. Ye, and S-J. Wei. 2017. “Value-added Gains and Job Oppor- tunities in Global Value Chains.” IDE Discussion Paper No.  668, IDE– JETRO, Chiba City, Japan. Mudambi, R. 2008. “Location, Control, and Innovation in Knowledge-Inten- sive Industries.” Journal of Economic Geography 8 (5): 699–725. OECD (Organisation for Economic Co-operation and Development). 2016. Bilateral Trade in Goods by Industry and End-use (BTDIxE) database, Inter- national Standard Industrial Classification of All Economic Activities (ISIC), Revision 3. Paris: OECD. Available at: http://stats.oecd.org/Index.aspx ?DataSetCode=BTDIXE_I3. OECD (Organisation for Economic Co-operation and Development), WTO (World Trade Organization), and World Bank. 2014. Global Value Chains: Challenges, Opportunities, and Implications for Policy. Report prepared for submission to the G20 Trade Ministers Meeting, Sydney, July 29. Pietrobelli, C., and R. Rabellotti. 2010. “Global Value Chains Meet Innova- tion Systems: Are There Learning Opportunities for Developing Coun- tries?” Working Paper Series 232, IDE, Chiba City, Japan. Richard, R., and J. Lopez-Gonzalez. 2013. “Supply-Chain Trade: A Portrait of Global Patterns and Several Testable Hypotheses.” NBER Working Paper No. 18957, NBER, Cambridge, MA. Shih, S. 1996. Me-Too is Not My Style: Challenge Difficulties, Break through Bottlenecks, Create Values. Taipei: The Acer Foundation. 68 • Measuring and Analyzing the Impact of GVCs on Economic Development Timmer, M., B. Los, R. Stehrer, and G. de Vries. 2016. An Anatomy of the World Bank. World Development Indicators (database). Washington, DC: Global Trade Slowdown based on the WIOD 2016 Release. Groningen World Bank. Available at: http://data.worldbank.org/data-catalog Growth and Development Centre Research Memorandum 162. Gronin- /world-development-indicators/. gen, Netherlands: University of Groningen. Xiao, H., T. Sun, B. Meng, and L. Cheng. 2017. “Complex Network Analysis UNCTAD (United Nations Conference on Trade and Development). 2013. for Characterizing Global Value Chains in Equipment Manufacturing.” —­ World Investment Report 2013­ Global Value Chains: Investment and PLoS ONE 12 (1): e0169549. Trade for Development. New York and Geneva: UN. Xing, Y., and H. Detert. 2010. “How the iPhone Widens the United States Wang Z., S. Wei, X. Yu, and K. Zhu. 2017a. “Measures of Participation in Trade Deficit with the People’s Republic of China.” Working Paper 257, Global Value Chain and Global Business Cycles.” NBER Working Paper Asian Development Bank Institute, Tokyo. No. 23222, NBER, Cambridge, MA. Ye, M., B. Meng, and S. Wei. 2015. “Measuring Smile Curves in Global ———. 2017b. “Characterizing Global Value Chains: Production Length Value Chains.” IDE Discussion Paper No. 530, IDE–JETRO, Chiba City, and Upstreamness.” NBER Working Paper No. 23261, NBER, Cam- Japan. bridge, MA. CHAPTER 3 From domestic to regional to global: Factory Africa and Factory Latin America? NADIM AHMAD AND ANNALISA PRIMI G lobal value chains (GVCs) have been drivers of on the surface, the various forms of upgrading have also pre- growth in developed and emerging economies for sented a conundrum to policymakers. The evidence reveals the many years, perhaps best characterized by China’s importance of having access to cheap and efficient imports for experience. Export-driven growth is about gen- exports. In most countries and industries around the world, the erating higher overall value added, employment, and income foreign content of exports has risen considerably over the past through more efficient (and, ideally, higher productivity) pro - two decades. But upgrading can also involve the development duction. The process of generating higher value added is typi- of strong domestic upstream supply chains to exporting firms. cally referred to as upgrading. But the scale of integration within In simple terms, therefore, the policy conundrum is whether to GVCs has varied, with many low-income countries, particularly in emphasize increasing the foreign content or the domestic con- Sub-­Saharan Africa, integrating only at the primary (commodity) tent of exports. part of the value chain, with little diversification or upgrading to This chapter provides a brief overview of upgrading and higher value-added activities. And unlike most other regions­ —­ GVC terminologies, providing insights on interpretability pit- particularly Europe, North America, and Southeast Asia­ —­ Sub-­ falls. It offers evidence of the complementarities between strong Saharan Africa and, to a lesser extent, South America show little domestic supply chains and imports and then demonstrates the intraregional integration. In part, that reflects thick borders that importance of strong regional value chains for integration at add to trade costs, especially in landlocked African economies, a global level. And to illustrate the complementarities, it ends but it also reflects a “spaghetti bowl” of regional trade agree - with examples of broad and targeted policies that countries are ments.1 Better trade facilitation measures­ —­such as establishing implementing for the motor vehicle value chain. a single window for customs clearance, reducing tariffs, improv- ing transport and logistics­—­ are policy levers that governments can pull to deepen regional and global connectivity within value What is upgrading? chains and to facilitate upgrading within firms. The development of the Organisation for Economic Co- The concept of upgrading has its origins in international trade operation and Development–World Trade Organization (OECD– theory, where it indicates a shift toward the production of higher WTO) Trade in Value-Added database, and similar initiatives value goods. But with the increasing international fragmentation such as the World Input-­ Output Database, have transformed of production, the definition has incorporated the notion that the ability to understand integration and assess the benefits of goods are produced through a combination of specific tasks integration into GVCs. But while the literature on GVCs has gen- within a value chain, each generating a proportion of the good’s erated a rich new vocabulary that describes the various forms overall value. This has given rise to the term “moving up the of upgrading, the terms can in turn be misunderstood. At least value chain,” whereby firms upgrade by engaging in a task within 69 70 • Measuring and Analyzing the Impact of GVCs on Economic Development the value chain that extracts a higher share of the overall value of FIGURE 3.1 The smile curve of the global value chain, the good (higher value added), typically referred to as functional 1970s and 2000s upgrading. Value added Functional upgrading Functional upgrading is usually associated with higher labor productivity, since the move to a higher value part of the chain Research and development Services typically (but not always) requires higher skills. Notwithstand- Global value chain in the 2000s ing the high correlation between productivity growth and profit Design Marketing growth, profit remains the primary driver for where firms position themselves in the value chain. From the perspective of the firm, Logistics: purchase Logistics upgrading may involve a move to a part of the value chain where Value chain in the 1970s relative labor productivity is lower but profits are higher. Indeed, Production a firm may take a lower overall part of the value of the final good Value chain at the end of the chain (even if overall sales of the final good Pre-production Pre-production Post-production activities remain unchanged). That is one reason why care is needed when intangible tangible activities intangible deriving messages on upgrading using data on the domestic value-added content of output. Source: Author’s analysis based on Shih 1996 and Gereffi, Humphrey, and Upgrading also has implications for social cohesion and over- Sturgeon 2005. all economic growth. Thus the country perspective on upgrading may differ from the firm perspective, a point often overlooked. Upgrading can result in higher profits and higher employment part of the value chain. Indeed, in many OECD economies, labor creation for the firm but lower overall productivity and lower productivity is typically higher in manufacturing (often the low overall GDP. For a country, however, the driver for functional value part of the smile curve) than in business services (typically upgrading is to increase GDP, as well as labor productivity and at the extreme ends; figure 3.2). employment. Government intervention to ensure that upgrading Functional upgrading goes beyond existing firms moving to occurs in a way that incentivizes the firm to upgrade to a higher different parts of the value chain. In a national context, it can also skilled (higher labor productivity) part of the value chain can thus occur as new firms enter the market, often through new supply affect outcomes. For example, high tariffs on imports of capi- chains driven by lead firms (generally foreign affiliates) that pro - tal goods could push firms to activities with low capital intensity vide (easier) indirect access to international markets for these (typically low labor productivity) and thus with lower domestic new (upstream) entrants. Additional value is thus created through value added in order to maximize profits. upstream domestic supply chains. Lead firms can also encourage Partly for these reasons, care is needed in interpreting the incumbents to upgrade through process and product upgrading “smile curve” developed by Acer’s CEO Stan Shih to illustrate facilitated by technology and human capital spillovers from the the position of Chinese Taipei in the electronics value chain. The lead firms. Typically, this process results in higher overall domes- smile curve accurately describes the decomposition of value of a tic value-added content of exports within a specific value chain given product into the underlying stages (tasks) of production (at as new entrants and incumbents, capitalizing on comparative least for typical manufactured products; figure 3.1). But it does advantages (such as proximity), displace less competitive foreign not follow that firms will necessarily seek to position themselves imports. This process highlights the one-time complementarity in tasks at the extreme ends of the curve, typically those that between importing for exports and eventually creating upstream extract a higher share of the overall value. supply chains. The same holds for the national perspective. Countries clearly The data point to this type of upgrading for textiles in China, would like firms to position themselves at the higher value ends although not unambiguously, as the data may also point to other of the curve, since these are typically the tasks associated with forms of upgrading, including the more general case of func- higher labor productivity, but other considerations are also in tional upgrading.2 For example, the foreign content of China’s play. Countries with a focus on higher social inclusion and lower textile exports fell from 43% in 1995 to 26% in 2011. Some of inequality, for example, may want firms to position themselves that content was displaced by the Chinese textiles industry, but in the higher employment part of the curve, particularly if that by far the biggest contributor was the Chinese service sector, is where they have a comparative advantage and if doing so which displaced upstream foreign services providers (figure 3.3). results in high volumes of output­ —­recall that where to position Indeed, the Chinese textile industry’s contribution to the value along the value chain is as much a volume game (sales) as a ratios of gross textile exports remained broadly steady (suggesting game (share of overall value). In addition, a low share of the over- limited classic functional upgrading in the firm or sector), but its all value of a product does not necessarily equate with low pro- share of domestic value added in textile exports fell from just ductivity. There are examples of specialized and capital-intensive under 50% in 1995 to just over 40% in 2011, as Chinese firms niche activities with high labor productivity in the manufacturing began to occupy other parts of the GVC for textiles. From domestic to regional to global: Factory Africa and Factory Latin America? • 71 FIGURE 3.2 Labor productivity: Manufacturing relative to business services in selected Organisation for Economic Co‑operation and Development countries, 2010 1.50 1.25 1.00 0.75 es y s ria d en m ry k ay ce p. ia ly nd an ar an Ita en ga iu Re at w an st ed m m rla lg nl or St Au ov un Fr en Sw ch er Fi Be N he Sl d H G D ze ite et C N Un Source: Author’s analysis based on Shih 1996 and Gereffi, Humphrey, and Sturgeon 2005. FIGURE 3.3 China’s exports of textiles, by origin of value added, 1991 and 2011 Percent 40 Foreign China 30 20 10 0 1995 2011 1995 2011 1995 2011 1995 2011 Agriculture Textiles Other industry Services Source: Author’s analysis based on Organisation for Economic Co-operation and Development–World Trade Orgnization Trade in Value-Added database. Product upgrading diversification of products (and relative unit value prices) within a Another mechanism for upgrading is by producing higher value particular product group and country. products (product upgrading), as the firm seeks to increase prof- its through sales of higher value products rather than moving to Process upgrading a different part of the value chain. This typically manifests itself Process upgrading typically refers to improved production meth- as higher domestic value-added content through price rather ods that more efficiently transform intermediate inputs into than displacement (of imports) effects, as well as higher labor final products, particularly through innovations in the produc- productivity. The aggregated Trade in Value-Added database– tion process or new technologies (see, for example, Humphrey type measures of trade make it difficult to observe this type of and Schmitz, 2000, 2002, 2004). In theory, this type of upgrad- upgrading. But analyses of detailed merchandise trade statistics ing should also generate higher domestic value content of pro- can provide insights­ —­for example, by looking at the (growing) duction and higher labor productivity, since fewer intermediate 72 • Measuring and Analyzing the Impact of GVCs on Economic Development inputs are needed, especially if the innovations are related to measure of geographic spillovers from neighboring countries knowledge-based capital that allows for higher rent extraction. (the distance-weighted domestic value added in final demand of Again, this can manifest itself as upgrading in upstream domestic partner countries) and a measure of domestic demand linkages, suppliers that respond to competition from foreign producers. which help control for the size of the economy (captured indi- rectly through the domestic value added used for final domestic Intersectoral upgrading consumption).5 Another common form of upgrading is intersectoral, extracting higher value by entering new product value chains. For example, Strong domestic supply chains and strong international Chinese Taipei used its competence in producing televisions to supply chains drive export growth make monitors and eventually (through functional upgrading) to Demand linkages with the domestic economy, proxied through make computers (Humphrey and Schmitz 2002). the domestic value added of a sector in domestic demand, is the most significant determinant of growing domestic value added in exports for both developed and emerging economies Integration for growth: Imports for exports (figure 3.4). But foreign value added used in the production of exports is the second most significant component in developed The ability of firms to organize production processes into dis- economies and the third most significant in emerging econo - crete tasks has transformed the nature of trade and the scope mies, clearly illustrating the complementary nature of imports for firms (and countries) to participate in global production net- for export growth. For example, in emerging economies a 1 per- works. This reorganization of global production has opened centage point increase in the import content of exports trans - opportunities not only for multinational companies and leading lates into roughly a 0.1 percentage point increase in the value exporting firms in advanced economies, but also for firms in of exported domestic value added. Distance to economic activ- emerging and developing economies. Firms in advanced econ- ity (measured as the distance-weighted domestic value added omies are able to outsource to more cost-competitive countries, in the final demand of partner countries) is also an important while emerging and developing economies can enter GVCs by determinant of value added in exports. But it is almost twice as taking advantage of a new tradable commodity in which they important in emerging economies as in developed economies, have comparative advantages­ —­ namely labor. possibly capturing the constraints imposed from less devel- This is intuitive for firms that are able to source cheaper inputs, oped transportation networks. Tariffs, even if low, also have an but concerns remain that the reallocation of resources induced impact in developed economies and marginally (albeit not sta- by such changes may work imperfectly. Although debate con- tistically significant) in emerging economies (see table A3.1.1 in tinues on the benefits of trade for economic growth, the grow- annex 3.1). ing body of evidence points to a positive relationship between increases in imported intermediates and increases in competi- Not all drivers affect emerging and developed economies tiveness and indeed in exports at a broader level. This positive equally association has been demonstrated to occur through two chan- There are also some differences in significant factors between nels: through the use of a greater variety of intermediates (also emerging and developed economies.6 For example, the pro - more competitively priced) and through technology transfers duction of more sophisticated products (even though this may embodied in the imported products, which is also seen in the capture only insertion in processing parts of the value chain) is greater boost to productivity through imports from developed associated with growing domestic value added in exports in economies (Bas and Strauss-Kahn 2014). Similarly, a positive rela- emerging economies only, while skill intensities are significant in tionship has been found between imports and GDP, though with developed economies only, likely reflecting the differing nature gains distributed unevenly across sectors (Kummritz 2014). of integration between the two types of economies (see figure Further evidence of a positive relationship comes from a 3.4). Increases in capital–labor ratios are also an important deter- study using OECD–WTO Trade in Value-Added database data minant in emerging economies but not in developed economies. on foreign and domestic value added embodied in exports that On the surface, this may point to low wages as an important relates changes in domestic value added in exports to struc- determinant of integration in emerging economies, but the result tural and policy factors (Kowalski and Lopez-Gonzalez 2016; see is more nuanced. annex 3.1 for a full description of the variables and data sources).3 Capital–labor ratios can also be loosely proxied by the The study controls for structural determinants using the ratio of inverse of unit labor costs, which in turn reflect the ratio of aver- capital to labor, the intensity of skill, and the country’s relative age compensation costs divided by average productivity.7 The productivity. The policy determinants are the quality of domes- covariance with productivity may partly explain why productivity tic institutions, revealed investment openness, and trade policy on its own was not a significant determinant for emerging econ- stance. To identify the role of foreign inputs, the study takes omies. But the key point is that it is not average wages alone foreign value added (by sector) to produce exports but with a that determine integration in emerging economies but the com- temporal lag to avoid mechanical associations or reverse cau- bination of wages and productivity. And the higher the unit labor sality with the dependent variable.4 The study also includes a costs (the lower the capital–labor ratio), the lower the degree of From domestic to regional to global: Factory Africa and Factory Latin America? • 73 FIGURE 3.4 Significant determinants of a change in domestic value added in exports for developed and emerging economies Domestic demand (log of value) Lagged foreign value added in industry exports (log) Skill intensity Capital–labor ratio (log) Relative output per worker Sophistication of exports Share of foreign direct investment stocks in GDP Tariffs charged (log) Rule of law Developed Emerging Distance to economic activity (log) -0.25 0.00 0.25 0.50 Standardized coefficient Source: Kowalski and Lopez-Gonzalez 2016. Note: The figure shows the standardized coefficients of the determinants of changes in domestic value added in exports across agriculture, manufacturing, and services. The regression results are in table A3.1.1 in annex 3.1. No significance was found for depth of free trade agreements, share of exports covered by free trade agreements, or concentration of exports. export-driven growth (and in turn integration). This result may and surprisingly this is also the case for share of foreign direct in part explain why economies with an abundance of unskilled investment (FDI) stocks in GDP, though that may reflect differ- cheap labor still struggle to integrate in GVCs, despite rising ences in the outward orientation of inward FDI (FDI in manufac- wages in other parts of the world. For example, despite a four- turing to serve export markets as opposed to FDI in services to fold increase in average wages in China between 2000 and 2010, serve domestic markets, including final demand). As for emerg- its unit labor costs (at the economy level) were little changed and ing and developing economies, tariffs on imports also act as a remained significantly below those in most economies in Africa drag on domestic value added in exports at the sectoral level, (table 3.1). And Sub-­ Saharan African economies that generally including the services sector, reflecting that in most economies saw little change in average wages between 2000 and 2010 still upstream services content accounts for around a third of the had high unit labor costs relative to other countries (figure 3.5). value added of manufactured exports. Puzzlingly, increasing the share of exports covered by free trade agreements does not Nor do the drivers affect all sectors equally appear to lead to increased exports of value added. A similar pattern emerges at the sectoral level, but the impor- tance of foreign inputs for manufacturing exports is more stark Domestic supply chains are an important stepping stone (figure 3.6), while domestic (demand) linkages are most impor- for improving participation in global value chains tant for the services sector (reflecting the importance of inte - An important result of Kowalski and Lopez-Gonzalez’s (2016) gration by services as upstream suppliers to manufacturers).8 study relating changes in domestic value added in exports to Perhaps not surprising, given the limited role of foreign interme- structural and policy factors is the co-incidence of domestic diates in services, foreign inputs are less important for services. demand (a proxy for internal domestic supply chains) and for- Structural factors such as relative output per worker are also eign inputs in export-driven growth, highlighting the comple - important, but skill intensity does not appear to be significant mentarity of the two for export growth. Further evidence of for services, though it is difficult to discount the possibility that this complementarity is provided by Beverelli and others’ (2016) this may to some extent reflect an aggregation effect that cannot study of the relationship between upstream domestic supply differentiate between underlying high-skilled workers (such as chains and the foreign value added of exports (as a measure of software developers) and low-skilled workers (such as cleaners) GVC participation). They found a robust relationship between within the industry grouping, as well as the different nature of domestic value chains and future participation in GVCs. The the underlying integration process. study estimated that a 1 percentage point increase in domestic Promoting the creation of more sophisticated products has integration raises GVC backward integration by 0.5% over the a positive effect only on manufacturing activities (not services), short run. 74 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE 3.1 Average wages and unit labor costs in manufacturing in selected developing and emerging economies, 2000 and 2010 2000 2010 Unit labor cost Unit labor cost Average wages (ratio of average wages Average wages (ratio of average wages Region and country (U.S. dollars) to GDP per capita) (U.S. dollars) to GDP per capita) Sub-­Saharan Africa Burundi — — 3,261 14.9 Cameroon 3,088 5.3 — — Ethiopia 771 6.3 807 2.4 Ghana 1,832 4.9 — — Kenya 2,118 5.2 2,854 3.6 Malawi 436 2.8 2,045 5.7 Mauritius 3,254 0.8 6,285 0.8 Senegal 3,680 7.8 6,450 6.5 South Africa 7,981 2.6 12,331 1.7 Tanzania 2,296 7.5 1,581 3.0 North Africa Egypt 2,028 1.3 3,453 1.2 Morocco 4,123 3.2 6,654 2.4 Tunisia 4,066 1.8 5,455 1.3 Latin America Brazil 5,822 1.6 10,918 1.0 Colombia 4,096 1.6 4,680 0.8 Mexico 8,048 1.2 7,310 0.8 Asia Bangladesh — — 680 1.6 China 1,016 1.1 4,770 1.1 India 1,356 3.0 2,619 1.8 Indonesia 929 1.2 1,897 0.6 Malaysia 4,405 1.1 6,548 0.7 Viet Nam — — 1,727 1.3 Eastern Europe Czech Republic 3,964 0.7 12,673 0.7 Latvia 3,689 1.1 9,191 0.8 Poland 5,829 1.1 10,162 0.8 Source: Ceglowski and others 2015. —­is not available. Note: ­ Regional value chains as enablers of global level. But another important takeaway message, often integration into global supply chains overlooked in the debate on GVCs, is that it matters where a country is located: it matters who its trading partners are, espe- The analysis so far offers two takeaway messages for countries cially how integrated the partners are into regional and global looking to drive growth through integration in global value value chains, and how far the country is from poles of economic chains. The first is that imports can be an important driver of activity (including markets). The composition of firms within an export growth. The second is that strong domestic supply chains economy also matters. In most economies, particularly emerging provide an important launching pad for integration at a more economies, the majority of firms are small or medium size. The From domestic to regional to global: Factory Africa and Factory Latin America? • 75 FIGURE 3.5 Evolution of unit labor costs and average wages, 2000 to 2010 Unit labor costs (ratio of average wages to GDP per capita) 2000 8 6 4 Morocco South Africa Tunisia 2 Latvia Colombia Brazil Malaysia Poland Mexico Egypt China Czech Republic 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 2010 8 6 4 Morocco 2 South Africa Egypt China Tunisia Brazil Latvia Mexico Czech Republic Colombia Malaysia Poland 0 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 Average wages ($) Source: Ceglowski and others 2015. Note: Unit labor costs are the ratio of average wages to per capita GDP. evidence points strongly to a lower likelihood of direct engage- provides strong evidence of increased integration in GVCs in most —­ ment in trade the smaller the firm, reflecting additional barriers­ economies based on foreign value added in exports, backward lower probability of financing, lower economies of scale, higher linkages, forward linkages, domestic value added in other coun- relative fixed costs in dealing with regulation, and so on. tries’ exports, and standard GVC participation indices (figure 3.7). The fact that geography matters, coupled with the fact that strong domestic supply chains are important enablers of integra- —­ Intraregional integration is unequal­ and poor in Africa tion into global supply chains, leads the debate toward regional and Latin America value chains as enablers. Currently the best statistical tool used Although the coverage of countries in the OECD–WTO Trade to measure GVC integration comprehensively is the OECD–WTO in Value-Added database reflects a significant proportion of Trade in Value-Added database, which has data on 63 countries. It world output and world trade, it remains patchy in many regions, 76 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3.6 Determinants of change in domestic value added in exports, by sector Lagged foreign value added in exports (log) Domestic demand Skill intensity Relative output per worker Sophistication of exports Capital labor ratio (log) Share of foreign direct investment stocks in GDP Index of depth of free trade agreement Share of exports covered by free trade agreements Concentration of exports Rule of law Agriculture Manufacturing Tariffs charged (log) Services Distance to economic activity -0.2 0.0 0.2 0.4 Standardized coefficient Source: Kowalski and Lopez-Gonzalez 2016. Note: The figure shows the standardized coefficients of the determinants of changes in the domestic value added in exports across agriculture, manufacturing, and services. The regression results are in table A3.1.1 in annex 3.1. FIGURE 3.7 Changes in measures of integration into global value chains between 1995 and 2011 for the 63 economies in the Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database Change (percentage points) 30 Change in backward participation Change in forward participation Change in backward + forward participation 20 10 0 –10 –20 rla ina Ki Isr m Po do el R rtu m os a l itz R ia Bu rla a lg nd Fr aria L ce or ia So Sl Bra y ut ove zil Ar A ia n a nl a n A p d de tra n ra lia st n C un ria Sa Ge m ia ud rm bia Vi bo an ei T t Ndia D ha am M ra y In ala bia ze R g D h R p. m . C rk C Ja hile sa d Tu lam la y se In d Ta dia Re ore ng ei of Ic Re y th el p. w d ld ro a t ia Ze hina g L a a N ng, uan d et C ia Si Cy nds C po s an re ili la a pp nd Un S ex es ite w ico Be tat n lg es ne ia Lu G It ia xe re aly o b e C om ga en ep a u a i A an Po rke a, ar e ic ge fric Fi tin C alt C i Ph Ire ad Fe us ai Au tio S e Sl m ec an C vak our us an Ko ith lan n e an N atv h n Es at ew on Sw ta n ol is do ys s ng a K u ip a iu or c e ng pr M in w an am p d ed he h M ar il S o T e H ne d ite N hi st ia C Un un on ss Ru Br H Source: Author’s analysis based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. From domestic to regional to global: Factory Africa and Factory Latin America? • 77 notably Africa and Latin America. That limits its ability to pro- The stylized fact that a limited ability to integrate has gone vide insights on the nature of regional value chains. And where hand in hand with limited income convergence can also be seen evidence does exist, it points strongly to very weak regional par- in measures of economic complexity, which provide a broad ticipation (intraregional trade) outside Asia, Europe, and North indication of a country’s upgrading (relative to other countries; America relative to extraregional trade (figure 3.8). Hausmann and others 2011).9 Most African economies show little For regions not covered, notably for Africa, conventional change in ranking on these measures over the last two decades (gross trade) statistics provide similar messages of weak regional (where 1 indicates the highest complexity and 124 the lowest). integration (figure 3.9). Notable exceptions are North African economies, reflecting, at Moreover, where integration does occur, it is very much at the least in part, their geography­ —­ their proximity to European mar- low-value end of GVCs for low-income countries, with exports of nat- kets and value chains (figure 3.12). ural resources a significant form of integration and imports of inter- The pattern is similar in Latin America and the Caribbean, mediate parts generally satisfying domestic demand (figure 3.10). with gains generally observed only in economies that improved their integration into North American value chains, such as Costa Poor integration is stifling convergence Rica (figure 3.13). Many countries that have integrated into GVCs have found them- This contrasts starkly with countries in Asia and former transi- selves “captive participants,” experiencing difficulties in scaling tion economies in Eastern Europe (figure 3.14). up as a result of being locked into low-value tasks or as providers There is a positive correlation between change in economic of commodities at the beginning of the value chain. With seem- complexity ranking over the last two decades (where a negative ingly limited ability to upgrade or diversify, they are often hos- entry reflects greater economic complexity) and change in the tage to price competition that keeps wages low or to the vaga- foreign content of exports for countries with a more than 5 per- ries of commodity prices (the resource curse). And this low-value centage point change in the foreign content of exports (figure form of integration appears to have, at least in part, inhibited 3.15).10 But for countries with a smaller change in foreign content, greater improvement in economic convergence and stymied the data point to a negative correlation. the upgrading process (figure 3.11). Most African economies, for Important here is the relative performance of countries in example, have experienced only a 0–2 percentage point increase regions not well covered in the Trade in Value-Added data- in GDP per capita in the last two decades relative to U.S. levels base and how representative they may be for their regions as (although in some cases this amounts to doubling relative GDP a whole: Argentina, Brazil, Chile, Colombia, and Costa Rica for per capita and sometimes even more, as in Angola’s case). Latin America and Saudi Arabia, South Africa, and Tunisia for the FIGURE 3.8 Intraregional and extraregional value chains, by region, for the 63 economies in the OECD–WTO Trade in Value-Added database, 1995 and 2011 Foreign value added content of gross exports as percent of total value added in exports 20 Extraregional Intraregional 15 10 5 0 1995 2011 1995 2011 1995 2011 1995 2011 Asia European Union 28 Latin America United States and Canadaa Source: Author’s analysis based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. China; Note: The regional classification is limited to countries in the Trade in Value-Added database. Asia includes Brunei Darussalam; Cambodia; Hong Kong, ­ India; Israel; Japan; the Republic of Korea; Malaysia; the Philippines; Saudi Arabia; Singapore; Chinese Taipei; Thailand; Turkey; and Viet Nam. Latin America includes Argentina, Brazil, Chile, Colombia, Costa Rica, and Mexico. a. A significant share of extraregional trade reflects trade with Mexico.  78 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3.9 Extraregional and intraregional trade in intermediates, 2014 Percent of total intermediates trade 100 Extraregional Intraregional 75 50 25 0 Africa Asia Asia European Union 28 Latin America Latin America (excluding China) (excluding Mexico) Source: Author’s analysis based on data from the UN Comtrade database for Africa and Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database for other regions. Note: Trade in intermediates is defined as total trade (the sum of gross exports and gross imports) in the sectors classified as primary and processed food and beverages destined mainly for industry, other industrial supplies, fuels and lubricants other than processed motor spirits, and parts and accessories for capital goods and transport equipment. The composition of macro-geographic (continental) regions follows the UN methodology (http://unstats.un.org/unsd/methods/ m49/m49regin.htm). Countries in the Trade in Value-Added database, by region, are as follows: Asia includes Brunei Darussalam; Cambodia; Hong Kong, China; India; Israel; Japan; the Republic of Korea; Malaysia; the Philippines; Saudi Arabia; Singapore; Chinese Taipei; Thailand; Turkey; and Viet Nam. Latin America includes Argentina, Brazil, Chile, Colombia, Costa Rica, and Mexico. FIGURE 3.10 Composition of trade in low-income countries by intermediate and final goods, 2000–13 $ (billions) 140 Exports Primary intermediate goods 120 Processed intermediate goods Final goods 100 Imports Primary intermediate goods 80 Processed intermediate goods Final goods 60 40 20 0 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 Source: Author’s analysis based on UN Comtrade database. From domestic to regional to global: Factory Africa and Factory Latin America? • 79 FIGURE 3.11 Convergence in income per capita and exports in Africa relative to the United States between 1995 and 2014 2014 export volume (index, 1995 = 1 relative to the United States) 5 Mozambique 4 Burkina Faso Chad 3 Ethiopia The Gambia 2 Angola Mauritius 1 Nigeria South Africa Algeria 0 –5 –4 –3 –2 –1 0 1 2 3 4 5 6 7 8 9 10 Change in GDP per capita relative to the United States (percentage points) Source: Author’s analysis based on World Development Indicators database 2016. FIGURE 3.12 Economic complexity rankings in Africa, 1995 FIGURE 3.13 Economic complexity rankings in Latin and 2014 America and the Caribbean, 1995 and 2014 Rank (1 = highest, 124 = lowest) Rank (1 = highest, 124 = lowest) Tunisia 1995 Mexico 1995 Egypt 2014 Panama 2014 Senegal Morocco Uruguay Kenya El Salvador Cameroon Costa Rica Zimbabwe Uganda Brazil Côte d’Ivoire Colombia Madagascar Jamaica Tanzania Mozambique Chile Gabon Guatemala Ethiopia Honduras Algeria Ghana Paraguay Malawi Cuba Congo, Rep. Trinidad and Tobago Libya Mauritania Peru Yemen Nicaragua Guinea Ecuador Sudan Venezuela, RB Nigeria Angola Bolivia, Plurinational State of 0 25 50 75 100 124 0 25 50 75 100 124 Source: Hausmann and others 2011. Source: Hausmann and others 2011. Note: Rankings are among 124 economies, with a ranking of 1 reflecting the Note: Rankings are among 124 economies, with a ranking of 1 reflecting the highest complexity and 124 the lowest. highest complexity and 124 the lowest. 80 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3.14 Economic complexity rankings in Asia and average in 2014, countries in Latin America and the Caribbean Central Europe, 1995 and 2014 (except for Mexico) exported 10 times more products within the Rank (1 = highest, 124 = lowest) region than to China, 7 times more to the European Union, and 2 1995 times more to the United States (table 3.2). Korea, Rep. Further differences emerge in Latin America and the Carib- 2014 China bean by the size of exporting firms. Small and medium-size firms (almost 15,000) export predominantly within the region (figure Malaysia 3.17). Firm-level customs data show that the number of large firms Philippines that exported globally fluctuated between 500 and 1,000 in 2011 (in Bolivia, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, and Thailand Uruguay). However, although increased exports by small and medi- India um-size firms can be an important driver of improved regional inte- gration (and then global integration) as well as of improved inclu- Indonesia siveness, the contribution of their exports remains limited because Viet Nam of their low share in the overall value of exports (around 6% in 2011, much lower than in more developed regions such as Europe).11 And Cambodia given the high concentration of commodity exports, the contribu- Czech Rep. tion of smaller firms as upstream suppliers to larger firms integrated within existing value chains is also likely to be limited, certainly when Hungary compared with other regions (OECD and World Bank 2015). Slovak Rep. Enhancing regional trade agreements for regional trade 0 25 50 75 100 124 A surprising result from the analysis by Kowalski and Lopez-­ Source: Hausmann and others 2011. Gonzalez (2016) was the negative relationship between the share Note: Rankings are among 124 economies, with a ranking of 1 reflecting the of exports covered by free trade agreements and value added highest complexity and 124 the lowest. in exports. A number of factors might explain this. For example, in emerging economies most extraregional trade is in commod - ities, so diverging price effects could play a role. For example, Middle East and Africa. Costa Rica and Tunisia, for example, have higher values of commodity exports to countries with which the seen a marked improvement in both integration and economic exporting country has no free trade agreement could create an complexity rankings, but this largely reflects integration though inverse correlation. In addition, the scope and depth of regional North American and European production chains and their prox- trade agreements matter. In some regions, regional trade agree- imity to those chains and markets. The same access for other ments may have only limited benefits, if they are not also part countries in their regions may not be as straightforward. of more comprehensive liberalization and facilitation policies, including multilateral and unilateral efforts. But there is scope for improved intraregional integration Despite a proliferation of free trade agreements and regional Of particular interest is the technological content of exports by trade agreements, nontariff barriers to trade remain high in region. As expected from interpreting figures 3.12 and 3.13, the Africa. Trade costs within Africa are only slightly lower than trade technological content of exports is generally lower in Africa and costs with the rest of the world, at 313–337% in ad valorem equiv- Latin America than in other regions (figure 3.16), explaining in alent (UNECA 2013). Indeed, as many as 10 African countries large part the poor performance in their economic complexity have higher trading costs with their intraregional partners than and, potentially, their economic convergence. But intraregional with the rest of the world. And in the median African country, integration, where it does occur, is typically in higher value (tech- document preparation to export or import takes 25% more time nology) production. Intraregional trade is a small share of activity than in the rest of the world, while customs procedures are 30% in these two regions, but it does point to the potential to improve more expensive (ESCAP and World Bank Trade Cost Database). regional integration by accelerating structural transformation and In the Asia–Pacific region, formal trade agreements may not to the ability of intraregional integration to serve as a launching have been a crucial driver of GVC trade at the intraregional level pad for greater global integration in higher value products. because economies are already connected through the regional For example, despite Africa’s abundance of primary com- production networks of multinational corporations. In addition, modities, they also account for an important share (35%) of the the effectiveness of regional trade agreements for exports continent’s imports, indicating missed opportunities for sourc- appears to depend on the level of development of the exporting ing commodities internally. Intra-Africa trade has grown only and importing economies. For example, regional trade agree - modestly, from 11.0% of total exports in 2002 to 15.7% in 2014, ments appear to have a greater impact for low-income countries emphasizing its considerable unrealized potential. The potential when exporting to high-income countries than when exporting is similar in Latin America and the Caribbean. On (unweighted) to another low-income country. From domestic to regional to global: Factory Africa and Factory Latin America? • 81 FIGURE 3.15 Correlation of change in economic complexity rankings and change in foreign value-added content of exports between 1995 and 2014 Change in foreign value added content of exports With foreign value-added increase of 5 percentage points or more 30 Cambodia Korea, Rep. 20 Hungary Turkey Poland Thailand Viet Nam India Bulgaria Malaysia 10 Australia Tunisia Greece Latvia South Africa Costa Rica Portugal 0 –60 –40 –20 0 20 40 With foreign value-added increase of less than 5 percentage points 5 Mexico Chile Brazil Romania Argentina Lithuania Russian Federation 0 Indonesia New Canada Saudi Arabia China Colombia Croatia Zealand Hong Kong, Estonia China Norway –5 Philippines –10 –60 –40 –20 0 20 40 Change in economic complexity ratio Improvement in economic competitiveness ranking, 1995–2014 Source: Hausmann and others 2011. Another possibility is that the multitude of overlapping free National experience with value chain trade agreements and regional trade agreements impede upgrading and integration: Automotive sector rather than promote exports by adding to the complexity of managing trade, particularly for small and medium-size firms, There is no single solution to GVC policymaking. Country-­ for which barriers to entry are already high. In general, higher specific factors shape how countries integrate into GVCs: where intraregional trade is associated with fewer overlaps of regional they are located, the size and relative income of their neighbors, trade agreements. For example, Europe, with the highest level their relative income, the structure of their economy, the scope of intraregional trade, also seems to have the simplest structure, and nature of trade agreements, and endowments of physical whereas Latin America and Africa, with poor intraregional trade, and human capital, to name but a few. So GVC policymaking have the most complex arrangements (figure 3.18). requires a whole supply chain approach, which is largely country 82 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3.16 Regional exports by share of technological intensity, 2014 Percent Africa 100 Commodities Commodity-based manufactures Low-tech 75 Medium-tech High-tech 50 25 0 Asia Latin Within the OECD America region Asia Asia (excluding China) 100 100 75 75 50 50 25 25 0 0 Africa Latin Within the OECD Africa Latin Within the OECD America region America region Latin America Latin America (excluding Mexico) 100 100 75 75 50 50 25 25 0 0 Africa Asia Within the OECD Africa Asia Within the OECD region region Source: UN Comtrade database. Note: The figure shows the cumulative total exports for each region between 2013 and 2014. The technological classification follows Lall 2000 and Aboal and others 2015. OECD group refers to members up to the end of 1993. From domestic to regional to global: Factory Africa and Factory Latin America? • 83 TABLE 3.2 Number of exported products by destination FIGURE 3.17 Number of exporters in Latin America and the from countries in Latin America and the Caribbean, 2014 Caribbean by main export destination, 2011 Destination region or country Small and medium-size firms Latin Latin America America 16 and European United 14 Source country Caribbean China Union States 12 Antigua and Barbuda 17 — 4 11 10 8 Argentina 3,358 359 1,488 1,333 European 6 United States Bahamas 17 4 8 45 Union 4 and Canada 2 Barbados 906 40 259 475 0 Belize 75 5 26 85 Bolivia, Plurinational State of 566 51 278 273 Brazil 3,779 1,402 2,937 2,786 Chile 2,932 327 1,472 1,291 Rest of the world ASEAN+6 Colombia 3,176 277 1,375 1,762 Costa Rica 2,791 273 1,033 1,690 Large firms Dominican Rep. 2,281 203 1,223 2,151 Latin America Ecuador 1,883 109 940 1,052 1.0 El Salvador 2,442 44 466 1,149 0.8 Guatemala 3,198 113 637 1,361 0.6 Guyana 471 29 94 294 European 0.4 United States Honduras 1,485 757 682 1,531 Union 0.2 and Canada Jamaica 470 38 252 337 0 Mexico 3,756 1,401 2,830 4,052 Nicaragua 1,837 52 304 923 Panama 289 32 66 156 Paraguay 968 63 408 287 Peru 3,034 249 1,599 1,772 Rest of the world ASEAN+6 Saint Lucia 355 6 188 848 Source: Economic Commission for Latin America and the Caribbean using Uruguay 1,367 108 786 453 data from national customs offices. Venezuela, RB 920 16 168 618 Note: Data cover exports from Bolivia, Chile, Costa Rica, Ecuador, El Salvador, Guatemala, and Uruguay. Source: Economic Commission for Latin America and the Caribbean using data from the UN Comtrade database. Note: A product is defined at the six-digit code level in the Harmonized and by tasks (research and development, design, testing, and System. assembly and production), with significant upstream chains. In  — is not available. all regions the automotive industry contributes no more than a third of overall final export value, less than services in all regions except Asia, where the automotive industry contributes mar- specific. That makes it useful to draw lessons from actual country ginally more (figure 3.19). In Latin America, services contribute experiences. This section does that by synthesizing the results of nearly twice as much (more than 40%) as the automotive sector. questionnaires developed by the OECD Development Centre to The high fragmentation in the industry has provided broad target policy measures in the automotive sector. scope for integration for a variety of countries­ —­ and not just Although the automotive industry is highly concentrated, with those with a significant motor vehicle industry. That, in turn, only a few countries (companies) contributing to global produc- shapes the policy tools for improving the nature and space of tion, its value chain is especially fragmented, both geographically integration. And in many countries­ —­especially those with 84 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 3.18 Selected regional and megaregional agreements, 2016 European Single Market Transatlantic Trade and Investment Partnership Eurasian Central American European Customs North Common Market Free Trade Union/ American Bolivarian Alliance Association Central Eurasian South Asian Free Trade for the Peoples EU28 European Economic Free Trade Agreement of Our America Free Trade Union Area Caribbean Agreement Community Association of Belt and Road Initiative Regional Southeast Asian Trans-Pacific Partnership Latin American Comprehensive Nations Integration Association Economic Partnership Gulf Cooperation Andean Council Community Pacific Central African Economic Alliance Economic and Community of Monetary West African Trans-Pacific Community States West African Strategic Economic Community of Economic and Partnership Latin American and MERCOSUR Monetary Union Caribbean States Arab East African Maghreb Community Union Southern Southern African African Development Customs Community Union Common Market for Eastern and Southern Africa Tripartite Free Trade Area Source: OECD and UNCTAD forthcoming. Note: The size of circles is proportional to the number of members that are parties to the agreement. Dashed lines indicate selected announced megaregional initiatives. FIGURE 3.19 Gross exports of motor vehicles and parts by region and origin of value added, 2011 $ (billions) By region of origin By sector of origin 250 250 200 200 150 150 100 100 50 50 0 0 NAFTA Europe East and Latin Other NAFTA Europe East and Latin Other Southeast Asia America Southeast Asia America NAFTA Europe East and Southeast Asia Motor vehicles Other manufacturing Services Other Latin America and Caribbean Other Source: Author’s analysis based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database 2015. Note: NAFTA is the North American Free Trade Agreement. From domestic to regional to global: Factory Africa and Factory Latin America? • 85 negligible automotive sectors, but also those with large automo- and Turkey), and the rest, with small automotive industries, are —­ tive sectors­ a strong policy focus is on the upstream part of the focusing on linkage opportunities through other activities (Chile, motor vehicle chain, where two-thirds to three-quarters of total Dominican Republic, Ethiopia, Ireland, Peru, and Singapore). value is created. Of the 15 countries that responded to the OECD question- Targeted programs naire, 5 are implementing targeted programs for the industry Even in these specific categories, countries adopt different (Brazil, Colombia, France, Morocco, and Uruguay), four follow approaches to improve growth (table 3.3), often in parallel with a horizontal approach (Costa Rica, Czech Republic, Mexico, broader multidimensional strategies (table 3.4). For example, TABLE 3.3 Main characteristics of targeted programs to promote the automotive industry in selected countries, 2014 Characteristic Brazil Colombia France Morocco Uruguay Program name Inovar-Auto Production Plan Automobile Pact for Industrial Automotive Industry Transformation Resurgence– Export Promotion Program Automotive Regime Responsible Ministry of Development, Ministry of Trade, Ministry of Economic Ministry of Industry, Ministry of Industry, Institution Industry and Foreign Trade Industry and Tourism Regeneration and Trade, Investment and Energy, and Mining E-economy E-economy Timeframe 2013–17 2009–32 Began in 2012, no 2009–15 1992 (expired in 2015 termination date following World Trade Organization rules) Objective Strengthening Strengthen national supply Achieve revenues Strengthen linkages Increase GDP by Promote exports national supply chain (reaching a minimum (including exports) of among local suppliers 12 billion dirhams and in certain industrial chain investment of 1% of gross at least $3.4 billion and create 70,000 new segments, mostly revenues net of taxes of exports of $1.1 billion jobs by 2015; setup focused on qualified companies) and create at least second- and third-tier MERCOSUR 33,000 jobs by 2032 factories Green targets Increase energy efficiency Develop affordable of vehicles (efficiency goal green vehicles of 1.82 megajoules per kilometer for all cars sold in the country by 2017) Innovation Increase research and Promote innovation Increase innovation development and engineering through the Center content capacities (0.5% of gross for Technological revenues from sales of goods Development of the and services, matching with Automotive Industry grants from the National Fund for Scientific and Technological Development) Territorial National initiative National initiative, National initiative National initiative, National initiative dimension in coordination in coordination with with territorial with regional local authorities dimension (Tanger, competitiveness Keintra, and commissions Casablanca) Budget — — 1.4 billion euros — — Monitoring and Brazilian Agency for National Planning No evaluation A monitoring No evaluation carried evaluation Industrial Development is Department is in foreseen committee with out or foreseen in charge of developing a charge of monitoring private and public monitoring system for the and evaluation stakeholders has been program established Links http://inovarauto.mdic.gov www.ptp.com.co www.redressement www.emergence .br/ -productif.gouv.fr .gov.ma/MMM /plan-soutien-a-filiere /Automobile/Pages -automobile /prqMaroc.aspx Source: Author’s compilation based on country responses to the OECD questionnaire, “Targeted Programmes to Promote the Automotive Industry.” —­is not available. Note: ­ 86 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE 3.4 Multi-institution and multidimensional policy mix targeted to the automotive industry, 2014 Type of incentive and Description/ country/responsible institution beneficiaries Conditions Innovation content Finance Fiscal incentives Brazil, Ministry of Industry and Foreign Multinational and domestic Minimum requirements of research and Development of domestic Trade companies development and investments in engineering technology; adoption of and business information technology foreign frontier technology Colombia, Bancoldex Domestic companies No No France, Ministry of Research All companies carrying out research No All innovation activities and development Morocco, Ministry of Economy and Total exemption for five years for Beneficiaries need to be located in special No Finance all companies located in special economic zones economic zones Uruguay Tax credit linked to export Local content requirement (20% of No performance national value added) Matching funds/grants Colombia, Innpulsa Colciencias Domestic companies, specific line for Cooperation among local suppliers Adaptation to domestic small and medium-size firms market France, Ministry of Economy All companies carrying out research No Future-oriented research and and development on future cars affordable green vehicles Morocco State contribution of up to 10% of Beneficiaries need to be located in special No total investment economic zones Skills Brazil, Ministry of Industry and Technical, vocational, and higher Cooperation among private sector, local — Education) education universities, and training institutes Colombia, National Learning Service — — — and Centre for Technological Development of Automotive Industry Morocco Creation of training institute for skills Partnerships with private sector — for the automotive sector; grants for training Business services Brazil, Brazilian Agency for Export Domestic and multinational The company should operate in Brazil (or — Promotion companies be willing to relocate) Colombia, Bancoldex Domestic companies — — Morocco, Industrial Platforms offer a — — — one-stop shop for business services Demand-side support Public procurement Brazil, Ministry of Planning and Agrarian Multinational and domestic Companies capable of giving after-sale Special incentives for Development companies assistance over all national territory adaptation to local markets Colombia, Agency for Efficient Domestic companies No Special incentives for Purchase adaptation to local markets France Domestic companies 25% of purchased cars are hybrid or Green cars electric Other France, Ministry of Environment Taxes on high emission vehicles and — Green cars fiscal incentives to buy green cars Standards Brazil, National Institute for Metrology, — — — Quality and Technology Source: Author’s compilation based on country responses to the OECD questionnaire, “Targeted Programmes to Promote the Automotive Industry.” —­is not available. Note: ­ From domestic to regional to global: Factory Africa and Factory Latin America? • 87 Mexico and Turkey, with relatively large automotive industries, within a broader horizontal strategy (Costa Rica, Czech Repub- are developing strong assembly capacities with an empha- lic, Mexico, and Turkey) rather than as a targeted strategy. In the sis on adding value though higher (quality) upstream domestic Czech Republic, for example, the National Incentive Scheme, value chains. Some countries are strengthening domestic indus- which covers manufacturing, technology, and business services, trial capacities, such as Inovar-Auto in Brazil,12 the Production sets out the conditions for attracting FDI­—­tax incentives, grants Transformation Program in Colombia, and the Plan Automobile for job creation and training, preferential rates for land and in France. Another approach is to capitalize on FDI; Morocco’s infrastructure use, and grants for capital investment. But it also target is to increase the attraction of FDI in second- and third- sets conditions for the investors­ —­notably job creation targets tier upstream operations, while Colombia is looking at devel- (table 3.5). oping domestic technological capabilities. Another approach In Mexico, a variety of horizontal programs can benefit the is to strengthen export capacities, as in Uruguay, by upgrading automotive sector (table 3.6). They include programs that sup - through free trade agreement (MERCOSUR) chains. port the development and uptake of information and commu- The approaches also differ in time horizons. The Colombian nication technologies (PRODIAT and PROSOFT) and incentives program is part of a long-term (2032) strategy of production for innovation (CONACYT), for strengthening local suppliers transformation. Brazil and Morocco follow multiyear planning, (NAFIN), and for attracting FDI and promoting regional develop - while France has no predetermined end date. Uruguay ended its ment (ProMexico). regime in 2015 in line with WTO requirements. Turkey introduced an investment incentives system in 2012 Countries have set job targets (Colombia and Morocco), export with the objectives of promoting production transformation and targets (Brazil, Colombia), or investment targets (Brazil). Brazil and specialization in higher value-added activities, increasing jobs, France have targets linked to “green” cars and sustainable devel- and reducing territorial disparities. The system targets both opment. Brazil and Colombia also target technological develop- domestic and foreign companies and includes four categories ment and innovation. Colombia, for example, has a new Center for of incentives by type of investment: general, regional (broken Technological Development of the Automotive Industry. down into six regions based on socioeconomic criteria), large scale, and strategic. Each includes a different mix of incen - Horizontal strategies tives (table 3.7). The different schemes are applied taking into Similarly heterogeneous strategies can be observed in coun- account the characteristics of the region in which the investment tries that identified the automotive industry as a priority area is made. TABLE 3.5 Policy mix of the National Incentive Scheme, Czech Republic, 2014 Policy mix • Corporate income tax relief • Partial corporate income tax relief • Job creation grants • Training and retraining grants • Site support • Cash grant on capital investment Beneficiaries Manufacturing firms Technology centers Business support services centers Conditions Regular ▼ Strategic ▼ Regular ▼ Strategic ▼ Regular ▼ Minimum investment Minimum investment Minimum investment Minimum investment Creation of at least of 100 million koruna of 500 million koruna of 10 million koruna of 200 million koruna 40 new jobs at software ($5 million) within three ($25 million) within ($0.5 million) within ($10 million) within development centers. years. This limit is three years. three years. three years. Creation of at least reduced in regions with Minimum investment Investors’ own equity Minimum investment 100 new jobs at other high unemployment. of 250 million koruna must equal at least half of 100 million koruna business support Investors’ own equity ($12.5 million) in new of the investment. ($5 million) in new services centers (shared must equal at least half machinery. machinery. services centers and Minimum investment the investment. high-tech repair The investor must create of 5 million koruna The investor must create centers). Minimum investment at least 500 new jobs. ($0.25 million) in new at least 120 new jobs. in new machinery machinery. of 50 million koruna The investor must create ($2.5 million). at least 40 new jobs. Eligible costs Long-term assets, when the value of machinery equals at least half the value of acquired assets. Maximum 40% of total eligible costs (30% in Southwest 40% of total eligible costs. state aid regions and for investment in low-tech sectors) Source: Author’s compilation based on country responses to the OECD questionnaire, “Targeted Programmes to Promote the Automotive Industry.” 