A World Bank Group Flagship Report © 2020 International Bank for Reconstruction and Development / The World Bank 1818 H Street NW, Washington, DC 20433 Telephone: 202-473-1000; Internet: www.worldbank.org Some rights reserved 2 3 4 22 21 20 19 This work is a product of the staff of The World Bank with external contributions. The findings, interpretations, and conclusions expressed in this work do not necessarily reflect the views of The World Bank, its Board of Executive Directors, or the governments they represent. The World Bank does 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 World Bank concerning the legal status of any territory or the endorsement or acceptance of such boundaries. Nothing herein shall constitute or be considered to be a limitation upon or waiver of the privileges and immunities of The World Bank, all of which are specifically reserved. Rights and Permissions This work is available under the Creative Commons Attribution 3.0 IGO license (CC BY 3.0 IGO) http://creativecommons.org/licenses/by/3.0/igo. Under the Creative Commons Attribution license, you are free to copy, distribute, transmit, and adapt this work, including for commercial purposes, under the following conditions: Attribution—Please cite the work as follows: World Bank. 2020. World Development Report 2020: Trading for Development in the Age of Global Value Chains. Washington, DC: World Bank. doi:10.1596/978-1-4648-1457-0. License: Creative Commons Attribution CC BY 3.0 IGO Translations—If you create a translation of this work, please add the following disclaimer along with the attribution: This translation was not created by The World Bank and should not be considered an official World Bank translation. The World Bank shall not be liable for any content or error in this translation. Adaptations—If you create an adaptation of this work, please add the following disclaimer along with the attribution: This is an adaptation of an original work by The World Bank. Views and opinions expressed in the adaptation are the sole responsibility of the author or authors of the adaptation and are not endorsed by the World Bank. Third-party content—The World Bank does not necessarily own each component of the content contained within the work. The World Bank therefore does not warrant that the use of any third-party-owned individual component or part contained in the work will not infringe on the rights of those third parties. The risk of claims resulting from such infringement rests solely with you. If you wish to re-use a component of the work, it is your responsibility to determine whether permission is needed for that re-use and to obtain permission from the copyright owner. Examples of components can include, but are not limited to, tables, figures, or images. All queries on rights and licenses should be addressed to World Bank Publications, The World Bank Group, 1818 H Street NW, Washington, DC 20433, USA; e-mail: pubrights@worldbank.org. ISSN, ISBN, e-ISBN, and DOI: Softcover ISSN: 0163-5085 ISBN: 978-1-4648-1457-0 e-ISBN: 978-1-4648-1495-2 DOI: 10.1596/978-1-4648-1457-0 Hardcover ISSN: 0163-5085 ISBN: 978-1-4648-1494-5 DOI: 10.1596/978-1-4648-1494-5 Cover image: The cover image is a screen capture of an interactive visualization depicting the flow of international trade, with each dot representing US$1 billion in value. The interactive map was created by data visualization expert Max Galka, from his Metrocosm blog: http:// metrocosm.com/map-international-trade/. Used with the permission of Max Galka; further permission required for reuse. Cover design: Kurt Niedermeier, Niedermeier Design, Seattle, Washington. Interior design: George Kokkinidis, Design Language, Brooklyn, New York, and Kurt Niedermeier, Niedermeier Design, Seattle, Washington. Library of Congress Control Number: 2019952802 Contents xi Foreword xiii Preface xv Acknowledgments xvii Abbreviations xx Part I: Overview 1 Overview 2 The expansion of GVCs could stall unless policy predictability is restored 3 GVCs boost incomes, create better jobs, and reduce poverty 3 The gains from GVCs are not equally shared, and GVCs can hurt the environment 4 New technologies on balance promote trade and GVCs 4 National policies can boost GVC participation 6 Other policies can help ensure GVC benefits are shared and sustainable 7 International cooperation supports beneficial GVC participation 8 Notes 9 References 12 Part II: Global value chains: What are they? 14 Chapter 1: The new face of trade 16 What is a global value chain? 19 The evolution of GVC participation 23 How are GVCs distributed across regions? 26 Which countries have accounted for most of the GVC expansion? 26 How are GVCs distributed across sectors? 