88 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE 3.6 Horizontal programs that can support the automotive sector in Mexico, 2014 Technological development Promoting Financing for Attracting foreign innovation suppliers direct investment Characteristic PRODIAT PROSOFT CONACYT NAFIN ProMexico Objectives Business services, Promoting use of Fund for technological Financing for working Financial incentives training, certifications, information and development and capital and fixed- to attract foreign and fiscal incentives to communication innovation asset acquisition direct investment and allow smaller firms to technologies to to suppliers of any promote national and become suppliers of increase productivity industry to promote regional development larger companies growth and increase local content Year of introduction 2009 2008 2009 — — Time of frame Three-year program, — Budget is approved Upon exhaustion of — but with annual annually by congress current budget budget approval Main responsible Ministry of Economy Ministry of Economy National Council NAFIN (development Ministry of Economy institution for Science and bank for small and Technology medium-size firms) Territorial dimensions Federal Federal Federal Federal Federal with matching resources from states Budget 200 million pesos 700 million pesos 4 billion pesos 500 million pesos — approved in 2014 approved in 2014 approved in 2014 ($385 million) ($154 million) ($54 million) ($308 million) Links www.economia.gob.mx www.prosoft.economia www.conacyt.mx www.nafin.com www.economia.gob.mx /mexico-emprende .gob.mx/ /index.php/fondos-y /portalnf/content /conoce-la-se /programas/7107 -apoyos/programa /sobre-nafinsa /programas-se -programa-para-el -de-estimulos-a-la /sala-de-prensa /informes-de -desarrollo-de-las -innovacion /boletin_15_14.html -evaluaciones-externas -industrias-de-alta /promexico -tecnologia-prodiat -para-el-ejercicio -fiscal-2012 Source: Author’s compilation based on country responses to the OECD questionnaire, “Targeted Programmes to Promote the Automotive Industry.” —­is not available. Note: ­ TABLE 3.7 Turkey’s investment incentives system, 2014 General Regional Large-scale Strategic investment investment investment investment Incentive incentives incentives incentives incentives Value-added tax exemption ✔ ✔ ✔ ✔ Customs duty exemption ✔ ✔ ✔ ✔ Tax reduction ✔ ✔ ✔ Social security premium support (employers’ share) ✔ ✔ ✔ Income tax withholding allowance ✔ ✔ ✔ Social security premium support (employees’ share) ✔ ✔ ✔ Land allocation ✔ ✔ ✔ Interest rate support ✔ ✔ Value-added tax refund (with minimum investment of 500 million ✔ Turkish lira) Source: Author’s compilation based on country responses to the OECD questionnaire, “Targeted Programmes to Promote the Automotive Industry.” From domestic to regional to global: Factory Africa and Factory Latin America? • 89 Broader policies supporting upstream integration Barriers to trade, whether at the border or behind the border, Even in countries with only small upstream providers to the can severely impede integration, but they are not the only obsta- automotive sector, such as Ireland, horizontal programs can be cles. Countries need strong domestic supply chains coupled important. Enterprise Ireland, the agency developing Irish enter- with unhindered access to imports. And it is no coincidence that prises in world markets, provides direct support (subject to EU economies making inroads in GVCs (as shown here for the auto- state aid guidelines) to foster high-potential startups by offering motive sector) focus policy as much on the improved function- research and development grants and tools for expansion, inter- ing of these domestic chains as on improved access to foreign nationalization, capacity development, and productivity. In addi- markets. Strong domestic chains are almost a precondition for tion, Ireland’s global sourcing strategy develops strong domes- sustainable and long-term success in GVCs. The spoils of export tic supplier chains to multinational enterprises in the country. success accrue to different sectors, most notably to small and In Chile and Peru, both net importers of vehicles, the focus is medium-size firms, which struggle for direct access to foreign on developing upstream capacities in copper, rubber products markets, especially in emerging economies. Domestic chains (tires), design, and textiles. also provide greater scope for functional upgrading. In many economies policy and partnerships nudge lead firms toward Special economic zones developing stronger competitiveness through technology trans- Several countries also reported using special economic zones, fers and training and greater scope for upstream incumbents to including Morocco, Costa Rica, the Dominican Republic and also upgrade through process and product upgrading. Colombia. Morocco has targeted the broader automotive But for many economies the road to success is not exclusively sector through the creation of integrated industrial platforms. in their own hands. Where the countries are matters as much Tanger Automotive City and Kenitra Automotive City are linked as who they are, and certainly for Sub-­ Saharan Africa and, to a to Tanger and Casablanca special economic zones, which host lesser extent, Latin America, distance from more developed and automotive assembly multinational corporations to create local higher income markets matters, especially for landlocked econo- clusters of competences. mies. And just as strong domestic chains matter for global inte- Costa Rica’s FDI attraction policy has been linked to special gration, so too do strong regional chains. For many economies, economic zones. It also promotes local suppliers (Costa Rica regional chains are a necessary intermediate step. Provee) and supports science and industry links to increase the It is possible, of course, that even without targeted action, innovation content of domestic companies. the benefits from GVCs will begin to trickle down­ —­in much the A variety of upstream suppliers to the automotive industry same way that GVCs are trickling inward in central China from the have established manufacturing operations in special economic coast. This could happen as countries in Africa and Latin Amer- zones in the Dominican Republic. In addition to fiscal incentives, ica that are on the periphery of richer regions begin to develop foreign companies are attracted to skilled human resources in as a result of larger spillovers southward from the European and disciplines related to the automotive industry, such as electri- North American poles of activity and westward from Asian poles cal and electronics manufacturing and injection molding. A free of activity. But there is no guarantee that this will happen, espe- trade regime, managed by the National Free Zones Council, cially with trade slowing and signs emerging that the GVC engine offers fiscal incentives to attract domestic and foreign compa- may be stalling, especially with growing calls for protectionism in nies that manufacture goods or provide export services. The richer markets and emerging signs of reshoring, and with auto- council also establishes links between companies inside and mation on the horizon. outside the special economic zones, and the government has Worryingly, the evidence suggests that new free trade agree- invested in promoting human capital in disciplines related to ments that overlap with existing arrangements may not improve the main industries operating in the special economic zones (in regional trade, especially if they are not broad in their liberaliza- partnership with the National Institute for Vocational Training). tion and facilitation policies. It is perhaps no coincidence that in The Dominican Republic is now shifting toward technology-­ Sub-­ Saharan Africa and Latin America, with significant shares of intensive sectors and higher value-added activities, including small and medium-size firms and relatively low regional integra- the automotive industry, capitalizing on preferential access to tion, overlapping agreements create a spaghetti bowl­ —­adding the U.S. and EU markets. barriers that many firms are ill equipped to deal with. In this respect, the more comprehensive multilateral agreements such as the Common Market for Eastern and Southern Africa–East Conclusions African Community–Southern African Development Community Tripartite Free Trade Area, with 26 African countries and 58% There is no unique solution or approach to capitalizing on GVCs. of the continent’s GDP, bode well. Similar arguments could be Who or where a company is matters almost as much as the what made for Asia, but the starting point here differs. Integration has and how, and indeed largely, determines how it integrates. Also been facilitated by significant FDI flows, drawn in part by lower clear, certainly from the upgrading stories in many economies, is unit labor costs, and significant poles of higher income, with mul- that companies are not necessarily static, and countries can take tinationals better equipped to handle the multiple layers of free many actions to overcome barriers to integration. trade agreements. 90 • Measuring and Analyzing the Impact of GVCs on Economic Development Improving regional integration may also help address com- Efforts to increase and preserve participation in GVCs in the petitiveness gaps that exacerbate those caused by geography years to come may face a more difficult economic environment, (and indeed costs related to poor infrastructure). This is espe- reinforcing the need to better understand the challenges, the cially important since entry to GVCs through cheap labor alone drivers of success, and the barriers that impede it. The diversity does not seem to be enough. What appears to matter is the of success stories and the diversity of failures point to pragma- combination of labor and productivity, in other words unit labor tism and a realization that there is no one-size-fits-all approach. costs. Despite, for example, the recent rise in China’s labor costs, Different pathways exist, each delivering specific results and its unit labor costs appear to have remained competitive with entailing a different balance of risks and gains, depending on the those of Sub-­ Saharan Africa and Latin America. It is important, characteristics of a country, including its market structure and therefore, to make inroads in improving productivity, particu- policy approach. For developing countries today, it is crucial to larly through FDI, bringing much needed capital, technology, look forward and anticipate changes in the global organization and know-how. But FDI has to be coupled with policies that can of production in order to adapt to the future of production and extract maximum spillovers through robust domestic supply services delivery. chains, including a more robust entrepreneurial environment. From domestic to regional to global: Factory Africa and Factory Latin America? • 91 ANNEX 3.1 Result, variables, and data sources for the study by Kowalski and Lopez-Gonzales (2016) The annex tables present information and detailed results from the Kowalski and Lopez-Gonzales (2016) model analysis of changes in domestic value added in exports. TABLE A3.1.1 Determinants of changes in the domestic value added in exports (standardized coefficients) (1) (2) (3) All Developed Emerging Variable countries economies economies Capital–labor ratio (log) 0.0739*** 0.0280 0.112*** (0.0163) (0.0195) (0.0299) Skill intensity 0.0928*** 0.118*** 0.844 (0.0354) (0.0343) (1.027) Relative output per worker 0.0978*** 0.0802*** –0.119 (0.0276) (0.0291) (0.138) Share of foreign direct investment stocks in GDP 0.00512*** 0.0103*** –0.00497 (0.00172) (0.00245) (0.00315) Rule of law –0.0250 0.0303 –0.0615** (0.0157) (0.0208) (0.0280) Lagged foreign value added in industry exports (log) 0.151*** 0.150*** 0.139*** (0.0177) (0.0196) (0.0349) Tariffs charged (log) –0.0507*** –0.0586*** –0.0131 (0.00794) (0.0105) (0.0114) Share of exports covered by free trade agreements –0.00930 0.00403 0.0256 (0.00635) (0.00689) (0.0186) Index of depth of free trade agreements 0.00222 –0.00134 –0.00414 (0.00581) (0.00669) (0.0120) Sophistication of exports 0.0257* 0.0119 0.0527** (0.0139) (0.0149) (0.0250) Concentration of exports –0.00507 –0.0171 0.0167 (0.00976) (0.0119) (0.0206) Domestic demand (log of value) 0.327*** 0.312*** 0.397*** (0.0276) (0.0322) (0.0734) Distance to economic activity (log) –0.130*** –0.105*** –0.195*** (0.0250) (0.0289) (0.0504) Constant –0.167*** –0.206*** –0.138 (0.0357) (0.0418) (0.226) Number of observations 10,882 7,394 3,488 R -squared 0.649 0.641 0.667 Number of repeating sections 1,838 1,250 588 * ** p < 0.01, ** p < 0.05, * p < 0.1.  Source: Kowalski and Lopez-Gonzales 2016. Note: Numbers in parentheses are robust standard errors. Regressions are at the sectoral level using a fixed-effects specification that restricts the variance to the country-sector dimension and thus captures the impact of changes in the independent variables on the dependent variable. This setup controls for time-invariant country-sector omitted variables. See table A3.1.2 for a description of the variables and table A3.1.3 for descriptive statistics. 92 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE A3.1.2 Description of variables Variable Description Source Domestic value added in exports (log) Domestic value added used by industry to OECD Inter-Country Input-Output Tables produce exports Capital–labor ratio (log) Aggregate economywide capital-to-labor ratio Penn World Tables Skill intensity High-skilled workers divided by low skilled International Labour Organization workers (aggregate) Relative output per worker Country output per worker divided by average International Labour Organization global output per worker Share of foreign direct investment stocks Aggregate share of foreign direct investment United Nations Conference on Trade and in GDP stock in country Development Rule of law Worldwide Governance Indicators Lagged foreign value added in industry Foreign value-added use by industry to produce OECD Inter-Country Input-Output Tables exports (log) exports Tariffs charged (log) Weighted average applied tariffs (weights from Trade Analysis Information System BACI data) Share of exports covered by free trade Weighted average trade covered by free trade Trade Analysis Information System and Design of agreements agreements if countries share an agreement all Trade Agreements Database their exports are considered to be covered Index of depth of free trade agreements Count of deep provisions in free trade agreements Design of Trade Agreements Database Sophistication of exports EXPY variable calculated following Hausmann- BACI Herfindahl indicators of concentration normalized Concentration of exports EXPY variable calculated following Hausmann- BACI Herfindahl indicators of concentration normalized Domestic demand (log of value) Domestic value added from industry that is OECD Inter-Country Input-Output Tables consumed domestically Distance to economic activity (log) Distances weighted domestic value added in Centre d’Etudes Prospectives et d’Informations consumption of other countries Internationales Geography Source: Kowalski and Lopez-Gonzales 2016. TABLE A3.1.3 Descriptive statistics Number of Standard Variable observations Mean deviation Minimum Maximum Domestic value added in exports (log) 14,302 6.58192 2.28415 –9.876943 13.53801 Capital–labor ratio (log) 14,518 11.70262 0.88856 8.13445 12.84867 Skill intensity 14,450 10.62023 43.35979 0.07450 358.80000 Relative output per worker 14,518 1.00000 0.60184 0.05052 3.59928 Share of foreign direct investment stocks in GDP 14,042 5.12193 7.57578 –16.40000 67.20000 Rule of law 12,036 0.79489 0.86737 –1.19000 1.98000 Lagged foreign value added in industry exports (log) 12,118 4.80752 2.81482 –12.79376 12.37180 Tariffs charged (log) 14,280 1.35067 0.75808 0.00000 3.34222 Share of exports covered by free trade agreements 11,628 0.56681 0.27757 0.00000 0.99390 Index of depth of free trade agreements 14,518 112.60660 116.49580 0.00000 366.00000 Sophistication of exports 14,076 9.69665 0.26274 8.26598 10.38535 Concentration of exports 11,832 0.05517 0.10275 0.00288 0.68158 Domestic demand (log of value) 14,756 11.81066 1.86974 7.47943 16.33693 Distance to economic activity (log) 14,518 13.40891 0.32243 12.43798 13.91208 Source: Kowalski and Lopez-Gonzales 2016. From domestic to regional to global: Factory Africa and Factory Latin America? • 93 Notes 8. See also Lopez-Gonzalez, Meliciani, and Savona (2015). 9. The economic complexity indicator provides a broad measure of the 1. According to the World Bank’s 2014 Logistics Performance Index, 6 of relative complexity of products and countries by ranking the diversity the 10 lowest ranked countries were in Africa: Republic of the Congo, of products produced by a country with products weighted by com- Djibouti, Democratic Republic of the Congo, Eritrea, Somalia, and plexity based on their ubiquity. Tracking movements over time can Sudan. therefore provide insights into the relative upgrading (in complexity 2. An additional complication should be recognized in attempting to of production) of countries. Unweighted values of diversity and ubiq- look for signs of functional upgrading (in the classic case where an uity are initially defined as follows, with Mcp = 1 if country c produces existing firm upgrades to a higher value part of the chain) using offi- product p, and Mcp = 0 otherwise: level) statistics. These data cial national aggregated (and not firm-­ Diversity = kc,0 = ∑  Mcp(3.1) typically aggregate firms on the basis of their core activity, usually p Ubiquity = kp,0 = ∑ Mcp(3.2) measured on the basis of their main source of value added. So firms c that engage in functional upgrading may appear in different indus- Weighted values are generated through an iterative procedure: 1 trial activity codes over two periods, hampering the ability to observe kc,n = ∑pMcp × kp,N–1(3.3) k c,0 their evolution. But this is not necessarily a complicating feature when 1 investigating functional upgrading at the national (rather the firm) kp,n = ∑ M × kc,N–1(3.4) kp,0 c cp level. 3. Timmer and others (2014) showed that emerging economies special- Inserting equation 3.4 into equation 3.3 gives: ize in capital-intensive activities, which suggests that financial devel- 1 1 opment can be important for GVC integration and upgrading. Harri- kc,N = ∑pMcp ∑  M  × kc’,N–2 (3.5) kc,0 kp,0 c’ c’p son, Lin, and Xu (2014) showed that the key factors explaining Africa’s disadvantage at the firm level are lack of infrastructure, low access to Mcp Mc’p kc,N = ∑c’ kc’,N–2 ∑p  (3.6) finance, and political competition. kc,0 kp,0 4. sector and year A fixed-effect model is used that controls for country-­ and it follows that: characteristics. This restricts the variance of the dependent variable ~ to temporal changes in domestic value added embodied in exports kc,N = ∑c’ Mcc’ kc’,N–2(3.7) and controls for sector and country effects that do not vary over time. where While this reduces the incidence of unobserved heterogeneity, con- ~ McpMc’p cern remains about possible correlations between lagged changes Mcc’ = ∑p (3.8) kc,0 kp,0 in the foreign value added used to produce exports and current changes in the specialization measures. If prior changes are correlated Equation 3.7 is satisfied when kc,N = kc,N–2 = 1. This is the eigen­ ~ with current changes driven by a common trend, the estimates will vector of Mcc’ associated with the largest eigenvalue. Since this eigen- be biased. Further checks to account for the dynamic nature of these vector is a vector of ones, it is not informative, and the eigenvector processes (through the use of a difference generalized method of associated with the second largest eigenvalue ρ is taken. This is the movements specification) confirmed the robustness of the results. eigenvector that captures the largest amount of variance in the system 5. This variable does not overlap with the dependent variable since it and is used as the measure of economic complexity. The economic captures value added engaged in different activities. complexity index (ECIc) for a given country c is therefore defined as: 6. Interestingly, positive changes in the rule of law reduce rather than ρc – < ρ > ECIc = (3.9) increase domestic value addition in emerging countries. Although this stdev (ρ) is at odds with the common perception that better institutions lead to where < > represents an average over all countries, stdev better economic outcomes, it likely reflects threshold effects. In other reflects the standard deviation of ρ over all countries, and ρc is the words, when considering the relatively low current rating of emerging ~ second-largest eigenvalue of Mcc’ . economies on the rule of law, a positive association of value-added exports with the rule of law may arise only after a certain threshold 10. Some care is needed in interpreting the relationships between for- has been reached. Perhaps surprisingly, the analysis reveals no cor- eign value-added content and economic complexity, especially for relation with inward FDI. This may reflect, at least in part, dispropor- determining any causality, partly reflecting the nature of economic tionate investment in many emerging economies in existing (rather —­ complexity measures­ for example, increased specializations in nat- than greenfield) natural resource activities, coupled with volatile price ural resource exports are likely to generate lower rankings of com- effects. plexity. In addition, the economic complexity rankings are based on 7. A simple view of the capital–labor ratio can be given by the share of gross trade statistics, so countries that integrate in low-value process- labor in overall value added relative to capital’s share. The capital–labor ing tasks at the end of complex products will, all other things equal, ratio (C/L) can be described simply as C/(C + L), where C is the return have higher economic complexity measures. Moreover, changes in to capital and L the return to labor and C + L = value added (GDP). Unit the foreign value-added content of exports are a far from perfect labor costs reflect average wages divided by average productivity, or measure of GVC integration. For example, countries that upgrade L/(C + L) or 1 – C/(C + L). through stronger upstream domestic content are likely to see declines 94 • Measuring and Analyzing the Impact of GVCs on Economic Development in their foreign content but not necessarily lower GVC integration, References which partly explains China’s position. Equally, the upper and lower bound ranking of countries necessarily invalidates a linear relationship Aboal, D., P. Garda, B. Lanzilotta, and M. Perera. 2015. “Innovation, Firm between the two measures, which explains the omission of the top Size, Technology Intensity, and Employment Generation: Evidence from 20 ranked countries in 1995 from the charts. Japan, Germany, and the Uruguayan Manufacturing Sector.” Emerging Markets Finance and Switzerland, for example, whose foreign content of exports increased Trade 51 (1): 3–26. 6–10  percentage points over the period, retained their rankings as Bas, M., and V. Strauss-Kahn. 2014. “Does Importing More Inputs Raise first, second, and third, almost throughout the period. Exports? Firm-Level Evidence from France.” Review of World Econom- 11. See, for example, the OECD–Eurostat Trade by Enterprise Char- ics 150 (2): 241–75. acteristics database (www.oecd.org/std/its/trade-by-enterprise Beverelli, C., R. B. Koopman, V. Kummritz, and S. Neumueller. 2017. -characteristics.htm). “Domestic Value Chains as Stepping Stones to Global Value Chain Inte- 12. In addition, Brazil implemented a targeted program in 2014 to facili- gration.” CTEI Working Paper No. 2017-04. Center for Trade and Eco- tate upstream integration (Productive Linkage Automotive Sector) by nomic Integration, Graduate Institute, Geneva. small and medium-size firms: the program includes targeted training Ceglowski, J., S. Golub, A. Mbaye, and V. Prasad. 2015. “Can Africa Com- by automakers to enhance the production and innovation capacities pete with China in Manufacturing? The Role of Relative Unit Labor of their suppliers. Mexico also introduced a technology development Costs.” Working Paper 201504, Development Policy Research Unit, program for industry in 2009 (PRODIAT) run by the Ministry of Econ- University of Cape Town, Cape Town, South Africa. omy to reduce the information gap between large companies and ESCAP (Economic and Social Commission for Asia and the Pacific) and potential domestic suppliers. The program also offers financial sup- World Bank. ESCAP-World Bank Trade Cost Database. Bangkok: port for certification to allow small and medium-size firms to operate ESCAP. Available at: http://www.unescap.org/resources/escap-world as subcontractors. -bank-trade-cost-database. Gereffi, G., J. Humphrey, and T. Sturgeon. 2005. “The Governance of Global Value Chains.” Review of International Political Economy 12 (1): 78–104. Harrison, A., J. Yifu Lin, and L. C. Xu. 2014. “Explaining Africa’s (Dis)Advan- tage.” World Development 63 (c): 59–77. Hausmann, R., C. Hidalgo, S. Bustos, M. Coscia, S. Chung, J. Kimenez, A. Simoes, and M. Yildirum. 2011. The Atlas of Economic Complex- ity: Mapping Paths to Prosperity. Cambridge, MA: Harvard University, Center for International Development. Kowalski, P., and J. Lopez-Gonzalez. Forthcoming. “Global Value Chains and Developing Countries: Drawing on Foreign Factors to Enhance Domestic Performance.” Kummritz, V. 2014. “Global Value Chains: Benefiting the Domestic Econ- omy.” Working Papers 2014–05, Centre for Trade and Economic Inte- gration, Geneva. Lall, S. 2000. “The Technological Structure and Performance of Develop- ing Country Manufactured Exports, 1995–1998.” Oxford Development Studies 28 (3): 337–69. Lopez-Gonzalez, J., M. Meliciani, and M. Savona. 2015. “When Linder Meets Hirschman: Inter-Industry Linkages and Global Value Chains in Business Services.” Working Paper Series 2015–20, Science Policy Research Unit, University of Sussex, Brighton, U.K. OECD (Organisation for Economic Co-operation and Development). Pro- ductivity Statistics (database). Paris: OECD. Available at: http://www .oecd.org/std/productivity-stats/. OECD (Organisation for Economic Co-operation and Development)– UNCTAD (United Nations Conference on Trade and Development. Forthcoming. Transforming Economies. Paris and Geneva. OECD (Organisation for Economic Co-operation and Development) and World Bank. 2015. Inclusive Global Value Chains: Policy Options in Trade and Complementary Areas for GVC Integration by Small and Medium Enterprises and Low-Income Developing Countries. Report prepared for submission to the G20 Trade Ministers Meeting, Istanbul, Turkey, 6 October 2015. From domestic to regional to global: Factory Africa and Factory Latin America? • 95 OECD (Organisation for Economic Co-operation and Development) UNECA (United Nations Economic Commission for Africa). 2013a. Eco- Added and WTO (World Trade Organization). 2015. Trade in Value-­ nomic Report on Africa 2013: Making the Most of Africa’s Commodities: (TiVA) database. Paris: OECD. Available at: http://www.oecd.org/sti Industrializing for Growth, Jobs and Economic Transformation. Addis /ind/measuringtradeinvalue-addedanoecd-wtojointinitiative.htm. Ababa, Ethiopia: UNECA. Shih, S. 1996. Me-Too is Not My Style: Challenge Difficulties, Break through ———. 2013b. Trade Facilitation: An African Perspective. Addis Ababa, Bottlenecks, Create Values. Taipei: The Acer Foundation. Ethiopia: UNECA. Timmer, M., A. A. Erumban, B. Los, R. Stehrer, and G. de Vries. 2014. “Slic- ———. 2013c. Assessing Regional Integration in Africa VI: Harmonizing ing Up Global Value Chains.” Journal of Economic Perspectives 28 (2): Policies to Transform the Trading Environment. Addis Ababa, Ethiopia: 99–118. UNECA. UN (United Nations). UN Commodity Trade Statistics Database (Comtrade). World Bank. 2016. World Development Indicators (database). Washington, New York: UN. Available at: https://comtrade.un.org. DC: World Bank. Available at: http://data.worldbank.org/data-catalog /world-development-indicators/. CHAPTER 4 Accumulated trade costs and their impact on domestic and international value chains HUBERT ESCAITH A ccording to trade analysts, trade costs­ —­together The decline in trade frictions stalled after the 2008–09 global with the relative size of the exporting and import- financial crisis (Escaith and Miroudot 2015), and the new empha- ing economies­ —­are among the main determinants ses are on reducing transaction costs and facilitating trade. In of bilateral trade patterns. More important from a the geographically fragmented production networks that have trade and development perspective, trade costs influence the emerged since the mid-1990s, trade in intermediate goods competitiveness of domestic firms on the international market accounts for more than half the volume of international transac- and the success of policies to join and move up global value tions. More than in traditional bilateral trade in final goods, trans- chains (GVCs). Although trade costs have declined over the past action costs (border and behind-the-border costs of trade) are decades, their relevance has increased with the surge of frag- crucial elements of the competitiveness of firms and partly deter- mented supply chains and the greater competition in a “small mine their ability to participate in global production networks. world” in which everybody cooperates and competes with These trade frictions are mainly an exogenous cost factor for the everybody. operators of international supply chains, who may mitigate the The reduction in transportation costs, the progressive decline negative impacts through leaner production management but in tariff duties and other customs barriers, and the progress in cannot alter the underlying causes. Facilitating trade remains information and communication technology connectivity have largely the domain of public action. “flattened the planet” by reducing transaction costs, which has Trade costs such as applied tariffs, transportation and insur- in turn contributed to the rapid expansion of global trade since ance costs, and other border taxes and fees are amplified as they 1985. After reviewing the domestic value added embodied in pass through the steps associated with modern supply chains. the final expenditure of markets of ultimate destination over This so-called cascade effect arises because trade costs accumu- time, Johnson (2014) and Johnson and Noguera (2016) identified late as intermediate goods are imported and then re-exported five stylized facts that explain the lessening of trade frictions.1 farther downstream, going through different processing nodes But as Anderson and van Wincoop (2004) noted, “The death of before reaching the final consumer. Thus, trade costs reduce the distance is exaggerated. Trade costs are large, even aside from gains from trade that countries expect from participating in GVCs. trade policy barriers and even between apparently highly inte- From the exporting firm perspective the financial impact of grated economies.” trade costs is magnified in the “trade in tasks” rationale that This chapter builds on Diakantoni and others (2017), Yu and others (2016), and Hayakawa, Laksanapanyakul, and Yoshimi (2016). The author wishes to rec- ognize the comments received on preliminary drafts at the two meetings of the Making Global Value Chains Work for Economic Development project in 2016 (Beijing in March and Washington in November) and the suggestions by the revisers on the final draft. 97 98 • Measuring and Analyzing the Impact of GVCs on Economic Development governs GVCs. In contrast to a large integrated firm concentrat- duties are specific (particularly for agricultural products), analysts ing most production processes under the same roof, specialized compute ad valorem equivalents. This chapter shows that tariffs processing firms that spread their manufacturing over multi- are not the biggest trade costs despite being the most visible, as ple locations need to recoup the associated trade cost, which for the iceberg. applies to the full value of the good, from the smaller fraction of For transportation costs the situation is more complex. In prac- value added at each productive stage. This larger relative weight tice, transportation costs depend on the nature of the good (per- of transaction expenses on the profitability of individual busi- ishable or not, bulky or not), the mode of transport, and the dis- ness operations explains why trade along GVCs is particularly tance between producers and consumers. Lewis (1994) identified exposed to trade costs. several factors besides freight costs that contribute to logistics This chapter measures international trade costs from the costs, including interest charges on goods awaiting shipment, on value chain perspective and reviews their implications at the goods in transit, and on goods held as safety stock, as well as the industry, national, and global levels. Trade frictions increase the loss, damage, or decay of goods between manufacture and sale. production cost 18% in a single stage of the value chain. Most of Because tariffs have become a less frequent barrier to trade, the additional expenses result from deficient logistic and trade the contribution of transportation to total trade costs­ —­ shipping facilitation conditions, many of which fall under the control of plus insurance­ —­has become more evident and more important. domestic policymakers. Trade costs are not only damaging for Hummels (2007) found that median transport expenditures were domestic firms willing to join GVCs, they also affect all trade half as much as tariff duties for U.S. imports in 1958, equal to partners and generate systemic losses. Using network analysis tariff duties in 1965 and three times higher than aggregate tariff that goes beyond the traditional bilateral dimension of interna- duties in 2004. tional trade, this chapter identifies where investment in trade Time matters, especially in GVC trade organized along com- facilitation would have the highest social returns from a multi- plex international supply chains. See, for example, Hayakawa, lateral perspective. Laksanapanyakul, and Yoshimi (2016), who concentrated on the time spent in the import process, including cargo handling and customs clearance. Those are the key components of the Tariffs, cascading transaction costs, and ad valorem time-related trade costs that are shifted onto the competitiveness import price of imported inputs. If those costs are passed on to the price of exports, the demand for these exported products Distance, transportation costs, and tariffs are only some of the fac- becomes smaller as time gaps lengthen. tors that affect trade costs; there are many others, some of them not directly measurable, such as uncertainty (see Anderson and Trade cost magnification and accumulation through van Wincoop 2004 or Ferrantino 2012 for a review of trade costs global value chains and border barriers and their measurement). One way of under- When manufacturing is geographically segmented and orga- standing these factors is to associate them with the set of frictions nized as an international production network, trade costs at each that tend to reduce trade. Samuelson (1954) depicts trade shrink- step of the production process are incorporated into production ing under the effect of frictions in the same way that an iceberg costs and passed on to the next step through a higher free-on- melts while moving through the sea. International economics has board value of the processed good. The trade costs propagate overwhelmingly relied on Samuelson’s hypothesis that frictions are through the supply chain, cascading from upstream to down- proportional to value (ad valorem “iceberg transport cost”). stream to the final consumers. An extensive literature has explored the influence of trade The impact of cascading transaction costs is amplified as costs, especially using the gravity model. Head and Mayer (2013) intermediate goods are further processed by importing countries showed that the magnitude of estimated elasticity of gross trade and then re-exported. If tariff accumulation is ultimately paid by in goods varies across studies depending on the sample and the final consumer, 3 tariff magnification relates to the processing methodology used but centers around –1. Baldwin and Taglioni firms’ financial returns (gross profits). GVC suppliers are mainly (2011) showed that for GVC trade the standard gravity model price takers, and high trade costs translate into reduced value used by most studies performs poorly when applied to bilateral added. For the processing firm at each step of the supply chain, flows where parts and components trade is important.2 Noguera the additional costs have to be compensated for out of the value (2012) applied a gravity model to trade in value added and found added generated by the fees the firm receives for processing the that the bilateral trade cost elasticity of value-added exports was imported inputs and re-exporting them to another GVC partici- about two-thirds that of gross exports and that bilateral value- pant. Unlike a fully integrated firm, which builds a product from added exports increased with both bilateral trade agreements stage A to Z and cashes in the full commercial value of the gross (a result also found in trade in final goods) and agreements with output, the processing firm can count on only the smaller share other countries. of value added it creates (its processing fees). Nominal tariffs are the most visible cross-border transaction It is thus important to measure the impact of trade costs not cost. Tariff duties increase the domestic price of tradable goods in proportion to the total value of the output (unlike the “ice- by adding a tax to their international or free market price. When berg” metaphor in conventional trade analysis) but in proportion Accumulated trade costs and their impact on domestic and international value chains • 99 to the value added generated at each step of the supply chain. rate on imported inputs as a share of free-on-board export value The latter value is often much smaller than the full commercial (trade-weighted average tariffs for intermediate inputs from the value of the intermediate good to which trade costs apply, so the other countries and regions that are used in the exporting coun- financial impact of trade costs on the processing firm’s competi- try to produce final good exports). The sum of the two tariffs is tiveness and profitability in a GVC context is said to be amplified. in column 4. To see how amplification affects the bottom line of an export- Column 5 provides the first-order accumulation effect of ing firm, take an export processing firm that uses imported inputs using imported intermediate inputs to produce exports. It rep - that cost a hypothetical value of 40 (excluding trade costs) to pro- resents the accumulation cost-push effect of the length of the duce a final good that sells on the international market for 100 supply chain increasing by a single processing step if tariffs were (table 4.1). The value added of 60 generated at international prices the only factor that augmented the trading costs. For instance, is split between employee remuneration (40) and gross profit one additional stage of production increases the trade costs of (20). If the processing firm is a price taker and the cost of labor is Viet Nam’s merchandise production by 80% of its standard tariff. exogenously fixed, any increase in trade costs (10 in the example) Column 6 reports the gross effective tariff rate on output, which will reduce gross profit. The impact of trade costs on the input equals the standard tariff rate in column 2 divided by the domes- procurement cost is magnified on what truly matters for the firm: tic content share (which is 1 minus column 2) and weighted by the share of value added that remains as gross profit, once other trade.4 Column 7 reports the implied magnification ratio due to production costs have been paid. In this example, an added trade the presence of vertical specialization. These effects are gener- cost of 25% leads to a reduction of 50% in gross profit. ally larger than the tariff accumulation factor in column 5. Obviously, this is a simple example, and the firm’s profitability The magnification effect worsens the impact of trade costs depends on many other factors, including returns to scale. The for low-income developing economies, because the share of firm should thus decide whether the higher volume of sales that domestic value added is usually lower in their manufactured may be expected from joining a GVC compensates for the lower exports than those of developed countries, and their trade costs profit margin per unit of output. While the exporting decision are higher. Considering that value added is used mainly for the depends on factors beyond the scope of this chapter, higher remuneration of employees and invested capital, higher-than-­ trade costs lower the probability of exporting compared with average trade costs result in lower salaries and reduced invest- selling on the domestic market. ment in order to maintain competitiveness at world market Koopman, Wang, and Wei (2014) illustrated the accumulation prices. So reducing tariffs and nontariff trade costs globally and amplification effects of tariffs based on actual data (table through multilateral agreement is fully consistent with the inter- 4.2). Column 1 reports the standard tariff on a country’s exports ests of developing economies because it lowers the cost of their (the trade-weighted tariff rate applied by a country’s trading GVC participation and improves their potential for upgrading. partners in ad valorem equivalent). Column 2 reports the share For domestic firms, lowering their trade costs on the import of imported content in final goods exports. These imported of intermediate inputs for domestic manufacturing production intermediate inputs are used to produce exports of final goods would greatly reduce the magnification effects, as demonstrated and thus incur multiple tariff charges. Column 3 reports the tariff in column 5. Lowering such costs in other countries would greatly reduce the effective tariff rate in their export markets, as seen in columns 6 and 7, because of the lower domestic value-added TABLE 4.1 Amplification effect of trade costs on value share in most developing countries’ manufacturing exports. added and profit margin Even if trade costs have decreased over the past decades as a result of technological progress and trade policies, their influ- Processing for export ence through cost accumulation and magnification is expected No trade With trade to become stronger as participation in GVCs increases, espe- Profit and costs costs costs cially in manufacturing industries. As discussed in chapter 2, the Imported intermediate input average length of total production shows a clear upward trend (free on board) 40 40 at the world level, especially after 2002. The relative importance Trade cost on inputs 0 10 of pure domestic production activities is diminishing, though the Value added 60 50 trend was temporarily interrupted by the global financial crisis, when the value added embodied in complex and simple GVC Labor 40 40 production-sharing activities had increased rapidly, until 2011. Profit 20 10 Further, the average production length of complex multistage Export price (free on board) 100 100 production-sharing arrangements increased by 0.36 between 2002 and 2011, much faster than the lengthening of production Source: Adapted from Diakantoni and others 2017. in simple production sharing and pure domestic production. Note: Simple example based on hypothetical values, for illustrative purposes Moreover, trade frictions remain substantial and are exposed to only. the return of protectionist sentiments. Using more recent 2011– 14 World Input-­ Output Database data, Timmer and others (2016) 100 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE 4.2 Magnification effect on tariff costs under global value chain trade, 2004 (2) (3) (6) Share of Tariff on Gross imported imported effective (1) content in inputs (4) (5) tariff rate (7) Standard final goods (% of free-on- Two-stage Accumulation in export Magnification tariff exports board export tariffs (1+3) effect market ratio Country/region (%) (%) value) (%) (4 ÷ 1) (%) (6 ÷ 1) Advanced economies Australia–New Zealand 15.55 0.13 0.34 15.89 1.02 27.00 1.74 Canada 1.60 0.38 0.24 1.84 1.15 7.05 4.41 Western European Union 6.16 0.12 0.24 6.40 1.04 12.09 1.96 Japan 6.22 0.12 0.05 6.27 1.01 11.19 1.80 United States 4.38 0.13 0.17 4.55 1.04 9.19 2.10 Emerging Asia China 6.17 0.29 1.91 8.08 1.31 21.42 3.47 Indonesia 7.53 0.30 1.34 8.87 1.18 24.39 3.24 Malaysia 3.55 0.46 2.11 5.66 1.59 20.93 5.90 Philippines 5.57 0.39 1.07 6.64 1.19 22.47 4.03 Thailand 8.16 0.40 4.23 12.39 1.52 36.54 4.48 Viet Nam 10.71 0.43 8.62 19.33 1.80 55.10 5.14 India 7.82 0.18 2.98 10.80 1.38 22.08 2.82 Other emerging economies Brazil 12.27 0.13 1.22 13.49 1.10 22.77 1.86 EU accession 2.41 0.34 0.55 2.96 1.23 12.67 5.26 Mexico 0.88 0.31 1.00 1.88 2.14 6.36 7.23 Russian Federation 5.36 0.18 1.61 6.97 1.30 17.23 3.21 South Africa 7.15 0.20 1.11 8.26 1.16 22.11 3.09 Source: Yu and others 2016. found that international fragmentation of production proba- An intuitive way of understanding this effect is to extend bly stalled in recent years, a slowdown that is reviewed in more table 4.1 to a case where a domestic firm chooses between sell- detail in chapter 2. ing on the domestic market (at a price that includes the effect of nominal protection) and exporting at the world price. While Transaction costs and domestic value added trade frictions lower the gross profits when the firm exports its This section focuses on the implication of trade costs for the pro- product, the same trade costs have the effect of increasing the duction function of industries, their operational costs, and ulti- profit margin when the product is sold (at a higher price) on the mately their gross margins. In a GVC environment where firms domestic market (table 4.3). trade in tasks or in such business functions as research and devel- The intuition behind the calculation of the net effect on opment, logistics, and manufacturing services, trade costs affect value added to the domestic price (called the effective rate of profitability and competitiveness. Trade in tasks is often called protection) is as follows. The theoretical referent is a neoclassic trade in value added, because what firms exchange in their busi- economy where countries have access to the same technologies ness-to-business transactions along GVCs are not products but and will choose the combination that best fits their resource value added. Monetary trade costs (tariffs, transportation, and endowments. In a frictionless trade environment the world price other financial costs identified by Lewis 1994) increase the price of a given product is unique. If the price charged by domestic of the value added on the domestic market, creating an anti-ex- producers is higher than the world price, consumers will shift port bias in a trade in tasks perspective. to imported products. Similarly, if the domestic price is lower, Accumulated trade costs and their impact on domestic and international value chains • 101 TABLE 4.3 Influence of trade cost on value added and Corden (1966). Their original formulation calculated effective profit margin, domestic versus export prices protection rates by deducting the additional production cost that manufacturers had to pay because of the tariff charged on Domestic Export market market tradable inputs from the additional benefit generated by selling their product at a price higher than the free-trade market price, No With No With trade trade trade trade thanks to the duties charged on competitive imports. The result Profit and costs costs costs costs costs is the rate of value added at domestic prices (selling price minus Imported intermediate input cost of intermediate inputs required for the production) and is (free on board) 40 40 40 40 compared with the hypothetical value added that would have resulted from the operation if no custom duties had been levied. Trade cost on inputs 0 10 0 10 In table 4.3 the effective rate of protection is 25%, correspond - Value added 60 75 60 50 ing to the additional gains (15) reported to the value added Labor 40 40 40 40 under free trade (60). Effective rates can be calculated because of the availability Profit 20 35 20 10 of international input-­output matrices, which are also used to Export price (free on board) .. .. 100 100 measure trade in value added, as in the Organisation for Eco- Domestic market price 100 125 .. .. nomic Co-operation and Development (OECD)–World Trade Organization (WTO) Trade in Value-Added database. As men- Source: Diakantoni and others 2017. tioned, the calculation also relies on the simplifying hypothesis Note: Example is based on hypothetical values, for illustrative purposes only. of perfect competition and substitutability between imported The ad valorem trade cost (25%) is the same for input and output products. and domestic products. Domestic industries are expected to raise their price in order to benefit from the additional costs due to tariff and freight costs applied to the imported goods that domestic producers would rather export their product and sell it compete with their products. In that situation international trans- at the higher world price. action costs influence the domestic price of all inputs, whether Because of trade costs, the domestic price of tradable prod- imported or domestically produced. This ad valorem increase in ucts is higher than the world price. Producers gain because they the price of competing goods is the extended tariff and trans- are able to sell at a higher price, with the markup correspond- port nominal protection. ing to the ad valorem trade cost. But they have to pay a higher When all applied tariffs are most-­ favored-­nation tariffs that price for the inputs used in production. That will be the case for do not discriminate between trading partners and when trans- imported inputs, but also­ —­and this is a key assumption of the portation costs are proportional to the value of the imported underlying model­ —­for the domestically produced goods. If a good, the extended tariff and transport effective protection is downstream firm producing a final good for the domestic market the difference between the nominal tariff and transport pro- is able to increase its prices in proportion to the nominal pro- tection enjoyed on the output minus the weighted average of tection received because of trade costs, this is also the case for tariff and transport paid directly (imported goods) or indirectly upstream firms producing intermediate inputs. (domestic goods) on the inputs required for production. The The net effect for a firm gives a higher rate of value added weights applied to the additional tariff and transport costs on per unit of output than the free-trade benchmark when the addi- inputs are derived from the technical coefficients of the input- tional cost of production is lower than the nominal protection output matrix.5 The extended tariff and transport effective pro - received (or a lower rate of value added per unit of output when tection rate is obtained by dividing this result by the value added the additional cost of production is higher than the nominal pro- that the industry would have enjoyed in the absence of tariff and tection received). Nominal protection in the domestic market transport costs. A formal presentation of the calculation is in for goods that are a firm’s inputs raises production costs and annex 4.1. thus provides negative incentives to export. The service sector is always on the losing side (there is no nominal protection on Impact on competitiveness and export-led growth strategies its output, but it has to pay an additional cost for its tradable When the tariff and trade cost schedules are flat, the extended inputs). So are consumers. But because services are usually less effective protection rate equals the nominal rate of tariff and tradable than goods, it is understood that service providers can transport protection. In table 4.3 that rate equals the rate of pass on the additional costs to their customers. nominal protection (25%). But it will differ when there is vari- ance in the tariff and nontariff trade costs, because some sec- Extended effective protection rates and the relative price of tors are more effectively protected than others. With tariff and value added transport cost escalation (most-­ nation tariffs rising with favored-­ The method used here to estimate the impact of trade costs the degree of processing or transportation and insurance costs across several countries and industries is adapted from the effec- increasing more than proportionally to the unit value of the tive protection rate theory introduced by Balassa (1965) and goods), downstream domestic industries producing final goods 102 • Measuring and Analyzing the Impact of GVCs on Economic Development for the domestic market will benefit from a higher effective pro - purchase inputs internally because their second-tier domestic tection on their value added. By contrast, upstream industries suppliers would not be able to benefit from the duty exemption. producing unprocessed inputs and basic parts and components Duty drawbacks and export processing zones compensate will have a low extended effective protection rate­ —­and possibly the exporting firm for the additional production costs caused by a negative one if the sum of tariff and transport margins paid on tariffs only when it uses imported inputs. Such a strategy effec- inputs is higher than the margin of protection received on the tively prices out domestic suppliers when nominal tariffs and output. trade costs are high. Second-tier national suppliers of a domes- Therefore, downstream industries registering a high extended tic exporting firm are usually not entitled to draw back the trade effective protection rate on their production will have little incen- margins paid on their imports. Even if they were able to do so tive to export because the rate of return from exporting is lower through some complex and arcane administrative mechanisms, than that from selling on the domestic market.6 Even upstream they would still be at a disadvantage when using nonimported industries supporting a negative effective protection rate will inputs (because nominal tariff and transport protection raises the still be better off selling on their domestic market, and the result domestic price of all tradable products, regardless of whether holds for all domestic firms, but the anti-export bias is stronger they are actually imported). The only possibilities for second-tier for highly protected industries. This hurtful effect of escalation is domestic suppliers to avoid tariff and transport costs would be particularly relevant for developing countries that want to diver- to use only imported inputs or to exert downward price pres- sify their export basket away from basic commodities. sure on their own domestic suppliers to recoup lost competi- Trade frictions reduce the competitiveness of domestic firms tiveness. While the anti-export bias is a well-known result from a in the most frequent situation where they are price takers and final-goods perspective, the anti-upgrading traditional trade-in-­ compete on the global market at international prices. When a corollary is new and relevant only from the vertical specialization domestic firm exports, it loses the additional benefit due to the perspective typical of GVCs, where a “buy” decision arising from nominal protection it receives on its output while still paying a make-or-buy assessment implies arbitraging between domes- the additional cost on inputs purchased domestically. The only tic and foreign suppliers. way to compensate for the additional costs and lower profits at export would be to reduce the value-added cost­ —­for example, Trade costs per sector and country by paying lower wages or retaining less profit.7 Diakantoni and others (2017) applied the extended effective This loss of cost competitiveness is particularly critical in a protection rate methodology by crossing OECD-WTO Trade in GVC context, when the customers on the export market are for- Value-­Added database data on 61 economies and the underlying eign lead firms that make their “make-or-buy” decisions as well OECD Inter-Country Input-­ Output Tables. The detailed tariff as their choice of offshore localization on the basis of tight cost data for 2006 and 2011 were sourced from the WTO. Nontariff and profit margins. For this reason, policymakers have devel- costs were taken from Duval, Saggu, and Utoktham (2015). These oped several strategies, from duty drawbacks (the exporter can trade costs do not proceed from a direct calculation but are indi- redeem the value of the tariff duties and other indirect taxes rectly derived from a gravity model applied to Trade in Value-­ paid on inputs used for exports) to free export processing zones Added database data: the trade frictions may result from a direct (industrial parks installed in fiscal enclaves). monetary cost (such as transportation, insurance, and other fees) Such schemes (duty drawbacks and export processing zones) as in the extended effective protection rate approach, but they fall short of providing a first-best policy when the policymakers’ may also arise from more subjective aspects, such as the ease or ultimate objective is to use GVCs as a path toward industrializa- difficulty of gathering relevant information and other nonmone - tion. Even if the typical arrangement in a supply chain contract is tary barriers (regulation, licensing), insecure contracts and weak- for the lead-firm or supply-chain manager to cover the interna- ness in trade governance leading to uncertainty, differences in tional costs of procurement, an exporting firm will still face the consumer taste, and so on. Nontariff trade measures are partic- higher cost of purchasing its inputs domestically. So the high ularly relevant for GVCs because they may constrain the produc- tariff and transport protection in place outside export process- tion process itself (box 4.1). The monetary component, according ing zones will limit the possibility of developing domestic interin- to the experts who build the database, is believed to account dustry links (second-tier domestic suppliers), even if a domestic for only one-third of these costs; this is the value retained in the firm can join an international supply chain.8 extended effective protection rate application. Take the most favorable case of a first-tier supplier operating The first effect of tariff and nontariff trade costs is to pro - from an export processing zone in an international supply chain tect domestic producers from competitive imported products where the foreign lead firm covers the costs of transportation by increasing the import price by a trade margin of 20% to the of the intermediate inputs and the re-export of the processed international price of competing imports (averaged across all good. In that situation the first-tier supplier does not have to pay sectors in 2011), 17% for nontariff costs and 3% for tariff costs, any transaction costs. Yet, even when duty drawbacks or tariff including 2.5% preference margin (table 4.4).9 Trade costs vary exemptions (as in export processing zones) correct for trade fric- by a factor larger than four between the highest (food prod- tions and allow domestic producers to purchase inputs at inter- ucts) and the lowest (mining). Ranked by trade costs, the top national prices, export-oriented firms still have a disincentive to five sectors are food products (35%), motor vehicles (27%), other Accumulated trade costs and their impact on domestic and international value chains • 103 BOX 4.1 Tariff and nontariff measures Among trade costs, nontariff measures have a specific to promote trade. The compliance of arm’s length suppliers role because they interfere with industrial norms that reg- with public and private norms (International Organization ulation may also consider trade enhancing. Nontariff mea - for Standardization standards on quality) substitutes for sures considered by the World Trade Organization (WTO) closer and more expensive lead-firm monitoring of the qual - concern mostly regulations and standards, which are dealt ity of traded intermediate inputs. By contrast, cumbersome with under sanitary and phytosanitary measures and the and unharmonized nontariff measures increase trade costs, technical barriers to trade. Nontariff measures are not only if only because they entail more complex customs proce- normative; they also include such quantitative measures as dures. When not harmonized, nontariff measures are there- safeguards, countervailing or antidumping measures, and fore expected to be trade-restrictive, especially for smaller other quantitative restrictions imposed against discrimina- firms or firms in less technologically advanced countries. tive policy measures by trading partners. Discussions of the protectionist nature of nontariff mea- With the lowering of tariff duties over the past decades, sures are ongoing. Attempts to assess the trade impacts awareness is growing that nontariff measures are imposing of nontariff measures have led to the development of new restrictions on trade, especially with the rising impor- “tariff equivalent” methods, which seek to estimate the tance of global sourcing within global value chains (GVCs). ad valorem tariff that would have a trade-restricting effect Since the global financial crisis of 2008–09 the WTO primary equal to the nontariff measure in question (Ferrantino 2012). monitoring and surveillance mechanism has been based on Adopting a specific GVC perspective, Ghodsi and Stehrer its periodic Trade Policy Reviews. Transparency mechanisms (2016) provided new ad valorem equivalents for nine types are also present in many regional trade agreements. Ing, of nontariff measures, capturing the effects of these policy Cadot, and Walz (2016) developed an index of nontariff mea- measures’ intensity across sectors, importers, and export- sure transparency, based on WTO notification requirements. ers. Interestingly, some providers (such as Canada) may They show that transparency varies positively with income actually benefit from what would be conceived as restrictive (except for non–Organisation for Economic Co-operation measures, while others (such as Bulgaria) incur larger losses. and Development high-income countries). The index also Less advanced countries may therefore be more affected by varies across regions, high in the Association of Southeast stringent nontariff measures. The effect is also differenti- Asian Nations (ASEAN) and low in Africa and the Middle ated by industry and by type of nontariff measure: technical East. It may not be a coincidence that ASEAN is much better barriers to trade improve the cost efficiency of the inputs inserted in GVCs than the two other regions and that the for the production of electrical and optical equipment, top  five countries are Austria, Denmark, Ireland, Germany, while sanitary and phytosanitary measures, tariffs, and aver- and Sweden, all economies with an important GVC sector. age bilateral trade-restrictiveness indices increase the costs In theory, tariffs are trade restrictions imposed to pro- of inputs for these industries. Ghodsi and Stehrer (2016) tect domestic producers, while nontariff measures are set concluded that regulated nontariff measures that enhance to protect domestic consumers. Moreover, the use of inter- information symmetries reduce trade costs and increase national standards by either exporters or importers is likely market efficiencies. transport equipment (24%), agriculture (22%), and textiles (21%). was about 5.5%, after preferential treatments were included. The bottom five are computers (17%), other nonmetallic mineral Even so, the distribution of costs is skewed toward the higher products (17%), chemicals (17%), pulp, paper, and publishing range (7% and above). And this calculation takes into account (16%), and mining (8%). Commodities or primary goods such as only the direct cost of trade margins on imported inputs and not mining, wood, or paper imported products face the lowest trade the indirect effect of also increasing the domestic market price of costs: tariffs are usually low, and the products are shipped in all products, regardless of whether they are imported. bulk, using sea freighters. At the other extreme, food products Factoring in the impact of trade costs on the imported and combine the disadvantages of being expensive to transport (as domestic cost of inputs requires computing extended effective for perishable products) and being heavily protected by tariffs. protection rates relative to a free-trade situation. This free-trade Considering that the protection received on output translates benchmark is not directly observable, but Diakantoni and others into an increase in the production cost for the users of those inter- (2017) used German industries as the international benchmark mediate products, the weight on competitiveness is substantial. because the German economy showed the lowest country/sector Using the technical coefficient of the OECD Inter-Country Input-­ trade costs in their sample. Comparisons with this benchmark Output Tables as weights, Diakantoni and others (2017) found that show that trade costs on inputs can greatly affect the compet- the additional production cost due to tariffs on imported inputs itiveness of industries. The average non-German motor vehicle 104 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE 4.4 Incidence of trade costs on output and input prices, 2006–11 Preferential Total including Nontariff Most-­favored-­nation tariff tariff preferences Outputs Inputs a Outputs Inputs a Outputs Inputs a Outputs Inputs a Change, Change, Change, Change, Change, Change, Change, Change, 2006–11 2006–11 2006–11 2006–11 2006–11 2006–11 2006–11 2006–11 (percent- (percent- (percent- (percent- (percent- (percent- (percent- (percent- 2011 age 2011 age 2011 age 2011 age 2011 age 2011 age 2011 age 2011 age Sector (%) points) (%) points) (%) points) (%) points) (%) points) (%) points) (%) points) (%) points) 001 Agriculture 16.1 0.4 2.9 0.1 11.7 –3.9 3.0 –0.7 5.6 –2.4 0.2 0.0 21.8 –2.0 3.1 0.0 002 Mining 7.3 0.1 2.2 0.0 0.6 –0.5 0.5 –0.1 0.8 –0.5 0.1 0.0 8.1 –0.4 2.3 0.0 003 Food 25.5 1.1 3.7 0.2 18.5 –2.2 6.3 –1.4 9.0 –3.7 0.5 –0.2 34.5 –2.6 4.2 0.1 004 Textiles 15.5 0.2 4.8 0.2 10.2 –1.5 2.9 –0.5 5.4 –1.9 0.5 –0.1 20.8 –1.7 5.3 0.1 005 Wood 18.2 0.2 3.0 0.1 4.3 –1.1 2.8 –0.7 2.5 –1.3 0.3 –0.1 20.7 –1.1 3.3 0.0 006 Pulp, paper 14.7 0.1 3.3 0.1 2.3 –1.0 1.3 –0.4 1.7 –0.8 0.2 –0.1 16.5 –0.7 3.5 –0.1 007 Coke, petroleum 19.0 0.2 7.0 0.2 2.8 –0.7 0.9 –0.3 1.3 –0.8 0.2 –0.1 20.3 –0.6 7.2 0.1 008 Chemicals 15.1 0.1 5.0 0.1 3.1 –0.6 1.6 –0.4 1.7 –0.7 0.3 –0.1 16.8 –0.6 5.3 0.0 009 Rubber, plastic 16.4 0.1 5.5 0.1 6.5 –1.3 2.1 –0.5 3.6 –1.5 0.4 –0.1 20.0 –1.3 5.8 0.0 010 Other mineral products 13.7 0.1 3.6 0.1 5.5 –0.1 1.4 –0.2 3.1 –1.2 0.2 –0.1 16.8 –1.0 3.8 0.0 011 Basic metals 16.9 0.0 6.1 0.1 2.3 –0.9 1.2 –0.4 1.5 –0.9 0.3 –0.1 18.4 –0.9 6.5 0.0 012 Metal products 15.2 0.0 5.0 0.1 4.8 –1.2 1.5 –0.5 3.0 –1.2 0.4 –0.2 18.2 –1.2 5.3 –0.1 013 Machinery not elsewhere classified 17.9 0.1 6.6 0.1 3.2 –0.6 1.7 –0.4 1.9 –0.7 0.4 –0.1 19.8 –0.5 7.0 0.0 014 Computer, electronic equipment 16.1 0.2 6.6 0.1 2.1 –0.4 1.3 –0.3 1.2 –0.6 0.4 –0.1 17.3 –0.4 7.0 0.0 015 Electrical machinery 17.9 0.2 6.6 0.2 4.3 –0.8 1.8 –0.4 2.6 –1.1 0.5 –0.2 20.5 –0.8 7.0 0.0 016 Motor vehicles 22.4 0.5 8.3 0.2 9.9 –2.0 3.1 –0.7 4.9 –1.9 0.7 –0.2 27.3 –1.4 9.0 0.0 017 Other transport 21.0 0.5 7.1 0.2 3.1 –0.3 1.8 –0.3 2.6 –1.1 0.4 –0.1 23.6 –0.5 7.6 0.0 018 Manufacturing not elsewhere classified 19.2 0.4 5.5 0.2 4.7 –1.3 1.8 –0.5 3.4 –1.2 0.4 –0.1 22.5 –0.8 5.9 0.1 b Average 17.1 0.2 5.2 0.1 5.5 –1.1 2.1 –0.5 3.1 –1.3 0.4 –0.1 20.2 –1.0 5.5 0.0 Source: Yu and others 2016. Note: Presents the main results obtained for 2011 and the changes observed since 2006, the initial and final year being imposed by data availability. a. Imported products only, using the 2011 technical coefficients of international input-­  output matrix as weights.  b. Simple average across countries or sectors. industry, a sector closely associated with GVCs, would register This loss of competitiveness varies by country according to a gross margin 27% lower than the benchmark firm. Benefiting trade costs (figure 4.1). The highest trade costs are in small devel- from drawbacks would reduce this loss, but the home industry oping economies (such as Cambodia and Costa Rica). Small devel- would still lag behind the international competitor by a margin oped countries can also face high costs when they are isolated of about 20% if it continued sourcing other inputs domestically. from the main markets, as for small islands (Malta and Cyprus Food industries also have little incentive to export: their value as well as New Zealand). Two factors may increase freight rates: added would be 18% lower than the benchmark (14% with draw- the geographic distance between main trading partners and the backs). When the industry relies heavily on imported inputs, as small size of individual shipments. Except for China, the econo - in the case of petroleum products, drawback schemes can yield mies facing the lowest import costs are all developed economies. an improvement of 10 percentage points.10 But this remains an Nominal tariff protection (as measured by most-­ favored-­ exception; on average, drawbacks improve the competitiveness nation tariffs on industry output) declined between 2006 and of domestic exporters by a margin of only 4–5 percentage points. 2011 in a majority of the countries surveyed. Tunisia, the Republic Accumulated trade costs and their impact on domestic and international value chains • 105 FIGURE 4.1 Ten countries with the highest and lowest trade cost in all sectors, 2011 Additional production cost due to trade costs as a share of production cost (%) Economies with the highest cost on output Economies with the lowest cost on output 90 90 80 80 70 70 60 60 50 50 40 40 30 30 20 20 10 10 0 0 –10 –10 a es y a ca a d ta ia us a d ce en n m m ly s a nd an ni bi di bi in pa an an Ita tv al iu do pr at Ri an ed ua Ch ra bo m m La rla M lg el al Ja St Cy ng a Fr iA lo Sw th er Ze Ic Be m st he d Co Ki Li G Co ud Ca ite et ew d Sa N Un ite N Un Nominal trade cost Nominal protection (most-favored-nation tariffs) Preferential margin Total cost Source: Diakantoni and others 2017. Note: Total trade cost includes the preference margin. of Korea, Argentina, Brazil, and India applied the highest nominal of them immediately (381 were gradually removed through 2015). protection in 2011­ —­ between 11% and 15%­ —­ and Hong Kong, Other trade costs were expected to be lowered, because customs China; China; Singapore; New Zealand; the United States; and procedures became simpler as importers would no longer need Brunei Darussalam the lowest, between 0% and 2.5%. certify compliance with preferential rules of origin. The incidence The cost of tariffs is much lower than other transaction costs, of the measure is sizable: in 2010 intermediate goods accounted estimated at an ad valorem tariff equivalent of 17%. Moreover, for 47.5% of Canada’s gross imports and capital goods for 18.9%. though tariffs have been decreasing, nontariff trade costs for Obviously, the trade cost reduction will also benefit other coun- inputs increased globally in 2011 over 2006, probably a result of tries by facilitating their access to the Canadian market. increased uncertainty in the post-crisis era (Escaith and Miroudot Ciuriak and Xiao (2014) calculated that such benefits exceeded 2015).11 Trade frictions would translate into an average increase of those from mutual tariff elimination under any of the major pref- 17–32% of the production cost in a single stage of the value chain. erential trade agreements that Canada had been pursuing. In Unless compensated for by savings on other aspects of produc- comparing estimated gains from unilateral liberalization and tion (either unsustainable ones such as low remuneration for labor preferential liberalization through trade agreements, they noted and investment or export subsidies) or by improving total factor that not only did the gains from the unilateral route come with- productivity, those higher costs reduce the international compet- out the trade distortions associated with regional trade agree- itiveness of the industries in these countries. As the next sections ments, but they were easier to realize since the question of rules show, reducing trade costs in one or several countries has impor- of origin and the use of preferences did not enter the equation. tant positive spillover effects on other trading partners. Focusing on tariffs, Yu and others (2016) used the Canadian example to study how partially or completely eliminating tariffs Extension and application to Canada and China on imported intermediate inputs can help a country integrate into To lower trade costs and improve the competitiveness of man- GVCs and enhance its trade competitiveness. They applied a com- ufacturers, the Canadian government decided in 2010 to unilat- putable general equilibrium model to quantitatively analyze the erally eliminate tariffs on a broad range of manufacturing inputs impact of intermediate inputs tariff reduction on reducing multi- and equipment. The elimination covered 1,541 tariff lines, most stage production costs, promoting GVC-related trade activities, 106 • Measuring and Analyzing the Impact of GVCs on Economic Development and accelerating the structural adjustment of China and the world The results illustrate the interdependency of national indus- economy under three policy scenarios. They looked at the impli- tries through trade in intermediate inputs and the importance cation for China of a similar initiative and explored three policy of reducing trade costs in as many lead economies as possi- scenarios to analyze the impact of liberalizing intermediate goods ble. The next section looks at the systemic effect of trade costs trade on the global economy. The first scenario is China’s unilat- and their spillover effects through close-knit interindustry trade eral elimination of tariffs on imported intermediate goods (the interactions. Canada scenario). The second covers a regional trade agreement between China and the Asian and East African countries included in China’s Belt and Road Initiative, reducing bilateral import tar- Cascading transaction costs in the world trade iffs on intermediate goods trade to zero but maintaining tariffs for network non–regional trade agreement countries. In the third scenario, all Group of 20 (G20) member countries completely eliminate tariffs By measuring the impact of trade costs on the effective value on intermediate goods imported from all countries. added, the extended effective protection rate measures the The first scenario­ —­ unilateral trade liberation of intermedi- magnification effect of tariff and transport costs on individual ate goods­ —­ would enhance China’s economic growth and trade firms’ value added and competitiveness. This section turns to the with the rest of the world. Relative to the baseline, China’s real entire international supply chain and examines trade costs as a GDP would increase 1.2%, its exports would rise 5.7%, and its cascading source of transborder cost-push transmission. imports would rise 6.6%. China’s unilateral trade liberalization on intermediate goods imports would generate a small pos- Accumulation of trade costs along international supply itive spillover effect, and the real GDP of the rest of the world chains would increase 0.01%, with exports expanding 0.17% and imports GVC trade is characterized by multiple border crossings that 0.25%. The second scenario­ —­bilateral tariff reduction on inter- generate double counting in traditional trade statistics because mediate goods trade between China and the Belt and Road processing goods will cross several borders before reaching region­ —­ would stimulate the economic growth and trade of sig- the final consumer. Correcting for this statistical bias was one of natory countries. If bilateral tariffs on intermediate goods were the initial objectives of measuring trade in value added. Double exempted completely, real GDP would increase 0.43% for China counting arises when goods in process cross successive borders. and 0.42% for the Belt and Road region. China’s imports would These successive border crossings open the door to potentially rise 2.8%, and its exports would rise 3.2%, and trade in the Belt explosive embodied tariffs along GVCs. and Road region would grow 1.5%. GDP would increase 0.43% for For example, Yi (2003), Ma and Van Assche (2010), and Fer- China and 0.40% for the other signatories. However, the bilateral rantino (2012) highlighted the nonlinearity in the way transaction trade agreement creates trade diversion and leads to some neg- costs negatively affect trade flows in a trade in tasks perspec- ative economic effects on nonmember countries: their exports tive. Ferrantino (2012) showed that when trade costs apply in would decline 0.04%, and their imports would decline 0.03%. proportion to the value of a good, the total cost of delivering If, as in the third scenario, G20 countries act jointly and unilat- the product through the supply chain down to the final consumer erally eliminate tariffs of intermediate goods imports, no diversion increases exponentially with the number of production stages. would take place, and the economic impact would be consider- For example, if the average ad valorem transaction cost is 10%, able. Real GDP would rise 1.35% for the European Union, 0.23% accumulated transaction costs in a five-stage supply chain lead for the United States, 0.61% for Japan, 0.56% for other advanced to an ad valorem tariff equivalent of 34%. Doubling the number G20 members (Canada and Australia), 1.87% for China, and 3.32% of stages by slicing up the supply chain more than doubles the for other new emerging/developing G20 members. Exports would total delivery costs, since the tariff equivalent is 75%. rise 3.1% for the European Union, 1.2% for the United States, Recent statistical advances on trade in value added and 4.3% for Japan, 1.9% for other advanced G20 members, 13.5% for related trade costs allow accumulation to be measured with China, and 8.6% for other new emerging/developing G20 mem- actual data. Rouzet and Miroudot (2013) formalized a measure of bers, and imports would rise 2.3% for the European Union, 1% the cumulative tariffs embodied in trade in intermediates along for the United States, 5.2% for Japan, 2.2% for other advanced international supply chains. Although nominal tariffs are low in G20 members, 13.4% for China, and 7.5% for other new emerging/ most OECD economies, indirect tariffs can add a major burden developing G20 members. According to the authors, trade liber- by the time a good reaches its final user. For example, products alization would help exploit the comparative advantage of those imported from India into the European Union have paid a series countries and facilitate their economic upgrading. For example, of tariffs totaling 3.7%, 52% of which is directly levied at the EU the value added of tertiary industries and their GDP share in the border and 48% of which results from duties on intermediate United States, European Union, and other advanced G20 coun- inputs imported by India at previous production stages. tries would rise substantially. The secondary industries in many Building on the pioneering ideas of Wang and others (2016), developing G20 countries would also grow much faster. For exam- who enhanced the analytical tools, Muradov (2016b) developed a ple, the value added of secondary industries in China would rise similar analytical framework that decomposes sector value added by $67 billion, and its GDP share by 0.35 percentage point. or value of its final products along various value chain paths and Accumulated trade costs and their impact on domestic and international value chains • 107 measures the length of each component.12 The decompositions of The largest indirect tariffs were for Indonesia (3.76% direct, 1.33% GVCs at the sector level reveal substantial variation in the length indirect), Australia (2.44%, 1.30%), Chinese Taipei (2.52%, 1.28%), and importance of the relevant parts of the value chain. Using the and Japan (1.39%, 1.28%). The indirect cost due to tariffs was output matrices behind the OECD–WTO Trade international input-­ higher than the direct one only in countries with low nominal pro- in Value-Added database, Muradov (2016b) found that, overall, tection: Luxembourg (0.18%, 1.02%), Malta (0.38%, 0.69%), the industries are moving downstream along the value chain in two- Russian Federation (0.73%, 1.27%), and Greece (0.72%, 0.92%). thirds of the 34 sectors. As in Johnson and Noguera (2016), the In practice, the accumulation effect is lower than the simple results show that GVCs are also gaining importance over domestic exponential formula suggested­ —­ for several reasons. The first value chains in both upstream and downstream directions. is the geography of supply chains. While the image of a chain Of special interest for this chapter, the GVC decompositions implicitly projects a succession of sequential steps, most supply allow the trade costs accumulated along GVCs to be estimated. chains are not linear but are defined by a hub and spoke pattern. Muradov (2016a) found that the direct impact of tariffs (paid on Figure 4.2 shows the topological differences between “spiders” imports) was almost always more significant than the accumu- and “snakes” types of GVC organization (Baldwin and Venables lated tariffs embodied in the cost of production of the products. 2010). FIGURE 4.2 The global value chain zoo: spiders, snakes, and hybrid “snikers” 1. SNAKE 2 3 4 1 5 6 Assembly 5 1 6 2. SPIDER Assembly 2 Final demand 3 4 5 4 3. Hybrid “SNIKER” Assembly 3 1 2 6 Source: Diakantoni and others 2017, based on Baldwin and Venables 2010. 108 • Measuring and Analyzing the Impact of GVCs on Economic Development In the spider first-tier suppliers of parts and components are Not only is the total accumulated trade cost bounded by GVC arranged around a central assembly plant that ships the end efficiency, but for a given structure of efficiency gains the length product to its final destination. Unbundling costs are lower in the of the GVC is negatively correlated with trade costs. As Yi (2003) hub and spoke configuration: inputs cross a border at most twice, showed, the relationship is not linear, and trade costs have to be once as a part and once embodied in final output. In a snake greatly reduced before GVCs start expanding. So it is unrealis- each task is embodied in goods during processing, which are tic to extrapolate accumulating trade costs downward on longer shipped again to the next production stage. At each stage the GVCs where ad valorem trade costs do not decrease. The net gross commercial value of the good in process increases, lead- result between the decrease in ad valorem trade costs (the exog - ing to cascading transaction costs. Diakantoni and others (2017) enous factor) and the resulting increase in GVC length may lead showed that the accumulated trade costs are greatly reduced in to relatively small increases in total accumulated costs ex post. —­ the spider. In real life, actual supply chains are “snikers”­ hybrids Consider a simple simulation exercise based on the hypoth- of spiders and snakes. esis that, for a given product, GVC expansion is endogenous to The other important mitigating factor identified by Diakan - trade costs (figure 4.3). When trade costs are above a certain toni and others (2017) is endogenous to the development of threshold, the length of the GVC measured in border crossings GVCs: supply chains can prosper and develop only when trade is 0: the places of production and consumption coincide, with- costs are low. And only when trade costs are below a certain out a border crossing. Only when trade costs fall beyond certain threshold will a lead firm find it profitable to internationally out- thresholds does it become profitable to shift part of the produc- source part of the production.13 GVCs are Coasian constructs tion to another country that offers efficiency gains larger than that exist only when the incremental benefit from improved the additional trade cost incurred. When trade costs are further complexity (GVC length) is higher than the increased transaction reduced, new outsourcing opportunities may increase produc- cost (box 4.2). tion efficiency by enlarging the supply chain. BOX 4.2 Transaction costs, trade, and foreign direct investment Ronald Coase posited that corporations exist to economize Similarly, firms engage in vertical foreign direct invest- on the transaction costs of markets. After they reach some ment (when they fragment production or slice up the value size, organizational complexity becomes overwhelming, chain because vertical foreign direct investment is motivated and the firm faces diseconomies of scale and scope. What by comparative advantages, unlike older horizontal foreign Coasian economists call transaction costs include all imped- direct investment, which is motivated by market access) iments to cooperation and encompass the trade costs dis- because of cost considerations arising from countries’ cussed here. One aspect of this question, the “make-or- factor cost differences (Alfaro and Chen 2017). Distance has buy” decision (vertical specialization), is central to the rise become less an obstacle for foreign direct investment, as it of GVCs and has been discussed from an international trade did for trade. Using U.S. foreign direct investment outflow perspective by Grossman and Helpman (2002, 2003, 2005), and inflow data for 2001 and 2010, they found that the share who analyzed the determinants of international outsourcing of U.S. outward foreign direct investment concentrated as a function of trade and transaction costs. As firms adopt within 5,000 kilometers fell from around 30% to around 20%. increasingly complex organization and sourcing strategies This change suggests an expansion of foreign direct and as global value chains grow in length and layers, lower investment flow across space in an era when transportation transaction costs become even more essential, especially and communication costs have declined. The trend may also when just-in-time management models make transport and reveal that firms are less risk-averse when it comes to invest- communication a critical component of competitiveness. ing overseas because they perceive that global economic Indeed, an increasingly important component of trans- governance, in particular competition policy, has improved. action cost, especially in GVCs, is information cost. Infor- A growing number of bilateral trade agreements have mation and communication technologies enable firms to included chapters on competition policy that allow foreign better monitor assets and operation (Head and Ries 2008), affiliates to compete on more equal footing with domestic communicate with foreign suppliers and customers (Old- firms. As discussed in chapters 7 and 8 of this book, busi- enski 2012), and substitute for the transfer of technology ness climate and contractual enforcement influence not only embodied in traded intermediates (Keller and Yeaple 2013). make-or-buy decisions, but also the choice between inter- An emerging strand of research analyzes the role of com- national outsourcing (arm’s length subcontracting) and off- munication costs in determining the patterns of trade and shoring through foreign direct investment when rules lack multinational activity. sufficient binding force for disciplining business practices. Accumulated trade costs and their impact on domestic and international value chains • 109 FIGURE 4.3 Ad valorem and accumulated trade costs in delays in importing have large effects on firms’ activities. For snakes and spiders one, the delays require importers to pay extra storage costs. Ad valorem trade cost (%) Further, particularly when producers use imported inputs, delays require producers to reconsider initial production schedules, 80 reducing their productivity. Djankov, Freund, and Pham (2010) found that each additional day a product was delayed by border formalities equated to adding 70 kilometers on average to the distance between trading partners. The effect was particularly Ad valorem 60 Accumulated, snake trade costs pronounced for time-sensitive agricultural goods, where a day’s delay reduced a country’s relative exports by 6%. Hayakawa, Laksanapanyakul, and Yoshima (2016) demonstrate that longer import time reduces export shipment frequency and 40 exports per shipment­ —­and thus total exports. A longer import time raises the marginal cost of production by lowering the pro- duction efficiency and increasing the storage cost. This raises the Accumulated, spider marginal cost and lowers the firm’s total operating profit. Given 20 that firms have to pay fixed costs for each export shipment, the total operating profit will not cover the total fixed costs unless the firm reduces the number of export shipments. As a result, firms that experience longer import times are more likely to 0 reduce the number of export shipments. 0 1 2 3 4 5 6 7 8 9 10 Length of supply chain (number of border crossings) In Thailand doubling the number of days to import would reduce total exports an estimated 3.3% and the number of Source: Diakantoni and others 2017. export shipments an estimated 2.9%.15 Import time also has a major effect on import patterns. Increased import time reduces import shipment frequencies but raises imports per shipment. When GVCs are spiders, the marginal decrease in the ad Specifically, doubling the number of days to import would valorem trade costs may compensate for the additional cost of reduce the number of import shipments an estimated 3.6% and further fragmenting the chain (at the third split). When GVCs increase imports per shipment an estimated 0.1%. As a result, are snakes, the decreasing ad valorem trade costs apply to an total imports would be reduced. In sum, the time spent in one increasing accumulated value of trade in intermediate goods. In stage has effects on both upstream and downstream stages in this case the mitigating effect of lower ad valorem trade costs is international production networks. not as strong as for spiders because the trade costs accumulate In a competing GVC market this has clear implications for exponentially (Ferrantino 2012). But the additional unit cost is upgrading. In apparel value chains the most profitable seg- decreasing, and the net effect becomes negative after a certain ments are in the fashion industry, which is known to run on a point (the seventh split).14 Established GVCs are also vulnerable high degree of uncertainty. With ever-changing trends it is even to a reversal in the decreasing trend in ad valorem trade costs. harder to predict the market and thus to forecast the required If the ad valorem trade costs start increasing again, the GVC raw materials and supplies: only suppliers able to switch produc- length is gradually shortened. tion rapidly and adapt to fast turnovers are likely to be consid- ered to supply this high value-added segment. Less flexible ones Cascading costs and trade facilitation: A world trade will remain confined to the high-volume–low-value segments, network perspective competing on low production costs. Monetary costs are only one of the many facets of trade costs. More generally, it is largely accepted by analysts that all The accumulation of trade frictions from beginning to end of pro- downstream final-good producers prefer timely delivery of duction networks goes against the raison d’être of GVCs, which (imported) intermediate inputs. Hummels and Schaur (2012) require participants to operate in time-critical decentralized sys- modelled exporters’ choices between fast-but-expensive air tems. To realize cost savings in production networks, intermedi- cargo and slow-but-cheap ocean cargo. Shorter delivery times of ate products must be worked on and shipped between produc- shipments lead to greater benefits because they allow importers tion locations and onward into retail distribution systems (and to optimize production flows. If final-good producers can receive then to the final consumer) as efficiently and quickly as possible. and use imported inputs exactly when they need those inputs, Time lost waiting at borders (and related costs of storage and they are assumed to be able to transition smoothly into the pro- the like) are deadweight economic costs within the network. The duction processes, realizing greater production efficiency. time required to import depends on various elements, such as In a trade network this bilateral effect is compounded the efficiency of cargo handling at ports. The import process because the efficient organization of production flows between also takes longer when customs physically inspect cargo. Such two trading partners also depends on the efficiency of upstream 110 • Measuring and Analyzing the Impact of GVCs on Economic Development and downstream GVC partners: the production chain will be as trade by 2.8–4.5% (WTO 2015; G20 TIWG 2016). While the direct swift as its slowest link. Bilateral trade frictions should therefore benefits of trade facilitation will be proportionally higher for be analyzed from a multiplayer perspective, including not only countries not well integrated into international trade because of the other bilateral trading partners, as in conventional gravity their high trade costs, the global benefits will be higher if key models, but also indirect participants that are farther upstream traders at the core of GVCs undertake trade facilitation invest- or downstream in the supply chain. Improving the effectiveness ments (see figure 4.5). Improving trade facilitation for econo - in processing trade with a minimum of frictions will not have the mies below the line would benefit the entire trade community same impact on the world trade network as improving logistic by reducing accumulated trade costs­ —­the farther from the line, and trade facilitation in a country playing the role of a GVC hub. the higher the expected benefits. Six countries (among the 61 Network and graph analysis applied to trade in intermediate in the Trade in Value-Added database) are particularly relevant inputs identifies key players by computing centrality indicators. If from this perspective: Indonesia, the Russian Federation, Brazil, a trading partner (a node or a vertex, in network analysis) “influ- India, China, and Italy. ences just one other node, who subsequently influences many The network approach also suggests that the global benefits other nodes (who themselves influence still more others), then will be higher when trade facilitation investments go to the key the first node in that chain is highly influential” (Borgatti 2005, GVC traders. As mentioned by Hayakawa, Laksanapanyakul, and p. 61). A player’s centrality is therefore a function of both its own Yoshimi (2016), trade costs often take the form of customs delays importance in the world trade economy and the centrality of the due to processing issues such as inconsistencies in Harmonized trading partners it is associated with. System codes between importers and customs, particularly when Trade in intermediate goods is organized along three large the correct applicable Harmonized System code is unclear for a regional clusters­ —­East Asia, centered on China; Europe, cen- product. Those issues can be solved without huge investment tered on Germany; and North America, centered on the United costs­—­for example, by implementing an advance ruling system States­ —­and dense extraregional exchanges (figure 4.4). The East that expedites the delivery of shipments because importers and Asia and Europe regional value chains include several smaller other related parties can inquire about tariff classifications and clusters organized around, for example, Japan and the United duty rates prior to import. Kingdom. To assess the contribution of each economy as a GVC trade facilitator, Diakantoni and others (2017) computed the PageRank Conclusions centrality indicator, which is a more robust centrality indicator than alternative specifications, for each partner. They then com- The accumulation and magnification effects of cascading trade pared the PageRank indicator with various trade and transporta- costs explain why complex GVCs cannot develop when those tion indicators, including the World Bank’s Logistics Performance costs are above a certain threshold (Yi 2003). When the produc- Index, the most appropriate for the purpose (figure 4.5). tion of a final good is fragmented across several countries, trade Trading partners are ranked according to their network cen- costs increase the purchase price of inputs, parts, and compo- trality and compared with their relative performance in timeli- nents. The additional production cost increases the sale price and ness (as measured by a trade facilitation index). An ideal situation is transmitted to the next production step. Those costs accumu- would be to have a perfect fit between GVC centrality and trade- late in the supply chain through a cascading effect and are ulti- cost efficiency. When that is not the case, the analysis identi- mately embodied into the higher price paid by the final consumer. fies where trade facilitation investments would have the largest Overall, trade frictions would translate into an average global impact. The hypothesis is that investments in upgrading increase of 18% of the production cost in a single stage of the trade-facilitation performance will have a large positive spillover value chain. Most of the trade frictions result from transporta- and be highly profitable for global welfare when they improve tion costs and deficient logistic and trade facilitation conditions: the situation of a key player. A perfect fit between centrality and their incidence is estimated at an ad valorem tariff equivalent of trade facilitation would show all countries aligned on the diago- 17%. While some of these nontariff costs are outside the realm nal, which is far from the case. There is a large mismatch between of national policymakers (as with geographic distance from the the quality of trade and transport facilitation and the role of each trading partner or sharing a common language), many fall under economy in the world trade network. the control of domestic policy (logistics performance, cost of The benefits of improving trade facilitation are usually mea- doing business, and so on). sured using the traditional bilateral trade perspective, which is Cascading trade costs not only penalize final consumers, they only part of the bigger GVC picture. The OECD has estimated also erode the competitiveness of domestic industries on inter- the bilateral benefits of reducing trade costs from full implemen- national markets and lower the effectiveness of export-led indus- tation of the WTO Trade Facilitation Agreement at 16.5% of total trialization strategies. Steep trade cost escalation creates a large costs for low-income countries, 17.4% for lower-middle-income anti-export bias on complex manufactured goods when value countries, 14.6% for upper-middle-income countries, and 11.8% added is the traded commodity. This bias creates additional for OECD countries. Together, these estimates imply that a 1% obstacles for export diversification and GVC upgrading. Besides reduction in trade costs has the potential to increase bilateral tariff and transportation, nonmonetary costs, particularly delays Accumulated trade costs and their impact on domestic and international value chains • 111 FIGURE 4.4 Graphical representation of trade in intermediate goods, 2011 Czech Republic Austria Hungary Sweden Poland France Ireland Netherlands Italy Russian Federation United Kingdom Germany Canada Belgium Turkey Switzerland Mexico United States Spain Korea, Rep. China Brazil Chinese Taipei Portugal Japan Singapore India Australia Hong Kong, China Israel South Africa Thailand Malaysia Indonesia Argentina Chile Viet Nam Source: Diakantoni and others 2017, based on the UN Comtrade database (https://comtrade.un.org). Note: Includes the 61 economies in the Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database and their most important bilateral gross trade flows. and uncertainty, are particularly relevant when the manufacture Trade costs vary by sector and country. Outside agriculture, of merchandise is fragmented across several countries. Delays in the costliest sectors, as measured with the extended effective a just-in-time business model disrupt the whole supply chain and protection margin, are motor vehicles, transport equipment, render the entire process inoperable. petroleum products, computers, and machinery. Primary sectors 112 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 4.5 PageRank scores and Logistics Performance Index values, 2011 PageRank centrality indicator 60 Costa Rica Iceland Latvia Luxembourg New Zealand Croatia Cambodia Tunisia Lithuania 50 Slovenia Estonia Bulgaria Colombia Greece Norway Portugal Saudi Arabia Slovak Republic Ireland 40 Philippines Finland Romania South Africa Israel Argentina Denmark Hungary Viet Nam 30 Chile Mexico Austria Malaysia Singapore Czech Republic Canada Australia Thailand Indonesia 20 Sweden Turkey Chinese Taipei Poland Hong Kong, China Switzerland Brazil Japan Spain India Russian Federation 10 Belgium United Kingdom Republic of Korea Netherlands France United States Italy Germany China 0 0 10 20 30 40 50 60 Logistics Performance Index Source: Diakantoni and others 2017. Note: The diagonal line shows equal ranking on the Logistics Performance Index and on PageRank. carry the lowest trade costs because they require few inputs price-out second-tier domestic firms. These mitigating policies in the production chain. Small and low-income countries tend are only second-best alternatives to fully fledged trade facilita- to suffer more from trade costs: Cambodia ranks as the most tion when it comes to deepening domestic interindustrial links. expensive country in additional trade costs. Reducing tariff and nontariff trade costs globally through multi- The smaller domestic value added share in developing econ- lateral agreements is thus fully consistent with the interests of omies’ manufactured exports, compared with that in developed developing economies because it lowers their cost of GVC countries’ economies, tends to amplify the impact of trade costs participation. through the magnification effect. From a trade and development Finally, in a production network, bilateral trade costs tell only perspective higher-than-average trade costs marginalize low- part of the story. In a close-knit network, competitiveness also income countries and prevent them from joining international depends on the costs faced by trading partners and by trade supply chains. They may still compete by further reducing the competitors. Poor trade facilitation among countries that rank wages paid to workers and the gross profit retained by the firm, highly in GVC trade (at or close to the heart of regional networks) but such a race to the bottom would severely limit their potential impose a systemic cost both to themselves and to the rest of for industrial and social upgrading. the trade community. The welfare benefits of the WTO Trade Many developing countries intend to lower their trade costs Facilitation Agreement from gains from trade will be enjoyed by by setting up duty drawback schemes and export processing the implementing economy, by its direct trading partners, and zones. But the effect is limited in time and scope, because they by the entire community. This magnified effect of trade facilita- compensate exporting firms for the additional production costs tion is directly attributable to the way trade costs accumulate in only when they use imported inputs. Such strategies tend to GVCs. Accumulated trade costs and their impact on domestic and international value chains • 113 ANNEX 4.1 Extended effective protection rates and the relative price of value added Effective protection rates, in their original formulation, are calcu- To analyze more precisely the impacts of trade costs on lated by deducting the additional production cost that manufac- competitiveness as well as some mitigating measures that the turers have to pay because of the tariff charged on tradable inputs exporting country could implement, it is important to distinguish from the additional benefit generated by selling their product at a between the costs of domestic (superscript h) and foreign inputs price higher than the free-trade market price, thanks to the duties (superscript f ). Extended effective protection rates can be writ- charged on competitive imports. The effective rate of protection ten as: (EEPR) on tradable good j is the difference between Vj, the value tj – [ ∑i (ti × afij) + ∑i (ti × ahij)] added obtained on the domestic market (with process influenced EEPRj = – 1(A4.1.4) by trade costs), and V *j, the value added that would be generated in 1 – ∑iaij the absence of policy and natural trade barriers, expressed as a pro- where afij is the intermediate consumption i from the foreign portion of the frictionless value added. It is given by the expression: country required to produce one unit of output j and ahij is the EEPRj = (Vj – V *j) / V *j(A4.1.1) intermediate consumption i from the home country required to produce one unit of output j. Substituting products for industries, equation A4.1.1 can be Even when duty drawbacks or tariff exemptions correct for output notation: expressed in standard input-­ trade frictions and allow domestic producers to purchase inputs pj × tj – ∑i (ti × aij) at international prices (as in export processing zones), export-­ EEPRj = – 1(A4.1.2) pj – ∑iaij oriented firms still have a disincentive to purchase inputs inter- nally from second-tier domestic suppliers, represented by the sum where pj is the nominal price of output j at the frictionless trade [∑i (ti × ahij)]. The first-tier domestic suppliers exporting their prod- price; aij are elements of the matrix A of technical coefficients in ucts to other participants in the international supply chain remain an input-­ output matrix at the frictionless trade price of inputs  i;16 at a disadvantage to their free-trade competitors (right side of tj is 1 + the rate of ad valorem tariff and transport nominal protec- equation A4.1.5) when they source some of their inputs from tion on sector j, where tj ≥ 1; and ti is 1 + the rate of ad valorem other local suppliers or outsource some of their tasks to them: nominal tariff and transport protection on inputs purchased (1 – [ ∑iafij + ∑i (ti × ahij)]) < (1 – ∑iaij) (A4.1.5) from sector i, where ti ≥ 1. i can be equal to j when a firm pur- chases inputs from other firms in the same sector of activity. In In other words, export processing zones or drawbacks price- an intercountry framework i also includes the partner dimension out domestic suppliers when nominal tariffs and trade costs are [c] because inputs from sector i might be domestic or imported. high. To summarize the main implications of the formal model, In the trade literature this expression is often simplified into: even in the absence of tariff and transport cost escalation and a flat extended effective protection rate, trade frictions reduce the t’j – ∑i (t’i × aij) EEPRj = (A4.1.3) competitiveness of domestic firms, most frequently when they 1 – ∑iaij are price takers and compete on the global market at interna- tional prices. where t ’i and t ’j are the rates of ad valorem protection, where t ’i > 0. 114 • Measuring and Analyzing the Impact of GVCs on Economic Development ANNEX 4.2 Measuring the length of global value chains and the number of border crossings The analysis of trade costs embodied in multistage international The length of the GVCs can be factored in by using geo- production processes is often carried out using international input-­ graphic distance or monetary transportation costs between two output models. The calculations have been made possible by the inter-related industries instead of counting production stages, as availability of the underlying input-­ output tables: Koopman and in the average propagation length formulation shown in equa- others (2010) estimate the cumulative effect of transportation and tion A4.2.2. This calculation was suggested by Los and Temur- tariff margins using Global Trade Analysis Project Multi-Country shoev (2012). Once the distance between the supplying firm and Input-­Output tables; Tamamura (2010) uses the Institute of Devel- its clients (dij) is known, a vector of input-weighted distance from oping Economies–Japan External Trade Organization international customers to suppliers provides a geographical distance (dij is in tables to estimate the impact of regional trade agreements. kilometers) or an economic one (if dij is in monetary terms). The Length is most often estimated using the concept of average distance covered by the global value chain between its initiation propagation length applied at the international level in Dietzen- and the final consumer is given by: bacher and Romero (2007) for major European countries and by D: (I – A’ ) –1 D*(A4.2.2) Inomata (2008) for Asia. The average propagation length rep- resents the average number of production stages lining up in where D* is a diagonal input-weighted matrix of supplier-to-­ every branch of all the given supply chains. It is a shorthand rep- client distance by industry and country.18 Miroudot and Nord- resentation for an industry’s level of fragmentation, which relies stöm (2015) adapt this methodology to measure the length of on weighting the distance index by successive powers k of the the external network of suppliers, sourcing the distance from the technical coefficient matrix A. A k s are regarded as progressive GeoDist database maintained by the Centre d’Études Prospec- impacts of the initial demand when supply chains are sliced at tives et d’Informations Internationales (Mayer and Zignago 2011). the k th stage of the production process. The average propaga- Another way of looking at GVC length is to estimate the tion length is defined as: number of border crossings from the first step (most upstream aij [A 2]ij [A 3]ij sector/country) to final demand. Such a decomposition also APL j –i = 1 × + 2 × + 3 × + … (A4.2.1) (lij – δij) (lij – δij) (lij – δij) allows an industry to be located in relation to its situation in the supply chain (upstream or downstream). Wang and others (2016) where lij is Leontief inverse coefficients [I – A] –1 and δij is a Kro- synthesize the various backward and forward measures by defin- necker delta product that is δij = 1 if i = j and δij = 0 otherwise. ing a GVC position index based on a thorough decomposition In other words, the average propagation length is the of the contribution of each production stage to the total value. weighted average of the number of production stages that an Their index measures the distance from any production stage impact from industry j goes through until it ultimately reaches between the final demand and the initial factor inputs in a pro - industry i, using the strength of an impact at each stage as a duction line by a combination of production links based on both weight. By construction, the average propagation length shows forward and backward links. The length of the international part a rapidly decaying effect after the second round, because of of supply chains (the one subject to cumulative tariff and trans- the low value of the nondiagonal coefficients in the Leontief portation costs) varies from country to country and sector to matrix. As a result, the value of the additional element of the sector. average propagation length suite falls rapidly to zero after the Muradov (2016a) proposes a new approach to quantify the second production stage. That the A k s tend toward zero when k accumulation of trade costs and the average number of back- increases is actually a condition for the suite to converge to the ward and forward border crossings. When the input-­ output Leontief inverse [I – A] –1, a result central to most GVC indicators. coefficients are calculated at basic prices (the most common situ- The smaller the value of extra-diagonal technical coefficients, the ation), trade costs can be integrated into the A matrix by adding faster is the convergence to zero. an additional row of trade margins. His method also relies on the This is particularly true for the international coefficients (those use of an alternative to the Leontief matrix to compute a “global” outside the bloc-diagonal matrices representing the domestic inverse, disaggregating ex ante (instead of ex post, as in other interindustrial exchanges) and reflects the fact that most coun- approaches) the diagonal and off-diagonal blocs corresponding tries are largely self-sufficient in intermediate inputs. Therefore, to, respectively, domestic and international transactions in the A the foreign component of A k (coefficients outside the bloc-diag- matrix. Diakantoni and others (2017) discuss the interpretation of onal of domestic industries) is rapidly insignificant from an eco - those trade margins when some of the trade costs are embodied nomic perspective when k increases.17 in domestic inputs. Accumulated trade costs and their impact on domestic and international value chains • 115 Notes transportation costs influence the domestic price of all inputs, includ- ing domestically produced ones (goods, but also services). 