30 A few large trading firms account for most GVC trade 34 Notes 34 References 36 Chapter 2: Drivers of participation 39 Factor endowments matter 46 Market size matters 49 Geography matters 53 Institutional quality matters 54 Transitioning up the GVC typology 59 Notes 60 References Contents | iii Part III: What are the effects of GVCs? 64  66 Chapter 3: Consequences for development 68 Economic growth 76 Employment 80 Poverty and shared prosperity 82 Distribution of gains 90 The gender gap 92 Taxation 93 Notes 96 References 102 Chapter 4: Macroeconomic implications 103 Synchronizing economic activity 105 Propagating shocks 106 Synchronizing inflation 107 Reducing the effect of devaluations 111 Mitigating trade diversion and increasing trade 112 The return of protectionism 115 Notes 115 References 118 Chapter 5: Impact on the environment 119 Scale effects of trade and growth 124 Changes in the composition of production 128 Relational GVCs and production techniques 130 Green goods 132 Notes 132 References 136 Chapter 6: Technological change 137 Declining trade costs 146 New products 147 Automation anxiety 153 Openness and innovation 153 Export-led industrialization 154 Notes 154 References Part IV: What domestic policies facilitate fruitful 158  participation? 160 Chapter 7: Policies to enhance participation 162 Facilitating participation 175 Policies to enhance benefits 186 Policies for upgrading 189 Notes 190 References 194 Chapter 8: Policies for inclusion and sustainability 195 Sharing the gains 202 Managing adjustment 204 Environmental sustainability 209 Notes 210 References iv | Contents Part V: How can international cooperation help? 214  216 Chapter 9: Cooperation on trade 218 The case for cooperation 221 Deepening trade cooperation 235 Notes 236 References 238 Chapter 10: Cooperation beyond trade 239 Taxes 242 Regulation 246 Competition policy 251 Infrastructure 254 Notes 255 References 259 Appendix A: Databases used in this Report 265 Appendix B: Glossary Boxes 1.1 17 Defining global value chains 3.5 79 GVC participation can lead to indirect 1.2 17 Measuring global value chains welfare improvements for women 1.3 22 Types of GVC participation 3.6 82 Does GVC participation lead to human capital accumulation? 1.4 30 A firm-level approach to GVCs 3.7 89 Home-based work in GVCs 2.1 38 Vietnam’s integration in the electronics GVC 4.1 105 The Japanese earthquake and the costs of supply chain disruptions 2.2 40 Modeling results on the drivers of GVC participation 4.2 108 Blunting the effects of devaluation on Turkey’s exports 2.3 45 Sharing suppliers: How foreign firms benefit domestic firms 4.3 110 Trade imbalances in using value- added data 2.4 45 How liberalizing trade and FDI helped China move up in GVCs 5.1 125 The ban on plastics by China disrupted the waste GVC 2.5 49 Trade preferences as catalytic aid? 5.2 127 Virtual water 2.6 55 PTAs and GVCs: The role of rules of origin 5.3 128 Toward sustainable fashion 2.7 57 Most important determinants of GVC 5.4 130 Demanding environmental standards participation, by taxonomy group and in GVC upstream firms region 6.1 139 Digital innovation and agricultural 3.1 71 Dynamic estimations of the trade relationship between GVC 6.2 143 GVC linkages and cross-border participation and per capita income connections between people move growth together 3.2 73 Mining GVCs: New opportunities 6.3 150 Fully automating the production of and old obstacles for local suppliers hearing aids from developing countries 6.4 152 Mexico and technological change 3.3 75 Assessing outcomes of GVC 7.1 164 Determinants of efficiency-seeking participation using event studies investment 3.4 76 Skills and upgrading in Cambodia’s 7.2 169 Foreign services firms in India’s apparel value chain manufacturing value chains Contents | v 7.3 176 Local content requirements are a 8.3 201 Transparency promotes compliance mismatch in the global auto industry with labor standards and improves 7.4 178 Supplier development programs working conditions help deliver inclusive, sustainable 8.4 206 Cost-effectiveness and equitability of GVCs environmental regulation 7.5 180 Building a workforce with industry- 8.5 208 Green industrial parks support specific skills: Penang Skills sustainable production and attract Development Centre better investors 7.6 183 Clarifying the terminology: SEZs 9.1 218 Special and differential treatment for versus industrial parks developing countries 7.7 184 Comparing SEZ experiences: China, 9.2 219 A story of the demise of most-favored- India, and Sub-Saharan Africa nation status foretold? 7.8 188 Costa Rica moves into the medical 9.3 231 The impact of Brexit on GVC trade devices GVC 9.4 234 How the African Continental Free 8.1 197 Taking advantage of comparative Trade Area can support integration advantage: Agribusiness GVCs deliver into GVCs more and better jobs in Côte d’Ivoire 10.1 252 International cooperation on and Rwanda transport infrastructure 8.2 199 A tale of two economic zones: Initiatives to promote women’s employment in garment GVCs in Bangladesh and Jordan Figures O.1 2 GVC trade grew rapidly in the 1990s 1.7 24 GVC activities increased globally and but stagnated after the 2008 global regionally from 1990 to 2015 financial crisis 1.8 25 Global production networks are O.2 3 GDP per capita grows most rapidly organized around three main regions, when countries break into limited 2018 manufacturing GVCs 1.9 26 A handful of countries drove global O.3 4 Automation in industrial countries GVC expansion from 1990 to 2015 has boosted imports from developing 1.10 27 GVC participation by sector, 1995 and countries 2011 O.4 5 Transitioning to more sophisticated 1.11 28 A handful of sectors drove global GVC participation in GVCs: Some expansion from 1995 to 2011 examples of national policy 1.12 29 Services are playing a growing role 1.1 16 Where do bicycles come from? in GVCs 1.2 19 GVC trade grew rapidly in the 1990s 1.13 29 GVCs expanded in both the but stagnated after the 2008 global agriculture and food industries from financial crisis 1990 to 2015 1.3 20 The ICT revolution spurred the 1.14 32 Firms that both import and export emergence of GVCs dominate GVC participation 1.4 20 From 1948 to 2016, tariffs dropped 1.15 33 Foreign direct investment thanks to multilateral and regional accompanied the fragmentation of trade agreements production from 1970 to 2018 1.5 21 Country transitions between B2.1.1 38 Vietnam’s backward GVC integration different types of GVC participation, increased from 2000 to 2015 as 1990–2015 tariffs declined and foreign direct 1.6 23 Average backward and forward GVC investment expanded participation across taxonomy groups vi | Contents B2.2.1 40 What explains backward and forward 3.8 79 Worldwide, GVC firms hire more GVC participation? women than non-GVC firms B2.2.2 42 What explains a country-sector’s 3.9 81 The boost to wages is largest in GVC participation levels and gross countries after they first enter limited exports? manufacturing GVCs 2.1 43 Countries specializing in limited 3.10 81 GVC participation is associated with manufacturing rely on low labor poverty reduction costs, and countries specializing in 3.11 82 In municipalities in Mexico, the commodities derive almost a fifth of expanded presence of GVC firms GDP from natural resources is more strongly associated with 2.2 43 Increases in labor costs and capital poverty reduction than the presence stock accompany upgrading in GVCs of firms that export only or import B2.3.1 45 In Bangladesh, local suppliers grew only as FDI grew from 1985 to 2003 3.12 83 In Vietnam, poverty reduction was B2.4.1 45 Domestic value added in exports greater in locations with a higher from China increased from 2000 to presence of GVC firms 2007 3.13 84 Rising income inequality is a greater 2.3 46 FDI increases and tariff declines problem for countries breaking accompany GVC upgrading into the innovation stages of GVC engagement 2.4 48 Manufacturing tariffs are high and preferential trading partners few in 3.14 84 A majority worldwide views trade countries connected to commodity and international business ties GVCs positively, but skepticism grew from 2002 to 2014 B2.5.1 50 Four stories of AGOA apparel exports from Africa 3.15 85 Increasing GVC participation is associated with rising markups 2.5 52 Connectivity is associated with in developed countries but falling specialization in more advanced markups in developing countries GVCs 3.16 85 In Ethiopia, firms entering GVCs 2.6 52 Improving customs and introducing experience greater declines in electronic systems are as important markups, 2000–2014 as infrastructure for African trade 3.17 86 GVCs have contributed to the B2.6.1 56 Mauritius’s exports of apparel to the declining labor share within United States, by origin of fabric, countries 2001–15 3.18 90 Women are more likely to be 3.1 69 GVC participation is associated with production workers and less likely to growth in exports and incomes own or manage GVC firms 3.2 69 GVC participation is associated with 3.19 91 Gender equality in business growth in productivity regulations ensures that women are 3.3 70 Firms that both export and import are more fairly rewarded more productive 3.20 92 Corporate income tax rates have B3.1.1 71 GVC trade is associated with larger declined by almost 50 percent since per capita income than non-GVC 1990 trade 3.21 93 As a share of GDP, non-OECD 3.4 72 GVC firms with relationships receive countries lose the most from profit more assistance shifting 3.5 74 GDP per capita grows most rapidly 4.1 104 In all income groups, countries’ when countries break into limited economic activity has become more manufacturing GVCs synchronized since the mid-1990s 3.6 77 In Ethiopia, GVC firms are relatively 4.2 104 Greater synchrony of economic more capital-intensive but their activity is associated with GVCs employment is increasing fastest 4.3 106 The synchrony of inflation increased 3.7 78 In Mexico, employment expansion between 1988 and 2010 is more strongly linked to GVC expansion than non-GVC trade Contents | vii 4.4 107 GVCs are associated with greater 6.4 146 Globally, the number and trade share inflation synchrony in some of new products