1. Measuring the input use and value-added contributions along the 9. The last year for which Trade in Value-Added database data were production chain from beginning to end since the mid-1970s, the available is 2011, and 2006 is the first one where preferential tariffs five stylized facts are the ratio of world value-added to gross exports were available for all trade partners on a comparable basis. (an indicator of GVC trade) has fallen over time, by roughly 10 per- 10. Even when the extended effective protection rate is negative, as in centage points; this ratio has fallen for manufacturing but has risen the mining sector, trade frictions still reduce the competitiveness of outside of manufacturing; changes have been heterogeneous across domestic firms when they compete on the global market at interna- countries, with fast growing countries seeing larger declines in the tional prices while still paying domestic prices for their inputs. ratio of their value-added to gross exports; declines in value added 11. The nontariff trade costs from Duval, Saggu, and Utoktham (2015) to export ratios have been larger for proximate partners that entered include factors other than freight and insurance costs. into regional trade agreements; and declines in value added to export 12. The paper is also of interest since it surveys the state of the art and ratios have been larger for country pairs that entered into regional brings together the results of alternative decompositions. The pre- trade agreements. sentation of these decomposition techniques, which rely on the inter- 2. The author shows that in the presence of trade in intermediates GDP output matrix and its mathematical properties, would national input-­ is not a good proxy for economic mass. As Noguera (2012) explains, require complex calculus. For example, the decomposition of the deriving a gravity equation for bilateral value-added trade is com- number of transactions along the downstream value chain in Muradov plicated by the nonlinear relationship between the value added and (2016b) results in as much as twelve indicators (see annex 4.2 for an final-good demands. Trade costs affect trade in value added through introduction). their effect both on bilateral gross trade and on production sharing 13. In practice, the lead firm may have strategic objectives in international arrangements, but also through the trade costs corresponding to outsourcing that go beyond pure cost-efficiency, but this chapter other pairs of countries in the supply chain. focuses only on value added and production costs. 3. In competitive markets GVC trade exists only when trade costs are 14. Intuitively, the existence of an inflection point can be explained as fol- lower than the efficiency gains of fragmenting the supply chain and lows: when trade costs are very high, accumulated cost is 0 because outsourcing the tasks. So by definition accumulating trade costs have no trade takes place; when trade is frictionless, accumulated trade an upper limit. In a competitive market where all efficiency gains are costs are also 0 because there are no trade costs. So, between these translated to the price of the final product, any increase in trade costs two extreme positions, accumulated trade costs should increase with will be paid by the consumer. In a semi-monopolistic market the effi- the length of the GVC up to a maximum, then decrease afterward. ciency gains will accrue mainly to the lead firm. 15. The authors use transaction-level export and import data from 2007 4. The effective tariff rate on output differs from the effective rate of pro- to 2011 that cover all commodity exports and imports in Thailand. tection as it is usually understood in trade analysis and is used later in 16. Input coefficients aij are calculated by dividing input values of goods the chapter. It contemplates only the nominal protection on output and services used in each industry by the industry’s corresponding but excludes the additional production cost on inputs. total output. That is, aij = zij / Xj , where zij is a value of good/service i 5. Input coefficients aij are calculated by dividing input values of goods purchased for the production in industry j, and Xj is the total output and services used in each industry by the industry’s corresponding of industry j. Thus, the coefficients represent the direct requirement of total output. That is, aij = zij / Xj , where zij is a value of good/service i inputs for producing just one unit of output of industry j. purchased for the production in industry j, and Xj is the total output output 17. This is true for the industry average, represented in an input-­ of industry j. Thus, the coefficients represent the direct requirement of matrix. This average is prone to aggregation bias, and export-ori- inputs for producing just one unit of output of industry j. ented industries may be much more reliant on imported inputs than 6. The exporting firm is considered to be a price taker that cannot the average domestic firm. This can be observed in the Trade in impose higher prices and will have to compete on the global market at Added database, which distinguishes several types of firms in Value-­ international prices. Incidentally, this result explains why small firms do China and Mexico. not export as much as large firms in the more realistic situation where 18. For each domestic industrial sector, an average distance to interna- some of the trade costs are not ad valorem fees but are sunk costs. tional supplier is calculated, weighting the distance to the suppli- 7. This tactic may be used to gain a contract, but it is not sustainable er’s country by its share in the total inputs imported by the domes- in the long term if the firm wishes to retain skilled staff or invest and tic industry. From a purely international trade perspective, domestic expand its production capacity. interindustry commerce should be set to 0 (the distance between two 8. The negative impact of high extended effective protection rates on domestic firms is nil), but this is an oversimplification for some devel- second-tier domestic suppliers and the perspective of GVC upgrad- oping countries, where most people live in coastal areas and inland ing in developing countries derives from the fact that tariff and transportation is more expensive than international freight. 116 • Measuring and Analyzing the Impact of GVCs on Economic Development References ———. 2005. “Outsourcing in a Global Economy.” Review of Economic Studies 72 (1): 135–59. Alfaro, L., and M. Chen. 2017. “Transportation Cost and the Geography of Hayakawa, K., N. Laksanapanyakul, and T. Yoshimi. 2016. “Effect of Import Foreign Investment.” In Handbook of International Trade and Transpor- Time on Export Pattern.” Institute of Developing Economies (IDE) Dis- tation, edited by B. Blonigen and W. Wilson. Cheltenham, U.K.: Edward cussion Paper No. 566, IDE–Japan External Trade Organization, Chiba, Elgar Publishing. Japan. Anderson, J., and E. van Wincoop. 2004. “Trade Costs.” National Bureau Head, K., and J. Ries. 2008. “FDI as an Outcome of the Market for Corpo- of Economic Research (NBER) Working Paper No. 10480, NBER, Cam- rate Control.” Journal of International Economics 74 (1): 2–20. bridge, MA. Head, K., and T. Mayer. 2013. “What Separates Us? Sources of Resistance Balassa, B. 1965. “Tariff Protection in Industrial Countries: An Evaluation.” to Globalization.” Canadian Journal of Economics 46 (4): 1196–1231. Journal of Political Economy 73 (6): 573–94. Hummels, D. 2007. “Transportation Costs and International Trade in the Baldwin, R., and A. Venables. 2010. “Spiders and Snakes: Offshoring and Second Era of Globalization.” Journal of Economic Perspectives 21 (3): Agglomeration in the Global Economy.” NBER Working Paper No. 131–54. 16611, NBER, Cambridge, MA. Hummels, D., and G. Schaur. 2012. “Time as a Trade Barrier.” NBER Work- Baldwin, R., and D. Taglioni. 2011. “Gravity Chains: Estimating Bilateral ing Paper No. 17758, NBER, Cambridge, MA. Trade Flows When Parts and Components Trade is Important.” NBER Ing, L.-Y., O. Cadot, and J. Walz. 2016. “Transparency in Non-Tariff Mea- Working Paper No. 16672, NBER, Cambridge, MA. sures: An International Comparison.” Fondation pour les Études et Borgatti, S. 2005. “Centrality and Network Flow.” Social Networks 27 (1): Recherches sur le Développement International (FERDI) Working Paper 55–71. 182, FERDI, Clermont-Ferrand, France. Ciuriak, D., and J. Xiao. 2014. Should Canada Unilaterally Adopt Global Inomata, S. 2008. “A New Measurement for International Fragmentation Free Trade? Ottawa: Canadian Council of Chief Executives. of the Production Process.” IDE Paper 175, IDE–JETRO, Chiba, Japan. Corden, W. M. 1966. “The Structure of a Tariff System and the Effective Johnson, R. 2014. “Five Facts about Value-Added Exports and Implications Protection Rate.” Journal of Political Economy 74 (3): 221–37. for Macroeconomics and Trade Research.” Journal of Economic Per- Diakantoni, A., H. Escaith, M. Roberts, and T. Verbeet. 2017. “Accumulat- spectives 28 (2): 119–42. ing Trade Costs and Competitiveness in Global Value Chains.” World Johnson, R., and G. Noguera. 2016. “A Portrait of Trade in Value Added Trade Organization (WTO) Working Paper Economic Research and Sta- over Four Decades.” NBER Working Paper No. 22974, NBER, Cam- tistics Division (ERSD) 2017–2, WTO, Geneva. bridge, MA. Dietzenbacher, E., and I. Romero. 2007. “Production Chains in an Inter- Keller, W., and S. R. Yeaple. 2013. “The Gravity of Knowledge.” American regional Framework: Identification by Means of Average Propagation Economic Review 103 (4): 1414–44. Lengths.” International Regional Science Review 30 (4): 362–83. Koopman, R., W. Powers, Z. Wang, and S-J. Wei. 2010. “Give Credit Where Djankov, S., C. Freund, and C-S. Pham. 2010. “Trading on Time.” The Credit Is Due: Tracing Value Added in Global Production Chains.” NBER Review of Economics and Statistics 92 (1): 166–73. Working Paper No. 16426, NBER, Cambridge, MA. Duval, Y., A. Saggu, and C. Utoktham. 2015. “Value Added Trade Costs in Koopman, R., Z. Wang, and S-J. Wei. 2014. “Tracing Value-Added and Goods and Services.” United Nations Economic and Social Commission Double Counting in Gross Exports.” American Economic Review 104 for Asia and the Pacific (ESCAP) Trade and Investment Working Paper (2): 459–94. 01, ESCAP, Bangkok. Lewis, D. 1994. “Freight Mode Choice: Air Transport Versus Ocean Trans- Escaith, H., and S. Miroudot. 2015. “World Trade and Income Remain port in the 1990’s’ Flight.” Cambridge, MA: Transportation Laboratory, Exposed to Gravity.” In The Global Trade Slowdown: A New Normal?, Massachusetts Institute of Technology. edited by B. Hoekman, 127–60. London: Centre for Economic Policy Los, B., and U. Temurshoev. 2012. “Distance-Based Measures of Global- Research. ization in a World with Fragmented Production.” Presentation at the Ferrantino, M. 2012. “Using Supply Chain Analysis to Examine the Costs World Input-Output Database (WIOD) Conference on Causes and Con- of Non-Tariff Measures (NTMs) and the Benefits of Trade Facilitation.” sequences of Globalization, Groningen, Netherlands, April 24–26. WTO Working Paper ERSD-2012–02, WTO, Geneva. Ma, A. C., and Ari Van Assche. 2010. “The Role of Trade Costs in Global G20 TIWG (Trade and Investment Working Group). 2016. “Discussion Production Networks Evidence from China’s Processing Trade Regime.” Paper on Monitoring and Reducing Trade Costs.” Second Meeting of World Bank Working Paper 5490, World Bank, Washington, DC. the Trade and Investment Working Group, Nanjing, China, April 4–5. Mayer, T., and S. Zignago. 2011. “Notes on CEPII’s Distances Measures: Ghodsi, M., and R. Stehrer. 2016. “Non-Tariff Measures Trickling through The GeoDist Database.” Centre d’Études Prospectives et d’Informa- Global Value Chains.” Productivity, Non-Tarrif Measures and Openness tions Internationales (CEPII) Working Paper 2011–25, CEPII, Paris. (PRONTO) Working Paper, Vienna Institute for International Economic Miroudot, S., and H. Nordström. 2015. “Made in the World.” European Studies, Vienna. University Institute (EUI) Robert Schuman Centre for Advanced Studies Grossman, G., and E. Helpman. 2002. “Integration versus Outsourcing in Working Paper 2015/60, EUI, Florence. Industry Equilibrium.” Quarterly Journal of Economics 117 (1): 85–120. Muradov, K. 2016a. “Trade Costs and Borders in the World of Global Value ———. 2003. “Outsourcing versus FDI in Industry Equilibrium.” Journal of Chains.” Nineteenth Annual Conference of Global Economic Analysis, the European Economic Association 1 (2–3): 317–27. Washington, DC, June 15–16. Accumulated trade costs and their impact on domestic and international value chains • 117 ———. 2016b. “Structure and Length of Value Chains.” Draft paper sub- Tamamura, C. 2010. “Cost Reduction Effects of ‘Pseudo FTAs’ in Asia: mitted to the workshop on Regional/Multi-Regional Input-Output Anal- Application of a Price Model Based on a Multilateral I/O table.” IDE ysis, Osnabrück, Germany. Discussion Paper No. 226, IDE–JETRO, Chiba, Japan. Noguera, G. 2012. “Trade Costs and Gravity for Gross and Value Added Timmer, M., B. Los, G. de Vries, and R. Stehrer. 2016. An Anatomy of the Trade.” Job Market Paper, University of California, Berkeley and Colum- Global Trade Slowdown based on the WIOD 2016 Release. Groningen bia University, Berkeley, CA. and New York, NY. Growth and Development Centre Research Memorandum 162. Gronin- OECD (Organisation for Economic Co-operation and Development) and gen, Netherlands: University of Groningen. WTO (World Trade Organization). n.d. Trade in Value-Added (TiVA) Wang, Z., S.-J. Wei, X. Yu, and K. Zhu. 2016. “Characterizing Global Value database. Paris: OECD. Available at: http://www.oecd.org/sti/ind Chains.” Working Paper 578, Stanford Center for International Devel- /measuringtradeinvalue-addedanoecd-wtojointinitiative.htm. opment, Stanford, CA. Oldenski, L. 2012. “Export Versus FDI and the Communication of Complex WTO (World Trade Organization). 2015. World Trade Report Information.” Journal of International Economics 87 (2): 312–22. —­ 2015­ Speeding up trade: benefits and challenges of implementing Rouzet, D., and S. Miroudot. 2013. “The Cumulative Impact of Trade Bar- the WTO Trade Facilitation Agreement. Geneva: WTO. riers Along the Value Chain: An Empirical Assessment Using the OECD Yi, K.-M. 2003. “Can Vertical Specialization Explain the Growth in World Inter-Country Input-Output Model.” Paper prepared for the 16th Annual Trade?” Journal of Political Economy 111 (1): 52–102. Conference on Global Economic Analysis, Shanghai, June 12–14. Yu, X., R. Wang, J. Yang, and Z. Wang. 2016. “Policy Options on Interme- Samuelson, P. 1954. “The Transfer Problem and the Transport Costs.” The diate Goods Trade under the Era of Global Value Chains.” University of Economic Journal 64 (254): 264–89. International Business and Economics, Beijing. CHAPTER 5 The middle-income trap and upgrading along global value chains JAKOB ENGEL AND DARIA TAGLIONI A fter acting as the primary drivers of global growth in show substantial upward mobility between 2000 and 2015, par- the years immediately following the 2007–08 global ticularly for middle-income countries, with 79 of 133 countries financial crisis, emerging and developing economies that were low or middle income in 2000 improving their income experienced a substantial decline. Having exceeded status and none declining (table 5.1). 4% a year from 2010 to 2014, their growth declined to 3.4% in After addressing some of the definitional issues, the chapter 2015 and to an expected 3.5% in 2016, with commodity produc- reviews recent cross-country and case-study literature on the ers projected to grow only 0.4% in 2016 (figure 5.1; World Bank middle-income trap­ —­its causes and its possible solutions. 2016a). The end of the commodity boom and concerns about It then links the middle-income trap to the emergence and financial stability in many emerging economies led Haldane growing significance of trade through global value chains (GVCs). (2015, p. 13) to argue that after the Anglo-Saxon crisis of 2008/09 The role of GVCs in trade and investment flows dominates aca- and the euro-area crisis of 2011/12, “we may now be entering demic and policy debates on trade and industrial development the early stages of Part Three of the [crisis] trilogy, the ‘Emerging (see Gereffi 2014; Taglioni and Winkler 2016). One characteristic Market’ crisis of 2015 onwards.” Many emerging economies face of GVC trade is the denationalizing of comparative advantage, high corporate debt and excess capacity, leaving them vulner- which could allow countries to industrialize by joining GVCs able to unexpected domestic or global events (IMF 2016). And rather than by building their own (Baldwin and Lopez-Gonzalez many of the world’s largest middle-income countries­ —­including 2015). So integration into GVCs has been widely viewed as a stra- Argentina, Brazil, China, Indonesia, Nigeria, and the Russian tegic pillar for developing countries to become more competi- Federation­ —­ have seen substantial slowdowns.1 In this economic tive, to develop the skills and human capital of their labor force, climate, gaining better understanding of different growth trajec- and to acquire technology to industrialize and move into higher tories and the obstacles middle-income countries face in sustain- value-added production. Whether such economic upgrading is ing rapid growth becomes all the more relevant. happening­ —­ and if so, where and how­ —­remains subject to much This chapter examines in greater depth the middle-income debate and speculation. trap­—­ in which high sustained growth becomes increasingly dif- This chapter surveys these two debates­ —­ on the middle- ficult once a country reaches GDP per capita of around $10,000. income trap phenomenon and on countries’ ability to grow and The term, coined by Gill and Kharas (2007) in relation to growth develop through GVC participation­ —­ and asks whether integra- prospects in Asia, remains ambiguous and is interpreted in var- tion into GVCs can help countries avoid a middle-income trap ious ways, producing different empirical findings and policy and, if so, why and under what circumstances. The primary focus recommendations. Indeed, a trap specific to middle-income is examining how the factors that are hypothesized to contrib - countries is disputed (see Pritchett and Summers 2014; Im and ute to growth slowdowns at middle income may also impede Rosenblatt 2013; Roy and others 2016), and the data tend to economic upgrading through GVCs. The literature is limited on 119 120 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 5.1 GDP per capita growth, by income group, 2006–15 10 Lower middle income 5 Low income Upper middle income High income 0 –5 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Source: World Development Indicators database. TABLE 5.1 Share of all countries in a given income group in 2000 and 2015 Percent Income group in 2015 Low Lower middle Upper middle High income income income income Total Low income 47.6 47.6 4.8 0.0 100 Lower middle income 0.0 37.7 58.5 3.8 100 Income group Upper middle income 0.0 0.0 51.4 48.6 100 in 2000 High income 0.0 0.0 0.0 100.0 100 Total 14.6 24.4 25.8 35.1 100 Source: World Bank County and Lending Group Classification (see https://datahelpdesk.worldbank.org/knowledgebase/articles/906519-world-bank-country -and-lending-groups). the relationship between the two debates, but empirical analy- coherent and applicable set of policy recommendations ses have found some evidence that GVC participation supports to address the causes of growth slowdowns and structural escape from the dynamics hypothesized in the middle-income stagnation. trap literature, albeit with substantial variation (Kummritz and • The need for developing countries to adapt to trade through others 2016; Boffa and others 2016). However, this chapter does globally integrated value chains in goods, services, and not assign causality; it instead asserts that while GVC integra- information presents a partial but important conceptual par- tion can support sustained high growth rates for middle-income adigm and policy framework to identify levers for middle- countries, a certain level of development and industrial com- income countries to converge with richer countries. The eco- plexity also tends to be a prerequisite for participation in more nomic complexity and institutional sophistication required sophisticated, higher value-added GVCs. to upgrade into higher value-added tasks and products over The chapter makes four central claims: time­—­ in the context of the emergence of globally integrated • The two debates have existed mostly independent of each lead firms­ —­are lacking in many middle-income countries. other, but they should be bridged. The factors that constrain • The institutional, macroeconomic, trade, and industrial policies GVC participation and upgrading provide a more granular required for successful GVC participation can also address eco- perspective of tasks, products, and industries­ —­and a more nomic stagnation among trapped middle-income countries. The middle-income trap and upgrading along global value chains • 121 • Emerging technological changes are likely to further complicate to change the economic structure from an input-driven growth the ability to develop by integrating into and upgrading within model into a productivity-driven growth model.” GVCs unless countries explicitly address the links between pro- Three approaches to assessing when a country is stuck in a duction and distribution and between economic and social middle-income trap have emerged: one on absolute conver- change. This informs a broad set of policy recommendations gence to high-income countries, one on relative convergence, —­ that­ while requiring more nuanced targeting and adapta- and one on structural change, going beyond income-related tion specific to each country and sector­ —­provide a promising measures of development. The approaches are not mutually framework for overcoming difficulties specific to middle-income exclusive, and even studies focused on assessing convergence countries in the age of automation and digitization. dynamics­ —­and in many cases not finding any unique middle- The chapter first goes into greater depth on the debates sur- income country trap­ —­generally acknowledge that specific rounding the middle-income trap and clarifies the main terms. It structural changes are required for middle-income countries to then provides a framework for viewing transitions from low- to increase their income. middle-income status and from middle- to high-income status through a GVC lens­ —­and the GVC-related factors that medi- Absolute convergence. Drawing on Hausmann and others’ (2005) ate these transitions. It then examines emerging technological definition of growth slowdowns, Eichengreen and others (2013) and economic factors and trends that are likely to make efforts stipulated three conditions for a growth slowdown to be clas- to escape the middle-income trap through participation in GVCs sified as a middle-income trap: a seven-year average growth more complex in the medium term and offers some potential rate of GDP per capita of at least 3.5% prior to the slowdown, a policy solutions. decline in the seven-year average growth rate of GDP per capita of at least 2 percentage points, and GDP per capita greater than $10,000 in 2005 international purchasing power parity prices. Definitions and implications of the middle- They find a bimodal middle-income trap at GDP per capita of income trap $10,000–$11,000 and $15,000–$16,000, suggesting that growth in middle-income countries slows in two main stages. The term “middle-income trap” was coined almost a decade ago by Gill and Kharas (2007), who discussed three transformations Relative convergence. Felipe and others (2012) focused on how that modern growth theory predicted middle-income countries long it took countries to cross income thresholds and defined a in East Asia would experience. First was the slowing and rever- lower-middle-income trap as a country failing to attain average sal of diversification as countries became more specialized in growth of income per capita of at least 4.7% a year and an upper- production and employment. Second was the declining impor- middle income trap as a country failing to attain average growth tance of investment and the acceleration of innovation. And third of income per capita of at least 3.5% a year. The relative approach was the shift in education systems to equip workers with the is exemplified by Aiyar and others (2013), who regressed growth skills not just to adjust to new technologies, but also to shape in GDP per capita on lagged income and measures of physical new products and processes. They noted that many Southeast and human capital to come up with a predicted growth rate. The Asian countries stagnated and failed to make the transition to residuals of this regression are defined as actual growth minus productivity-driven growth. While the term middle-income trap estimated growth, and a slowdown takes place when a substan- was novel, the concept was not­ —­drawing on earlier work on low- tial deviation in actual versus expected growth is sustained over level equilibrium traps (Nelson 1956), poverty traps (Leibenstein 10 years. Robertson and Ye (2013) likewise used the growth rate 1962; Aazariadis and Drazen 1990; Kraay and Raddatz 2007), and of income relative to the United States as their dependent vari- globalization’s missing middle (Garrett 2004). able. Similarly, Huang (2016) defined this process of stagnation as an economy’s ability to continue to grow more rapidly than the What is the middle-income trap, does it exist, and how United States after reaching middle-income status. Furthermore, can it be measured? not even the World Bank and International Monetary Fund defi- A large and growing body of literature focuses on whether the nition of a middle-income country is considered a helpful bench- term is useful for examining the problems facing industrializing mark by all researchers: Aiyar and others (2013) and Roy and others countries. As Gill and Kharas (2015) noted, after 10 years and (2016) assigned a country middle-income status if its GDP range more than 300 articles the term remains poorly defined and was 15–50% of U.S. income, depending on the specification. backed by almost no formal modeling, with very few exceptions (Agénor and Canuto 2015; Dabús and others 2016). However, two Structural change. A third approach, while not contradicting the dominant definitions of the middle-income trap have emerged. relative and absolute convergence approaches, focuses less on At its most basic, the trap is seen as sustained economic stagna- quantitative measures of growth slowdowns and more on the tion. Egawa (2013, p. 2) argues that it is “a situation in which an structure of the country’s economy and on processes of trans- MIC [middle-income country] falls into economic stagnation and formation. Dingemans (2016, p. 644) defined Chile’s middle- becomes unable to advance its economy to a high-income level income trap as the country’s “inability to (incrementally) diversify for certain reasons specific to MICs” related to “a delay or failure and enhance its export trade.” Ohno (2009, p. 1) argued that the 122 • Measuring and Analyzing the Impact of GVCs on Economic Development defining characteristic of the middle-income trap is a country’s growth trajectories that “do not conform to one clear pattern failure “to build a national mindset and institutions that encour- that can be easily characterized as a ‘trap.’” age constant upgrading of its human capital.” Ohno divided the More recently, Roy and others (2016) found little value in the catching-up industrialization process into four stages and iden- middle-income trap as an empirical phenomenon. Using var- tified a middle-income trap as a glass ceiling in manufacturing ious measures of convergence based on catching up with rich between stages two and three (figure 5.2). In stages one through countries either as a group or with the United States and based three foreign direct investment is critical to promote and sus- on both the country and individuals as the unit of measurement tain growth. Ohno found that Viet Nam’s growth in the past two (accounting for and assuming away distributional changes within decades was driven largely by liberalization and large inflows of countries), they found that while economic divergence was a external purchasing power. Rigg and others (2014) took a more dominating global phenomenon before the 1980s, there is strong sociological approach in their analysis of Thailand and argued evidence for economic convergence globally since. They then that a middle-income trap can be assessed by how individuals tested whether middle-income countries were negative outliers and households negotiate­ —­ —­ or do not­ the skills/employment within an unconditional convergence framework that included transition. all countries and only middle- and high-income countries2­ —­and found no evidence for either form of middle-income trap. Useful? While most researchers find at least some value in the concept, Pritchett and Summers (2014) demonstrated empiri- What are the main identified causes of middle-income cally that there has been little continuity in growth performance traps, and which countries are affected? historically and found that growth declines are more likely to be The substantial definitional issues and differing empirical results sudden and large than gradual and small. Thus, what others may complicate the notion of a clearly demarcated middle-income perceive as the middle-income trap is more likely to be a regres- trap. Can something unique about industrialization processes sion to the mean. Im and Rosenblatt (2013, p. 25) rejected the for present-day middle-income countries be generalized? Here middle-income trap concept arguing that “MICs [middle-income there is greater convergence, even among the skeptics. The lit- countries] do not really look that different in terms of transitions erature assessing the causes of the trap differentiate between across the inter-country distribution of income” and display structural causes and policy-related and institutional causes. FIGURE 5.2 Ohno’s stages of catch-up industrialization Creativity Stage four Technology absorption Full capability Stage three in innovation and product Management design as Agglomeration and technology global leader Arrival of Stage two mastered, can manufacturing produce high- Japan, foreign direct Have quality goods United States, investment supporting European Union industries, but Republic Stage one still under of Korea, foreign Chinese Taipei Simple manufacturing guidance Stage zero under foreign guidance Thailand, Malaysia Monoculture, subsistence Viet Nam agriculture, Glass ceiling for ASEAN countries aid dependency (middle-income trap) Source: Ohno 2009. The middle-income trap and upgrading along global value chains • 123 Moreover, some researchers distinguish between factors that environment, dependency and sex ratios, the share of gross cause the trap and those that perpetuate it (see, for example, capital inflows, investment public debt in GDP, output diver- Toh 2013). However, there is a general consensus that the trap sification, agriculture and service shares, a country’s GDP- reflects a country’s inability to shift its growth strategy and eco - weighted distance, its degree of output diversification, nomic structure toward sustained high growth rates. whether it is involved in a war or civil conflict, and whether it In reviewing the previous decade of debate, Gill and Kharas has a tropical climate. (2015) argued that the middle-income trap occurs when rapidly • Bulman and others (2014) used pooled regressions on middle- growing countries with rising wages have tried to sustain an income countries and found that escapees from the trap had economy based on labor-intensive manufacturing and export-led higher growth at all relative incomes, higher total factor pro- growth. But as their competitive advantages decline, they have ductivity growth, faster transformations toward industry, been unable to find alternative sources of demand to replace better macroeconomic management, and consistently more exports. This has been exacerbated by the declining elasticity of export orientation. Furthermore, countries with high second - trade to growth in recent years­ —­as well as by rising global com- ary and tertiary education and with a larger share of high-tech petition, increasing currency and balance-sheet risk due to the products in exports are less likely to fall into the trap. heightened international financial flows, and for some countries, The results, while methodologically distinctive and using dif- a lack of the requisite infrastructure while prematurely attempt- fering control variables, provide some consistency in their focus ing to become knowledge economies. on demography, equity, the macroeconomic framework, and­ —­ Differentiating between structural change and convergence most prominently­ —­ the export structure. In other words there is (whether relative or absolute) has implications from a policy per- some consensus that factors seen as important for long-run eco- spective. As Paus (2014) noted, even within an income conver- nomic development are important for middle-income countries gence framework there can still be no capability convergence. to sustain GDP growth. This is a useful contribution, but as Paus Jankowska and others (2012) explicitly framed this as an issue (2014, p. 25) noted, “it is not clear what these findings mean for of structural transformation, with Latin America unable to com- policymaking.” pensate for the decreasing labor share in agriculture through its Trade and export diversification is central to numerous recent manufacturing sector, but with the Asian newly industrializing analyses. Felipe and others (2012) compared the export struc- countries developing modern sectors in which productivity is ture of countries in the trap across variables related to their abil- both higher than in the traditional sector and sufficiently labor- ity to structurally transform and found that escapees had more intensive to transmit the gains to a sizable share of the labor sophisticated and diversified export baskets than did nonescap - force. ees. This is also supported by country case studies. Dingemans Glawe and Wagner (2016) pointed to two primary theoret- (2016) found that the lack of diversification in Chile was caused ical arguments to explain the trap. The first draws on Arthur in large part by the country’s inability to promote innovation and Lewis’s dual-sector model of the economy and sees the trap as develop more complex export products. Paus (2014) likewise saw reflecting a country’s inability to continue boosting productivity the main challenge for Latin America as addressing the disjunc- by shifting workers from agriculture to industry. The latter argu- ture between global competitive pressures and the slow process ment, derived mostly from more recent developments in growth for firms to learn and countries to implement capability-enhanc- theory, focuses on a country’s ability (or lack thereof) to imitate ing policies. Rigg and others (2014) identified the primary failure foreign technologies and develop comparative advantages of inadequate structural transformation at three distinct levels in new export products. Agénor and Canuto (2015), in broad in Thailand: government’s inability to develop the population’s strokes, attempted to model and extend this line of thinking, human capital, firms’ failure to develop human capital or exploit arguing that knowledge network effects to developing advanced what already exists, and individuals’ unwillingness to develop skills and infrastructure allow countries to evade a lower-growth human capital and embrace opportunities away from their home equilibrium that they see as equivalent to the middle-income villages. The trap is “as much personal as it is structural and insti- trap. tutional” (p. 196). Several studies using absolute and relative convergence defi- Several researches have focused on social and demographic nitions have determined the impact of variables that either are factors. Egawa (2013) saw the worsening income distribution correlated with or causally contribute to the trap: as a primary engine of stagnation, while Ozturk (2016) empha- • Eichengreen and others (2013) used a sample of present-day sized the presence and size of the middle class. Panther and developed countries and found that correlates and determi- Flechtner (2015) took the relevance of inequality a step further nants of growth slowdowns were more likely in economies by examining domestic and international inequality as political with high old-age dependency ratios, high investment rates, economy drivers of the trap, using a large sample of compar- and undervalued real exchange rates. ative qualitative case analyses. For national inequality multiple • Aiyar and others (2013) examined 42 variables in seven cate - intersecting inequalities (income, ownership of assets, access gories using a weighted average least-squares approach and to power) result in low institutional quality, which in turn pre- found the following to be significant determinants of falling vents the adoption of policies that may be opposed by vested into the trap: rule of law, size of government, the regulatory interests and would allow the country to transition to a more 124 • Measuring and Analyzing the Impact of GVCs on Economic Development productive economy. For international inequality the ability of nine-year periods. So definitively stating which countries are now countries to benefit from globalization and the proliferation of in the middle-income trap is an imprecise science. multinational corporations and GVCs are influenced by distribu- tions of power and income at the global level. Ito (2016) made How can countries escape from the trap? this focus on institutions as the mediating factor for countries The broad array of causes for countries entering and becoming either escaping or stuck in the trap more explicit by arguing stuck in an alleged middle-income trap include macroeconomic that countries sort themselves into three equilibria (low-income, and microeconomic factors related to industrial structure, trade middle-income trap, and middle-income convergence toward profile, demographics, income distribution, macroeconomic high-income), depending on their willingness to carry out fun- management, and the quality of institutions. So how have coun- damental economic and structural reforms. Much recent work tries in the past escaped from middle-income status? And what deals either explicitly (Woo 2012; Huang 2016) or implicitly with lessons might this hold for countries today? In absolute terms China (World Bank 2013; Pritchett and Summers 2014). In most many if not all of today’s high-income countries were arguably of these China-focused analyses governance and institutions are stuck in some sort of middle-income trap in the 20th century particularly relevant. (table 5.2; Im and Rosenblatt 2013). Both the definitions and the causes inform the classifica- It is clear by the objective criteria for escaping the middle- tion of countries in the middle-income trap. Some researchers income trap (graduating from middle-income country status) that focus on individual countries­—­Egawa (2013) on Malaysia, China, numerous countries, particularly in East Asia and Central and and Thailand, Dingemans (2016) on Chile, and Ohno (2009) on Eastern Europe, have escaped whatever trap dynamics middle- Malaysia and Thailand. Other researchers take a multicountry income status might entail. Bulman and others (2014) described approach and come to different conclusions. For example, Felipe this process as part of a momentum hypothesis, where past and others (2012) considered 35 of 52 countries to be stuck in escapees achieved strong growth in one period, followed by the trap. Aiyar and others (2013) used a “trap map” based on strong growth in the subsequent period. But Im and Rosenblatt the seven factors and 42 variables to determine countries most (2013, p. 25) are cautious about this inevitability approach, argu- at risk. Panther and Flechtner (2015) examined which countries ing that attempts to grow at rates higher than 7% could lead to have achieved convergence with the United States over discreet “unsustainable polices that eventually create the ‘trap’-like pat- eight- to-nine-year time periods. tern of dismal growth that MICs [middle-income countries] are Annex 5.1 illustrates the results of the three cross-country trying to avoid in the first place.” So gradualism that focuses on studies that address different points in time. Aiyar and others overcoming the institutional factors inhibiting growth might be (2013) examined whether countries risk falling into the middle- more promising. Roy and others’ (2016) analysis of the structural income trap in the future. Felipe and others (2012) analyzed break in the 1980s when convergence started attributed this to countries now in the trap. Panther and Flechtner (2015) assessed the sharper focus on macroeconomic stability in the 1990s and whether countries that were middle-income countries at some on the transformational changes that the spread of information point in the past 40 years have managed to converge toward the and communication technologies engendered in developed average GDP per capita in high-income countries over discreet economies. TABLE 5.2 Countries that have escaped the middle-income trap Reference Definition of escape from middle-income trap Countries that have escaped Agénor and Canuto Middle-income countries in the 1960s that became Equatorial Guinea; Greece; Hong Kong, China; Ireland; Israel; Japan; (2015) high-income countries by 2008 Republic of Korea; Mauritius; Portugal; Puerto Rico; Singapore; Spain; and Chinese Taipei Bulman and others Middle-income countries in 1960 that became high- Greece; Hong Kong, China; Ireland; Japan; Republic of Korea; (2014) income countries by 2009 Puerto Rico; Seychelles; Singapore; Spain; and Chinese Taipei. Cyprus and Portugal, still classified as middle income in 2009, are considered on the verge of escaping Felipe and others Crossing from lower-middle-income status in at No clear list of past escapees, but at their current pace China, (2012) most 28 years and from upper-middle-income status Bulgaria, Poland, and Thailand should be able to escape the upper- in at most 14 years. middle-income trap if they sustain their income per capita growth Im and Rosenblatt Middle-income countries in 1950 that have since Austria; Estonia; Finland; Germany; Greece; Hong Kong, China; (2013) become high-income countries (though the authors Ireland; Israel; Italy; Japan; Republic of Korea; Singapore; Slovenia; reject the concept of a middle-income trap) Spain; and Chinese Taipei Jankowska and others Countries that have attained income convergence Hong Kong, China; Republic of Korea; Singapore; and Chinese (2012) with high-income countries Taipei The middle-income trap and upgrading along global value chains • 125 The experiences of these countries center policy recommen- and capabilities transferable from existing industries. A central dations primarily on structural, industrial, and trade policies as lesson from these past industrialization processes is learning how well as social policy: to produce and export more complex products­ —­ a finding that • Macroprudential policies limit the buildup of excessive capital also emerges from the complex systems analysis literature.3 inflows to cushion impacts of potential sudden stops. How- Ohno (2009) proposed that Viet Nam develop a proactive ever, Aiyar and others (2013) see an important role for mea- industrial policy to internalize skills and technology, develop sures to enhance regional integration, infrastructure invest- effective public–private partnerships, and deepen industrial ments, and deregulation in areas where private sector activity knowledge. Focusing primarily on structural characteristics is excessively stifled. Their threat map aims to provide an ana- related to trade, industry, and labor market transformations, lytical tool to assess where these issues may be at play (see Felipe and others (2012) argued that the most direct strategy for table A5.1.1 in annex 5.1). a middle-income country to become a high-income country is to • Developing knowledge network externalities could link indi- acquire a revealed comparative advantage in sophisticated and viduals’ skill attainment and access to public infrastructure well-connected products. The focus on linking education to (pri- (Agénor and Canuto 2015). marily horizontal) industrial development objectives is a logical • Skilled workers are needed to move up the value chain from corollary for numerous researchers. Pantner and Flechtner (2015) low value-added industries to develop higher value-added pointed to mobilizing talent through education and providing activities (Eichengreen and others 2013). this talent with the right incentives to assimilate best-practice • To avoid the middle-income trap, China, like past escapees, technologies and organizational routines to adapt and apply will need to upgrade its industrial structure through new them to cutting-edge technologies. Rigg and others (2014) industries with higher levels of technology (Huang 2016). This linked this more explicitly to countries moving up the value chain will require differentiating between state-owned enterprises through re-training and re-skilling and sustained investments in and non-state-owned enterprises and between product and upper secondary and tertiary education. factor markets. Researchers that focus on the role of inequality in driving trap- Jankowska and others (2012) compared the experience of Latin like dynamics for middle-income countries see addressing dispar- American countries with the Asian newly industrialized countries ities as the central component. Egawa (2013) argued that policies using a product-space methodology and suggested that diversi- need to address urban–rural disparities, providing benefits for fying to new products is central to emulating the experience of low-income individuals, fiscal redistributive reforms, transfers, the newly industrialized countries (figure 5.3). In these countries as well as equalizing education opportunities. Kahras and Kohli new production was sequentially developed in industries such as (2011) see social programs and a change in the policymaking iron, steel, machinery, and electronics through workers with skills mindset that targets the middle class as essential to avoiding the FIGURE 5.3 Product space maps of Peru and the Republic of Korea in 2009 Peru Republic of Korea Source: Jankowska and others 2012. Note: The product space methodology provides a map of all traded goods displaying relative proximity or similarity between products. The colors on the map represent the Leamer classification, which categorizes products according to labor, capital, and other resource intensiveness. The black squares indicate prod - ucts in which the country has a revealed comparative advantage. 126 • Measuring and Analyzing the Impact of GVCs on Economic Development middle-income trap. All this helps countries mediate three critical and what factors condition countries’ ability to upgrade in transitions­ —­from diversification to specialization in production, GVCs? And what is the empirical relationship between GVC from physical accumulation of factors to productivity-led growth, participation and the middle-income trap? It then provides a and from centralized to decentralized economic management. conceptual framework for viewing income transitions through a This in turn leads to a focus on institutions. Panther and Flecht- firm-level GVC lens. ner (2015) used a two-level model of the relationship between inequality and the middle-income trap to argue that at the inter- An overview of global value chain participation and national level ensuring a certain level of domestic equality medi- economic upgrading ates the benefits of global integration for growth. At the domes- Driven by lower transport, information, and communication costs, tic level a focus on economic (over political) equality is central to technological improvements, and lower barriers to the move- catch-up policies. At the global level having some independence ment of goods and capital, global patterns of trade and produc- in policy-setting from dominant external powers is essential for tion have changed dramatically over the past decades. Trade is convergence when paired with export diversification. Dinge - now characterized by the growth and increasing dominance of mans (2016) saw the bidirectional relationship between structural vertically integrated multinational firms with fragmented value change and economic development as driven by institutional chains stretching across borders. The internationalization (and change. Chang (2011) argued that increased wealth intensifies particularly regionalization) of global production and the devel- the demand for and provision of higher quality institutions and opment of value-chain trade in both goods and services have new political actors who demand and shape them. In that sense, changed the prospects for countries to benefit from trade. In this moving from a more state-­ centered approach to export devel- context, understanding a country’s current participation in value opment, not just export promotion, is essential. chains is central to ensuring that its industrial and trade policies Gill and Kharas’s (2015) assessment of 10 years of literature on can facilitate sustainable productivity gains and increased qual- the middle-income trap focused on the need for policymakers to ity employment in higher value-added sectors. For developing manage a transition to more mature institutions so that capital countries this creates opportunities to upgrade into new higher investments remain efficient even after growth moves from pro - productivity tasks and activities and to integrate into global ductivity growth stemming from intersectoral resource realloca- production networks. But according to some researchers, this tions to intrasectoral catch-up technological growth. is often less an issue of catching up than of fitting into existing A new World Bank study (2016b, forthcoming) on Poland’s GVCs (Whittaker and others forthcoming). recent high and stable growth sees this as being due in part The value chain concept in the industrial organization litera- to the country’s institutions transforming in parallel with firms ture (Porter 1985) has become ever more central to understand - increasing in sophistication and complexity­ —­including provid - ing and analyzing the interfirm and intrafirm dynamics and gov- ing better foundations for resolving conflict, enforcing contracts, ernance of value-chain trade (Gereffi and others 2005). And in and implementing antitrust and competition laws. This has been trade economics it has increasingly become the dominant frame- bolstered by rapid integration into the EU bloc, boosting pro- work to understanding the second unbundling of globalization ductivity through increased trade openness, investment and (Baldwin 2006; see Antràs and Rossi-Hansberg 2009 and Ahmad talent, increased domestic competition and regulatory harmoni- 2013 for overviews of this literature). Viewing trade through a zation, and more certainty through commitments to EU institu- GVC framework involves four paradigm shifts for trade policy tions. In comparing new high-income countries (Chile, the Czech (Catteneo and others 2013): Republic, Hungary, the Republic of Korea, Poland, and the Slovak • It implies a move toward a global (or at least regional) view of Republic) with trapped middle-income countries (Brazil, Mexico, policy. Romania, and Turkey), the study found that while in the late • It requires a shift from entire industries to narrower tasks and 1990s these countries were fairly similar in many aspects, by 2015 business functions. barriers to entrepreneurship, trade, and investment were much • It requires assessing a country’s competitiveness not as lower in the new high-income countries, which also had lower endowments and stocks but as flows in which GVCs are the perceived economic and political risk. primary channel enabling transfers. • It implies a change from focusing on tariffs as the most rele- vant obstacles to trade to focusing on behind-the-border bar- The relevance of global value chain trade for riers and regulatory measures.4 understanding the middle-income trap Viewing trade this way requires a revised approach to mea- suring and analyzing cross-border and cross-industry flows, with While the role of trade through GVCs remains more implicit value-added trade becoming a more relevant measure of trade than explicit in discussions of the middle-income trap, the sig - flows within GVCs. For individual countries becoming competitive nificance of producing and trading higher value-added goods in specific components and tasks to participate in globalized pro- is a central lesson from the literature. Establishing this link is the duction networks and in turn generate more value domestically focus here. This section first addresses three questions. What is over time becomes increasingly important (Taglioni and Winkler meant by GVC participation and upgrading? How is it measured, 2016). So using multiregion input-­ output tables allows for tracking The middle-income trap and upgrading along global value chains • 127 use at the sectoral level and for differentiating between transac- FIGURE 5.4 Achieving functional, product, and tions in intermediate and final goods. This has informed a growing intersectoral upgrading through skills, capital, and literature on the development of value added in trade (see Hum- process upgrading mels and others 2001, Koopman and others 2014, and Johnson and Noguera 2012) and a growing number of indicators and indi- ces (Fally 2012, Antras and Chor 2013, and Wang and others 2016). Functional Measuring value-added trade through multiregion input-­ upgrading output databases has clear benefits (Ahmad 2013): • It makes it possible to understand a country’s actual indus- trial structure and international links among sectors in order High to formulate targeted policies and strategies. value-added production I n t e r gra d • It provides evidence of how nontariff measures or regulatory in g up ad t gr d u c changes affect upstream and downstream producers. se i • It offers better analytical tools to anticipate the impact of o ro ct ng ra l P potential shocks. up • It enables calculations of the trade’s job content and impact on ecosystem services, thanks to satellite accounts of employ- ment and environmental indicators. However, multiregion input-­ output tables have some limita- tions. They cannot measure the links among service sectors very accurately. And they are subject to two simplifying assumptions: Capital ad the proportionality assumption, that all products (for export and u gr ing p domestic use) have the same import content, and the homo- geneity assumption, that the use of inputs is uniform among all firms in a sector. Value- The new data make it possible to quantify economic upgrad- added growth in g upg ing through GVC participation. Drawing on earlier work by Hum- ad s gro c e s S k r a di phrey (2004), Taglioni and Winkler (2016) differentiated four ill s ng P r types of economic upgrading based on skills, capabilities, and up comparative advantage. Process upgrading is based on effi- ciency gains and productivity improvements. Product upgrading entails moving into more sophisticated products in an existing value chain. Functional upgrading involves increasing the value- added share by moving toward more sophisticated tasks. And intersectoral upgrading involves moving into new value chains and ca kills pa with higher value-added shares (figure 5.4). ts en bil The ability of firms to upgrade is determined by improving Curr ities workers’ skills, improving firms’ absorptive capacity and technol- Tasks ogy, and increasing productivity in existing tasks. Lead firms set e w it h ta g detailed specifications and requirements that exceed local norms an m co and create opportunities for improving capabilities, technolo- p ar v a tiv e a d gies, and assets. But this is not always the case: the complexity of GVCs and the power dynamics within their governance structures can lead to stagnation or downgrading (Rossi 2013; Blažek 2015). Source: Taglioni and Winkler 2016. The empirical relationship between global value chain integration and the middle-income trap A broad literature on the factors likely to influence a country’s technology spillovers, minimum scale achievements that amplify ability to upgrade in GVCs is based primarily on case studies, pro-­competitive effects, and labor market effects, including the with few econometric analyses (until recently). But open econo- demand for skilled workers and their training as well as turn- mies tend to grow faster and have higher incomes than do closed over when trained workers move to local firms (figure 5.5).5 The economies (Wacziarg and Welch 2008; Gill and Kharas 2015). individual channels have complex and frequent intermediating GVC participation can lead to higher output, productivity, effects on each other. and value added through five main transmission channels: back- Three main factors link value-chain integration to productiv- ward and forward links, pro-­ competitive market restructuring, ity: foreign direct investment, exporting, and importing inputs 128 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 5.5 Transmission channels from global value chain participation to the domestic economy Technology Diffusion spillovers effect Demonstration Availability effect and quality Pro- effect Backward competitive and forward market links restructuring Pro-competition Demand effect effect Demonstration Assistance effect effect Amplification of Minimum pro-competition Domestic Demand effect Training effect Labor scale effect impact of GVC Labor turnover markets achievements Sustainability participation effect effect Source: Taglioni and Winkler 2016. (Kummritz and others 2016). For foreign direct investment the a 10% drop in input tariffs leads to a 12% improvement in pro - impact of spillovers on productivity is not conclusive (Görg and ductivity for importing firms. Bas (2012) showed that for a sample Greenaway 2004, Paus and Gallagher 2008). For the link between of Argentinian firms, input tariffs facilitate entry into export mar- exporting and economic upgrading, Bernard and Jensen (1999) kets. MacGarvie (2006), drawing on French trade and citation demonstrated that exporters outperform nonexporters in the data, and Bøler and others (2015), using a sample of Norwegian same sector and country in productivity, skills, and wages. Is firms, found importers to be more innovative and profitable. this self-selection or learning by exporting? For self-selection Skills are relevant for importing and complementary to it. Koren the assumption is that only more productive firms are able to and C ­ sillag (2011) showed that importing more sophisticated absorb additional trade costs. The learning by exporting liter- machines requires higher skills to operate them and increases ature argues that exporting improves the productivity of firms the returns to skills. over time, with the most robust findings for developing countries To test whether GVC participation has enabled countries to and nascent industries. Recent research questions the robust- upgrade economically, Kummritz and others (2016) used foreign ness of these early learning by exporting studies (Clerides and value added in exports and domestic value added re-exported others 1998), but Lileevea and Treffler (2007) found learning by by third countries as respective measures of backward and for- exporting effects for Canada, and Fernandes and Isgut (2005) ward GVC integration and domestic value added generated by found them for Colombia. a specific sector as the measure of economic upgrading. They Research on the link between importing inputs and produc- tested the impact of national characteristics that may be associ- tivity focuses on developed countries. Importing can improve ated with economic upgrading via GVC participation: infrastruc- key aspects of competitiveness through three main feedback ture, connectivity, investment and trade policy, business climate loops: productivity, innovation, and skills. Easier access to and institutions, financial and labor markets, skills and education, imports tends to improve firm productivity. Grossman and Rossi-­ innovation and product standards, as well as labor, social, and Hansberg (2006) showed that offshoring can entail productivity environmental standards. They found that overall GVC integra- gains similar to technological progress for offshoring countries tion increases a country’s domestic value added. Splitting the through lower input costs. Amiti and Konings (2007) showed that sample into income groups did not substantially change results, The middle-income trap and upgrading along global value chains • 129 though GVC integration as a buyer (through foreign value added) services both types of integration­ forward and backward­ —­ —­ is more significant at lower incomes and selling into GVCs has lead to similar GDP increases. more impact at higher incomes. For countries buying from GVCs These studies suffer from two main limitations. First, because air freight infrastructure and road network quality are particu- of the lack of value-added trade data prior to 1990, they permit larly important. Connectivity, education and skills, and standards analysis for only the last 20 years, while much of the middle-­ compliance are most important for countries selling into GVCs. income-trap literature goes back 50 years or more. Second, they The researchers concluded that the policy areas thought to be do not specify the conditions for specific types of institutions significant for economic upgrading in GVCs largely have the and policies to produce greater gains from GVC participation. expected impact. The next section addresses both limitations in part from a theo - The correlation between GVC integration and GDP per capita retical perspective. depends on income status and the type of integration (figure 5.6; Boffa and others 2016). GVC integration increases GDP per Viewing income transitions through a global value chain capita, but the gains diminish as income increases. Similarly, lens: bringing in the firm perspective growth in output per capita is highest for lower income groups. The previous section showed that while integrating into GVCs Some channels for GVC integration depend on industry similar- is associated with sustained growth and development, doing ity, with links assumed to be easier when trade is intraindustry. so may become more difficult and complex at higher incomes. Manufacturing leads to higher GDP gains for buyers, but for This section explicitly adopts a GVC firm-level lens to the income FIGURE 5.6 Growth of global value chain integration and GDP per capita by income category GDP per capita Buyer perspective Low income Lower middle income Upper middle income High income 3 2 1 0 –1 –2 –3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 Foreign value added Seller perspective Low income Lower middle income Upper middle income High income 3 2 1 0 –1 –2 –3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 –3 –2 –1 0 1 2 3 Foreign value added Source: Boffa and others 2016. 130 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 5.7 How global value chain determinants evolve as the engagement in global value chain changes Mature GVC engagement process engagement GVC engagement stage, domestic value added in exports, foreign value added in exports Buying Producing Selling • Organizational • Full alignment of own capital as core and buyer procurement capability strategies • Frontier research and development • Deep relationship with own suppliers • Widely recognized brand Coordination • Deep relationship with of processes • Functional and tech institutions intersectoral upgrading • Acquiring capabilities Upgrade • Input quality in intangible capital • Capital deepening • Firm organizational Learning and managerial practices Automation • Adaptive research and by doing • Learning about development (to seamlessly demand fit inputs to own product) • Adaptive research and development, product and process upgrading • Skills for efficient use of Connect • Establishing proper access to inputs intermediates • Development of • Learning through investment capabilities imitation Define own product scope Proto-connect • Overcome low-value transaction • Overcome low-value transaction constraints on the buying (importing) side constraints on the selling (exporting) side GVC engagement intensity on the buying side GVC engagement intensity on the selling side Source: Mariscal and Taglioni 2017. transitions from low to middle to high through a series of dia- tightly allocating capabilities into specific steps on the diagram, grams laying out these transitions. most capabilities are shared between elements. The diagram is Mariscal and Taglioni (2017) proposed a framework that views somewhat selective in reporting the most relevant determinants, firms’ connection to GVCs as a dynamic process for the rele - since they may vary by industry and GVC. vance of capabilities to evolve in a continuum (figure 5.7). The Engagement starts with the proto-connecting stage, when first dimension (x-axis) contains the buying, producing, and sell- reaching a minimum scale of transactions is key in both the ing aspects of integration in GVCs, and the second dimension buying and the selling functions. Intermediaries that play a (y-axis) illustrates the degree of GVC engagement, from proto-­ matching role can help firms move the first steps toward inter- connecting to connecting to upgrading to mature engagement. national engagement. The large intermediaries may themselves The orange lines indicate the growing intensity of engagement become companies engaging in GVCs, as traders or by gradu- on the buying side, and the blue lines indicate the growing inten- ally adding value by expanding into processing for some of the sity of engagement on the selling side. More sophisticated issues immediate upstream or downstream functions. tend to appear on the upper section of the diagram, where The pure connection stage is mostly about meeting minimum upgrading meets buying, producing, and selling in increasingly requirements­ —­a few basic capabilities that allow the firm to con- nuanced and complex ways. Connecting usually starts on the nect to either a foreign market or a lead firm. Basic capabilities buying side: firms that correctly evaluate their core capabilities (such as production or managerial abilities and cheap access to and have an effective sourcing strategy are more likely to suc- key inputs of the production process) and the ability to correctly cessfully engage in GVCs. However, the sequence of engage- evaluate and leverage the firm’s core competences are crucial in ment from simpler to more sophisticated, and from buying to connecting to GVCs. From a buying perspective this means that selling capabilities, is intended to be illustrative. Rather than the firm can streamline its processes and product scope while The middle-income trap and upgrading along global value chains • 131 complementing production with proper and effective access to potential and virtuous feedback loops through direct and indi- input markets. From a selling perspective the key capabilities rect exposure to new ways of managing and organizing produc- are aligning its goals to those of actual and potential buyers tion. Meanwhile product complexity also increases, with prod - and modulating its processes to fit seamlessly in their produc- ucts both more elaborate and containing more value added. tion processes. At this stage engagement in GVCs is not a robust The evolutionary process in GVCs is unlikely to take place in situation but is unstable and subject to market forces that may a vacuum. As firms transition from proto-connecting to connect- exclude the firm from continuing the connection. ing to upgrading to mature engagement, a parallel process of Once a firm overcomes the challenges of pure connection, development takes place in the hosting economy. The domestic its learning processes and absorptive capacities become more economy will likely have evolved to development stages where fundamental­ —­learning by exporting, learning from selling to an ecosystem of firms starts building up, populating the middle global buyers, and learning to connect decisions on the prod- size. The determinants of firm growth will also evolve, increas- uct scope to the available importing possibilities. At this stage ingly a function of firm capabilities rather than institutional demand-side elements also acquire importance, as the firm needs idiosyncrasies. to make its product known and valued. Efforts will also be made to Reflecting their comparative advantages, firms in low-income accumulate customers, although competition will be based mostly countries will tend to be engaged in GVCs in industries such as on price rather than quality. As the process continues and deep- agriculture and manufacturing, where complexity is limited and ens, process innovation and product innovation will start to matter price competition is more common than nonprice competition greatly. Relatedly, firms’ organization of skills will change. Middle (table 5.3). In these industries buyer–seller relations tend to be management will start to become more important than produc- either at arm’s length or captive. They are captive when the lead tion, and the firm will increasingly focus on its core competences firms are technologically very dependent on suppliers, transac- and learn to spin off tasks that are not its comparative advantage. tions are highly codified, and supplier competence is low. Firm In the final stage firms upgrade toward the most complex size is not a constraint at this stage, so small firms can easily stages of GVC production. Firms are now in direct relation with, engage. Once countries graduate to middle-income status, their or themselves become, lead firms, turnkey suppliers, trading firms start integrating in GVCs with functions in advanced man- platforms, or global buyers. The relationships between buyers ufacturing and modern professional services, including prepro- and sellers are seldom the result of market interactions but are duction and postproduction high-value added services. In these geared to modular (or even captive) interaction. Firms connect GVCs buyer–seller relations tend to be more relational, captive, to the most technologically relevant buyers with good learning or hierarchical, with substantial know-how transfers. Participating TABLE 5.3 Trajectories in global value chain engagement Global value chain engagement stage Low-income status Middle-income status High-income status Industry complexity Simple Intermediate Complex Typical specialization Commodity production in Advanced manufacturing, Organization capital, coordination and agriculture, light manufacturing, agri‑business and services research and development in complex low value-added services agri-business, manufacturing and services, branding Typical market structure Predominantly small firms Some large, missing middle size in Complex market structure with several and average firm size the market likely lead firms and conglomerates and a large and dynamic fringe of small to medium- size firms that interact in complex ways Buyer–seller relational Market relations (or captive, if Relational or hierarchical Highly modular and complex vertical dependence and supplier competence is low, and horizontal relationships of governance in global transactions highly codified, and interdependence value chains technological dependence high) Typical firm structure Few organizational layers, narrow Mid-complexity organization and Large firm or conglomerate, and pool of skills set of capabilities, workforce firm structure quantitatively important middle and distribution highly skewed toward higher management and research and production functions development staff relative to production functions Mode of competition Price-to-quality competitiveness Increasingly diversified, nonprice Based purely on brand and value added competitiveness features in highly specialized areas at the technology frontier Source: Adapted from Mariscal and Taglioni 2017. 132 • Measuring and Analyzing the Impact of GVCs on Economic Development firms tend to be medium to large, particularly in manufacturing There is no one way to optimally sequence policies, since (Cusolito and others 2016). Firm growth is driven by productivity considerations are context-specific. Yet some general regular- and capabilities rather than by rent positions. And competition ities in policies are likely to matter at different stages of GVC among firms is increasingly based on nonprice features, such engagement and development (Taglioni and Winkler 2016). At as quality, customization, and responsiveness and timeliness in the initial stages of GVC engagement, policies are best directed delivery. toward facilitating efficient use of resources and factors of pro - Once countries reach high-income status, their firms’ engage - duction and encouraging competition through broad market ment in GVCs will likely be predominantly specialized in coor- access (table 5.4). As development takes place and GVC engage - dination and high-value added services, such as research and ment deepens, the institutional setup should focus on helping development and branding. Firms are primarily buyers of inputs firms thrive in a complex world­ —­ with both imports and exports and components and sellers to end markets­ —­or engaged in from and to multiple countries and sourcing and selling to multi- modular relationships. Their comparative advantage is based on national corporations (domestic and foreign). offering highly specialized products at the technology frontier. When a country reaches middle-income status, institutions Institutions are central to these processes. Consider three key can help leverage GVC engagement for development by foster- aspects. First, coordination among different levels of govern- ing skill building, innovation, and efficient access to capital; by ments has to ensure that policies are not done and undone at dif- including deep provisions in agreements with key trade partners; ferent levels or that competition among regions does not erode by supporting the engagement of more local firms and workers the fiscal base. Coordination becomes increasingly important at in the GVC network; and by focusing on structural reforms that higher levels of development. Second, predictability in policy raise domestic labor productivity and skills. As countries target implementation also matters proportionally with the level of high-income status, building institutions that allow for contracts development. Uncertainty could erode good current incentives to be more complete and for administrative burdens to be lower. by exposing firms to unnecessary risk. For example, uncertainty They also need to ensure high confidence in the institutions, busi- in trade openness policies may freeze the formation of buyer– ness friendliness, and stable policies. Labor market–­ enhancing supplier links as firms find it optimal to wait before engaging in outcomes for workers at home and more equitable distributions investments that lose all their value unless variable trade costs of opportunities and outcomes create social support for a reform are actually reduced. Third, policies should be well sequenced. agenda aimed at strengthening a country’s GVC participation. For example, opening to foreign direct investment without actu- Climate-smart policy prescriptions can mitigate the challenges for ally developing basic infrastructure and institutions is unlikely to firms from climatic disruptions. Frederick’s (2016a, 2016b) work on generate much investment or many jobs. It may be beneficial to the apparel GVC in China illustrates this dynamic (box 5.1). consider gradual increases in competition so that foreign firms do not eradicate all domestic firms and capture all economic Participating in global value chain trade in the context of a rents. To put in place policies conducive to real domestic com- rapidly changing world of industrial production and work petition is a sensible requirement before opening to trade or The positive and significant relationship between GDP per capita foreign direct investment. Another, more classic example is the and integration into GVCs raises questions about the gains of coordination between foreign direct investment and trade pol- GVC trade for workers in countries at the middle-high income icies. Since multinational corporations are import-intensive, threshold. Over the past few years numerous reports and papers opening to foreign direct investment to create jobs will not work have investigated the impact of technological change on produc- unless foreign companies can also have access to the foreign tion, trade, and labor markets (West 2015; Oxford Martin School services and intermediate goods they require. 2016; Chui and others 2016; WEF 2016; KPMG 2016; Autor 2015; TABLE 5.4 The institutional dimension of global value chain engagement Low income Middle income High income Attract foreign direct investment Expand and strengthen participation Ensure sustainability and transform global and facilitate domestic firm entry in global value chains, including value chain participation into inclusive growth Global value chain into global value chains promoting economic upgrading and engagement objective densification, and strengthening domestic firms’ absorptive capacity Facilitate efficient use of resources Foster skills-building and efficient Target contract completeness, lowering and factors of production and access to capital and include deep of administrative burden, high degree of Institutional setup encourage competition through provisions in agreements with key confidence in institutions, business friendliness, broad market access trade partners stable policies, labor market–enhancing outcomes, and climate-smart policies Source: Adapted from Taglioni and Winkler 2016. The middle-income trap and upgrading along global value chains • 133 BOX 5.1 Lessons from China for apparel upgrading China has been remarkably successful in the apparel indus- • Investments in connectivity through sourcing offices try, improving all key areas and growing rapidly. Fredrick in Hong Kong, China, and near airports in mainland (2016) argued that it was able to upgrade (process, product, China that focused on having buyers both close and end-market, function, and intersectoral) by having a dis- comfortable. tinct value proposition that entailed affordability, reliability, • Continuous innovation and modifications in production and the development of a good reputation; a broad range processes to keep costs low and keep up with emerging of product categories; full-package offerings; and “good- competitors. enough” compliance. She attributes this to six factors: • Developing long-term relationships that built on • Industrial polices geared toward upgrading and assess- improvements in quality, speed, and price competition. ing global dynamics, including grants and loans for tech- • Correctly assessing the tipping points for buyers possi- nological upgrading, incentives for machinery to make bly inclined to change suppliers and not exceeding them. more advanced products while removing incentives for While China’s success cannot be directly replicated, it lower-value products; targeting of non-EU and non-U.S. does have numerous lessons for other countries, including markets as well as the domestic market; investments in the importance of developing long-term relationships and textile capabilities and sustained technological invest- ties with foreign firms, partnering with Chinese firms both ments and growth in key material segments; a “go-out” to transfer skills and knowledge and to access the Chinese policy to encourage investment of low value-added seg- market, targeting emerging markets without highly devel- ments in neighboring countries; and support for domes- oped buyer–supplier relationships, targeting niches within tic brand development. a highly diversified industry, and taking advantage of rising • A functional division of labor between sales and produc- labor costs in China to undercut these suppliers. tion, using agents and intermediates to promote manu- facturers and link up with global buyers. Source: Frederick 2016. Beaudry and others 2016, Eden and Gaggi 2015; Morikawa 2016; skills and tacit knowledge for using new equipment and think- Pikos and Thomsen 2016). Researchers have focused on the rapid ing computationally and analytically and high levels of technical technological advances in automation, big data analytics, and and engineering knowledge. For many middle-income countries digitization. They have also looked at manufacturing responses this will require a fundamental upgrade of education systems, to climate change and other environmental- and resource-­ research institutions, and innovation systems. So the already related risks, including transitions toward additive manufacturing diminishing advantage that labor-abundant, low-wage countries through three-dimensional printing technologies. And they see possess for low-skill manufacturing is likely to diminish further. the growth of the circular economy as likely to require manufac- Second, re-shoring production to developed economies­ —­ turers to design products for several cycles of disassembly and given the need for highly skilled workers and, more important, reuse. the ability to automate many tasks­ —­is likely to become even GVCs are characterized by four features: customized produc- greater in coming years, reinforced by the rapidly growing polit- tion; sequential production decisions going from the buyer to the ical backlash against globalization and rising economic nation- suppliers; high contracting costs; and global matching of goods, alism in many western countries. Some 70% of clients surveyed services, production teams, and ideas (Antràs 2015). All four point in a recent study believe that automation and developments in to the substantial power that multinational corporations coordi- three-dimensional printing will encourage companies to move nating GVCs have in selecting where to geographically locate their manufacturing closer to home, with North America seen as individual production tasks. Technological improvements are having the most to gain from this trend and China the most to likely in each of these cases to increase both the sophistication of lose (Oxford Martin School 2016). The jobs of 77% of workers in buyer demands and the supplier capabilities to meet them. A full China and 69% of workers in India are at risk because of auto - exploration of these issues is beyond the scope of this chapter, mation (World Bank 2016c). In this context the rapidly growing but given their implications for the relationship between GVC par- importance of trade in data and information, even within produc- ticipation and declining economic growth and structural stagna- tion and manufacturing, is likely to further increase the modu- tion that many middle-income countries experienced, it is worth larity of work processes and to bypass all but the most sophisti- addressing two aspects of these medium-term developments. cated middle-income countries. First, the workforce skills required to manufacture even Together, these issues are likely to reinforce concerns of pre- unsophisticated products is likely to increase substantially, requir- mature deindustrialization, with countries running out of industri- ing not only higher levels of education but also cross-domain alization opportunities sooner and at lower income than earlier 134 • Measuring and Analyzing the Impact of GVCs on Economic Development industrializers (Rodrik 2016). The trend may have hit Latin Amer- alone inevitable, about growth slowdowns at specific incomes. ican middle-income countries, both economically and in risks But problems related to the structural transformation of indus- for political stability and democratization. Only recently have tries are quite specific to middle-income countries, and this firms and governments in developed and developing economies more limited understanding of a middle-income trap is analyt- come to terms with the fact that the GVC revolution required a ically more tractable. The closely linked debates on GVCs and fundamental rethinking of trade and, more broadly, industrial middle-income traps both strongly point to developing coun- development. These new disruptive technological changes will tries’ need to adapt to a world of global trade and investment again require new policies and strategies to adapt. operating through globally integrated value chains in goods, This points to the challenges for ensuring that the gains from services, and information. This presents a partial but important GVC trade for industrializing countries actually benefit work- conceptual paradigm for addressing many middle-income coun- ers and households­ —­ to the recent and emerging ever-more-­ tries’ inability to converge with Organisation for Economic Co- complex aspects of the political economy of globalization, par- operation and Development economies. The macroeconomic, ticularly for industrializing countries. What is needed? First is a trade, and industrial policies required for successful GVC partic- better understanding of what automation is and what globaliza- ipation can play an important role in the factors that have con- —­ tion is­ since narratives have profound political consequences. tributed to economic stagnation both in the recent past and in Second is a sharper focus on the distributional impacts of GVC present-day middle-income countries. There is a reason why trade, on adjustment costs, and on displacement­ —­renewing many countries find it difficult to graduate to high-income status: attention to labor market impacts and to the risks of downgrad- capabilities, policies, investment decisions, and institutional pro- ing within GVCs for certain workers even as countries upgrade cesses become highly complex. As economic forces interact in overall. multidimensional, unpredictable, and dynamic ways, it is often difficult for institutions to capture such complexity, adapt swiftly, and set policy priorities. Moreover, many challenges­ —­and thus Conclusion and policy implications solutions­ —­ are unique to the respective country, sector, and commodity, so adopting previously successful strategies may not Can integration into GVCs help countries avoid a middle-income help. Indeed, emerging technological changes are likely to fur- trap? And if so, through what channels and under what circum- ther complicate countries’ ability to integrate into and upgrade stances? And how do the factors hypothesized to contribute within GVCs. to growth slowdowns at the middle-income level also impede Even so, policy recommendations can be formulated. First, economic upgrading through GVCs. Inevitably, in bridging two policymakers and companies in the digital era­ —­in developed and issues for which even definitions are heavily disputed, review- developing countries alike­ —­will have to focus on the key features ing the relevant theoretical literature and empirical analysis may of the 21st century economy. This includes addressing the inter- have created more shadow than light. play between technological (digital) innovation and globalization In discussing the middle-income trap, it probably helps to (increased connectivity and GVCs) and creating an environment move away from the deterministic framing that the concept conducive to diversification, innovation, and productivity. Second, can assume, particularly in the eyes of policymakers. The evi- attention to the macroeconomic, social, and environmental sus- dence is fairly robust that there is nothing overly probable, let tainability of a GVC-led development model is also in order. The middle-income trap and upgrading along global value chains • 135 ANNEX 5.1 Attempts to quantify the middle-income trap TABLE A5.1.1 Country status relative to indicators associated with middle-income trap Felipe and others (2012) Panther and Flechtner (2015) Aiyar and others (2013) LMIC trap = not crossed the lower-middle-income Share of four total time periods “Trap Map” based on segment in at most 28 years (1976–2009) during which GDP seven key factors UMIC trap = not crossed per capita growth exceeded (higher score signifies greater the upper-middle-income high-income country average Country risk of middle-income trap) segment in at most 14 years (%) Albania LMIC trap 75 Algeria 13 LMIC trap 100 Argentina 12 100 Bolivia, Plurinational State of 12 LMIC trap 0 Botswana LMIC trap 75 Brazil 8 LMIC trap 75 Bulgaria 100 Cambodia 100 Chile 7 100 China 2 100 Colombia 6 LMIC trap 0 Congo, Rep. LMIC trap 75 Costa Rica 5 75 Czech Republic 100 Cyprus 75 Dominican Republic LMIC trap 75 Ecuador 9 LMIC trap 25 Egypt 8 LMIC trap 75 El Salvador 2 LMIC trap 50 Equatorial Guinea 50 Estonia 100 Gabon LMIC trap 25 Greece 75 Guatemala 11 LMIC trap 25 Honduras 11 50 Hong Kong, China 100 Hungary 100 India 4 100 Indonesia 1 100 Iran 4 LMIC trap 75 Ireland 75 (continued) 136 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE A5.1.1 Country status relative to indicators associated with middle-income trap (continued) Felipe and others (2012) Panther and Flechtner (2015) Aiyar and others (2013) LMIC trap = not crossed the lower-middle-income Share of four total time periods “Trap Map” based on segment in at most 28 years (1976–2009) during which GDP seven key factors UMIC trap = not crossed per capita growth exceeded (higher score signifies greater the upper-middle-income high-income country average Country risk of middle-income trap) segment in at most 14 years (%) Jamaica LMIC trap 0 Jordan 11 LMIC trap 50 Korea, Rep. 100 Lebanon LMIC trap 100 Libya LMIC trap 50 Malaysia 5 UMIC trap 100 Mauritius 100 Mexico 6 50 Morocco 9 LMIC trap 100 Namibia LMIC trap 25 Nicaragua 50 Oman 100 Panama 13 LMIC trap 50 Paraguay 10 LMIC trap 25 Peru 12 LMIC trap 75 Philippines 10 LMIC trap 25 Poland 100 Portugal 50 Romania LMIC trap 75 Russian Federation 100 Saudi Arabia UMIC trap 25 Singapore 100 South Africa LMIC trap 25 Sri Lanka LMIC trap 75 Swaziland LMIC trap 50 Syria UMIC trap 100 Thailand 5 100 Tunisia 10 LMIC trap 75 Turkey 3 Ukraine 100 United Arab Emirates 50 Uruguay 8 UMIC trap 75 Venezuela, RB UMIC trap 100 Viet Nam 7 50 Yemen, Rep. LMIC trap 0 The middle-income trap and upgrading along global value chains • 137 Notes References 1. Middle-income countries are defined by the World Bank as having a Agénor, P. R., and O. Canuto. 2015. “Middle-Income Growth Traps.” GDP per capita of $1,046–$12,735 in 2014. Countries with a higher Research in Economics 69 (4): 641–60. GDP per capita are classified as high-income countries, and countries Ahmad, N. 2013. “Estimating Trade in Value-Added: Why and How?” In below $1,046 are classified as low-income countries. Upper-middle-in- Global value chains in a changing world, edited by D. Elms and P. Low, come countries have a GDP per capita of $4,126–$12,735, and lower- 85–105. Geneva: World Trade Organization. middle-income countries have a GDP per capita of $1,046–$4,125. Aiyar, S., R. Duval, D. Puy, Y. Wu, and L. Zhang. 2013. “Growth Slowdowns 2. This dual interpretation of the middle-income trap is explained as fol- and the Middle-Income Trap.” International Monetary Fund (IMF) Work- lows. The first is that middle-income countries start growing slower ing Paper WP/13/71, IMF, Washington, DC. than the average country conditional on their income; the second is Amiti, M., and J. Konings. 2007. “Trade Liberalization, Intermediate Inputs, that while it may be easy to become a middle-income country, it is and Productivity: Evidence from Indonesia.” The American Economic difficult to move beyond it. Review 97 (5): 1611–38. 3. This is also central to much of the work linking complexity economics Antràs, P. 2015. Global Production: Firms, Contracts and Trade Structure. to economic development (Pugliese and others 2015). While address- Princeton: Princeton University Press. ing the literature’s perceived deficiencies on poverty traps and on Antràs, P., and D. Chor. 2013. “Organizing the Global Value Chain.” Econo- processes of economic development more broadly, the researchers metrica 81 (6): 2127–2204. pointed particularly to the complexity of economic systems at the Antràs, P., and E. Rossi-Hansberg. 2009. “Organizations and Trade.” Annual outset of industrialization, viewing this a dynamic process where com- Review of Economics 1 (1): 43–64. reinforcing production capabilities and product demand plex network-­ Autor, D. 2015. “Why Are There Still So Many Jobs? The History and Future emerges. Using a new measure of complexity, they found that more-­ of Workplace Automation.” The Journal of Economic Perspectives 29 complex economies face lower barriers (in differentiated and more-­ (3): 3–30. GDP per capita) when starting the transition toward industrialization. Azariadis, C., and A. Drazen. 1990. “Threshold Externalities in Economic 4. There have even been discussions of multilateral trade liberalization Development.” The Quarterly Journal of Economics 105 (2): 501–26. through specific value chains rather than through trade policy issues Baldwin, R. 2006. “Globalisation: The Great Unbundling(s).” Economic (the case thus far; Hoekman and Jackson 2013). Council of Finland 20 (3): 5–47. 5. In broad strokes this mirrors the findings of the empirical literature of Baldwin, R., and J. Lopez-Gonzalez. 2015. “Supply-Chain Trade: A Portrait transmission channels for trade and foreign direct investment. of Global Patterns and Several Testable Hypotheses.” The World Econ- omy 38 (11): 1682–1721. Bas, M. 2012. “Input-Trade Liberalization and Firm Export Decisions: Evidence from Argentina.” Journal of Development Economics 97 (2): 481–93. Beaudry, P., D. A. Green, and B. M. Sand. 2016. “The Great Reversal in the Demand for Skill and Cognitive Tasks.” Journal of Labor Economics 34 (S1): S199–S247. Bernard, A. B., and J. B. Jensen. 1999. “Exceptional Exporter Performance: Cause, Effect, or Both?” Journal of international economics 47 (1): 1–25. Blažek, J. 2015. “Towards a Typology of Repositioning Strategies of GVC/ GPN Suppliers: The Case of Functional Upgrading and Downgrading.” Journal of Economic Geography 16 (4): 1–21. Boffa, M., V. Kumritz, G. Santoni, D. Taglioni, and D. Winkler. 2016. “Over- coming the Middle-Income Trap: The Role of GVC Integration for Climbing-Up the Income Ladder.” University of Geneva. Bøler, E. A., A. Moxnes, and K. H. Ulltveit-Moe. 2015. “R&D, International Sourcing, and the Joint Impact on Firm Performance.” The American Economic Review 105 (12): 3704–3739. Bulman, D., M. Eden, and H. Nguyen. 2014. “Transitioning from Low-In- come Growth to High-Income Growth. Is There a Middle Income Trap?” World Bank Policy Research Working Paper No. 7104, World Bank, Washington DC. 
 Cattaneo, O., G. Gereffi, S. Miroudot, and D. Taglioni. 2013. “Joining, Upgrading and Being Competitive in Global Value Chains: A Strategic Framework.” World Bank Policy Research Working Paper No. 6406, World Bank, Washington DC. Chang, H. 2011. “Institutions and Economic Development: Theory, Policy and History.” Journal of Institutional Economics 7 (4): 473–98. 138 • Measuring and Analyzing the Impact of GVCs on Economic Development Chui, M., J. Manyika, and M. Miremadi. 2016. “Where Machines Could Görg, H., and D. Greenaway. 2004. “Much Ado about Nothing? Do Replace Humans – and Where They Can’t (Yet).” McKinsey Quarterly Domestic Firms Really Benefit from Foreign Direct Investment?” The (July): 1–12. World Bank Research Observer 19 (2): 171–97. Clerides, S., S. Lach, and J. Tybout. 1998. “Is Learning-by-Exporting Import- Grossman, G. M, and E. Rossi-Hansberg. 2006. “Trading Tasks: A Simple ant? Micro-Dynamic Evidence from Colombia, Mexico and Morocco.” Theory of Offshoring.” American Economic Review 98 (5): 1978–1977. Quarterly Journal of Economic 113 (3): 903–47. Haldane, A. 2015. “How Low Can You Go?” Speech delivered at the Por- Cusolito, A. P., R. Safadi, and D. Taglioni. 2016. Inclusive Global Value tadown Chamber of Commerce, Portadown, Northern Ireland, Septem- Chains: Policy Options for Small and Medium Enterprises and Low-In- ber 18. come Countries. Washington, DC: World Bank. Hausmann, R., L. Pritchett, and D. Rodrik. 2005. “Growth Accelerations.” Dabús, C., F. Tohmé, and M. Á. Caraballo. 2016. “A Middle Income Trap in Journal of Economic Growth 10 (4): 303–29. a Small Open Economy: Modeling the Argentinean Case.” Economic Hoekman, B., and S. Jackson. 2013. Reinvigorating the trade policy Modelling 53 (C): 436–44. agenda: Think supply chain!. VoxEU, January 23. Dingemans, A. 2016. “Trying to Stay Ahead of the Curve in Chile’s Eco- Huang, Y. 2016. “Can China Escape the Middle-Income Trap?” China Eco- nomic Development: Exploring a Way out of the Middle-Income Trap nomic Journal 9 (1): 17–33. through Pragmatic Export Development.” Development Policy Review: Hummels, D., J. Ishii, and K. M. Yi. 2001. “The Nature and Growth of Ver- 34 (5): 643–69. tical Specialization in World Trade.” Journal of international Economics Eden, M., and P. Gaggl. 2015. “On the Welfare Implications of Automa- 54 (1): 75–96. tion.” World Bank Policy Research Working Paper No. 7487, World Humphrey, J. 2004. “Upgrading in Global Value Chains.” World Commis- Bank, Washington, DC. sion on the Social Dimensions of Globalization Working Paper No. 28, Egawa, A. 2013. “Will Income Inequality Cause a Middle-Income Trap in International Labour Office, Geneva. Asia?” Bruegel Working Paper 2013/06, Breugel, Brussels. Im, F. G., and D. Rosenblatt. 2013. “Middle-Income Traps: A Conceptual Eichengreen, B., D. Park, and K. Shin. 2013. “Growth Slowdown Redux: New and Empirical Survey.” World Bank Policy Research Working Paper No. Evidence on the Middle-Income Trap.” National Bureau of Economic 6594, World Bank, Washington, DC. Research (NBER) Working Paper No. 18673, NBER, Cambridge, MA. IMF (International Monetary Fund). 2016. Global Financial Stability Report: Fally, T. 2012. “Production Staging: Measurement and Facts.” Working Fostering Stability in a Low-Growth, Low-Rate Era. Washington, DC: IMF. paper, University of Colorado-Boulder, Boulder, CO. Ito, T. 2016. “Growth Convergence and the Middle Income Trap.” Center income Felipe, J., A. Abdon, and U. Kumar. 2012. “Tracking the Middle-­ on Japanese Economy and Business Working Papers No. 349, Colum- Trap: What is It, Who Is in It, and Why?” Working Paper No. 715, Levy bia Business School, New York. On-Hudson, NY. Economics Institute of Bard College, Annandale-­ Jankowska, A., A. J. Nagengast, and J. R. Perea. 2012. “The Middle Income Fernandes, A. M., and A. Isgut. 2005. “Learning-by-Doing, Learning-by-Ex- Trap: Comparing Asian and Latin American Experiences.” Organisation porting, and Productivity: Evidence from Colombia.” World Bank Policy for Economic Co-operation and Development (OECD) Development Research Working Paper No. 3544, World Bank, Washington, DC. Center Policy Insights No. 96, OECD, Paris. Frederick, S. 2016a. “Benchmarking South Asia in the Global Apparel Johnson, R. C., and G. Noguera. 2012. “Accounting for Intermediates: Pro- Industry.” In Stiches to Riches? Apparel Employment, Trade, and Eco- duction Sharing and Trade in Value Added.” Journal of international nomic Development in South Asia, edited by G. Lopez-Acevedo and R. Economics 86 (2): 224–36. Robertson, 39–76. Washington, DC: World Bank. Kharas, H., and H. Kohli. 2011. “What is the Middle Income Trap, Why do ———. 2016b. “Upgrading in the Apparel and Electronics Global Value Countries Fall into It, and How Can It Be Avoided?” Global Journal of Chains (GVCs): A Multi-Layered Approach: China Country Case.” Back- Emerging Market Economies 3 (3): 281–89. 
 ground paper, United Nations Industrial Development Organization, Koopman, R., Z. Wang, and S. J. Wei. 2014. “Tracing Value-Added and Vienna. Double Counting in Gross Exports.” The American Economic Review Garrett, G. 2004. “Globalization’s Missing Middle.” Foreign Affairs 83 (6): 104 (2): 459–94. 84–96. Koren, M., and M. Csillag. 2011. “Machines and Machinists: Capital-Skill Gereffi, G. 2014. “Global Value Chains in a Post-Washington Consensus Complementarity from an International Trade Perspective.” Centre for World.” Review of International Political Economy 21 (1): 9–37. Economic Policy Research (CEPR) Discussion Paper No. DP8317, CEPR, Gereffi, G., J. Humphrey, and T. Sturgeon. 2005. “The Governance of London. Global Value Chains.” Review of International Political Economy 12 (1): KPMG. 2016. Future State 2030: The global megatrends shaping govern- 78–104. ments. London: KPMG Gill, I., and H. Kharas. 2007. An East Asian Renaissance: Ideas for Economic Kraay, A., and C. Raddatz. 2007. “Poverty Traps, Aid, and Growth.” Journal Growth. Washington, DC: World Bank. of Development Economics 82 (2): 315–47. ———. 2015. “Middle Income Traps: A Conceptual and Empirical Survey.” Kummritz, V., D. Taglioni, and D. Winkler. 2016. “Economic Upgrading World Bank Policy Research Working Paper No. 6594, World Bank, through Global Value Chain Participation: Which Policies Increase the Washington DC. Value Added Gains?” Policy Research Working Paper No. WPS 8007, Glawe, L., and H. Wagner. 2016. “The Middle-Income Trap–Definitions, World Bank, Washington, DC. Theories and Countries Concerned: A Literature Survey.” MPRA Paper Leibenstein, H. 1962. Economic Backwardness and Economic Growth: Studies 71196. University of Hagen, Germany. in the Theory of Economic Development. New York: John Wiley and Sons. The middle-income trap and upgrading along global value chains • 139 Lileeva, A., and D. Treffler. 2007. “Improved Access to Foreign Markets Rodrik, D. 2016. “Premature Deindustrialization.” Journal of Economic Raises Plant-Level Productivity…for Some Plants.” NBER Working Paper Growth 21 (1): 1–33. No. 13297, NBER, Washington, DC. Rossi, A. 2013. “Does Economic Upgrading Lead to Social Upgrading in MacGarvie, M. 2006. “Do Firms Learn from International Trade?” Review of Global Production Networks? Evidence from Morocco.” World Devel- Economics and Statistics 88 (1): 46–60. opment 46 (C): 223–33. Mariscal, A., and D. Taglioni. 2017. “GVCs as a Source of Firm Capabili- Roy, S., M. Kessler, and A. Subramanian. 2016. “Glimpsing the End of Eco- ties.” World Bank, Washington, DC. nomic History? Unconditional Convergence and the Missing Middle Morikawa, M. 2016. “The Effects of Artificial Intelligence and Robotics Income Trap.” Center for Global Development (CGD) Working Paper on Business and Employment: Evidence from a Survey on Japan.” 438, CGD, Washington, DC. Research Institute of Economy, Trade and Industry (RIETI) Research Taglioni, D., and D. Winkler. 2016. Making Global Value Chains Work for Papers No. 16E066, RIETI, Tokyo. Development. Washington, DC: World Bank. Nelson, R. R. 1956. “A Theory of The Low-Level Equilibrium Trap In Underde- Toh, T. V. 2013. “The Middle-Income Trap: Issues for Members of the Asso- veloped Economies.” The American Economic Review 46 (5): 894–908. ciation of Southeast Asian Nations.” Asian Development Bank Institute Ohno, K. 2009. “Avoiding the Middle-Income Trap: Renovating Industrial (ADBI) Working Paper Series No. 421, ADBI, Tokyo. Policy Formulation in Vietnam.” Association of Southeast Asian Nations Wacziarg, R., and K. H. Welch. 2008. “Trade Liberalization and Growth: Economic Bulletin 26 (1): 25–43. New Evidence.” The World Bank Economic Review 22 (2): 187–231. Oxford Martin School. 2016. Technology at Work v2.0: The Future is Not Wang, Z., S. J. Wei, X. Yu, and K. Zhu. 2016. “Characterizing Global Value What It Used to Be. Oxford, U.K.: Oxford Martin School. Chains.” Stanford Center for International Development (SCID) Working Ozturk, A. 2016. “Examining the Economic Growth and the Middle-Income Paper 578, SCID, Stanford, CA. Trap from the Perspective of the Middle Class.” International Business West, D. M. 2015. What Happens if Robots Take the Jobs? The Impact of Review 25 (3): 726–38. Emerging Technologies on Employment and Public Policy. Washington, Panther, S., and S. Flechtner. 2015. “Global and Domestic Inequalities and DC: Brookings Institution Centre for Technology Innovation. the Political Economy of the Middle-Income Trap.” Paper prepared for WEF (World Economic Forum). 2016. The Future of Jobs: Employment, the 27th Annual Society for the Advancement of Socio-Economics Con- Skills and the Workforce Strategy for the Fourth Industrial Revolution. ference, London, July 2–4. Global Challenge Insight Report. Geneva: World Economic Forum. Paus, E. 2014. “Latin America and the Middle Income Trap.” United Whittaker, H. D., T. Sturgeon, T. Okita, and T. Zhu. Forthcoming. Com- Nations Economic Commission for Latin America and the Caribbean pressed Development. Oxford, U.K.: Oxford University Press. Financing for Development Series No. 250, United Nations, Santiago. Woo, W. T. 2012. “China Meets the Middle-Income Trap: The Large Pot- Paus, E., and K. Gallagher. 2008. “Missing Links: Foreign Investment and holes in the Road to Catching-Up.” Journal of Chinese Economic and Industrial Development in Costa Rica and Mexico.” Studies in Compar- Business Studies 10 (4): 313–36. ative International Development 43 (1): 53–80. World Bank. 2013. China 2030: Building a Modern, Harmonious, and Cre- Pikos, A. K., and S. L. Thomsen. 2016. “Rising Work Complexity but ative Society. Washington, DC: World Bank. Decreasing Returns.” IZA – Institute of Labor Economics Discussion ———. 2016a. Global Economic Prospects: Divergences and Risks. Wash- Paper No. 9878. Bonn: IZA. ington, DC: World Bank. Porter, M. E. 1985. Competitive Advantage: Creating and Sustaining Supe- ———. 2016b. The Role of Global Value Chains in the Rise of Poland and rior Performance. New York: Simon and Schuster. the New High-Income Economies. Washington, DC: World Bank. Pritchett, L., and L. Summers. 2014. “Asiaphoria Meets Regression to the ———. 2016c. World Development Report 2016: Digital Dividends. Wash- Mean.” NBER Working Paper No. 20573, NBER, Washington, DC. ington, DC: World Bank. Pugliese, E., G. L. Chiarotti, A. Zaccaria, and L. Pietronero. 2015. “Complex ———. n.d. World Bank County and Lending Groups Classification (data- Economies Have a Lateral Escape from The Poverty Trap.” PLoS ONE base). Washington, DC: World Bank. Available at: https://datahelpdesk 12 (1): e0168540. .worldbank.org/knowledgebase/articles/906519-world-bank-country Rigg, J., B. Promphaking, and A. Le Mare. 2014. “Personalizing the Mid- -and-lending-groups. dle-Income Trap: An Inter-Generational Migrant View from Rural Thai- ———. n.d. World Development Indicators (database). Washington, DC: land.” World Development 59 (7): 184–98. World Bank. Available at: http://data.worldbank.org/data-catalog Robertson, P. E., and L. Ye. 2013. “On the Existence of a Middle Income /world-development-indicators. Trap.” Economics Discussion Paper No. 13.12, University of Western Australia, Perth. CHAPTER 6 Services trade and global value chains CECILIA HEUSER AND AADITYA MATTOO I n Moliere’s The Bourgeois Gentleman, M. Jourdain discovers GVC­ —­such as importing to export­ —­ apart from value chains that that he has “been speaking prose all my life, and didn’t even involve a single international transaction, an import or export know it!” We suspect that many of us have been working on of intermediate goods. It may well be true that multiple inter- global value chains (GVCs) without quite knowing it. A value national transactions along a value chain have economic impli- chain comprises “the full range of activities that are required cations that are qualitatively distinct from (the sum of) a series to bring a product from its conception, through its design, its of single international transactions, even though this has not yet sourced raw materials and intermediate inputs, its marketing, been established empirically. However, it may also be true that its distribution and its support to the final consumer.”1 A GVC the implications of even single international transactions along emerges when these activities are undertaken by entities based in a value chain are economically important and policy relevant. So or from different countries. Several studies of international trade in this chapter adopts a broader view of GVCs that also includes services inputs and of foreign direct investment in business serv- single international transactions, consistent with the approach in ices are thus potentially relevant to an examination of GVCs. chapter 2, covering both simple GVCs, with one border crossing, Two questions arise. What makes a value chain a GVC? And and complex GVCs, with two or more border crossings (Wang, why do services merit special consideration in a discussion of Wei, and Zhu 2017).2 GVCs? The second question can be reformulated as two questions: The first question has sometimes been reformulated as “How Should services be examined separately from goods in a discus- many ‘borders’ does a value chain have to cross to qualify as a sion of GVCs? And should they be treated as a single broad cate- GVC?” (perhaps analogously to a recent Nobel laureate’s ques- gory, the same way goods are? Our answer to both questions: In tion, “How many roads must a man walk down, before you call some respects no, in others yes. him a man?”). In the GVC case, some analysts have suggested a In some ways, services play a role similar to that of goods in fairly precise answer. For example, the widely used GVC partic- GVCs, whether they are meant for final consumption or as inputs ipation index proposed in Koopman and others (2010) is deter- in the production of goods or other services. In fact, the most mined by the foreign value added embodied in the gross exports detailed analysis of the role of services in value chains­ —­drawing of a given country and the domestic value added embodied in on the new world input-output tables and value-added trade the gross exports of third countries. This definition would set a databases­ —­ relates only to situations in which services are traded This chapter draws on three background papers by Van der Marel and Sáez (2016), Miroudot (2016), and Liu and others (2017). The authors thank Laura Alfaro, Maggie Chen, Nick Hope, Gary Hufbauer, Bradford Jensen, Przemyslaw Kowalski, Joscelyn Magdeleine, Andreas Maurer, Marcel Timmer, Zhi Wang, Zhigang Yao, and participants at the Making Global Value Chains Work for Economic Development workshop in Washington, DC, November 2016, for helpful comments. 141 142 • Measuring and Analyzing the Impact of GVCs on Economic Development in a manner akin to how goods are traded. However, services Fourth, services markets have seen considerable unilateral deserve special attention for four reasons, relating to how they liberalization that has facilitated the emergence of GVCs with are transacted, how they affect downstream sectors, how they services as both inputs and outputs. Unfortunately, international are regulated, and how international cooperation can contribute cooperation has striven to replicate mostly the goods model of to integrating national markets. “reciprocal market opening,” which has so far delivered little First, the notion of GVCs involving services needs to encom- incremental liberalization. Because the impediments are differ- pass a broader range of transactions. GVCs need to encompass ent for services-related GVCs, international cooperation needs to not only transactions crossing borders, but also transactions take a different form. In particular, much more could be achieved within countries between national and foreign entities. While through a greater emphasis on regulatory cooperation. there is good reason to take a similar broad view of GVCs involv- The first section below sets the stage by describing the role ing only goods, the case is overwhelming for services because of services in GVCs, and the second presents the methods and focusing only on cross-border trade would ignore the large datasets currently used to measure this role. The third examines share of international transactions in services that takes place the patterns that emerge, showing the absolute and growing through consumers traveling to other countries (consumption importance of services in GVCs and suggests possible reasons. abroad, or mode 2 in World Trade Organization [WTO] parlance); The fourth discusses the implications of services’ presence in commercial presence (through foreign investment, or mode 3 GVCs, particularly for total factor productivity and patterns of in the WTO); and the presence of natural persons (temporary comparative advantage. The fifth presents policy implications of immigrants, or mode 4 in the WTO).3 Even though the ability to the evidence on the impediments to services being part of GVCs, measure the role in GVCs of international services transactions and the sixth argues that these impediments are most effectively through commercial presence is limited, ways have been found addressed through new forms of regulatory cooperation. to estimate their economic impact.4 Second, the argument that services can have a substantial economic impact because they are vital inputs into produc- The role of services in global value chains ing downstream goods and services may not seem a sufficient reason for separate consideration. After all, goods such as com- In some ways services play a role analogous to that of goods. But puters are also vital inputs. But two features of services seem to their roles are also different, in that services facilitate the emer- merit special focus. One is that the very existence of GVCs is due gence of GVCs in a way that goods do not. Services can be seen as to improvements in such services as transport, communication, elements in GVCs that are different from the typical cross-border and computing (or information and communication technology or arm’s length trade usually analyzed in the case of goods.6 services) that have made it possible to fragment and coordinate production globally. Another is the growing evidence that when Services global value chains GVCs include finance, communications, transport, and profes- The emergence of GVCs has increased the opportunities for sional and other business services in favorable price–quality bun- international specialization not only in final goods and their parts, dles and diverse varieties, firms perform better. These services but also in services and services tasks. In many instances, serv- enable firms to invest in new business opportunities and better ices represent the end stage of a GVC, with services firms choos- production technology, to exploit economies of scale by con- ing to source their inputs internationally. For example, financial centrating production in fewer locations, to efficiently manage services providers have outsourced and offshored their back-of- inventories, and to make coordinated decisions with their sup- fice data-management and analytical tasks, architects their basic pliers and customers. The result can be increased total factor design tasks, and doctors the reading of radiological images. In productivity and shifts in the pattern of comparative advantage. each case, direct interaction with the client is by locally based Third, services also differ at least qualitatively from goods in services providers. It has been suggested that in fragmented the nature of the policies that both inhibit or encourage the emer- production processes of services, value is sometimes created gence of services-related GVCs. Border measures such as tariffs differently from how it is created in goods value chains. Instead are much less relevant for services trade than for goods trade, of following a linear value chain, in which products move sequen- and behind-the-border regulatory measures are much more rel- tially from upstream to downstream, adding value at each stage evant. Some examples: Cross-border trade in international trans- (a “snake” formation in the terminology of Baldwin and Ven- port services is impeded by the exclusion of third-country provid- ables 2013), value creation in services value chains may occur as ers and by quantitative restrictions in bilateral agreements. Trade a network of activities, such as platform-based communication through commercial presence in banking and communication or transportation networks (a “spider” formation in Baldwin and services must confront restrictions on foreign ownership and reg- Venables). In such cases, multiple parts come together to add ulatory requirements that can be discretionary and discriminatory. value simultaneously in forming a final product or component­ The presence of foreign professionals is prevented by restrictive —­ or through alternative models, such as facilitated user net- visa and work permit rules as well as by a refusal to recognize works (which create value by linking customers, as in insurance their qualifications and licenses. And trade in all data-intensive or banking services) and solution shops (which create value by services is threatened by diverging national privacy laws.5 solving customer problems; Miroudot 2016). Services trade and global value chains • 143 Services as links in global value chains How services participation in global value One reason to consider some services in GVCs separately from chains is measured goods is their role in enabling GVCs to emerge. The international fragmentation of production was driven partly by changes in For a long time, measuring trade in services took a back seat transport, logistics, and information and communication tech- because data on trade in goods was more extensive and readily nology services. In particular, lower costs and improvements in available. More recently, measuring trade in services has received these services made it possible for firms to manage production impetus from the new prominence in international policy and processes that are geographically split (Jones and Kierzkow- negotiating agendas given to liberalizing trade in services, the ski 2001a, 2001b). Even though the anecdotal evidence itself is increased importance of services in GVCs, and the availability of compelling, a serious gap in the literature is not having rigorous multicountry input-output tables. empirical evidence on how improved access to these connecting services across space and time has facilitated the emergence of Statistics on trade in value added cover services as links, GVCs. outsourced inputs, or final products in global value chains As long as services are final products or inputs supplied by other Services as outsourced inputs in global value chains firms in the production process, input-output tables can help Besides their role as links between different stages of value identify their contribution to value added in output or exports chains, services often are important inputs in the production pro- (Francois and Woerz 2008; Nordås 2008). cess of manufacturing goods and services. For example, a value The starting point for analyzing the contribution of services chain may start with research, design, and engineering activities to GVCs is the decomposition of value added in exports by its that are clearly services inputs when they are outsourced. At the origin. Following seminal work by Hummels, Ishii, and Yi (2001), other end of the value chain are other services such as marketing an expanding literature has proposed alternative decomposi- and distribution that are also important stages in ensuring that a tions of trade in value added and measures of participation in product reaches the consumer. Therefore, services are not only GVCs (Koopman, Wang, and Wei 2014; Foster-­ McGregor and support functions that enable GVCs, but they are also crucial Stehrer 2013; Los, Timmer, and de Vries 2016). As in Miroudot inputs in key stages of production. (2016), the analysis here relies on the calculations published by Yet being vital inputs into goods and services production by the Organisation for Co-operation and Development (OECD)– itself does not constitute sufficient reason to consider services WTO Trade in Value-Added database, which features several separately from goods in analyzing GVCs. After all, goods are indicators that account for services value added in trade (OECD also vital inputs. The fact that access to some services of suffi- 2013). ciently high quality, low price, and diversity may matter for firm The main indicator is the total value added (VA) of the services performance is an empirical question akin to the role of infor- sector embodied in gross exports of industry i and country c as mation and communication technology goods in determining a percentage of total gross exports of i of country c, (SERV VAc,i). performance. But the fact that access to services inputs comes It is calculated as: through foreign direct investment and the movement of people ∑ j ∈S Vc,j (B c,c ) ji EXGR c,i /EXGR c,i + SERV VA c,i =  more often than it does for goods inputs warrants considering ∑ p ∑ j ∈S Vp,j (B p,c ) ji EXGR c,p,i /EXGR c,i(1) services separately and through a broader view of GVCs. The rel- ative importance of digital delivery in services also requires mod- where Vp,j is the value-added share of services industry j in coun- ifying the traditional customs-mediated and customs-measured try p; B is the global Leontief inverse of the intercountry input- role of international transactions within GVCs. output matrix, B = (I – A) –1, and therefore its ji ‑th element (Bp,c)ji represents the total requirements of j from p to produce a unit of Services as in-house inputs in global value chains i in country c ; EXGRc,p,i is gross exports from country c to country Another feature of services as inputs arises in a notion of GVCs p for any given industry i; and EXGRc,i is total gross exports for that goes beyond arm’s length market-based transactions to country c and industry i. The first term then represents all direct functions within the firm. It is common for firms to develop their and embodied domestic services value added in the exports of own support services in house, such as research and develop- product i from country c, and the second term represents all for- ment activities or information technology capacity. This means eign domestic services value added embodied in the exports of that services are produced not only by services firms, but also by product i from country c. manufacturing firms (Kelle 2013), which often export a variety of The services content of gross exports can then be decom- headquarters services to their affiliates. Some analysts argue that posed into a domestic and a foreign part, and the domestic part this “servicification” inside firms may need to be considered for a can be further decomposed into the direct domestic services full assessment of the impact of services on trade and value cre- industry value added content of gross exports (the value added ation (Miroudot 2016). But as noted below, this deconstruction from the exporting services sectors), the indirect domestic of activities within firms, when taken to the limit, may blur the services content of gross exports (the domestic services value distinction between goods and services because all tasks could added embodied in other exporting industries), and the re-im- be considered services. ported domestic services value-added content of gross exports 144 • Measuring and Analyzing the Impact of GVCs on Economic Development (domestic services value added in imported intermediate inputs industry. While such a framework is not currently available, ini- used in exports; figure 6.1).7 This decomposition can also be tiatives in this direction are under way by the U.S. Bureau of Eco- analyzed by services type. A similar approach can be used in nomic Analysis (BEA) and the OECD (Fetzer and Strassner 2015; measuring the services value added embodied in foreign final OECD 2015). In both cases the proposed approach is to link demand. existing supply-use tables to ownership and trade data­ —­ such as Value-added trade statistics based on multicountry input- OECD Inter-Country Input-Output Tables to the Activity of Mul- output tables are a starting point for understanding the impor- tinational Enterprises Database, or BEA input-output accounts tance of services inputs in GVCs, but these statistics cannot fully to BEA surveys and U.S. Internal Revenue Service Statistics of capture all services that are relevant to fragmented production Income data. processes. In particular, services traded through WTO mode 3 will not be accurately identified as foreign services in traditional Services as in-house inputs in global value chains value-added measures. When services inputs are supplied in-house, value-added analy- sis does not capture their contribution to GVCs. Additional infor- Existing statistics on trade in value added do not fully mation is needed to identify these services activities within each capture services traded though commercial presence production process. Since the supply of services through commercial presence One way to address this issue is to consider business functions, abroad is an important way of conducting international trans- in order to distinguish between the primary or core activity of the actions in services (mode 3–commercial presence), the distinc- firm and support functions such as research and development, tion between foreign- and domestic-owned firms is particularly sales, marketing, or information technology services. But statistics relevant for services. Accurately assessing the contribution of on business functions have only recently started to be collected in services to GVCs requires that the ownership status of the firm some national surveys (Sturgeon and others 2013).9 Alternatively, that originates the value added affect whether that contribution labor force surveys can be used to identify business functions by is classified as domestic or foreign, as this is bound to inform matching occupation classifications to business functions (Timmer, services trade policy.8 Stehrer, and de Vries 2015). Each industry is assigned one business A GVC accounting framework that allows for this sort of dis- function to describe its core activity, which usually covers occupa- tinction between firms can be similar to the global supply-use/ tions directly related to the production process. All other business input-output tables commonly used now, such as OECD functions are classified as support activities (or secondary busi- Inter-Country Input-Output Tables, but it would also need to ness functions) and can be regarded as services activities if they identify foreign or domestic ownership for each commodity/ would be classified as services if outsourced. This approach may supplement the usual trade in value added analysis, providing a sense of the role of in-house services in GVCs.10 FIGURE 6.1 Deconstructing services value added in gross While it is theoretically possible to determine a single core func- exports tion per industry, it can be difficult in practice to establish what is not a support function. Even if it were feasible to distinguish between tasks that would be services if they had taken place at arm’s length Services value added in exports c,i (such as bookkeeping) and other tasks that are intrinsically manu- facturing or agricultural in nature (such as wood processing), such distinctions slide on a slippery slope and raise the question: What is not a service? Ultimately, almost any task can be conceived of as an arm’s length service. For example, one could either directly employ a worker in the horticultural sector or buy “fruit-picking” services Foreign Domestic from an individual or a firm. So this deconstruction of the firm into services services constituent tasks could reduce each firm into a bundle of services. value added value added in exports c,i in exports c,i The emerging patterns of services in global value chains This section presents some stylized facts about the increasing Direct Indirect Re-imported domestic services domestic services role of services in GVCs­ —­first in aggregate and then across domestic services value added value added value added countries and industries. It then discusses why this might be hap- in exports c,i in exports c,i in exports c,i pening. The evidence is obtained by computing the value-added measures described in the previous section using the OECD Source: Authors’ construction. Trade in Value-Added database. Some results are also presented to illustrate the potential relevance of in-house services in GVCs. Services trade and global value chains • 145 FIGURE 6.2 Gross exports of goods and services as a FIGURE 6.3 Value-added exports of goods and services as percentage of total world gross exports, 1980, 1995, and a percentage of total world value-added exports, 1980, 2009 1995, and 2009 Percent Percent 100 Goods 100 Goods Services Services 75 75 50 50 25 25 0 0 1980 1995 2009 1980 1995 2009 Source: Authors’ calculations based on Johnson and Noguera 2016. Source: Authors’ calculations based on Johnson and Noguera 2016. The patterns of services in GVCs country as well, with no major differences between developed or developing, high-technology or low-technology, or high-wage or The share of services in value-added trade is large and increasing low-wage countries (Baldwin, Forslid, and Ito 2015). Multiple studies have found that the share of services in trade in value added is both large (significantly larger than the share of The increasing share of services in value-added trade was services in gross trade) and increasing (OECD, WTO, and World driven by services embodied in exports Bank Group 2014). While services as a share of total world gross Now take a closer look at the recent evolution of services value exports have remained around 20% since 1980, in value-added added in exports, decomposed into direct and indirect domestic terms they have increased from below 30% to more than 40% value added and foreign value added (figure 6.4). While direct (compare figures 6.2 and 6.3). For Asia, this pattern holds by exported value added shows a notable increase from 1995–2011, FIGURE 6.4 Share of direct, indirect, and foreign services value added in world gross exports, 1995–2011 $ (billions) 10 Direct domestic Indirect domestic Foreign 8 6 4 2 0 1995 2000 2005 2008 2009 2010 2011 Source: Miroudot 2016 based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. 146 • Measuring and Analyzing the Impact of GVCs on Economic Development more than 65% of the growth of services value added in exports The share of services value added in exports varies was due to an increase in services embodied in other exports. significantly across countries Both domestic and foreign embodied services grew, but the for- Decomposition of services value added in exports by country in eign services value-added component grew the most. 2011 supports the general observations that the share of services Figures 6.2–6.4 suggest that an increasing part of manufacturing value added is high and that embodied services explain a large exports corresponds to services value added and that a growing share of the total for all countries (figure 6.5). However, countries share of these “additional” services is being sourced abroad. How- differ in their share of services value added in exports, ranging from ever, since these measures are based on cross-border trade, value 35% in Chile to close to 90% in Luxembourg. These differences added from foreign services traded through commercial presence is reflect some specialization patterns: countries on the bottom of the counted in the “domestic” category, as previously mentioned. The figure specialize in exports of commodities (Chile, Norway) or man- apparent shift toward foreign services in the decomposition of serv- ufactured goods (Republic of Korea, Mexico), while economies on ices value added probably understates what actually happened. the top are services exporters. Countries specialized in services also FIGURE 6.5 Direct and indirect domestic services value added and foreign services value added in gross exports, by country, 2011 Luxembourg Netherlands Ireland Belgium Denmark Greece United Kingdom France Switzerland Latvia Israel Lithuania Estonia Sweden Spain Iceland Portugal India New Zealand Austria Costa Rica United States Italy Finland Turkey Slovenia Poland Hungary Germany Japan Bulgaria Brazil Philippines Czech Rep. Slovak Rep. Australia Canada South Africa Thailand Russian Federation China Malaysia Romania Korea, Rep. Direct domestic Norway Indirect domestic Viet Nam Foreign Mexico Chile 0 25 50 75 100 Percent Source: Miroudot 2016 based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. Services trade and global value chains • 147 have more indirect (domestic and foreign) services value added in in part by an increase in services value added embodied in exports because services are mainly produced with other services. exported manufactures. The growing importance of services in the economy has been The share of services value added in exports also varies a matter of discussion for a long time. Bhagwati (1984) set out the across industries main reasons: “splintering” (outsourcing services formerly pro- The share of services value added in exports ranges from 11% in vided in-house by manufacturing firms), the high-income elastic- mining to 38% in chemicals and motor vehicles (figure 6.6). Cau- ity of the demand for services, and relative price shifts due to tion in interpreting these results is advised, since value-added the lower growth of productivity in services than in goods. Splin- measures are based on input-output tables defined by arm’s tering was seen as the spontaneous result of the specialization length transactions and thus exclude services provided in-house. opportunities arising from growth and technical change. Mining exports are in many cases driven by large state-owned More recently, the increasing importance of services within enterprises that are likely to provide most services in-house (Mir- manufactured goods­ —­ —­ servicification, as it has been called­ has oudot 2016). For example, in Australia, where that is not the case, been extensively documented, in line with the evidence in fig- the share of services value added in exports is 24%. ures 6.2–6.4.11 While it has not yet been possible to empirically The decomposition of services value added by type of serv- establish the cause of servicification, Baldwin, Forslid, and Ito ices seems similar across manufacturing industries (see figure (2015) explore its potential sources in a way that echoes Bhag- 6.6). Distribution represents about a third of services value wati (1984). These include reclassification, task-composition added in exports, as do business services, which includes tele - shifts in connecting services and final goods, and task–relative communications services, computer services, professional serv- price shifts. ices, research and development services, consulting, advertising and marketing services, technical testing services, and environ- Reclassification. Over the past decades, many of the services tra- mental services. The last third is split among transport, finance, ditionally sourced in-house by manufacturing firms, and thus clas- and other services (a category covering construction, hotels and sified as manufacturing, began to be sourced at arm’s length and restaurants, government services, health and education, enter- classified accordingly as services. The servicification that arises tainment, and audio-visual services). from this reclassification can happen even if there is no change in products, production process, or relative price of inputs. This Why is the share of services in value-added exports argument corresponds to Bhagwati’s “splintering” reason. increasing? Figures 6.2–6.4 reveal an increasing share of services in value Task-composition shift: connecting services. The emergence added exports, suggesting that the increase may be explained of GVCs requires connections among geographically separate FIGURE 6.6 Decomposition of services value added in world gross exports, by manufacturing industry, 2011 Mining Agriculture Coke, petroleum Fabricated metals Total manufacturing Utilities Nonmetallic minerals Machinery Basic metals Other transport Wood Electrical machinery Other manufacturing ICT and electronics Rubber and plastics Textiles and apparel Food products Paper, print, publishing Motor vehicles Chemicals 0 10 20 30 40 Percent Distribution Transport Finance Business services Other Source: Miroudot 2016 based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. 148 • Measuring and Analyzing the Impact of GVCs on Economic Development production units, which typically involve services links. These The limits of value-added analysis: In-house services play links (including telecommunications, transportation, and mailing) a large role in manufacturing contribute to the value added embodied in the final good. So As mentioned earlier, services enter GVCs not only as outsourced outsourcing and offshoring tend to increase the share of services inputs or final products, but also as inputs frequently provided in a final good’s value added. in-house, which traditional value-added measures do not cap- ture. Illustrating how relevant this omission can be, Miroudot Task-composition shift: changes in final goods. The second (2016) matched occupation classifications and business functions task-composition shift arises from changes in the nature of the for 37 countries over 1995–2013. This decomposition of jobs final manufactured goods. For example, today’s cars contain embodied in manufacturing according to business function by software, which comes from the services sector. Similarly, many industry reveals considerable variation across industries, with other manufactured goods have become more intensive in serv- employment in core activities (operations) at more than 90% in ices. This argument resembles the argument that the income agriculture, but at only about 33% in coke and petroleum (figure elasticity of demand for services is high, but in this case techno- 6.7). This variability carries over to differences across countries logical progress is enhancing the services content and the qual- as well. On average, the core activities of manufacturing firms ity of manufactured goods. account for only 50% of employment, meaning that half the employees of manufacturing firms perform services activities. Task–relative price shift. For a variety of reasons, including the This suggests that the services value added embodied in man- need for coordination and face-to-face interaction, offshoring ufacturing, and thus the contribution of services to GVCs, could tasks tends to be easier for intermediate goods than for interme- be much higher than that captured by traditional value-added diate services. Since the decision to offshore a task is typically measures. driven by cost-reduction motives, there is a natural tendency for offshoring to reduce the relative price of the offshored tasks. If most of the offshored tasks are typically performed by the man- The implications of services in global value chains ufacturing sector, then offshoring would­ —­ in a mechanical way­ —­raise services value added in final manufactured goods. This As seen, services constitute the vital connecting links of value argument is a variant of the differential productivity growth chains as well as a range of inputs sourced either at arm’s length reason but is being driven by differences between goods and or in-house. These value chains are “global” not just when trans- services in cost-reducing opportunities through offshoring. actions cross international boundaries, but also when consumers FIGURE 6.7 Decomposition of jobs embodied in gross manufacturing exports, by business function by industry, 2011 Utilities Other manufacturing Transport equipment Electronics Machinery Metals Nonmetallic minerals Rubber and plastics Chemicals Coke, petroleum Paper, print, publishing Wood Textiles and apparel Food products Mining Agriculture 0 25 50 75 100 Percent Research and development, engineering, and related technical services Operations Transport, logistics, and distribution Marketing, sales, and after-sales services Horizontal support activities Source: Miroudot 2016. Services trade and global value chains • 149 or providers do so, especially by establishing a commercial pres- To make this information amenable to econometric analysis, the ence abroad. authors aggregated it into time-varying reform indexes. They Growth theories have emphasized that trade in intermediate then related the total factor productivity of about 4,000 manu- goods and services generally improves the allocation of capital facturing firms to the state of liberalization in services sectors, and labor across sectors and countries (Jones 2011). The litera- taking into account other aspects of openness, such as tariffs on ture helps draw out the implications of services in GVCs for two output and intermediate inputs as well as foreign direct invest- key aspects of economic performance: the growth of productiv- ment in final and intermediate goods sectors. ity and the evolution of comparative advantage. The results suggest that pro-­ competitive reforms in banking, The literature uses a framework based on Rajan and Zingales transport, insurance, and telecommunications boosted the pro- (1998) and first applied to services broadly in Arnold, Javorcik, ductivity of both foreign and locally owned manufacturing firms. and Mattoo (2011) in a study of the Czech Republic. The typi- A one-standard-deviation increase in the aggregated index of cal study examines whether increased access to specific foreign services liberalization resulted in a productivity increase of 11.7% services enhances performance in downstream sectors whose for domestic firms and 13.2% for foreign enterprises. The largest production is relatively intensive in those services. The usual additional effect was for transport reforms, followed by telecom- specification takes the following form: munications and banking reforms. Several other studies show that access to low-cost and Yi,t = α + β × services linkagej,t–1 + Xj,t–1 × π + δi + γt + εi,t high-quality (domestic or foreign) producer services can promote and productivity and economic growth (Hoekman and Mattoo 2008). services linkagej,t = ∑kaj,k × accessk,t Using firm-level data for the Czech Republic for 1998–2003, Arnold, Javorcik, and Mattoo (2011) found a positive effect on where i is the firm, j is the sector, and X is a matrix of sector-level the productivity of domestic firms in downstream manufacturing control variables; Yi,t is the outcome of interest (productivity, as a result of services sector reforms leading to greater foreign comparative advantage); and services linkage, the key explan- direct investment. Using the annual manufacturing survey of Chil- atory variable, is the interaction between a measure of a spe- ean firms, Fernandes and Paunov (2012) found a positive effect of cific sector’s dependence on services inputs and a measure of substantial foreign direct investment inflows in producer services services access (which could be related, for example, to serv- sectors on the total factor productivity of Chilean manufactur- ices policy reforms or foreign direct investment inflows).12 The ing firms. Their findings also suggest that services foreign direct hypothesis is that sectors using specific services more intensively investment fosters innovation in manufacturing and offers oppor- benefit more from the reform of those services. This general tunities for laggard firms to catch up with industry leaders.13 framework becomes clearer in the examples discussed below. These benefits arise not just from foreign investment but also from cross-border trade in services. For example, Amiti and Wei For productivity (2009a) found that services offshoring by high-income countries India offers a powerful example of the benefits of greater partic- tends to raise the productivity of their manufacturing sector.14 ipation in manufacturing value chains by foreign services firms. To investigate whether regulations in domestic services Conventional explanations of the modest resurgence of Indian markets have an effect on industries that rely on GVC linkages manufacturing since the early 1990s have focused on policy in services to generate value added, Van der Marel and Sáez reforms in manufacturing industries. But a key factor lies outside (2016) differed from the earlier studies by looking at all down- manufacturing and in the services sector. Reforms in the 1990s stream sectors rather than just manufacturing. They examined visibly transformed services sectors, with greater openness and the impact on domestic value added rather than on productivity improved regulation leading to dramatic growth in domestic and and on the link to upstream services sectors through backward foreign investment. Indian manufacturing firms were no longer at foreign and domestic linkages. Their key interactive variables the mercy of inefficient public monopolies but could now source were a product of the foreign and domestic backward linkages services from a wide range of domestic and foreign providers of services for each downstream sector with services-specific operating in an increasingly competitive environment. As a regulatory policies. Both entry restrictions and policies affecting result, they had access to better, newer, more reliable, and more the operations of firms matter, but the strength of their impact diverse business services. depends on the type of backward linkages that are more impor- These improvements enhanced firms’ ability to invest in new tant for the industry in question. Industries that rely more on business opportunities and better production technology, to backward foreign linkages of services are adversely affected by exploit economies of scale by concentrating production in fewer entry barrier regulations, while those that rely more on backward locations, to efficiently manage inventories, and to coordinate domestic linkages of services are more sensitive to behind-the- decisions with suppliers and customers. border regulations. To analyze the link between services reforms and manufac- turing productivity in India, Arnold and others (2016) collected For comparative advantage detailed information on the pace of reform across Indian serv- Since a large part of goods trade includes trade in embodied ices sectors, with a focus on entry and operational restrictions. services, the development of the domestic services sector and 150 • Measuring and Analyzing the Impact of GVCs on Economic Development access to imported services inputs can be expected to influence manufacturing sectors, the key explanatory variable is the inter- comparative advantage in manufacturing trade. The impact of action between a measure of the development of financial (or services development on manufacturing trade is not straight- business) services and the financial (or business) services inten- forward. Since services are used as inputs in the production of sity of each manufacturing sector. Domestic services develop- manufactured goods, their development can increase manufac- ment has a mixed effect on the revealed comparative advantage turing production. But since services and manufacturing com- of manufacturing exports: services development reduces the pete for resources, the development of services can be at the revealed comparative advantage of manufacturing exports in expense of manufacturing. For example, the development of the manufacturing sectors with low embodied services but increases services sector has drawn resources away from manufacturing it in sectors with a high degree of embodied services (figure 6.8). not just in industrial countries such as the United Kingdom and Liu and others also considered the role of services imports the United States, but also in developing countries such as India in overcoming the limitations of domestic services markets. In (see, for example, Kochar and others 2006). countries with less developed services, manufacturing exports Some early studies examined the link between services benefit more from access to foreign services inputs. Such a as inputs in manufacturing and the pattern of manufacturing bypass effect is also discussed in a theoretical model by Ju and exports using single national input-output tables. For example, Wei (2010), which derives the conditions for financial globaliza- Francois and Woerz (2008) found significant and strong positive tion to serve as a substitute for reforms of domestic financial effects of increased business services openness (greater imports) systems. These results suggest that lower services trade barriers on some industries. Their reliance on national data means that can help developing countries bypass inefficient domestic serv- inputs cannot be broken down according to their origins, and ices provision and promote their manufacturing exports through services inputs are mismeasured due to two-way trade in inter- intersectoral linkages. mediate products. More recently, Stehrer, Foster, and de Vries (2012), Timmer and others (2013), and Liu and others (2017) used Direct and indirect value-added exports of services the newly constructed international input-output tables to more The patterns of direct and indirect domestic value-added precisely measure the embodied services and indirect trade exports of services for financial and business services reveal how through other sectors. goods and services value chains and comparative advantage Liu and others (2017) focused on two key services sectors: evolve (figures 6.9 and 6.10). The horizontal axis measures direct financial services and business services. Well-functioning finan- value-added exports of services and the vertical axis measures cial sectors are critical in mobilizing resources, stimulating invest- indirect domestic value-added exports of services (the value- ment, and helping firms (and households) manage risk. Business added exports of services embodied in exports of goods). Lines services cover a variety of critical activities, from software con- representing the median shares divide the countries into groups sulting and data processing to management consultancy, engi- occupying four quadrants. neering, and research and development. Intensive use of these For financial services, figure 6.9 shows: modern services can help manufacturing firms increase pro - • In the bottom left quadrant are Greece, the Russian Feder- ductivity, reduce the cost of doing business, expand their input ation, and Turkey. The low competitiveness of financial serv- choices, differentiate their products from those of competitors, ices in these countries is reflected in the low share of direct and strengthen their after-sale customer services. But these are exports and the low level of embodied exports­ —­which could the services that most strongly provoke deindustrialization con- reflect the low financial services intensity of goods produc- cerns, such as financial services in industrial countries such as the tion, the reliance primarily on imported financial services, or United Kingdom and the United States and business services in both. developing countries such as India and the Philippines. • In the top left quadrant are China, India, and the Republic of Liu and others (2017) quantified the indirect role of services Korea, which are not yet sufficiently competitive to export in international trade in goods and construct new measures of financial services directly but which do export a significant revealed comparative advantage based on domestic value added share indirectly. That goods sectors in these countries rely in gross exports. Embodied services in manufacturing sectors significantly on domestically produced financial services were computed using a method developed by Koopman, Wang, could be because financial services in these countries have and Wei (2014) and Wang, Wei, and Zhu (2013) that generalizes reached an intermediate level of development at which they the vertical specialization measures proposed by Hummels, Ishii, can compete in the domestic market but not yet internation- and Yi (2001). Revealed comparative advantage is calculated ally. It could also be that restrictions on cross-border imports based on domestic value added in gross exports, as in Koop - in these countries oblige goods producers to use domesti- man, Wang, and Wei (2014) and Wang, Wei, and Zhu (2013), who cally produced services. improved on the traditional (Balassa 1965) measure of revealed • In the top right quadrant are Austria, Belgium, and the Neth- comparative advantage by taking into account both domestic erlands, whose more developed financial services sector production sharing and international production sharing.15 exports both directly and indirectly. In Liu and others’ econometric analysis of the impact of • In the bottom right quadrant are such “mature deindustrializ- services development on revealed comparative advantage in ers” as the United Kingdom and the United States, where the Services trade and global value chains • 151 FIGURE 6.8 Financial development and revealed comparative advantage, by degree of embodied services, 2005 Revealed comparative advantage based on domestic value added in gross exports Basic and fabricated metals (low embodied financial services) 2.0 1.5 1.0 0.5 0.0 0.0 0.5 1.0 1.5 2.0 Bank private credit / GDP Food, beverages, and tobacco (high embodied financial services) 3.0 2.5 2.0 1.5 1.0 0.5 0.0 0.0 0.5 1.0 1.5 2.0 Bank private credit / GDP Source: Liu and others 2017. Note: The figures show a negative association between the manufacturing revealed comparative advantage and a measure of financial development for a sector with low embodied financial services, but a positive association for a sector with high embodied financial services. Includes data for all Organisation for Eco - nomic Co-operation and Development members except Chile, Iceland, Israel, New Zealand, Norway, and Switzerland as well as data for Brazil, Bulgaria, China, Cyprus, India, Indonesia, Lithuania, Malta, Romania, Russian Federation, Chinese Taipei, and rest of the world. domestic tangible industries have shrunk in importance and country is not competitive in producing these services, it nei- financial services are mostly exported directly. ther exports them, nor do its goods sectors import them. When For business services, figure 6.10 shows two interesting a country is competitive, it exports both directly and indirectly. differences: • In the bottom right quadrant, India offers an example of “pre - • Given the greater cross-border tradability or openness to trade mature deindustrialization,” where direct exports of busi- of business services, there is less scope for an intermediate ness services are high but indirect exports are low, perhaps stage (for countries to populate the top left quadrant). When a because of the relative weakness of goods sectors. 152 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 6.9 Direct and indirect value added exports of financial services as shares of GDP, 1995–2009 Indirect value added / GDP 0.20 Median Belgium 0.15 China Netherlands Korea, Rep. Austria 0.10 India Median United Kingdom 0.05 United States Turkey Greece Russian Federation 0.00 0.00 0.05 0.10 0.15 0.20 Direct value added / GDP Source: Author’s calculations based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. Note: Includes data for all Organisation for Economic Co-operation and Development members except Chile, Iceland, Israel, Ireland, Luxembourg, New Zealand, Norway, and Switzerland as well as data for Brazil, Bulgaria, China, Cyprus, India, Indonesia, Lithuania, Malta, Romania, and Russian Federation. FIGURE 6.10 Direct and indirect value-added exports of business services as shares of GDP, 1995–2009 Indirect value added / GDP 0.6 Median 0.5 0.4 0.3 0.2 Median 0.1 India 0.0 0.0 0.1 0.2 0.3 0.4 0.5 0.6 Direct value added / GDP Source: Author’s calculations based on Organisation for Economic Co-operation and Development–World Trade Organization Trade in Value-Added database. Note: Includes data for all Organisation for Economic Co-operation and Development members except Chile, Iceland, Israel, New Zealand, Norway, and Switzer- land as well as data for Brazil, Bulgaria, China, Cyprus, India, Indonesia, Lithuania, Malta, Romania, and Russian Federation. Services trade and global value chains • 153 Policy implications numerous second-generation restrictions remain on entry, own- ership, and operations. Even where there is little explicit discrim- Two types of policy issues inhibit an enhanced role for services in ination against foreign providers, market access is often unpre- goods value chains and the emergence of services value chains: dictable because the allocation of new licenses remains opaque explicit restrictions on foreign services and services suppliers, and highly discretionary in many countries. Regulatory discretion and regulatory divergence across jurisdictions. Explicit restric- is accentuated by a lack of accountability in a number of coun- tions on linking services, such as telecommunications, inhibit the tries where regulators are not required to provide reasons for emergence of all GVCs. Restrictions in other services­ —­such as rejecting a license application or where foreign providers do not finance, business, education, and health services­ —­ either block have the right to appeal regulatory decisions. the emergence of GVCs involving services or increase the associ- Across regions some of the fastest growing countries in Asia ated transaction costs. Regulatory divergence reduces the inter- and the oil-rich Gulf states have restrictive policies on services, compatibility of goods and services and services components while some of the poorest countries are remarkably open­ —­as that is needed to enable fragmenting production across jurisdic- measured by the Services Trade Restrictions Index, which takes tions. This divergence is one reason that GVCs have been slow values from 0 for completely open regimes to 100 for completely to emerge in sectors such as education and health. Regulatory closed (map 6.1).17 Across sectors, professional services and cooperation is necessary not just to address regulatory diver- transportation are among the most protected industries in both gence, but also to facilitate the removal of explicit restrictions. industrial and developing countries, while retail, telecommunica- tions, and even finance tend to be more open (figure 6.11).18 National policy barriers to international trade in services Unlike the rich information on policies affecting trade in goods, International cooperation and services in global value chains information remains limited on policies affecting trade in serv- Much of the openness in services markets and the emergence ices.16 The World Bank Services Trade Restrictions Database of services-related GVCs has come from unilateral liberalization. reveals interesting policy patterns. Although public monopolies Services trade negotiations have generated an abundance of are now rare, and few services markets are completely closed, rules and commitments that have enhanced the transparency MAP 6.1 Restrictiveness of services trade policy, 2008–10 Services Trade Retrictions Index 0.0–10.0 (1) 10.0–20.0 (37) 20.0–30.0 (31) 30.0–40.0 (11) 40.0–50.0 (13) 50.0–60.0 (5) 60.0–70.0 (4) 70.0–80.0 (0) 80.0–100.0 (1) No data (105) Source: Borchert, Gootiiz, and Mattoo 2014. Note: This figure compares the restrictiveness of services trade policy across countries based on the World Bank Services Trade Restrictions Index, which ranges from 0 (completely open) to 100 (completely closed), The World Bank Services Trade Restrictions Database covers 103 countries (79 developing) and financial, basic telecommunications, transport, distribution, and selected professional services. Data were collected between 2008 and 2010. 154 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 6.11 Services trade restrictiveness by services sector and region Professional services Transportation Europe and Central Asia Retail Telecommunications Financial OECD Latin America and Caribbean Sub-Saharan Africa East Asia and Pacific Middle East and North Africa South Asia Gulf Cooperation Council 0 20 40 60 80 Services Trade Restrictions Index Source: Borchert, Gootiiz, and Mattoo 2014. Note: This figure compares the restrictiveness of services trade policy across countries based on the World Bank Services Trade Restrictions Index, which ranges from 0 (completely open) to 100 (completely closed). Data were collected between 2008 and 2010. and security of market access but have not produced much Then, developing country policymakers would be reassured that additional liberalization. One reason is a form of “negotiating any regulatory inadequacies that could undermine the benefits tunnel vision,” which has led to a focus on reciprocal market of liberalization would be diagnosed and remedied before any opening rather than on creating the regulatory preconditions market-opening commitments take effect. for liberalization. More could be achieved if negotiations offered A second obvious reason for cooperation is that regulatory regulators the opportunity not just to tie their hands (through divergence segments markets. Firms must fulfill the regulations agreed-on commitments) but also to secure assistance to deal of each market separately­ —­ such as financial and accounting with problems they cannot solve on their own.19 standards­ —­which reduces the scope for exploiting economies of One obvious reason for international cooperation is that scale and the intensity of competition in each segment. Regula- poorer developing countries do not always have regulatory insti- tory convergence through harmonization or mutual recognition tutions equipped to deal with international competition. These of regulations­ —­or a combination of the two, as has tended to countries would participate meaningfully in negotiations that happen in the European Union­ —­creates an integrated market for offered an opportunity not merely to make binding commitments competition, economies of scale, and GVCs to flourish. but also to mobilize assistance for national regulatory reform. A less obvious case for regulatory cooperation is that even Market-opening negotiations now take their course in the WTO countries with sound national regulatory institutions can find it or in regional fora with only ad hoc links to international assis- difficult to address market failures related to services trade that tance for regulatory reform, including that from institutions such originate outside their jurisdiction. A country will be reluctant to as the International Monetary Fund and the World Bank. It would open its financial markets unless it is confident that it can pre - help establish a credible mechanism for regulatory assistance vent market instability and protect its consumers from unsound to support liberalization commitments by developing countries. foreign financial institutions, to let its citizens’ data flow to other Services trade and global value chains • 155 jurisdictions unless it is reasonably certain that that data will be binding corporate rules that also ensure data protection consis- kept secure, or to open its transport and Internet-based services tent with the directive. markets unless it is convinced that the gains from liberalization The U.S.-EU Safe Harbor Framework, recently supplanted will not be appropriated by international oligopolies. In some by the so called Privacy Shield Agreement, was developed in cases, such as the supply of services through locally incorporated response to the absence of a finding that the United States pro - subsidiaries, the importing country can, at least in principle, deal vides adequate data protection.21 The European Commission unilaterally with market failure because the provider is in its juris- recognized the Safe Harbor Framework Privacy Principles as pro - diction. But doing so requires adequate regulatory capacity and viding adequate protection for personal data transfers from the could increase the costs of trade by fragmenting markets (say, by European Union to approximately 3,000 companies in the United requiring local capital adequacy or local servers). In other cases, States that have signed up to the principles.22 A key difference such as cross-border banking, transport, or data-processing between the Safe Harbor Framework and the EU Data Privacy services, addressing market failure efficiently requires the coop - Directive adequacy standards is that the Safe Harbor Framework eration of the regulator in the exporting country. recognizes the self-regulatory approach with U.S. government A solution to these problems is the assumption of obligations enforcement as an effective means of guaranteeing that personal not just by importing countries, but also by exporting countries data from the European Union will be accorded privacy protec- when negative externalities are transmitted through exports of tion consistent with the data privacy principles agreed under services. Regional and multilateral negotiations are now struc- the Safe Harbor Framework. Under the Safe Harbor Framework, tured in a way that requires importing countries alone to make U.S. organizations can either join a self-regulatory privacy pro- binding commitments to market opening, regardless of the con- gram that adheres to the safe harbor principles or self-certify ditions in, or cooperative efforts by, source countries. Instead, (most common) to the U.S. Department of Commerce that they market access commitments by importing countries could be are complying with the principles. The U.S. Department of Com- made transparently and predictably conditional on the fulfillment merce reviews every self-­ certification and annual recertification of specific conditions by exporting countries. These exporter submission it receives from companies. The U.S. Federal Trade commitments need not be in the context of trade agreements Commission enforces the Safe Harbor Framework against com- but could be secured in other existing or new fora for inter- panies that self-certify as being in compliance.23 national regulatory cooperation. Then, regulators in import- The exporting country commitments embodied in the Safe ing countries would be reassured that exporting countries will Harbor Framework have played a crucial role in allowing data flows cooperate to protect their consumers’ privacy, financial security, between the European Union and the United States even though and well-being from the consequences of international market some concerns were expressed about its operation and effective- failures. ness. Some of these have been addressed in the recent Privacy Shield Agreement, but some shortcomings remain (Hufbauer and An example of exporting country regulatory commitments: Jung 2016). A major remaining problem that can affect the emer- data flows gence of GVCs is that the agreement applies only to EU–U.S. data International data flows provide an example of how such export- transfers and so is not useful for companies that want to transfer ing country commitments work. By allowing communication and data globally­—­that is, to establish a globally accessible database coordination of production across countries, such commitments or a global human resources information system. In sum, the Safe have probably been the most important reason for the emer- Harbor Framework is an example of remarkably effective, yet gence of GVCs. Governments are taking different approaches imperfect, dynamic regulatory cooperation. to regulating personal data collected by private enterprises. The European Union has the world’s most comprehensive legal data protection regime, the Data Privacy Directive adopted in 1995, Conclusions which it plans to develop further.20 The Data Privacy Directive makes it illegal to transfer personal This chapter has illustrated the role of services in GVCs, draw- data outside the European Union unless the European Commis- ing on selected evidence. For many purposes, services can be sion has found that the country receiving the personal data pro- treated analogously to goods in both the measurement and the vides adequate protection. In the absence of an adequacy deci- analysis of GVCs. And that is what existing trade in value added sion, data can be transferred to a third country under so-called databases and the literature that relies on them have tended to derogations, the main ones being consent of the data subject, do. Even though the share of services in trade in value added when the transfer is necessary for the performance of a contract varies across countries and industries, it is generally high (and between the data subject and the controller or is necessary on rising) and considerably larger than the share of services in gross important public interest grounds. The directive also allows for trade. While directly exported value added has increased in a cross-border transfer pursuant to a contract between the con- recent years, close to two-thirds of the growth of services value troller and the processor that guarantees the same protection of added in exports is due to an increase in services embodied in the personal data as under the directive. A global conglomer- exports of other sectors­—­ particularly foreign services, revealing ate can transfer data among its units where it has implemented the growing importance of GVCs. 156 • Measuring and Analyzing the Impact of GVCs on Economic Development The reasons for these developments are variants of the older Perhaps what really matters is not what a person makes but arguments for why the share of services in GDP tends to grow: what the person does. For a long time, notions of economic the splintering or outsourcing of services activities from manu- performance have been closely tied to economic sectors­ —­ facturing firms; the growing importance in a GVC world of con- manufacturing, agriculture, and services. In a world of frag- necting services such as telecommunications and transport; the mented production these distinctions are hard to sustain and growing services component in sophisticated manufacturing may not be economically meaningful. Instead, the focus could goods, such as software in cars; and the increase in the prices be on the implications of performing certain tasks. Do product of services tasks relative to manufacturing tasks because man- design and marketing offer greater scope for innovation and ufacturing tasks are easier to offshore to lower cost locations. learning-by-doing and thus for productivity growth than product However, there is little empirical evidence for these arguments, assembly? Such task-based analysis­ —­ perhaps initially focusing and understanding the reason for these developments should be on occupational structures­ —­ could be more help than the tradi- an area for future research. tional sector-based analysis in comprehending the implications For services GVCs there are good reasons to look beyond the for individuals and countries of the new international division of traditional arm’s length cross-border trade data, which ignore the labor. large share of international transactions in services that take place Finally, some policies both inhibit and encourage the emer- through commercial presence for foreign direct investment. Ini- gence of services-related GVCs. Even though most services tiatives to remedy the commercial presence gap are being taken markets are much more open today, thanks to unilateral liber- by the U.S. Bureau of Economic Analysis and the OECD. alization, services reforms remain incomplete, and barriers to Some evidence shows that the emergence of GVCs through domestic and foreign competition persist. Most of the policy foreign direct investment in services can affect downstream sec- barriers to competition and to foreign direct investment are tors. Improved access to finance, communications, transport, and not in goods but in services. For example, countries in South- other services, either through general reform or through reform east Asia that have reaped huge benefits from the liberalization of foreign direct investment, enhances manufacturing firms’ pro - of trade and investment in goods continue to maintain restric- ductivity and other aspects of the performance of downstream tions on foreign presence in services. Trade in transport services, firms. The development of domestic services sectors and access in particular, remains impeded in both industrial and develop- to foreign services can also shift the pattern of comparative ing countries by the exclusion of third-country providers and by advantage. Preliminary evidence suggests that trade in value quantitative restrictions in bilateral agreements. added data could help in understanding dynamic structural International cooperation in services has attempted to rep- change and deindustrialization­ —­areas that merit more analysis. licate the goods model of reciprocal market opening, but so Some have called for developing a notion of GVCs that goes far that approach has delivered little incremental liberalization. beyond arm’s length market-based transactions to functions Much more could be achieved through a greater emphasis on within the firm. It may be feasible to distinguish between tasks regulatory cooperation. First, and most obviously, greater regu- that would have been services if they had taken place at arm’s latory convergence­ —­as in prudential regulation-intensive finan- length (such as bookkeeping) and other tasks that are intrinsically cial, health, education, and professional services­ —­ is needed agricultural or manufacturing in nature (such as wood process- to create more-­ integrated markets in which competition, econ- ing). But such distinctions slide on a slippery slope: ultimately, omies of scale, and GVCs can develop. Second, credible regu- almost any task can potentially be conceived of as an arm’s length latory commitments by exporting countries to safeguard the service. A horticultural laborer can be hired as a worker in a hor- interests of consumers in importing countries­ —­as for deposit ticultural firm, or the laborer’s “fruit-picking” services can be pur- protection when capital flows internationally or privacy when chased from an individual or a firm. This kind of deconstruction of data flow internationally­ —­ could also induce greater liberaliza- a firm into its constituent tasks could reduce each firm to a bundle tion of explicit barriers to international transactions by providing of services regardless of what it ultimately produces. importing countries with the regulatory reassurance they need. Services trade and global value chains • 157 Notes and Sato (2014) showed that since the 1990s the share of value added in manufactured products in Asia has shifted decisively away from 1. https://globalvaluechains.org/concept-tools. manufacturing and toward services. 2. A definition that does not include single international transactions 12. The services linkage variable can be interpreted as a weighted aver- also creates a slight awkwardness in the treatment of transactions age across sectors of the access measure of interest, with the weights located at the beginning or at the end of linear value chains. Even indicating the sensitivity of sector j to input k (weight αj,k can, for if they do not qualify as part of GVCs, the hypothesis must be that example, be the share of k in total inputs of j). the history and the future matter. That is, a final import that involves Saharan Africa (Arnold, 13. Similar results have been found for Sub-­ border crossings at early stages of the value chain and an initial export Mattoo, and Narciso 2008) and Indonesia (Duggan, Rahardja, and that will cross other borders at later stages of the value chain should Varela 2013). have different implications from those that do not. 14. While services offshoring has both positive and negative effects on 3. For example, mode 3 trade exports represented on average 67% of domestic employment, Amiti and Wei (2009b) showed, at least for total U.S. exports of services between 2009 and 2014 according to the United States, that it tends to enhance domestic employment on data from the Bureau of Economic Analysis. average. 4. Mode 2 trade is captured in the balance of payments statistics cat- 15. Revealed comparative advantage based on gross exports (used as a egory “travel” but with limited disaggregation into sectors. Sales of dependent variable) can cause an endogeneity problem because the services by foreign natural persons too are also largely covered in embodied services (used as explanatory variables) are part of gross balance of payments statistics but are not identified separately from manufacturing exports. Liu and others (2017) avoided this problem cross-border trade. Data on mode 3 trade in services are not part of because manufacturing revealed comparative advantage is based on balance of payments statistics but are collected separately by some the value added created by the factors employed in manufacturing countries, such as the United States and the European Union. Efforts sectors, excluding the embodied services in gross exports contrib- to improve the measurement of mode 3 trade are discussed later in uted by the factors employed in services sectors. the chapter. 16. Two major initiatives to address this gap in information are in the 5. Restrictive policies of this sort are inherently difficult to identify and OECD and in a collaborative project between the World Bank and the measure. Two initiatives provide evidence on these restrictions: one in WTO. This section describes information on trade policies for serv- the OECD and another in a collaborative project between the World ices contained in the earlier World Bank Services Trade Restrictions Bank and the WTO. Some evidence regarding these measures is pre- Database. This database covers 103 countries, five major services sented later in the chapter. —­ sectors­ financial services, basic telecommunications, transport, dis- 6. This is not an exhaustive list of how services participate in GVCs. Other —­ tribution, and selected professional services­ and the relevant modes relevant aspects, such as bundling goods and services and services as of services delivery. This information, collected in 2008–10, has been creating activities, are discussed in detail in Miroudot (2016). value-­ subsequently updated only for some countries. Even though there is 7. See OECD (2013) for details on the algebraic definitions of these evidence of few major policy changes in the last few years, the data components. presented here are best seen as indicating broad patterns rather than 8. More generally, firm ownership, whether domestic or foreign, has the precise current situation in specific countries. been found to be a relevant dimension of firm heterogeneity, with for- 17. The analysis assesses policy regimes in their entirety and assigns them eign-owned firms often associated with greater exports, higher pro- to one of the following five principal categories: completely open (that intensive use of imported intermediates, and different ductivity, more-­ is, no restrictions at all), completely closed (that is, no entry allowed at patterns of value added (Fetzer and Strassner 2015). This in itself all), virtually open but with minor restrictions, virtually closed but with makes firm ownership a relevant dimension to understand participa- limited opportunities to enter and operate, and a residual “intermedi- tion in GVCs and the effects it may have in an economy. ate” category of regimes that allow entry and operations but impose 9. Examples include the 2010 National Organizations Survey in the restrictions that are neither trivial nor stringent. It is convenient to United States (Brown, Sturgeon, and Lane 2014) and two Eurostat sur- assign a value to each of these five categories of regimes on a scale of veys on the international sourcing of business functions by enterprises decreasing openness from 0 to 1 with intervals of 0.25. (Nielsen 2008). 18. There is some evidence of recent reform in services sectors. For 10. Statistics on business functions based on labor force surveys should example, a lifting of foreign equity caps, partly or fully, is observed be interpreted carefully since comparability across countries may be in Indonesia (air transport, architecture, engineering, telecommunica- affected by statistical conventions in the construction of input-output tions, distribution services, audiovisual services, and logistics), China tables. For example, data collected at the enterprise level and the (distribution services and maritime transport), and India (air transport, establishment level could yield different information for identical pro- insurance, and broadcasting). Mexico has opened the telecommuni- cesses on what is provided in-house and what is outsourced. cations sector and introduced procompetitive reforms. 11. For example, Lodefalk (2013) showed that services embedded in 19. The capture of regulatory barriers by established services providers Swedish manufactured goods account for a major and increasing may also partly explain the limited progress on this front. share of Sweden’s services exports. Similar evidence is presented for 20. As a directive, implementation of the Data Privacy Directive is left to other European countries in Boddin and Henze (2014), Crozet and EU member states, which vary widely in their enforcement. The Euro- Milet (2014), Kelle (2013), and Kelle and Kleinert (2010). Baldwin, Ito, pean Commission is seeking to update it as a regulation. 158 • Measuring and Analyzing the Impact of GVCs on Economic Development 21. According to a 1999 opinion from the Article 29 Working Party, the References U.S. approach was seen as not providing adequate protection in all cases for personal data transferred from the European Union. Amiti, M., and S.-J. Wei. 2009a. “Service Offshoring and Productivity: Evi- 22. The Safe Harbor Framework consists of seven principles that reflect dence from the US.” The World Economy 32 (2): 203–20. the key elements of the EU Data Protection Directive. The main ones ———. 2009b. “Does Service Offshoring Lead to Job Losses? Evidence are commitments to give European data subjects notice that a U.S. from the United States.” In International Trade in Services and Intan- entity is processing their data; to limit onward transfers of data to gibles in the Era of Globalization, edited by M. Reinsdorf and M. J. countries that also subscribe to the Safe Harbor principles or are sub- Slaughter, 227–43. Chicago: University of Chicago Press. ject of an adequacy finding; to take reasonable steps to protect per- Arnold, J., A. Mattoo, and G. Narciso. 2008. “Services Inputs and Firm Pro- sonal data from loss or misuse; to process personal data only for the Saharan Africa: Evidence from Firm-Level Data.”  Jour- ductivity in Sub-­ expressed purposes the organization intends to use it; to give Euro- nal of African Economies 17 (4): 578–99. pean data subjects access to their personal information and the ability Arnold, J., B. Javorcik, and A. Mattoo. 2011. “Does Services Liberalization to correct, amend, or delete inaccurate information; and to commit Benefit Manufacturing Firms? Evidence from the Czech Republic.” Jour- to enforce the principles and give European data subjects access to nal of International Economics 85 (1): 136–46. affordable enforcement mechanisms. Arnold, J., B. Javorcik, M. Lipscomb, and A. Mattoo. 2016. “Services 23. To date, the U.S. Federal Trade Commission has brought 10 Safe Reform and Manufacturing Performance: Evidence from India.”  The Harbor–related enforcement actions. The agency acts on referrals Economic Journal 126 (590): 1–39. from EU data protection authorities and from third-party private dis- Balassa, B. 1965. “Trade Liberalization and ‘Revealed’ Comparative Advan- pute resolution providers, as well as on its own. tage.” The Manchester School 33 (2): 99–123. Baldwin, R., and A. J. Venables. 2013. “Spiders and Snakes: Offshoring and Agglomeration in the Global Economy.” Journal of International Eco- nomics 90 (2): 245–54. Baldwin, R., R. Forslid, and T. Ito. 2015. “Unveiling the Evolving Sources of Value Added in Exports.” IDE–JETRO Joint Research Program Series No. 161. Baldwin, R., T. Ito, and H. Sato. 2014. “Portrait of Factory Asia: Production —­ Network in Asia and its Implication for Growth­ The ‘Smile Curve.’” Joint Research Program Series 159, Institute of Developing Economies– Japan External Trade Organization, Chiba, Japan. Bhagwati, J. N. 1984. “Splintering and Disembodiment of Services and Developing Nations.” The World Economy 7 (2): 133–44. Boddin, D., and P. Henze. 2014. “International Trade and the Servitization of Manufacturing: Evidence from German Micro Data.” Working Paper, European Trade Study Group, Münich, Germany. Borchert, I., B. Gootiiz, and A. Mattoo. 2014. “Policy Barriers to Interna- tional Trade in Services: Evidence from a New Database.” The World Bank Economic Review 28 (1): 162–88. Brown, C., T. Sturgeon, and J. Lane. 2014. “Using a Business Function Framework to Examine Outsourcing and Offshoring by US Organiza- tions.” Institute for Research on Labor and Employment (IRLE) Working Paper No. 121–14, IRLE, Berkeley, CA. Crozet, M., and E. Milet. 2014. “The Servitization of French Manufacturing Firms.” Centre d’Études Prospectives et d’Informations Internationales (CEPII) Working Paper 2014–10, CEPII, Paris. Duggan, V., S. Rahardja, and G. J. Varela. 2013. “Service Sector Reform and Manufacturing Productivity: Evidence from Indonesia.” Policy Research Working Paper No. 6349, World Bank, Washington, DC. Fernandes, A., and C. Paunov. 2012. “Foreign Direct Investment in Services and Manufacturing Productivity: Evidence for Chile.”  Journal of Devel- opment Economics 97 (2): 305–21. Fetzer, J., and E. Strassner. 2015. “Identifying Heterogeneity in the Pro- duction Components of Globally Engaged Business Enterprises in the United States.” U.S. Department of Commerce, Bureau of Economic Analysis, International Economic Accounts, Washington, DC. Services trade and global value chains • 159 Foster-McGregor, N., and R. Stehrer. 2013. “Value Added Content of Trade: Miroudot, S. 2016. “Services in Global Value Chains: From Inputs to Val- A Comprehensive Approach.” Economics Letters 120 (2): 354–57. ue-Creating Activities.” Organisation for Economic Co-operation and Francois, J., and J. Woerz. 2008. “Producer Services, Manufacturing Linkages, Development (OECD) Trade Policy Paper, OECD, Paris. and Trade.” Journal of Industry, Competition and Trade 8 (3–4): 199–229. Nielsen, P. B., ed. 2008. International Sourcing: Moving Business Functions Hoekman, B., and A. Mattoo. 2008. “Services Trade and Growth.” Policy Abroad. Copenhagen: Statistics Denmark. Research Working Paper No. 4461, World Bank, Washington, DC. Nordås, H. 2008. “The Impact of Services Trade Liberalisation on Trade in Hufbauer, G. C., and E. Jung. 2016. “The US-EU Privacy Shield Pact: A Non-Agricultural Products.” OECD Trade Policy Paper No. 81, OECD, Work in Progress.” Policy Brief 16–12, Peterson Institute for Interna- Paris. tional Economics, Washington, DC. OECD (Organisation for Economic Co-operation and Development). 2013. Hummels, D., J. Ishii, and K.-M. Yi. 2001. “The Nature and Growth of Ver- Interconnected Economies: Benefiting from Global Value Chains. Paris: tical Specialization in World Trade.” Journal of international Econom- OECD. ics 54 (1): 75–96. ———. 2015. “OECD Expert Group on Extended Supply-Use Tables. Draft: Johnson, R. C., and G. Noguera. 2016. “A Portrait of Trade in Value Added Terms of Reference.” over Four Decades.” National Bureau of Economic Research (NBER) OECD (Organisation for Economic Co-operation and Development) and Working Paper No. 22974, NBER, Cambridge, MA. Added (TiVA) WTO (World Trade Organization). 2015. Trade in Value-­ Jones, C. 2011. “Intermediate Goods and Weak Links in the Theory of Eco- database. Paris: OECD. Available at: http://www.oecd.org/sti/ind nomic Development.” American Economic Journal: Macroeconomics 3 /measuringtradeinvalue-addedanoecd-wtojointinitiative.htm. (2): 1–28. OECD (Organisation for Economic Co-operation and Development), WTO Jones, R, and H. Kierzkowski. 2001a. “A Framework for Fragmentation.” In (World Trade Organization), and World Bank Group. 2014. “Global Fragmentation: New production patterns in the world economy, edited Value Chains: Challenges, Opportunities, and Implications for Policy”. by S. W. Arndt and H. Kierzkowski, 17–34. Oxford, U.K.: Oxford Univer- Report prepared for submissions to the G20 Trade Ministers Meeting, sity Press. Sydney, Australia, 19 July 2014. ———. “Horizontal Aspects of Vertical Fragmentation.” 2001b. In Global Rajan, R., and L. Zingales. 1998. “Financial Dependence and Growth.” production and trade in East Asia, edited by L. K. Cheng and H. Kierz- American Economic Review 88 (3): 559–87. kowski, 33–51. New York: Springer. Stehrer, R., N. Foster, and G. de Vries. 2012. “Value Added and Factors Ju, J, and S.-J. Wei. 2010. “Domestic Institutions and the Bypass Effect of in Trade: A Comprehensive Approach.” World Input-Output Database Financial Globalization.” American Economic Journal: Economic Policy Working Paper No. 7, European Commission, Brussels. 2 (4): 173–204. Sturgeon, T., P. B. Nielsen, G. Linden, G. Gereffi, and C. Brown. 2013. Kelle, M. 2013. “Crossing Industry Borders: German Manufacturers as Ser- “Direct Measurement of Global Value Chains: Collecting Product- and vices Exporters.” The World Economy 36 (12): 1494–1515. Firm-Level Statistics on Value Added and Business Function Outsourcing Kelle, M., and J. Kleinert. 2010. “German Firms in Service Trade.” Applied and Offshoring.” In Trade in Value Added: Developing New Measures of Economics Quarterly 56 (1): 51–71. Cross-Border Trade, edited by A. Mattoo, Z. Wang, and S.-J. Wei, 291– Kochar, K., U. Kumar, R. Rajan, A. Subramanian, and I. Tokatlidis. 2006. 321. London: Centre for Economic Policy Research and the World Bank. “India’s Pattern of Development: What Happened, What Follows.” Timmer, M., B. Los, R. Stehrer, and G. de Vries. 2013. “Fragmentation, NBER Working Paper No. 12023, NBER, Cambridge, MA. Incomes and Jobs: An Analysis of European Competitiveness.” Eco- Koopman, R., W. Powers, Z. Wang, and S.-J. Wei. 2010. “Give Credit Where nomic Policy 28 (76): 613–61. Credit is Due: Tracing Value Added in Global Production Chains.” NBER Timmer, M., R. Stehrer, and G. de Vries. 2015. “Occupations in Global Working Paper No. 16426, NBER, Cambridge, MA. Value Chains: Patterns of International Specialisation.” OECD Trade Koopman, R., Z. Wang, and S.-J. Wei. 2014. “Tracing Value-Added Policy Paper, OECD, Paris. and Double Counting in Gross Exports.”  The American Economic Van der Marel, E., and S. Sáez. 2016. “Servicification, Regulation and Eco- Review 104 (2): 459–94. nomic Performance in GVCs.” World Bank Background Paper prepared Liu, X., A. Mattoo, Z. Wang, and S.-J. Wei. 2017. “Services Development for the conference, “Making Global Value Chains Work for Economic and Comparative Advantage in Manufacturing.” Development and Shared Prosperity,” Beijing, March 17–18. Lodefalk, M. 2013. “Servicification of Manufacturing – Evidence from Wang, Z., S.-J. Wei, and K. Zhu. 2013. “Quantifying International Produc- Sweden.”  International Journal of Economics and Business Research 6 tion Sharing at the Bilateral and Sector Levels.” NBER Working Paper (1): 87–113. No. 19677, NBER, Cambridge, MA. Los, B., M. Timmer, and G. de Vries. 2016. “Tracing Value-Added and ———. 2017. “Measures of Participation in Global Value Chains and Global Double Counting in Gross Exports: Comment.”  American Economic Business Cycles.” NBER Working Paper No. 23222, NBER, Cambridge, Review 106 (7): 1958–66. MA. CHAPTER 7 Institutional quality and participation in global value chains DAVID DOLLAR AND MATTHEW KIDDER O ne way to think about products that have com- domestic industries that use higher technologies and employ plex value chains is that they are contract-intensive higher skilled workers. The textbook Ricardian solution would be goods. That is, they often involve many exchanges for the government to sponsor investment in technologies that among different firms, each facing some risk of boost the productivity of desirable industries, thus giving these contract nonperformance by others in the chain. This chap- industries a Ricardian comparative advantage. The implication is ter reviews research on global value chains (GVCs) showing that the product in this industry could then be exported, thus that, other things equal, countries with better institutions such expanding the country’s GVC participation in this industry. But as stronger property rights and rule of law participate more in it is not clear that such simple plans for economic development GVCs.1 It investigates whether this finding holds up within coun- will prove fruitful without a careful consideration of economic tries. Using China as an example, it finds that Chinese cities that institutions. Institutions can resolve or worsen distortions arising score better on property rights and government efficiency are from asymmetric information, and classical models remain silent more likely to have firms involved in GVCs. It also applies the on this point. findings on institutions and GVCs to African economies, which To promote GVC participation in industries that drive eco- have only a small role in GVCs, finding that many African econo - nomic development, policymakers have to improve domestic mies have weak institutions or neighbors with weak institutions, institutions and take regional initiatives to improve the institu- an important reason for their small role in GVCs. tions of neighboring countries. This chapter thus studies how Ever since Ricardian trade theory emerged in 1817, econo- domestic institutional quality and the institutional quality in mists have considered that relative comparative advantage in neighboring countries influence the integration of domestic productivity should promote specialization within a country. At industries in GVCs. It looks at new empirical research on institu- the firm level these classical theories describe why firms might tions and GVC participation and on the effects of institutions in participate in GVCs. For example, a firm in China may have a neighboring countries. Finally, it examines Africa’s involvement in global comparative advantage in production processes that are GVCs to show what is holding back its participation. labor intensive but perhaps a comparative disadvantage in other stages of production. The firm should offshore the processes in which it has a comparative disadvantage and export the labor- Institutions and participation in global intensive portion of production. value chains But these classical frameworks are inadequate for craft- ing policies to deepen GVC participation because they do not The business services industry in China illustrates the challenges account for asymmetric information. Consider another exam- that institutions create in expanding global competitiveness in a ple. Policymakers may want to advance the competitiveness of sector. Chinese policymakers would like to promote the global 161 162 • Measuring and Analyzing the Impact of GVCs on Economic Development competitiveness of industries that are more skill intensive and chapter shows the importance of the U.S. legal system in driving higher on the value chain. The business services industry has the gap between Chinese and U.S. businesses in this sector. By both of these desired qualities. But it is unclear which policies providing strong protection for contract disputes, the U.S. legal will build the comparative advantage and deepen GVC participa- system contributes to the U.S. comparative advantage, which tion in business services. Classical trade models might suggest makes the United States a leading exporter of business services. that Chinese business services should already have a compara- The case of business services can be extended to other indus- tive advantage. The Chinese economy has an abundance of the tries. The importance of legal institutions increases as products primary production factors of business services: both a strong become more differentiated. In manufacturing the underlying information technology infrastructure and a large, educated logic is that there is a thinner market for differentiated goods work force. Despite these strengths, its business services sector than for commodities. Thin markets, with fewer potential buyers, is smaller than that in a developed country such as the United lead to what is called a “hold-up,” where there is an incentive to States. renegotiate terms after production has begun. For example, the Looking at the share of information, computing, and other completed engine of a Boeing jet may be more difficult to sell business services as a share of total value added for China and than a shipment of agricultural products if the buyer decides to the United States over 1985–2005 reveals a clear gap (figure 7.1). cancel the order after production is complete. This moral hazard This gap should puzzle policymakers who apply classical rea- problem leads to market inefficiencies if legal institutions cannot soning. Chinese business services should already have a com- enforce contracts. The legal system can reduce this problem if parative advantage, but the industry is still small and largely the court can transfer legal ownership of real property, thus forc- underdeveloped. ing the defaulting party to pay. This chapter sheds additional light on such puzzles by show- Another challenge is that contracts are not complete, in the ing that institutions are fundamentally important, both for com- sense that a contract cannot specify an agreement for every pos- parative advantage and for deepening GVC participation in sible contingency. Thus, in many conditions even the best con- industries that produce a more complex and customized prod - tract will not provide a remedy. A well-functioning legal system uct. The output of business services fits both categories. The can alleviate contract incompleteness if the system equitably protects rights. Both contract enforcement and equitable pro- tection of rights thus interact at an industry level with the con- FIGURE 7.1 Value added in business services as a tracting intensity of the industry. The more differentiated and percentage of GDP in China and the United States, contract-intensive the product, the more severe is the asymmet- 1985–2005 ric information problem in the absence of equitable protection Percent of 2005 GDP and credible enforcement. Looking at the 10 most and 10 least contract-intensive man- 15 United States ufacturing industries, as estimated by Nunn (2007), reveals that the most contract-intensive industries generally have a finished product that incorporates a higher level of technology and thus is higher in economic development (table 7.1). Given this distri- bution of institutional sensitivity across industries, it is not sur- 10 prising that institutions have a strong influence on economic growth and development (Robinson, Acemoglu, and Johnson 2005). Policymakers in resource-rich economies and developing economies naturally want to make contract-sensitive industries more globally competitive. In addition to legal systems, financial institutions are funda- 5 China mental to deepening global competitiveness and GVC partici- pation. And as with legal institutions, how important financial institutions are in fostering GVC participation differs across industries. Funding projects in contract-intensive industries can be constrained by asymmetric information, just as in the goods 0 market, partly because asymmetric information in the goods 1985 1990 1995 2000 2006 market causes uncertainty in the returns of the investors who Source: EU KLEMS database and the associated CIP database (http://euklems own capital in these industries. Indeed, property rights protec- .net/index.html, accessed 1 September 2016). tion is a more binding constraint on investment than external Note: The figure shows value added in KLEMS (K-capital, L-labor, E-energy, access to finance (Johnson, McMillan, and Woodruff 2002). In M-materials, and S-purchased services) industries 71–74 as a share of total addition, transparency and consistency in accounting methods value added in the economy. in estimating returns on investment can be more important for projects with complex transactions than for simple exchanges. Institutional quality and participation in global value chains • 163 TABLE 7.1 Ten most and ten least contract-intensive it can differentiate trade flows that cross borders multiple times manufacturing industries (complex GVC participation) from those that cross only once (simple GVC participation). Rank Most contract intensive Least contract intensive At the industry level, Dollar, Ge, and Yu (2016) found a pos- 1 Photographic equipment Poultry processing itive correlation between GVC participation and all measures manufacturing of institutional quality (see table A7.1.1 in annex 7.1). Country 2 Compressor manufacturing Flour milling measures of institutional quality, taken from the Worldwide Gov- 3 Analytical laboratory instrument Petroleum refining ernance Indicators data set of Kaufmann, Kraay, and Mastruzzi manufacturing (2010), include rule of law, government effectiveness, political 4 Engine equipment manufacturing Corn milling stability, regulatory quality, and absence of violence/terrorism. The key finding is that industries more sensitive to institutions 5 Electronic component Aluminum sheet manufacturing manufacturing have higher participation in complex GVCs in countries that have better institutions. This result appears to be robust against sev- 6 Packaging machinery Aluminum production manufacturing eral different statistical specifications and holds for all measures of institutional quality. The relationship is less robust for simple 7 Book publishers Fertilizer manufacturing flows that cross only one border.2 8 Breweries Rice milling There is no support at the industry level for rule of law or gov- 9 Musical instrument manufacturing Primary nonferrous metal ernment effectiveness significantly affecting simple GVC flows, but Dollar, Ge, and Yu did not rule this possibility out. The point 10 Aircraft engine part manufacturing Tobacco drying estimates from the statistical exercise are still positive for rule Source: Nunn 2007. of law and government effectiveness on simple GVC participa- tion, but the standard errors are large enough to make these estimates insignificant. In other words, both rule of law and gov- The role of informal institutions such as social networks is ernment effectiveness could be important even for simple GVC somewhat different from that of formal institutions. Informal participation, but the study is inconclusive on this point. Quali- institutions can benefit groups that are less likely to participate tatively, it is clear that the relationship between institutions and in GVCs. And they often arise where formal institutions cannot GVCs is statistically more established for complex GVC participa- resolve asymmetric information problems (Leff 1978; McMillan tion than for simple participation. and Woodruff 1999; Bigsten and others 2000; Rauch and Trin- The second fact that Dollar, Ge, and Yu (2016) discovered is dade 2002; and Guiso, Sapienza, and Zingales 2009). But they that complex GVC flows tend to be exported to countries that are not perfect substitutes. And they do not follow the rule that have worse institutions. They found that the effects of institu- the institution be equally shared by all agents in an economy. If tional quality on GVC development is completely opposite in informal institutions such as social networks favor a group that is upstream source countries than in downstream direct importing less likely to participate in GVCs, they could reduce GVC partici- countries. Direct importers with weak institutional quality show pation at the industry level. a faster growth in GVC production linkages with their upstream The expansion of GVCs across international borders encoun- suppliers. This may support Jones (2011), who found that the ters meaningful discontinuities for both countries and industries: availability of intermediate goods is positively related to eco- institutions vary across countries, while sensitivity to these insti- nomic development. Economic development is positively related tutions varies across industries. To understand how institutions to institutional quality. affect the pattern of GVC participation, Dollar, Ge, and Yu (2016) Recall the finding from Nunn (2007) that higher technology bring industry-  and firm-level evidence to this question. The industries that produce a more specialized product are more industry data trace interactions in the value chain through input- sensitive to institutional quality. Thus, poor institutions can output linkages, while the firm data help in understanding the constrict the domestic production of these type of intermedi- determinants of GVC participation. ate goods. With fewer domestic intermediate goods available, domestic firms in developing countries can have an incentive Industry evidence to find foreign intermediates. This gives a clear channel for Institutional quality at the country level is positively related to poor institutions to be positively related to foreign sourcing of participation in more complex GVCs. The distinction between upstream intermediates. But Dollar, Ge, and Yu (2016) did not complex and simple trade flows is made possible by new mea- explicitly identify this channel, so other channels may be driving sures of GVC participation developed by Wang and others (2016) faster GVC growth in importers with weak institutional quality. and Wang, Wei, and Zhu (2015), which build on Koopman, Wang, One channel could be processing trade, but again this channel and Wei (2014). This new industry–country measure spans 35 was not explicitly identified. industries and 41 countries from 1996 to 2011. The measure gives So, complex GVC participation in contract-intensive indus- a rich picture of GVC participation by decomposing trade flows tries is significantly influenced by the quality of domestic institu- based on value added. One benefit of this decomposition is that tions. Countries with weaker institutions deepen their upstream 164 • Measuring and Analyzing the Impact of GVCs on Economic Development GVC participation to countries with better institutions. And the that foreign buyers may have some sensitivity to contract growth of GVC participation is positively related to better insti- enforcement and may be more willing to buy from firms in areas tutions. These findings suggest that institutional quality is an with stronger rule of law. important determinant of an industry’s ability to fragment its Among firm characteristics ownership type directly influences production processes across international borders. GVC participation and interacts with institutional quality. The highest probability of GVC participation is in firms with foreign Firm-level evidence for China on domestic institutions ownership, followed by those with private and corporate owner- and firm and city characteristics ship. State ownership significantly lowers the probability of GVC Firm data allow modeling the mechanism that drives a firm’s participation. Further, how local institutions affect GVC partici- selection into different types of GVC categories. This analysis pation also depends on ownership type. The decision of state- can help policymakers create policies that increase participation owned enterprises to participate in GVCs is not significantly in GVCs. There is strong evidence that local institutions play a affected by local institutional quality. But there is strong evi- significant role in firms’ participation in GVCs. This suggests that dence that the decision of foreign firms to participate in GVCs is local governments have considerable scope to affect participa- sensitive to domestic institutions in China, showing that domes- tion in GVCs both by directly influencing other economic institu- tic institutional quality is more binding for foreign firms than for tions and by indirectly providing support by building appropriate state firms. One explanation is that state firms can have stronger infrastructure. informal institutions than foreign firms. For example, they may Dollar, Ge, and Yu (2016) evaluated the firm’s participation be able to lean on political connections to manage contract dis- choice using a cross-section of 11,709 firms in 120 Chinese cities putes in their favor, whereas foreign firms must rely on the legal that were surveyed in the World Bank Enterprise Survey. The system. Thus, informal institutions can crowd out GVC participa- dataset allows for measurement of the sourcing and exporting tion to the extent that informal institutions explain different par- behavior of firms. Thus, it is possible to tell whether a firm uses ticipation rates by ownership type. imports in production and whether a firm exports final products. At the city level there is evidence in China that lower transpor- The two categories create four subcategories of firms: tation costs, lower labor costs, higher economic development, • Those that use domestic inputs strictly for domestic and higher innovation are all positively correlated with higher firm consumption. GVC participation. Lower transportation costs are often achieved • Those that use imports for domestic consumption. through technology and infrastructure investments, which may • Those that use imports for foreign consumption (export be necessary, if not sufficient, to develop deeper GVC partici- production). pation. Even the best trade channel into an undeveloped region • Those that use domestic inputs for export production. may do little to promote GVC participation if other aspects of the These categories define types of participation in GVCs. Com- economic environment do not support such participation. How- plex participation, in this context, can be thought of as firms that ever, high transportation costs can certainly restrict GVC partici- import intermediate goods and export their products. pation, especially in resource-based economies where infrastruc- The data also have several measures of institutional quality ture still lags behind that in developed countries. and other firm and city characteristics that can be used to deter- The positive correlation of GVC participation with lower labor mine the effect of those characteristics on the probability that a costs should be weighed carefully against comparative advan- plant will participate in GVC production. The institutional quality tage in China. China has an abundance of labor and a compar- characteristics include contract enforcement, access to credit, ative advantage in industries intensive in low-skilled labor. So customs efficiency, and government intervention.3 These mea- lower wages should be correlated with higher GVC participation sures can be thought of as proxies for legal, financial, trade, and in China. But it cannot be concluded that lowering wages in a government institutions. Likewise, firm characteristics such as sector is generally helpful in promoting GVC participation or productivity, capital intensity, and size, as well as city characteris- comparative advantage. tics such as transportation infrastructure and economic develop- In sum: Less government intervention, higher customs effi- ment, can be evaluated to determine the most important predic- ciency, better contract enforcement, and more access to bank tors of GVC participation. loans significantly increase the probability that firms will partic- On the quality of local institutions in the Chinese cities stud- ipate in GVCs. ied, the main finding is that firms have a higher probability of participating in GVCs the more contracts are enforceable, the Institutions in neighboring countries less the government intervenes, the more efficient the customs That a country or locality’s institutions affect its participation in processes (see table A7.1.2 in annex 7.1), and the better access GVCs is an intuitive result. What may be more surprising is the firms have to credit. But there is some variation in the relation- evidence, summarized here, that neighboring countries’ institu- ship between institutions and the type of GVC participation. For tions matter for GVC participation. In particular, neighboring-­ example, better contract enforcement increases the probability country institutions may have more impact on the efficiency of that a firm will participate in exporting, but the effect is not sig- business-to-business linkages in industries that tend to be more nificant for plants that import but do not export. This suggests intensive in contracts. Institutional quality and participation in global value chains • 165 The more differentiated or tailored a good, the thinner is the quality in a country and in that country’s neighbors. Most coun- market for the good and the more severe is the hold-up prob- tries are near the diagonal line, but there is variation, with some lem.4 In other words, highly differentiated goods are more sen- countries below the line. For example, Hong Kong, China; Sin- sitive to the institutional environment. In manufacturing the 747 gapore; Norway; Finland; Israel; and Chile have neighbors with passenger aircraft, Boeing uses reaction engines, which are not weaker rule of law than their own. This could be a weak link for available in open markets and which do not have reference prices value chains when some parts of a productive process can be established by its manufacturers (firms like Rolls Royce and Gen- outsourced to nearby locations. eral Electric). The price, quantity, and especially the characteris- Using local and neighbors’ institutions, Miranda and Wagner tics are determined in a usually incomplete negotiation between (2015) calculated the average revealed comparative advantage Boeing and the engine supplier because the engine is a differen- by country and industry, separating high contract-intensive tiated good adapted to the model of aircraft. In the language of industries (high share of differentiated inputs) and low contract-­ Williamson (1975), there is fundamental transformation that cre - intensive industries, and countries having neighbors with high ates a specific relationship. Thus, the quality of domestic institu- rule of law (as a measure of contract institutional quality) and tions is a direct channel that can affect these contract-intensive those having neighbors with low rule of law. Having neighbors industries more than less contract-intensive industries, such as with weak contract enforcement reduces exports in contract-in- undifferentiated agricultural products. tensive industries (figure 7.3). On top of the direct channel, Miranda and Wagner (2015) Miranda and Wagner (2015) also found that local institutions show that neighbors’ institutions could also matter, over and explain more or less the same amount of variation as the sum of above the effect of own-country institutions. They followed the physical and human capital (see table A7.1.3 in annex 7.1). They original work by Nunn (2007) but also included the role of neigh- also explored what makes neighbors’ institutions more relevant. boring country institutions, previously missing from the analysis. Countries that share a common language and common colo- Own and neighbors’ institutions are related, but they can have nial history would be expected to do more business together meaningful differences (figure 7.2). For Chile the neighboring in contract-­intensive sectors than countries that do not share rule of law measure is the average of rule of law of Argentina, those links. The more similar the countries, the more scope Bolivia, and Peru, weighted by their respective GDPs. The diag- for “nearsourcing”­ —­outsourcing tasks to nearby countries or onal line in figure 7.2 indicates equality between the institutional regions. But precisely in that context, having neighbors with FIGURE 7.2 Relationship of own and neighboring countries’ judicial quality (rule of law) Neighbors’ judicial quality 1.0 0.8 Singapore 0.6 Israel Finland Chile Hong Kong, Norway 0.4 China 0.2 0.0 0.0 0.2 0.4 0.6 0.8 1.0 Local judicial quality Source: Miranda and Wagner 2015 on local and neighboring countries. The rule of law index is from the Worldwide Governance Indicators 1998, using Nunn 2007 data and GeoDist data from the Institute for Research on the International Economy. Note: Countries in the figure are those with a common land border with their neighbors. The diagonal line indicates equality between the institutional quality in a country and in that country’s neighbors. 166 • Measuring and Analyzing the Impact of GVCs on Economic Development FIGURE 7.3 Average revealed comparative advantage In sum, there seems to be a systematic relationship between across industries and countries, controlling for local what a country produces and the ability of its neighbors to institutional effects enforce contracts. The analysis here focuses mostly on the lack Revealed comparative advantage of contract enforcement upstream, since it uses sectors that have contract-intensive procurement and so they are sensitive to 1.5 Low contract intensity suppliers with poor contract enforcement. Some tests show that High contract intensity downstream contract enforcement could also be important, as in —­ sectors with more contract-intensive output­ say, because of the need to customize the product before selling it, which requires certainty. 1.0 African involvement in global value chains African economies have had little involvement in GVCs (Dollar 2016). One useful measure of position in the value chain is the 0.5 share of imported value added in a country’s exports, a reflec- tion of economies’ integration with each other and with the global economy (figure 7.4). For advanced economies one-third of exports were attributed to imported inputs in 2008–12, up from one-quarter in 1991–95. For low-income economies and 0.0 emerging market economies other than in Sub-­ Saharan Africa, Neighbors with Neighbors with strong institutions weak institutions the average was 21–22% in 2008–12, up from 17–18% in 1991–95. Among developing economies Poland and Viet Nam are stand- Source: Miranda and Wagner 2015. outs, with imported inputs accounting for more than one-third of their export value. Saharan African economies fall below About two-thirds of Sub-­ weak institutions appears even more binding for business, since the average value-chain position for developing countries based firms would otherwise be connecting with each other much more on the value of their exports derived from imports (see figure frequently. 7.4). Oil exporters such as Angola, Chad, Nigeria, and South Saharan African and comparator countries, 2008–12 FIGURE 7.4 Depth of integration in global value chains of Sub-­ Percent of foreign value added in exports 50 OIl exporters Nonoil resource-intensive countries Rest of Sub-Saharan Africa 40 Comparator countries 1991–95 average 30 20 10 0 Pr nd Se so e yc tho Ca au les V us h e ts ia am a Bu ibia i nz a Th Li ia Si Ga eria Le ia To e Bu K go Zi a F a So b so Af e fri Ni a M an ger as . Er ar Za rea M bia B i D ga n .R a G ep. ne a l bi li Co G ue Ca , R a d’ on G ire N on ut Ch a Su d Ba An an la la Ch h In a Po dia Em pin Vie and va rg co am ec ar s om ts s te ero . Rw nd aw ga ag ep Cô m ep ed m trie ie am Ma N an Ta and in y ric em nd Se ine o n i in Le cip Et erd on ut abw o, U eni s h a Bo iop an ra b er ng go on ke c bo riti de rk en m a ng ha d q o ab d ila M el er m m Ad e g t N ru it ad R al w e b ig l Iv u ín nc ing un h an waz h c S oz So lA M é ng ra lo m nt Co ve To Ce de o Sã e m co -in w Lo Source: IMF 2015. Institutional quality and participation in global value chains • 167 Sudan have almost no imported value added in their exports and Indicators project of the World Bank publishes a Rule of Law fall on the far right side of the figure. To some extent these econ- Index that “captures perceptions of the extent to which agents omies are subject to “Dutch disease,” with resource exporters have confidence in and abide by the rules of society, and in par- tending to have high wages and appreciated exchange rates that ticular the quality of contract enforcement, property rights, the make it difficult for them to diversify their exports. But this is only police, and the courts, as well as the likelihood of crime and vio- a partial explanation. These countries are all relatively poor, and lence” (Kaufmann, Kraay, and Mastruzzi 2010, p.  4). All but six their oil production is not sufficient to make their citizens wealthy. African countries are below average for the world (figure 7.5), On the left of the figure are countries with deeper GVC inte - not too surprising because there is a clear relationship between gration, but they tend to be resource-poor economies with small GDP per capita and the quality of economic institutions, and populations (Cabo Verde, Lesotho, Mauritius, São Tomé and African countries are relatively poor and, on average, have weak Principe, Seychelles, and Swaziland). Countries with advanced institutions. tourism industries that rely on high-value-added imported inputs But there is a lot of dispersion across African countries. will also show up as having deep involvement in GVCs. Ethiopia The regression line in figure 7.5 shows the typical relationship is an interesting case of a populous yet resource-poor country between per capita income and the rule of law. Countries above with a high degree of GVC integration, which has grown substan- the line have unexpectedly good institutions for their level of tially since 1995. development; countries below the line have unexpectedly poor What accounts for Africa’s low involvement in GVCs? One institutions. This is important because countries generally com- factor, as mentioned, is Dutch disease. But many countries in pete with other countries at similar levels of development. If one Africa are not resource rich, and yet they still have low involve- country has good institutions among its cohort, it can expect to ment in GVCs. And even where Dutch disease is an explanation, attract more investment and entrepreneurship and grow faster. it should not prevent the development of modern manufacturing In fact, having a robust rule of law relative to level of develop- and services sectors. A key issue in most African economies is ment is closely correlated with faster growth. Among the large deficient infrastructure: unreliable power, poor roads and high- countries in Africa with broadly similar GDP per capita, Ethiopia, ways, and inefficient ports. Tanzania, and Uganda are well above the regression line, Kenya Another issue is economic governance. As discussed, well-­ sits right on the line, and Sudan and Nigeria fall well below the developed economic institutions, such as property rights and line. Thus, while the countries are at similar levels of develop- the rule of law, have significant positive effects on develop - ment, Ethiopia, Tanzania, and Uganda (especially the last two) ment and on participation in GVCs. The Worldwide Governance have better economic institutions. FIGURE 7.5 Institutional quality relative to development level in African countries: Association of Rule of Law Index and GDP per capita, 2010 Rule of law index (mean = 0) 2 1 Egypt South Africa 0 Uganda Morocco Tanzania Ethiopia –1 Kenya Congo, Dem. Rep. Sudan Nigeria –2 5 6 7 8 9 10 11 12 GDP per capita (log) Source: World Bank Worldwide Governance Indicators (www.govindicators.org); Feenstra and others 2015. Note: The figure plots the Rule of Law Index for 146 countries in 2010 against per capita GDP measured in purchasing power parity (log scale). By design, the index has a mean of zero and a standard deviation across countries of 1.0. Labeled countries are the 10 most populous African countries. 168 • Measuring and Analyzing the Impact of GVCs on Economic Development The International Monetary Fund’s Africa Economic Outlook Saharan FIGURE 7.6 Potential increase in trade in Sub-­ for 2015 includes an estimate of the effect of improvements in the Africa from improvements in investment climate investment climate on African exports. The thought experiment Percent adjusts different indicators from the average for Sub-­ Saharan Africa to the average for the rest of the world. The investment cli- 50 mate indicators are an index of infrastructure, credit to the private sector (a measure of financial sector depth and efficiency), a rule of law index, and the level of import tariffs (indicating how open 40 or closed an economy is to world trade). The biggest potential gain­—­ a 42% increase in exports­ —­ comes from improving infra- structure (figure 7.6). Credit to the private sector and rule of law 30 are also important, accounting for potential increases in exports of 29% and 28%. African economies already have relatively low trade barriers, so reducing import tariffs to the average for the 20 rest of the world boosts exports only 14%. The specific estimates in this kind of empirical exercise should be taken with caution. But the general point is valid: Africa could expand its involvement in 10 global trade, including GVCs, through improvements in its invest- ment climate, including infrastructure development, stronger financial sectors, and improved property rights and rule of law. Of particular relevance for Africa are neighborhood effects, 0 Infrastructure Domestic Rule Tariffs or the problem of reforming countries that have some or all credit of law neighbors with poor institutions. Some African economies have improved their economic institutions, and they tend to be the Source: IMF 2015. ones with faster growth and some initial involvement in GVCs. Saharan Africa’s trade if the variable moves from Note: Percent increase in Sub-­ But in many cases, reformers are bordered by neighbors with Saharan Africa to the average for the rest of the world. the average for Sub-­ poor institutions (map 7.1). There is a pocket of reform in East- ern Africa, including Ethiopia, Kenya, Rwanda, Tanzania, and Uganda. But these countries are neighbored by Central African Republic, Democratic Republic of Congo, Eritrea, Somalia, and MAP 7.1 Rule of law across Africa Sudan­ —­ all with poor institutions. In Western Africa, Ghana and Senegal are relative bright spots, but Guinea-Bissau and Nige- ria drag down the neighborhood. The Southern part of Africa around South Africa is another bright spot. Policy implications The quality of institutions affects comparative advantage and thus participation in GVCs. The stages of the production process differ in the extent to which they use simple labor, skilled labor, and capital (factor intensity). Activities also differ in their contract intensity. Producing a complete, homogeneous product with no imported content has simple or no contract intensity; growing and exporting bananas is an example. At the other extreme is producing a specialized part for a sophisticated electronics product. If done at arm’s length, the purchasing firm has to have considerable confidence in the contract. Developing countries in general have a lot of simple labor relative to skilled labor and capital. Given those factor endow- Below the mean Above the mean ments, which tend to change only slowly over time, the quality of institutions determines whether countries export simple, undif- Source: Worldwide Governance Indicators 2014. ferentiated products or whether they can embed their produc- Note: Countries are color-coded to reflect how strong or weak their institu - tion in more sophisticated value chains. Being involved in value tions are on an index of property rights and rule of law, rescaled to have a chains, in turn, will accelerate technological upgrading, skill mean of zero and a standard deviation of 1.0. Institutional quality and participation in global value chains • 169 development, and capital accumulation. The quality of institu- shortcuts that can enhance GVC participation. As seen in ear- tions relative to development level is crucial. lier chapters, deep trade agreements enhance GVC participa- A general policy recommendation from this analysis is that tion, probably because those agreements target specific insti- developing countries need to improve their institutions­—­ provide tutional bottlenecks­ —­such as improving customs administration equitable protection of rights, increase the enforceability of con- and strengthening property rights and legal recourse. Deep tracts, require more transparency, adopt anticorruption mea- agreements are going to be most powerful if several countries sures, make customs processes efficient, and encourage financial in a region all participate, improving neighbors’ institutions. The deepening. The focus should be on reducing transaction costs Trans-Pacific Partnership had the potential to play this role in the so that a country’s firms can easily join GVCs. Asia–Pacific region, and the U.S. abandonment of the agreement Improving institutions across the board is a big challenge, is a setback. So far China has not shown much interest in deep of course, and takes time. So, it is worth considering some agreements, but that may change. 170 • Measuring and Analyzing the Impact of GVCs on Economic Development ANNEX 7.1 Results for Dollar, Ge, and Yu (2016) and Miranda and Wagner (2015) TABLE A7.1.1 Summary of Dollar, Ge, and Yu (2016) industry results Variable (1) (2) (3) (4) 0.040*** Insdepict × Regulatoryct (0.010) 0.042*** Insdepict × Stabilityct (0.008) 0.034*** Insdepict × Govct (0.010) 0.026*** Insdepict × Lawct (0.009) 0.0023*** 0.0023*** 0.0026*** 0.0025*** K-ratioict × ln(K-endowct) (0.0001) (0.0001) (0.0001) (0.0001) 0.066*** 0.066*** 0.065*** 0.065*** Skill-ratioict × ln(Skill-endowct) (0.012) (0.012) (0.012) R -squared 0.716 0.716 0.716 0.716 Fixed effects (country and industry and year) Yes Yes Yes Yes Number of observations 8,995 8,995 8,995 8,995  *Significant at 10% level; **significant at 5% level; significant at 1% level. Note: Dependent variable is GVCPict (GVC participation index in industry i by country c at time t ). Coefficients are reported. Numbers in brackets are robust standard errors. The estimated equation is GVCPict = β 0 + β1insdepictinsct + β3kratio × kendow + β2Skill ratioict × Skill endowct + αi + γc + δt + εict, where insdepict is ict ct a measure of institutional sensitivity and insct is one of 4 measures of country level institutional quality (regulatory, stability, government effectiveness, and rule of level measures of industrial intensities. The coefficient of interest law). Capital endowments and skill endowments are also interacted with industry- and country-­ is β1, which is reported in the first four rows of this table. More details can be found in Dollar, Ge, and Yu 2016. Institutional quality and participation in global value chains • 171 TABLE A7.1.2 Summary of Dollar, Ge, and Yu (2016) firm results (1) (2) (3) (4) (5) (6) (7) (8) (9) (10) (11) (12) Variable D2E I2P I2E D2E I2P I2E D2E I2P I2E D2E I2P I2E Government intervention 0.437*** 1.359*** 1.638*** Custom efficiency 0.428*** 1.318*** 1.563*** Contract enforcement 0.406*** 1.347*** 1.622*** Access to finance 0.487*** 1.360*** 1.655*** Ownership structure State-owned enterprises –0.499*** –0.209 –0.572*** –0.51*** –0.210 –0.570*** –0.490*** –0.215 –0.59*** –0.424*** –0.213 –0.585*** Collective-owned enterprises –0.769*** –0.714*** –0.928*** –0.78*** –0.732*** –0.973*** –0.775*** –0.72*** –0.93*** –0.756*** –0.710*** –0.921*** Private 0.042 –0.068 –0.097 0.050 –0.065 –0.085 0.0190 –0.072 –0.105 0.002 –0.062 –0.084 Hong Kong, China; Macao, China; and Chinese Taipei 0.437*** 1.359*** 1.638*** 0.428*** 1.318*** 1.563*** 0.406*** 1.347*** 1.622*** 0.487*** 1.360*** 1.655*** Foreign 1.046*** 1.621*** 2.322*** 1.030*** 1.599*** 2.281*** 1.026*** 1.605*** 2.296*** 1.070*** 1.611*** 2.310*** Firm characteristics Size 0.431*** 0.434*** 0.814*** 0.424*** 0.426*** 0.801*** 0.424*** 0.426*** 0.803*** 0.407*** 0.423*** 0.797*** Age 0.011 –0.004 0.021 0.010 –0.007 0.008 0.008 –0.006 0.012 –0.0003 –0.007 0.011 Capital –0.054** 0.334*** 0.194*** –0.053** 0.336*** 0.197*** –0.058** 0.332*** 0.190*** –0.061** 0.334*** 0.192*** City characteristics GDP per capita –0.033 0.628*** 0.448*** –0.080 0.554*** 0.303*** –0.090 0.574*** 0.350*** –0.028 0.587*** 0.380*** City wage 0.348* –0.420** 0.088 0.309* –0.471** 0.037 0.540*** –0.339 0.277* 0.126 –0.444** 0.049 Research and development share 1.202*** 1.176*** –0.166 1.188*** 1.193*** –0.052 0.852*** 1.045*** –0.512* 0.158 1.069*** –0.415 Transport cost –0.253*** –0.138*** –0.363*** –0.26*** –0.106** –0.294*** –0.237*** –0.15*** –0.37*** –0.269*** –0.163*** –0.405*** *Significant at 10% level; **significant at 5% level; significant at 1% level.  Note: See Dollar, Ge, and Yu 2016 for technical notes. Standard errors are omitted. 172 • Measuring and Analyzing the Impact of GVCs on Economic Development TABLE A7.1.3 Summary of Miranda and Wagner (2015) main Log of exports from country in industry Variable (1) (2) (3) (4) (5) Neighbor’s judicial quality × 0.159*** 0.206*** 0.252*** 0.140*** 0.244*** contract intensity (ziQN c ) (0.494) (0.621) (0.675) (0.559) (0.708 Local judicial quality × 0.200*** 0.212*** 0.220*** 0.161*** 0.196*** contract intensity (ziQc ) (0.395) (0.558) (0.605) (0.469) (0.612) Other determinants of comparative advantage No No No Yes Yes Skill and capital interaction No No Yes No Yes Constant Yes Yes Yes Yes Yes Fixed effects (country and industry) Yes Yes Yes Yes Yes Number of observations 18,383 8,148 8,148 12,934 7,988 R2 0.736 0.770 0.772 0.776 0.774 *Significant at 10% level; **significant at 5% level; significant at 1% level.  Note: Dependent variable is ln xic (natural log of exports in industry i by country c to all other countries). Standardized beta coefficients are reported. Numbers in brackets are robust standard errors. The estimated equation is ln xic = ac + ai + βziQc + βNziQN c   + γXci + γNX N ci + εci, with Xci as a vector that includes another determinants of comparative advantage and skill and capital interaction for the local country and neighbors (with superscript N ). All variables except fixed ef- fects are interactions between at least one industry-level variable and at least one country-level variable. All neighboring variables (with superscript N ) consist of interactions with country-level variables but refer to neighboring countries of country c, measured as a weighted average by neighbor’s GDP. More details can be found in Miranda and Wagner 2015. Institutional quality and participation in global value chains • 173 Notes Jäger, K. 2016. “EU KLEMS Growth and Productivity Accounts 2016 —­ release­ Description of Methodology and General Notes.” The 1. This chapter summarizes the research findings of two background Conference Board Europe. Available at: http://www.euklems papers. A paper by Dollar, Ge, and Yu (2016) evaluates the impact of .net/TCB/2016/Metholology_EU%20KLEMS_2016.pdf. domestic institutions on global value chain (GVC) participation and is Johnson, S., J. McMillan, and C. Woodruff. 2002. “Property Rights and sponsored by the Research Institute for Global Value Chains at the Uni- Finance.” American Economic Review 22 (6): 1335–56. versity of International Business and Economics in Beijing. The second Jones, C. 2011. “Intermediate Goods and Weak Links in the Theory of Eco- background paper by Miranda and Wagner (2015) examines the impact nomic Development.” American Economic Journal 3 (2): 1–28. of foreign institutions in neighboring countries on domestic GVC partici- Kaufmann, D., A. Kraay, and M. Mastruzzi. 2010. “The Worldwide Gover- pation and is sponsored by the Inter-American Development Bank. nance Indicators: Methodology and Analytical Issues.” Policy Research 2. The estimates of this regression are reported in Dollar, Ge, and Yu Working Paper No. 5430, World Bank, Washington, DC. (2016). Koopman, R., Z. Wang, and S.-J. Wei. 2014. “Tracing Value-Added and 3. Customs efficiency is a measure of the time needed for goods to clear Double Counting in Gross Exports.” The American Economic Review customs. 104 (2): 459–94. 4. A measure from Nunn (2007) is used that captures whether an indus- Leff, N. 1978. “Industrial Organization and Entrepreneurship in the Devel- trial sector is especially sensitive to contracts. Nunn (2007) focuses oping Countries: The Economic Groups.” Economic Development and on the share of an industry’s inputs that are differentiated. This is Cultural Change 26 (4): 661–75. implemented using the input-output matrix of a sector and identi- Mayer, T., and S. Zignago. 2011. “Notes on CEPII’s Distances Measures: fying which sectors tend to have more inputs that are traded on a The GeoDist Database.” Centre d’Études Prospectives et d’Informa- business relation as opposed to an input that can bilateral business-to-­ tions Internationales (CEPII) Working Paper 2011–25, CEPII, Paris. be bought in an arm’s length transaction in a formal exchange. The McMillan, J., and C. Woodruff. 1999. “Interfirm Relationships and Informal distinction among different goods comes from the classification by Credit in Vietnam.” Quarterly Journal of Economics 114 (4): 1285–320. Rauch (1999) according to whether inputs are traded in open markets Miranda, R., and R. Wagner. 2015. “Neighboring Institutions Matter for the with referenced prices or not. The case without is interpreted as a dif- Competitiveness of Your Value Chain.” Global Value Chain Develop- ferentiated good. ment Report Background Paper, World Bank, Washington, DC. Nunn, N. 2007. “Relationship-Specificity, Incomplete Contracts, and the Pattern of Trade.” Quarterly Journal of Economics 122 (2): 569–600. References Rauch, J. 1999. “Networks versus Markets in International Trade.” Journal of International Economics, 48 (1): 7–35. Bigsten, A., P. Collier, S. Dercon, M. Fafchamps, B. Gauthier, J. W. Gun- Rauch, J., and V. Trindade. 2002. “Ethnic Chinese Networks in International ning, A. Oduro, R. Oostendorp, C. Patillo, M. Soderbom, F. Teal, and A. Trade.” Review of Economics and Statistics 84 (1): 116–30. Zeufack. 2000. “Contract Flexibility and Dispute Resolution in African Robinson, J., D. Acemoglu, and S. Johnson. 2005. “Institutions as a Fun- Manufacturing.” Journal of Development Studies 36 (4): 1–17. damental Cause of Long-Run Growth.” Handbook of Economic Growth Dollar, D. 2016. China’s Engagement with Africa: From Natural Resources 1A: 386–472. to Human Resources. John L. Thornton China Center Monographs. Wang, Z., S.-J. Wei, and K. Zhu. 2015. “Quantifying International Produc- Washington, DC: Brookings Institution. tion Sharing at the Bilateral and Sector Levels.” NBER Working Paper Dollar, D., Y. Ge, and X. Yu. 2016. “Institutions and Participation in Global No. 19677, NBER, Cambridge, MA. Value Chains.” Global Value Chain Development Report Background Wang, Z., S.-J. Wei, X. Yu, and K. Zhu. 2016. “Characterizing Global Value Paper, World Bank, Washington, DC. Chains.” Working Paper No. 578, Stanford Center for International Feenstra, R. C., R. Inklaar and M. P. Timmer. 2015. “The Next Generation Development, Stanford, CA. of the Penn World Table.” American Economic Review, 105(10), 3150– Williamson, O. E. 1975. Markets and Hierarchies: Analysis and Antitrust 3182. Available at: www.ggdc.net/pwt. Implications. New York: Free Press. Guiso, L., P. Sapienza, and L. Zingales. 2009. “Cultural Biases in Economic World Bank. 1998. Worldwide Governance Indicators (database). Wash- Exchange?” Quarterly Journal of Economics 124 (3): 1095–131. ington, DC: World Bank. Available at: http://info.worldbank.org IMF (International Monetary Fund). 2015. Regional Economic Outlook: Sub-­ /governance/wgi/index.aspx#home. Saharan Africa, Navigating Headwinds. World Economic and Financial ———. 2014. Worldwide Governance Indicators (database). Washington, Surveys. Washington, DC: IMF. DC: World Bank. Available at: http://info.worldbank.org/governance /wgi/index.aspx#home. CHAPTER 8 Preferential trade agreements and global value chains: Theory, evidence, and open questions MICHELE RUTA T wo phenomena have characterized the trade and • What is the outlook of the relationship between preferential trade policy landscape since the early 1990s. The trade agreements and GVCs going forward? rise of global value chains (GVCs)­—­the denationaliza- This chapter contributes to the large literature on preferen- tion of production­—­ has changed international trade, tial trade agreements (such as Limao 2016) in several ways. First, with trade in parts and components increasing almost six times based on new World Bank data, it documents how preferential between 1990 and 2015, faster than the 4.5 times for other forms trade agreements have deepened over time and how this evo- of trade. On the policy side, preferential trade agreements are lution is associated with the rise of GVCs. Second, it reviews the increasing in number and deepening in content.1 Their number theoretical literature on the rationale for the relationship between surged from 50 in 1990 to close to 280 in 2015. These agree - preferential trade agreements and GVCs and outlines avenues for ments are also deepening, in the sense that they cover an future research. Third, it discusses empirical research suggest- expanding set of policy areas, such as investment and competi- ing that deep agreements boost GVC integration and showing tion policy, that go well beyond the traditional focus of preferen- how this impact differs across country groups. Finally, it presents tial trade agreements, such as tariffs. a simple framework for thinking about the relationship between This chapter analyzes the relationship between preferential preferential trade agreements and GVCs going forward. trade agreements, particularly “deep” preferential trade agree- While more work is needed, several findings emerge from this ments, and GVCs. The goal is to answer six policy-­ relevant review. New data on the content of trade agreements and on questions: participation in GVCs point to a strong positive correlation, with • How have preferential trade agreements evolved? deeper agreements associated with more intense GVC relation- • In a world with GVCs, why do countries sign preferential trade ships. Economic theory identifies several explanations for this agreements? relationship, ranging from the need to internalize cross-border • Do preferential trade agreements increase GVC integration? policy spillovers to the benefits of stronger commitments in poli- • How does the content of preferential trade agreements affect cies that affect GVC participation. Econometric analysis confirms GVC trade? that deep preferential trade agreements boost participation in • How do GVCs affect the choice of preferential trade agree- GVCs, suggesting that trade agreements can be an effective ment partners? tool for policymakers to anchor national producers to global and The author thanks his coauthors of the papers on which this chapter is in part based: Claudia Hofmann, Alen Mulabdic, Alberto Osnago, and Nadia Rocha. Aaditya Mattoo, Alen Mulabdic, Zhiguo Xiao, and seminar He is also grateful to Zhi Wang for sharing his data and to Andrew Crosby, Michael Ferrantino, ­ participants at the two preparatory conferences in Beijing and Washington, DC, for comments. 175 176 • Measuring and Analyzing the Impact of GVCs on Economic Development regional production processes. Going forward, the future of the The database documents the changing content of preferential relationship between preferential trade agreements and GVCs trade agreements. A growing number of trade agreements cover will depend on continuing trust in the willingness of other part- more than 20 policy areas, a majority of newly signed preferential ners to preserve an open trading system. trade agreements cover 10–20 policy areas, and a minority focus on fewer than 10 policy areas (figure 8.1). The new database also allows looking in detail at the content Evolution of preferential trade agreements of trade agreements. In addition to tariff reductions, more than half the preferential trade agreements in the database include New evidence on the evolution of preferential trade agree- legally enforceable regulations on some policy areas that fall ments offers a basis for discussing the relationship between under the current mandate of the WTO (figure 8.2). These pro - trade agreements and GVCs. The number of preferential trade visions, referred to as “WTO-plus” or “WTO+” in the literature, agreements has increased dramatically in the last quarter cen- include customs regulations, export taxes, antidumping mea- tury, from 50 trade agreements in force and notified to the World sures, countervailing duty measures, technical barriers to trade, Trade Organization (WTO) in 1990 to 279 at the end of 2015.2 and sanitary and phytosanitary standards. Provisions outside the This dramatic change has spurred debate among researchers3 WTO mandate (usually called “WTO-extra” or “WTO-X”) include and policymakers on the rationale for preferential arrange- a wide-ranging set of policy areas, from investment to environ- ments; their impact on the trade flows, growth, and welfare of mental laws and nuclear safety. The inclusion of these provisions member and nonmember countries; and their relationship with in preferential trade agreements and their legal enforceability the broader system of global trade governance. varies widely by policy area (figure 8.3). Often overlooked in the literature on trade agreements is that Preferential trade agreement provisions can also be disag- their content­—­as well as their number­ —­ has changed over time. gregated in different ways depending on the question under Before the 1990s, trade arrangements involved mostly tariff reduc- investigation. Following Hofmann, Onsnago, and Ruta (2017), tions, but more recent preferential trade agreements include other preferential trade agreement provisions are divided here into policy provisions as well. Two recent studies document how several core and noncore. Core provisions are identified in the literature trade agreements cover regulatory areas such as services, invest- as economically more meaningful (Baldwin 2008; Damuri 2012) ment, competition policy, intellectual property rights protection, and include the set of WTO-plus provisions and four WTO-extra and others (Horn, Mavroidis, and Sapir 2010; WTO 2011). Building provisions (competition policy, investment, movement of capi- on the methodology in these studies, Hofmann, Osnago, and Ruta tal, and intellectual property rights protection) that appear fre- (2017) collected information on all preferential trade agreements in quently in preferential trade agreements. Almost 90% of agree - force and notified to the WTO in 2015. Their new database contains ments include at least one of the core WTO-extra provisions, information on the inclusion and legal enforceability of 52 policy and one third of preferential trade agreements include all core areas in 279 preferential trade agreements among 189 countries.4 WTO-extra provisions (see figures 8.2 and 8.3). FIGURE 8.1 The number and content of preferential trade agreements, 1951–2015 Number of policy areas covered by newly signed agreements in each year Number of agreements 35 More than 20 policy areas 350 10–20 policy areas 30 Fewer than 10 policy areas 300 Not in force Cumulative number of agreements 25 250 20 200 15 150 10 100 5 50 0 0 1951 1955 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Source: World Bank Content of Deep Trade Agreements Database (http://data.worldbank.org/data-catalog/deep-trade-agreements). In C te om lle Ta ct pe rif ua tit lp M io fs 0 50 100 150 200 250 ro ov n on e In po pe in 0 50 100 150 200 250 300 rty me ve lic Ta du rig nt stm y rif st fs ria En ht of c en on lg vi s p ap t r ag oo In on rot ita ds fo m ec l ric Trade Organization. rm en tio ul a ta n tu Included Re Vi tion l la ra w Number of agreements Number of agreements gi sa lg on an soc s oo al d iet ds La co as y b op ylu Re or se m e m C a Legally enforceable Ag ra us Ec ar rk r tio to on ch et icu n m om an reg ltu d s u re Ex In ic p tec lat du o po st icy o l hn ion rt ria d lo ta l c ia gy xe Ed oo lo s uc pe gue An at ra Te Pu ion ti ch tid bl a En on ni um ic nd e ca pi ad t rg lb ng Sm m rai y ar al C in ni rie la on ist ng rs ra nd C sum S t Sa to m u l er ta ion ni ed u t p tis ta tra iu ral ro tic ry de m co te s /p -s c hy iz op tio to Po e e era n sa lit nte tio ni ic rp n ta al r ry Fi So dia ise na c l s G nc ial ogu en St ia m e er at l a at e ss te al ai Ap ist rs a Tr Ag d A ad re pr ox n Ta nc Ag C e em im Da o tio tic xa e re ou in FIGURE 8.2 “WTO-plus” policy areas in preferential trade agreements, 2015 at ta rru n em nt Se ent FIGURE 8.3 “WTO-extra” policy areas in preferential trade agreements, 2015 io p pt er rv o n ro io en va ic n of te n of t o ili es le cti In n T n g In gi o no sla n te lle rad m va tio ea ct e- su tio He n ua Re re n al p th l P la t s Ill olic ro ed pe A H icit ies rty sp um d a ru Pu Ri ec gh ts Au n r gs bl di igh ic ts M o ts vi pr oc on s St at ur Source: World Bank Content of Deep Trade Agreements Database (http://data.worldbank.org/data-catalog/deep-trade-agreements). Source: World Bank Content of Deep Trade Agreements Database (http://data.worldbank.org/data-catalog/deep-trade-agreements). Ill ey M ual eg la in e em tra en al un in Ag im de g re di ng t m rin em ig g en ra en te N Te tio In t o rp uc rro n ve n ris C lea ris st Tr es iv Included il r s m m ad en e pr af ot et t M -R ec y e tio ea lat n su ed re Note: WTO-extra refers to provisions on some policy areas in preferential trade agreements that fall outside the current mandate of the World Trade Organization. s Preferential trade agreements and global value chains • 177 Note: WTO-plus refers to legally enforceable regulations on some policy areas in preferential trade agreements that fall under the current mandate of the World Legally enforceable 178 • Measuring and Analyzing the Impact of GVCs on Economic Development The new data also reveal the changing depth of preferen- FIGURE 8.4 Relationship between depth of preferential tial trade agreements. Hofmann, Osnago, and Ruta (2017) con- trade agreements and global value chain trade structed synthetic indexes of depth, which measure the cover- Density age of policy areas in preferential trade agreements. The first 0.125 index of depth, referred to as “total depth,” is the simple count of (legally enforceable) provisions in a preferential trade agree- Medium depth ment. Total depth increased from an average of around 8 provi- 0.100 sions in the 1990s to more than 17 in 2010–15. An index of “core Low depth depth” can be constructed by counting how many core provi- High depth sions are included and legally enforceable in a preferential trade agreement. Core depth increased from around 7 provisions in 0.075 the 1990s to almost 14 in 2010–15. Principal component analy- sis can produce a third index of depth that accounts for most No preferential trade agreeement of the variability in the data. Principal component analysis depth 0.050 increased from around 1 in the 1990s to 2.8 in 2010–15. The wide country coverage of the new data allows for analy- sis of the heterogeneity of deep preferential trade agreements 0.025 across regions and incomes. Europe has the highest number of signed preferential trade agreements, and these preferential trade agreements are the deepest mainly because of the Euro- 0.000 pean Community Treaty and the subsequent EU enlargements. –10 0 10 20 The average total depth of EU agreements is 25 provisions. Deep Imports of parts and components (log) preferential trade agreements are also common for members of the European Free Trade Association (average of 23 policy pro - Source: Hoffman, Osnago, and Ruta 2017. visions), Japan (21), and the Republic of Korea (20). Preferential Note: Depth of preferential trade agreements is measured by a density index trade agreements signed between developed and developing for number of policy areas covered, where 0 means. Global value chains are countries (North–South preferential trade agreements) include measured by the log of imports of parts and components. on average almost as many provisions (20) as North–North pref- erential trade agreements (22). But legal enforceability is gener- ally weaker in North–South preferential trade agreements than in Economic theory and evidence suggest a relationship between North–North agreements. And South–South preferential trade cross-border production and shallow preferential trade agree- agreements, with an average total depth of 13 provisions, tend ments. For instance, Blanchard and Matschke (2015) estimated to be shallower than other preferential trade agreements. that a 10% increase in U.S. foreign affiliate exports to the United States is associated with a 4 percentage point increase in the rate of preferential duty-free access. Intuitively, firms that offshore Global value chains and the rationale for trade production are more likely to lobby for lower tariffs on products agreements re-imported into the U.S. market. Similarly, domestic firms may choose to locate production stages in another preferential trade What is the rationale for trade agreements, particularly deep agreement member under the expectation that tariffs on re-­ agreements, in a world with GVCs? An extensive literature has imported goods will be lower. examined the motives for trade policy cooperation and the design Deep agreements go beyond traditional market access issues of trade agreements in a traditional setting, where production is and include disciplines such as investment, competition policy, entirely national and not fragmented internationally.5 The focus and harmonization of product regulations. The new empirical generally is on cooperation on tariffs, consistent with the idea that evidence on the relationship between preferential trade agree- the main problem that trade agreements solve is to internalize the ment depth and GVC trade is the core of the next sections. Here, terms-of-trade externality created by unilateral tariffs. But there is this relationship is discussed from a theoretical point of view a positive correlation between GVC trade (measured as trade in (Antràs and Staiger 2012; Baldwin 2008; WTO 2011; Ederington parts and components) and the depth of trade agreements (mea- and Ruta 2016). sured by the number of policy areas covered by the agreements; A simple way to explain the correlation between GVC trade figure 8.4). This relationship indicates that the rationale for trade and depth of preferential trade agreements is that certain agreements may be more complex in the context of GVCs than in behind-the-border policies need to be disciplined in trade settings where production is not fragmented internationally. agreements for GVCs to operate efficiently. First, the unbun- Lawrence (1996) first introduced the notion of “shallow” and dling of stages of production across borders creates new forms “deep” trade agreements. Shallow agreements focus on tariffs of cross-border policy spillovers beyond the traditional terms- and other border measures that directly affect market access. of-trade externality. Second, governments may face credibility Preferential trade agreements and global value chains • 179 problems for behind-the-border measures in the context of involvement, but data are limited to a small sample of countries GVCs. And third, the costs created by coordination externalities (41) and time coverage is limited to 1995–2011.7 (such as heterogeneous regulations) may be higher in the pres- The empirical approach is based on the standard augmented ence of cross-border production. These spillovers and credibility gravity model, which has been widely used to assess the impact concerns generate demand for deeper forms of integration. of preferential trade agreements on trade flows (see Baier and Despite the rich set of arguments in the literature, many Bergstrand 2007). Unlike in the standard approach, which uses aspects of the relationship between deep preferential trade a dummy variable to identify the presence of a trade agreement agreements and GVCs have not been incorporated in formal between a country pair, Osnago, Rocha, and Ruta (2016) used the models. Foremost is the fundamental question of the role that three measures of the “depth” of preferential trade agreements deep agreements play in the presence of GVCs. For instance, discussed earlier (total depth, core depth, and principal compo- Antràs and Staiger (2012) show that behind-the-border policies nent analysis depth). Their regressions of the impact of preferen- create cross-border spillovers when production is internationally tial trade agreement depth on GVC trade included a set of fixed fragmented. While they indicate that deep provisions in prefer- effects and control for various determinants of bilateral trade. ential trade agreements may allow governments to internalize Signing deep agreements has a large and positive impact on these externalities, their model does not provide a formal treat- GVC trade (figure 8.5). Adding a provision to a preferential trade ment of this point. Similarly, the commitment rationale for deep agreement increases bilateral trade in parts and components agreements has been formalized only for specific provisions 1.5% and re-exported value added 0.4%. This means that signing (such as domestic subsidies in Brou and Ruta 2013), and this has the deepest preferential trade agreement in the sample doubles not been done in a GVC context. Finally, studies of the harmo - trade in parts and components and increases re-­ exported value nization of standards and other forms of regulatory cooperation added about 22%. (such as Costinot 2008) generally rely on traditional trade models The analysis of the impact of preferential trade agreements that assume that production is purely domestic. on GVC trade presents two difficult econometric challenges. Several other interesting questions are also still open. One set of questions relates to the content of deep agreements. A large trade literature has recently investigated the role of insti- FIGURE 8.5 The impact of deep preferential trade tutions in shaping the international organization of production agreements on two types of global value chain trade (Antràs 2015). Osnago, Rocha, and Ruta (2015) find evidence that Marginal trade impact the content of deep preferential trade agreements affects deci- sions on foreign direct investment, suggesting that the role of 0.15 Trade in value added specific provisions in shaping GVCs may be relevant. But more Trade in parts and components work is needed to understand the specific channels. A second area relates to the role of preferential as opposed to multilateral deep integration. Why is deep integration generally taking place in preferential trade agreements? How are preferential trade 0.10 agreement partners selected in a GVC context? (These ques- tions are returned to below.) Do deep agreements promote global value 0.05 chains? The relationship between GVCs and preferential trade agree- ments runs in both directions.6 An important policy question concerns how much trade agreements, particularly deep pref- 0.00 erential trade agreements, can boost GVC integration. Osnago, Total depth Core depth Principal Rocha, and Ruta (2016) used a gravity model to exploit the new component World Bank data on the content of preferential trade agree- analysis depth ments, using the three measures of the “depth” of preferential trade agreements discussed earlier (total depth, core depth, and Source: Osnago, Rocha, and Ruta 2016. principal component analysis depth). Bilateral GVC integration is Note: Total depth is the simple count of (legally enforceable) provisions in a measured in two ways: trade in parts and components and trade preferential trade agreement. Core depth is a count of how many core provi- in value added. Data on trade in parts and components have the sions are included and legally enforceable in a preferential trade agreement. advantage of being available for a larger set of countries and Principal component analysis depth is an index that accounts for most of the years covered by the new dataset on preferential trade agree- variability in the data. ments. Trade in value added is a more precise measure of GVC 180 • Measuring and Analyzing the Impact of GVCs on Economic Development The first is that, as first noted in Johnson and Noguera (2014) it considers two dimensions of heterogeneity: splitting the pro- and Noguera (2012), value-added trade depends not only on visions into different categories (WTO-plus and WTO-extra) and bilateral trade costs, but also on trade costs with third coun- dividing preferential trade agreements by the level of develop- tries. The second challenge is the endogeneity of GVC trade and ment of country pairs (North–North, North–South, and South– preferential trade agreements. Osnago, Rocha, and Ruta (2016) South). These extensions allow for investigation of the types of attempted to address both problems. provisions that drive the relationship between deep agreements First, to take into consideration the indirect effects that pref- and GVCs among different sets of countries. erential trade agreements by third countries may have on GVC Countries can have different reasons for signing trade agree- trade of other countries, the depth variable of interest needs to ments, depending on, among other things, the level of liberaliza- be weighted to take into account the international input-output tion already achieved. World Bank data show that North–North structure. Following the methodology proposed by Noguera preferential trade agreements and North–South preferential (2012), Osnago, Rocha, and Ruta (2016) found that accounting for trade agreements tend to have similar depth (number of provi- the depth of third-country agreements increases the impact of sions covered by the agreement) and South–South preferential preferential trade agreements on GVCs. trade agreements are on average shallower. In addition, the prev- The second empirical concern is endogeneity. Deep preferen- alence of WTO-plus and WTO-extra provisions varies according tial trade agreements may stimulate the creation of GVCs by pro- to the level of development of the signatories of the agreement. viding common disciplines that allow internalizing cross-border North–North and North–South preferential trade agreements policy spillovers and address credibility problems. But countries tend to have more WTO-extra provisions, though for North– already involved in GVCs may be more likely to sign deep pref- South agreements they are less likely to be legally enforceable erential trade agreements because cross-border production cre- (figure 8.6). South–South agreements tend to focus on WTO-plus ates a demand for deep provisions. The fixed-effect approach issues, which are the more traditional trade policy areas. partially controls for this reverse causality since it compares There is no formal theory to guide the analysis of differential country pairs before and after a preferential trade agreement is effects of deep preferential trade agreements across countries’ signed. But other time-varying country pair characteristics may level of development, but a plausible argument is that deep pref- not be controlled for. Osnago, Rocha, and Ruta (2016) adopted erential trade agreements matter for developed and developing an instrumental variable approach to address this type of endog- economies for different reasons. With trade among developed eneity, using as an instrument for the depth of the preferential countries already largely liberalized and domestic institutions trade agreement between country i and country j the (weighted) robust, North–North deep agreements aim mostly to internal- average depth of all the agreements signed by countries i and j ize cross-border policy spillovers. Deep agreements have addi- with any other country, excluding the agreements they have in tional roles for developing countries, since trade generally faces common. Results of the analysis confirm the relevance of deep higher barriers and domestic institutions are weaker relative to agreements in boosting GVC trade. advanced economies. North–South deep trade agreements offer An alternative approach for determining the importance of an anchor to boost GVC participation of developing countries by deep preferential trade agreements for GVCs is to look at the providing a commitment device for border and behind-the-bor- effect of depth on different sectors. The effect of deep prefer- der policies. Since tariffs between developing countries are often ential trade agreements should be stronger in sectors that are still high, South–South preferential trade agreements affect GVC more integrated in GVCs. To test this, Osnago, Rocha, and Ruta participation mostly through traditional trade liberalization. (2016) exploited the decomposition of gross exports into value- To investigate the effect of the content of preferential trade added components available for 13 manufacturing sectors in the agreements depending on countries’ level of development, World Input-Output Database for 1995–2011. They augmented a Osnago, Rocha, and Ruta (2016) ran three regressions on differ- sector-level gravity regression with an interaction term between ent groups of countries: North–North, North–South, and South– depth and an index of vertical specialization. The coefficient of South. To allow for a broader sample of countries, data on trade the interaction term is consistently positive and significant across in parts and components are used in these regressions to mea- specifications and using different variables of depth and provi- sure the extent of GVC integration among country pairs. In each sions. The results suggest that deep preferential trade agree- regression, the key explanatory variable is the number of WTO- ments have a larger impact on GVC-intensive sectors.8 plus and WTO-extra provisions in a preferential trade agreement rather than preferential trade agreement depth. The content of preferential trade agreements matters for GVC Global value chains and the content of integration, and the impact varies by countries’ level of devel- preferential trade agreements opment. WTO-plus provisions, such as tariff reduction, drive the effect of deep preferential trade agreements on South–South This section digs further into the relationship between deep trade in parts and components. Each additional WTO-plus pro- preferential trade agreements and GVCs and explores empiri- vision boosts South–South GVC integration 8.3%. WTO-extra cally potential heterogeneity in the effects of deep preferential provisions, such as investment and competition policy, drive trade agreements.9 Following Osnago, Rocha, and Ruta (2016), the effects of North–South trade in parts and components. An Preferential trade agreements and global value chains • 181 FIGURE 8.6 The depth of preferential trade agreements by the type or provisions and the development level of members Number of provisions WTO-plus WTO-extra 12 12 10 10 8 8 6 6 4 4 2 2 0 0 North–North North–South South–South North–North North–South South–South Average number of WTO-plus provisions Average number of WTO-extra provisions Average number of legally enforceable WTO-plus provisions Average number of legally enforceable WTO-extra provisions Source: Osnago, Rocha, and Ruta 2016. Note: WTO-plus provisions include legally enforceable regulations on policy areas that are under the current mandate of World Trade Organization (WTO) cus- toms regulations, such as export taxes, antidumping measures, and technical barriers to trade. WTO-extra provisions are outside the WTO mandate and include a wide-ranging set of policy areas, from investment to environmental laws and nuclear safety. additional WTO-extra provision in a North–South preferential trade agreement at the expense of other countries not included trade agreement increases GVC integration 4.3%. in the agreement. This would be the case where exports from members displace exports from nonmembers. Second, coun- tries may have noneconomic reasons to sign trade agreements, Global value chains and the choice of because preferential trade agreements can strengthen security preferential trade agreement partners ties or work as a building block for political integration. As these arguments are beggar-thy-neighbor or noneconomic, preferen- From a normative perspective the issue is whether the interna- tial trade agreements are an inefficient substitute for multilateral tional fragmentation of production changes the merits of region- trade liberalization from an economic point of view. alism relative to multilateralism. From a positive perspective the GVCs alter this logic by creating new rationales for preferen- question is whether the presence of GVCs (or the possibility of tial trade agreements: the unbundling of stages of production anchoring a country to them) changes the way countries select across borders creates new forms of international policy spill- their trading partners. This section briefly looks at both issues overs and time-consistency problems. These in turn generate from a theoretical perspective and then applies the question to demand for deeper forms of integration. For deep agreements China. involving behind-­ the-­border policies, a tradeoff arises between The debate on the merits of regionalism versus multilateral­ism economies of scale and heterogeneity of preferences.10 This dates back at least to Viner (1950). In traditional models, where tradeoff is well known in the public economics literature that production is entirely national and tariffs are the sole instrument deals with fiscal federalism (Oates 1999). While noneconomic of trade policy, preferential trade agreements are suboptimal to arguments and new beggar-­ neighbor gains (such as a “rule thy-­ a multilateral agreement from a global welfare perspective. How- of law” externality) may still drive the decisions to form prefer- ever, preferential arrangements may still be efficient from the ential trade agreements, smaller groups can be efficient from perspective of an individual country, for both economic and non- an economic point of view as they efficiently trade off the costs economic reasons. First, countries may benefit from a preferential and benefits of deep integration. As argued in WTO (2011), deep 182 • Measuring and Analyzing the Impact of GVCs on Economic Development preferential trade agreements may complement rather than sub- complementarities are important in the choice of preferential stitute for the multilateral trading system because they allow for trade agreement partners for China. coordinating or harmonizing policies that could not be coordi- nated or harmonized at the global level. From a positive perspective the literature on shallow prefer- The future of the relationship between deep ential trade agreements struggled with the notion of the ideal agreements and global value chains trade partner (Schiff and Winters 2003). Two main sets of eco- nomic characteristics increase the benefits of forming a prefer- The past 25 years have been a period of deepening trade agree - ential trade agreement with a specific partner. The first is trade ments and growing intensity of GVCs. Will this trend continue intensity, which suggests that the two countries are “natural” in the next quarter century? Given the growing backlash to glo - trade partners. Characteristics such as geographic proximity balization in advanced economies, this is no longer a rhetorical that increase trade intensity among partners make it more con- question. There are reasons for optimism, as GVCs and prefer- venient to reduce bilateral tariffs. The second has to do with ential trade agreements reinforce each other and make slipping comparative advantage: complementarities in production or backward less likely. But the future of this relationship should not consumption increase the benefit of forming a preferential trade be taken for granted because cross-border production decisions agreement. depend on expectations concerning trading partners’ future Are these characteristics relevant in the context of GVCs? trade policies. Negative expectations could result in a rever- The answer is not obvious, and the literature is not yet devel- sal of the current trends toward GVC expansion and deeper oped. Some characteristics still matter. For instance, proximity integration. may be important in selecting preferential trade agreement Some observers have argued that the current globalization partners because face-to-face communication is relevant to backlash has similarities to the backlash of the early 20th century managing supply chains. Similarly, comparative advantage and that this may lead to a prisoner’s dilemma, where countries can be defined at the task level, with complementarities­ —­say, escalate protectionism even though it is not in their interest, as between different stages of production­ —­guiding the choice of in the 1930s. This pessimistic view of the future of trade relations preferential trade agreement partners. But other characteristics does not acknowledge that the production structure and trade would appear to matter too, such as cross-country differences policy landscape of today are very different from those of the in policy preferences. If GVCs require deep agreements to early 20th century. function smoothly, ideal preferential trade agreement partners Trade agreements have stimulated the creation of GVCs by should not have policy preferences that are too different, since internalizing cross-border policy externalities, lowering trade this would increase the cost of coordinating and harmonizing costs, and providing deeper common disciplines that facilitate policies. the operation of economic activities spanning multiple borders. There is a new focus in the literature on the experience of In turn, GVCs have changed the political economy of trade policy, China in choosing preferential trade agreement partners. To discouraging protectionism and creating a demand for deep inte- characterize preferential trade agreements from the point of gration. The higher the domestic content of foreign-produced view of GVCs, Cheng and others (2016) borrowed the “smile final goods, the lower the tariffs set by governments (Blanchard, curve.” The horizontal axis represents a continuum of tasks or Bown, and Johnson 2016) and the higher the GVC trade with stages of GVC from upstream to downstream covering research partners, the deeper the agreements countries sign (Orefice and and development, intermediates, assembly, processing, market- Rocha 2014). This two-way relationship between GVCs and pref- ing, and after-sale services. The vertical axis depicts the value erential trade agreements supports the view that trade disinte- added generated from various tasks or stages. Based on this gration (protectionism, undoing trade agreements) is unlikely. notion, Cheng and others (2016) defined vertical preferential Despite these dramatic changes, however, the future of the trade agreements as agreements driven by comparative advan- relationship between preferential trade agreements and GVCs tage at the task level­—­ agreements formed as a result of the ver- should not be taken for granted. GVCs are the result of firms’ tical division of labor along the supply chain, with member econ- investment and sourcing decisions, which are endogenous omies locating at different GVC positions. because they depend on expectations of future trade policies. With this framework in mind, Cheng and others (2016) asked If firms expect a change in future trade policy, they will take this whether China’s preferential trade agreements exploit com- into account in their decisions, possibly leading them to rena- plementarities in production along the supply chain. China has tionalize (part of) their production processes. In this context, concluded and is implementing 13 preferential trade agree- expectations can lead to multiple equilibria and give rise to coor- ments involving 21 individual economies11 and is negotiating or dination failures. More than the well-known prisoner’s dilemma, has proposed 11 other bilateral and regional preferential trade the current situation may be described as a trust dilemma (or a agreements, along with the 16-member Regional Comprehen- coordination game) where what is rational to choose depends on sive Economic Partnership. By quantifying China’s GVC link- beliefs about what others will do. age with its preferential trade agreement and non–preferential A simple game illustrates the trust dilemma that may charac- trade agreement partners, Cheng and others found that GVC terize the relationship between preferential trade agreements Preferential trade agreements and global value chains • 183 and GVCs­ —­referred to here as the trust dilemma of deep inte - TABLE 8.1 The trust dilemma of deep integration gration (table 8.1). Consider two players, Home and Foreign, Global value National and assume that each has two strategies. They can opt for deep chains and deep production and no agreements and GVCs or choose national production and no Players: Home, Foreign agreements agreement trade agreement. Each player chooses an action without knowing Global value chains and the choice of the other. If a player chooses to maintain an inter- deep agreements 2, 2 0, 1 national production process and a deep agreement, it needs the National production and cooperation of the partner to succeed. Choosing national pro- no agreement 1, 0 1, 1 duction and no trade agreement, by contrast, requires no coop- eration with the other player but also leads to lower welfare. Source: Author’s elaboration. The trust dilemma of deep integration has two pure-­ s trategy Nash equilibria. The first is the upper-left corner of table 8.1, where Home and Foreign cooperate (2,2); the second is the groups of like-minded countries. But many important questions lower-right corner, where the two players defect and choose on the relationship between preferential trade agreements and not to cooperate (1,1). With global welfare inferior in the no- GVCs remain open. One is on the content (or, equivalently, the cooperation strategy, this equilibrium can be described as a efficient design) of deep preferential trade agreements; another coordination failure. Importantly, this equilibrium can be the is on the optimal choice of preferential trade agreement part- result of a self-fulfilling prophecy in that it can be triggered by ners. The literature has investigated these questions in models the belief that the other player will not choose to cooperate. of shallow agreements and national production, but not for deep While only an example, the game shows why continuing trust agreements and GVCs. in the willingness of others to cooperate is essential to the future Second, thanks to the new data on the content of prefer- of the relationship between preferential trade agreements and ential trade agreements and on measures of GVC integration, GVCs. In the past 25 years governments signed deep agree - some progress has been made in illuminating the extent of the ments, and firms fragmented production internationally. These relationship between preferential trade agreements and GVCs. decisions reinforced each other and sustained a cooperative Recent evidence shows that deep preferential trade agree- equilibrium (the upper-left corner). In the next 25 years changing ments boosts GVC integration and that undoing this depth is expectations for the course of policy could lead to a reversal and likely to hurt GVCs. The content of preferential trade agree- result in an inferior equilibrium where production is progressively ments also matters: WTO-extra provisions are key drivers of renationalized and trade agreements undone (the lower-right GVCs for North–South preferential trade agreements, while corner). This coordination failure can be avoided as long as firms’ WTO-plus provisions are important for South–South GVC inte- expectations of future trade policy does not induce them to opt gration. And an analysis of China’s trade agreements indicates for national production, with policymakers offering protection that the choice of the “right” preferential trade agreement and undoing trade agreements. partners is affected by a country’s GVC position, stressing the importance of comparative advantage at the task level among other factors. Conclusions On the future of GVCs and deep agreements, there are rea- sons for optimism and reasons for concern. In the past 25 years New World Bank data on the content of trade agreements show governments signed deep preferential trade agreements and that preferential trade agreements are becoming deeper. First, firms fragmented production. These decisions reinforced each economic theory indicates that preferential trade agreements other and sustained a cooperative equilibrium. In the next 25 and GVC integration are related, as the smooth functioning of years changing expectations for the course of policy could lead cross-border production activities calls for the regulation of to a reversal and result in an inferior equilibrium where pro- behind-the-border policy areas. Theory also points out that, in a duction is progressively renationalized and trade agreements GVC context, preferential trade agreements and the multilateral undone. Continuing trust in the willingness of others to cooper- trade system generally complement each other because some ate to preserve an open system is essential to the future of the policy areas can be more efficiently regulated within smaller relationship between preferential trade agreements and GVCs. 184 • Measuring and Analyzing the Impact of GVCs on Economic Development Notes References 1. This chapter uses the term “preferential trade agreements” rather Antràs, P. 2015. Global Production: Firms, Contracts, and Trade Structure. than “regional trade agreements” since some of these agreements Princeton, NJ: Princeton University Press. are not necessarily between countries within the same region or in Antràs, P., and R. Staiger. 2012. “Offshoring and the Role of Trade Agree- regional proximity. ments.” American Economic Review 102 (7): 3140–83. 2. This section draws on Hofmann, Osnago, and Ruta (2017). Bagwell, K, C. P. Bown, and R. W. Staiger. 2015. “Is the WTO Passé?” 3. See Freund and Ornelas (2010), WTO (2011), and Limao (2016) for National Bureau of Economic Research (NBER) Working Paper No. recent surveys of the literature on preferential trade agreements. 21303, NBER, Cambridge, MA. 4. This database offers the most comprehensive and up-to-date data avail- Bagwell, K., and R. W. Staiger. 2016. “The Design of Trade Agreements.” In able on the number of trade agreements, countries, and policy areas Handbook of Commercial Policy, Vol. 1A, edited by K. Bagwell and R. covered. The database is freely available on the World Bank website at Staiger, 435–529. Amsterdam: North Holland. http://data.worldbank.org/data-catalog/deep-trade-agreements. Baier S. L., and J. H. Bergstrand. 2007. “Do Free Trade Agreements Actu- 5. See Maggi (2014), Bagwell, Bown, and Staiger (2015), Bagwell and ally Increase Members’ International Trade?” Journal of International Staiger (2016), and Grossman (2016) for recent reviews. Economics 71 (1): 72–95. 6. This section is based on Osnago, Rocha, and Ruta (2016). Baldwin, R. 2008. “Big-Think Regionalism: A Critical Survey.” NBER Work- 7. Data on trade in parts and components come from Comtrade, while ing Paper No. 14056, NBER, Cambridge, MA. the data on trade in value added are based on the decomposition by Blanchard, E., C. Bown, and R. Johnson. 2016. “Global Supply Chains and Wang, Wei, and Zhu (2016) and come from the World Input-Output Trade Policy.” NBER Working Paper No. 21883, NBER, Cambridge, MA. Database. Blanchard, E., and X. Matschke. 2015. “U.S. Multinationals and Preferential 8. This section looked at the impact of preferential trade agreements in Market Access.” The Review of Economics and Statistics 97 (4): 839–54. boosting GVC trade. A related question is whether the undoing of a Brou, D., and M. Ruta. 2013. “A Commitment Theory of Subsidy Agree- preferential trade agreement would negatively affect GVCs. Mulab- ments.” The B.E. Journal of Economic Analysis & Policy 13 (1): 239–70. dic, Osnago, and Ruta (2017) studied the effect that EU membership Cheng, D., X. Wang, Z. Xiao, and W. Yao. 2016. “How Does the Selection had on GVC and overall trade of the United Kingdom, most notably of FTA Partner(s) Matter in the Context of GVCs? The Experience of with its European partners, and then used this information to assess China.” Working Paper, Fudan University, Shanghai. the future of U.K.–EU trade under different scenarios. Costinot, A. 2008. “A Comparative Institutional Analysis of Agreements on 9. This section is based on Osnago, Rocha, and Ruta (2016). Product Standards.” Journal of International Economics 75 (1): 197–213. 10. Maggi (2014) discusses a similar rationale. Specifically, bargaining fric- Damuri, Y. R. 2012. “21st Century Regionalism and Production Sharing tions may be higher for negotiations that involve many countries and Practice.” Center for Trade and Economic Integration (CTEI) Working complex issues. For this reason, deep provisions may be more effi- Paper No. 2012–4, CTEI, Geneva. ciently negotiated in a preferential trade agreement or in an agree- EC (European Commission). World Input-Output Database (WIOD). Brus- ment involving a subset of members within the WTO, such as a pluri- sels: EC. Available at: http://www.wiod.org/home. lateral or critical-mass agreement. Ederington, J., and M. Ruta. 2016. “Non-Tariff Measures and the World 11. The preferential trade agreements are with Australia; Chile; Costa Trading System.” In Handbook of Commercial Policy, Vol. 1B, edited by Rica; Hong Kong, China; Iceland; Macao, China; New Zealand; Paki- K. Bagwell and R. Staiger. Amsterdam: North Holland. stan; Peru; the Republic of Korea; Singapore; Switzerland; and the Freund, C., and E. Ornelas. 2010. “Regional Trade Agreements.” Annual 10-member Association of Southeast Asian Nations (Brunei, Cambo- Review of Economics 2 (1): 139–66. dia, Indonesia, Lao PDR, Malaysia, Myanmar, Philippines, Singapore, Grossman, G. 2016. “The Purpose of Trade Agreements.” In Handbook of Thailand, and Viet Nam). Commercial Policy, Vol. 1A, edited by K. Bagwell and R. Staiger, 379– 434. Amsterdam: North Holland. Hofmann, C., A. Osnago, and M. Ruta. 2017. “Horizontal Depth: A New Database on the Content of Deep Agreements.” Policy Research Work- ing Paper No. 7981, World Bank, Washington, DC. Horn, H., P. C. Mavroidis, and A. Sapir. 2010. “Beyond the WTO? An Anat- omy of EU and US Preferential Trade Agreements.” The World Econ- omy 33 (11): 1565–88. Johnson, R., and G. Noguera. 2014. “A Portrait of Trade in Value Added over Four Decades.” Working Paper, Dartmouth College, Hanover, NH. Lawrence, R. Z. 1996. Regionalism, Multilateralism, and Deeper Integration. Washington, DC: Brookings Institution. Limao, N. 2016. “Preferential Trade Agreements.” In Handbook of Com- mercial Policy, Vol. 1B, edited by K. Bagwell and R. Staiger, 279–367. Amsterdam: North Holland. Preferential trade agreements and global value chains • 185 Maggi, G. 2014. “International Trade Agreements.” In Handbook of Inter- ———. 2016. “Deep Trade Agreements and Global Value Chains.” Work- national Economics, edited by G. Gopinath, E. Helpman, and K. Rogoff, ing Paper, World Bank, Washington, DC. 317–90. Amsterdam: Elsevier. Schiff, M., and L. A. Winters. 2003. Regional Integration and Development. Mulabdic, A., A. Osnago, and M. Ruta. 2017. “Deep Integration and UK– Washington, DC: World Bank. EU Trade Relations.” Policy Research Working Paper No. 7947, World UN (United Nations). UN Commodity Trade Statistics Database (Comtrade). Bank, Washington, DC. New York: UN. Available at: https://comtrade.un.org/. Noguera, G. 2012. “Trade Costs and Gravity for Gross and Value Added Viner, J. 1950. The Customs Union Issue. New York: Carnegie Endowment Trade.” Working Paper, University of Warwick, Coventry, U.K. for International Peace. Oates, W. E. 1999. “An Essay on Fiscal Federalism.” Journal of Economic Wang, Z., S.-J. Wei, and K. Zhu. 2016. “Quantifying International Produc- Literature 37 (3): 1120–49. tion Sharing at the Bilateral and Sector Levels.” Working Paper, Colum- Orefice, G., and R. Rocha. 2014. “Deep Integration and Production Net- bia University, New York. works: An Empirical Analysis.” The World Economy 37 (1): 106–36. —­ WTO (World Trade Organization). 2011. World Trade Report 2011­ The Osnago, A., N. Rocha, and M. Ruta. 2015. “Deep Trade Agreements and existence to coher- WTO and preferential trade agreements: From co-­ Vertical FDI: The Devil is in the Details.” Policy Research Working Paper ence. Geneva: WTO. No. 7464, World Bank, Washington, DC. APPENDIX 1 Authors’ conferences First Authors’ Conference: Background Papers Beijing, March 17–18, 2016 Organized by the Research Institute for Global Value Chains at University of International Business and Economics and China Development Research Foundation Co-sponsored by Bill & Melinda Gates Foundation Thursday, March 17 8:00 a.m.–8:30 a.m. Opening remarks Speakers Dr. Anabel Gonzalez, Senior Director of Trade & Competitiveness Global Practice, World Bank Dr. Hubert Escaith, Chief Statistician, WTO Dr. Nadim Ahmad, Chief of Trade & Competitiveness Statistics Division, OECD Dr. David Dollar, Senior Fellow, Brookings Institution Fang Jin, Deputy Secretary General, China Development Research Foundation Professor Zhao Zhongxiu, Vice President, UIBE 8:30 a.m.–9:30 a.m. Characterizing global value chains Speakers Wang Zhi, UIBE, Wei Shangjin, ADB, Yu Xinding, UIBE, and Zhu Kunfu, UIBE Discussant Satoshi Inomata, IDE–JETRO 9:30 a.m.–10:30 a.m. Global value chains and their domestic foundations Speakers Cosimo Beverelli, Robert B. Koopman, Simon Neumueller, and Victor Kummritz, WTO Discussant Meng Bo and Jiyoung Kim, IDE–JETRO 10:45 a.m.–11:45 a.m. Estimation of cumulative trade cost along global value-chains Speakers Antonia Diakantoni, Hubert Escaith, Michael Roberts, and Thomas Verbeet, WTO Discussant Nadim Ahmad, OECD 11:45 a.m.–12:45 p.m. Taxation reform on intermediate imports and its implications for structure adjustment of Chinese economy­ —­a CGE model–based analysis Speakers Wang Fei and Pei Jianso, UIBE, He Jianwu, DRC Discussant David Dollar, Brookings institution 187 188 • Measuring and Analyzing the Impact of GVCs on Economic Development 2:00 p.m.–3:00 p.m. Measuring smile curves in global value chains Speakers Ye Ming, Fudan University, Meng Bo, IDE–JETRO and Wei Shangjin, ADB Discussant Ju Jiandong, Shanghai University of Economics and Finance 3:00 p.m.–4:00 p.m. Participation of developing countries in global value chains Speaker Przemyslaw Kowalski, Javier Lopez Gonzalez, Alexandros Ragoussis, and Cristian Ugarte, OECD Discussant Joseph Mariasingham, ADB 4:15 p.m.–5:15 p.m. Middle income trap and GVCs Speakers Gianluca Santoni, Daria Taglioni, Deborah Winkler, World Bank, and Victor Kummritz, WTO Discussant Li Zhigang, ADB 5:15 p.m.–6:15 p.m. Production transformation in emerging economies Speakers Nadim Ahmad and Annalisa Primi, OECD Discussant Yu Jiantuo, CRDF 6:30 p.m.–7:30 p.m. Dinner Participant All Friday, March 18 8:00 a.m.– 9:00 a.m. Services trade and GVCs Speakers Erik van der Marel, ECIPE, and Sebastian Saez, World Bank Discussant Cosimo Beverelli, WTO 9:00 a.m.–10:00 a.m. Services in Global Value Chains: From Inputs to Value-Creating Activities Speakers Sebastien Miroudot, OECD Discussant Li Shantong, DRC 10:10 a.m.–11:10 a.m. Preferential Trade Agreements and Global Value Chains Speakers Alen Mulabdic, Alberto Osnago, Michele Ruta, World Bank, Nadia Rocha, WTO Discussant Andrew Crosby, ICTSD 11:10 a.m.–12:10 p.m. How does the Selection of FTA Partner(s) Matter in the Context of GVCs? The Experience of China Speakers Cheng Dazhong, Fudan University, Wang Xinkui, Shanghai WTO Center, Xiao Zhiguo, Fudan University, Yao Weiqun, Shanghai WTO Center Discussant Robert B. Koopman, WTO 12:15 a.m.–2:00 p.m. ICTSD - WEF Joint Launching of the E15 Report on “Strengthening the Global Trade and Investment System in the 21st Century” in partnership with Caixin Insight Group and UIBE/RIGVCs Participant All 2:00 p.m.–3:00 p.m. Micro structure of global imbalance and the development of global value chains Speakers Yang Jun, UIBE, Li Xin, Beijing Normal University, Wang Zhi, UIBE Discussant Nick Hope, Stanford Center for International Development 3:00 p.m.–4:00 p.m. Local investment climates and participation in global value chains Speakers David Dollar, Brookings, Ge Ying and Yu Xinding, UIBE Discussant Hubert Escaith, WTO 4:10 p.m.–5:10 p.m. Neighboring institutions matter for the competitiveness of your value chain Speaker Rodrigo Wagner, University of Chile Discussant Ma Hong, Tsinghua University 5:10 p.m.-6:00 p.m. Keynote speech: The forces driving the future of supply chains and the tradable part of the global economy Speaker Michael Spence, the Noble Prize laureate in Economics Authors’ conferences • 189 6:00 p.m.–6:30 p.m. Closing remarks Speakers Dr. David Dollar, Senior Fellow, Brookings Institution Dr. Anabel Gonzalez, Senior Director of Trade & Competitiveness Global Practice, World Bank Robert B. Koopman, Chief Economist, WTO Dr. Nadim Ahmad, Chief of Trade & Competitiveness, OECD Dr. Ricardo Melendez-Ortiz, CEO, ICTSD Wang Zhi, Professor and Director, RIGVC Second Authors’ Conference: Chapters Washington, November 28–29, 2016 Monday, November 28 8:30 a.m.–9:00 a.m. Opening Presenters Anabel Gonzalez, WBG and Robert Koopman, WTO 9:00 a.m.–10:20 a.m. Chapter 1 Historical review of the development of GVCs and analytical frameworks Presenter Satoshi Inomata, IDE–JETRO Discussant Juan Blyde, IADB and Robert Koopman, WTO 10:40 a.m.–12:00 p.m. Chapter 2 Recent trends in global trade and GVCs Presenters Zhi Wang, UIBE, Bo Meng, IDE–JETRO Discussants Gaaitzen de Vries, University of Groningen and Deborah Winkler, WBG 1:30 p.m.–2:50 p.m. Chapter 3 Accumulated trade costs and their impact on the development of GVCs Presenter Hubert Escaith, WTO Discussants Yu Xinding, UIBE and Jose G Reis, WBG 2:50 p.m.–4:10 p.m. Chapter 4 GVCs and the development agenda Presenter Nadim Ahmad, OECD Discussants Nadia Rocha, WBG and Heiwai Tang, John Hopkins University 4:30 p.m.–5:50 p.m. Chapter 5 The “middle-income trap” and upgrading along GVCs Presenter Daria Taglioni, WBG Discussants Gary Gereffi, Duke University and Qi Yinan, UIBE Tuesday, November 29 9:00 a.m.–10:20 a.m. Chapter 6 Services trade liberalization and GVCs Presenters Aaditya Mattoo and Cecilia Heuser, WBG Discussants Przemyslaw Kowalski, OECD and J. Bradford Jensen, Georgetown University 10:40 a.m.–12:00 p.m. Chapter 7 Local investment climates, institutional quality, and GVCs Presenter David Dollar, Brookings Institution Discussants Cosimo Beverelli, WTO and Rodrigo A. Wagner, Tufts University 1:30 p.m.–2:50 p.m. Chapter 8 Regional trade agreements and GVCs Presenter Michele Ruta, WBG Discussants Dazhong Cheng, Fudan University and Michael Ferrantino, WBG 3:30 p.m.–6:00 p.m. Executive summary Presenters David Dollar, Brookings Institution and Zhi Wang, UIBE Discussants All