increased from 1996 countries to 2017 4.5 107 Trade in intermediate inputs 6.5 147 Robot adoption is greater in high- increased the weight of global factors income countries and in sectors in in inflation formation from 1983 to which tasks are easily automated 2006 6.6 149 Automation in industrial countries 4.6 108 Export boosts tend to coincide with has boosted imports from developing import boosts—more now than 30 countries years ago 6.7 150 Trade in hearing aids increased with 4.7 109 With GVCs, devaluations can have the adoption of 3D printing in 2007 complex consequences 6.8 150 Higher robot density is associated 4.8 109 GVCs dampen the reaction of export with lower shares of income for labor volumes to currency movements 6.9 151 Change in U.S. employment in robot- B4.3.1 110 Computing bilateral trade balance intensive industries, by occupation, in gross exports or in value-added 1990–2010 exports matters B6.4.1 152 Automation reduces the wage 4.9 114 The multilateral dimension of the employment of high school graduates U.S.–China trade war 7.1 163 Manufacturing labor costs are out 4.10 115 Impact of U.S. tariffs on imports from of line with national income levels China in Sub-Saharan Africa but not in 5.1 120 The complexity of producing the Bangladesh Pedego Conveyor electric commuter B7.1.1 165 MNCs involved in efficiency-seeking bike in Vietnam with parts from all FDI are more selective over the world 7.2 166 Better-quality investment promotion 5.2 121 Production-related CO2 emissions agencies attract more FDI inflows drop in countries that recently 7.3 167 Nontariff measure use increases by transitioned into advanced GVCs and development status innovation hubs 7.4 171 Shipping delays matter more for 5.3 122 International shipping emissions are products with complex value chains increasing 7.5 171 Customs reform can reduce delay 5.4 124 The world produced 50 million metric and expand imports: Evidence from tons of e-waste in 2018 Albania 5.5 126 U.S. output has increasingly shifted 7.6 172 Contract enforcement intensity is away from polluting goods, but higher in services sectors: Evidence imports have done so even faster from the United States B5.3.1 128 Swedish lead firms in apparel and 7.7 173 Share of “other business services” in textiles produce a lot of value added intermediate inputs is low in poor with little CO2, and their suppliers countries produce a lot of CO2 with little value added 7.8 174 Certification had long-lasting effects on quality in Mali’s cotton sector 6.1 142 Large platform companies are concentrated in North America and 7.9 175 Subsidies account for more than Asia half of distortionary trade policy instruments worldwide B6.2.1 143 Relationship of exports and GVC participation to online foreign 7.10 177 The share of locally supplied inputs connections in GVCs varies by sector and country 6.2 144 From 2013 to 2015, U.S. exports 7.11 179 Lack of financing impedes low- to Latin America through eBay income country suppliers the most increased after the introduction of from entering or moving up in GVCs machine translation 7.12 181 Managerial know-how is associated 6.3 145 Effects of an e-commerce program with greater GVC participation in on the number of buyers and online Mexico transactions in Chinese villages viii | Contents 7.13 187 Different policy priorities underpin 9.6 225 Low-income countries are penalized the transitions between types of GVC by tariff escalation both at home and participation in their destination markets B7.8.1 188 Costa Rica’s medical device exports 9.7 226 A majority of PTAs protect have increased in volume and investors from discrimination and sophistication since 2000 expropriation B8.3.1 201 Working conditions improved in 9.8 228 Agricultural producer support apparel sector firms participating in converged across some high-income the ILO-IFC Better Work Vietnam and lower-income countries from program 2000 to 2017 8.1 203 Denmark invests more than other 9.9 230 Deep trade agreements are associated OECD countries to support workers with GVC integration 8.2 204 Industrial development was uneven B9.3.1 231 Trade impacts of membership in across regions of Mexico during the the European Union on the United period of strong international market Kingdom and other EU members integration B9.4.1 235 AfCFTA members benefit from B8.5.1 208 The number of eco-industrial parks reductions in tariffs, nontariff grew rapidly from 1985 to 2015 measures, and implementation of B9.2.1 219 Shifts in trade shares and changes in the World Trade Organization’s Trade policy stances of the United Kingdom Facilitation Agreement and the United States since 1800 10.1 243 Regulatory commitments by 9.1 221 Attitudes toward trade differ in the exporters can be exchanged for sluggish North and the dynamic import liberalization commitments South 10.2 244 Countries’ restrictions on data flows 9.2 221 Corporate tax rates and personal increased from 2006 to 2016 income tax rates for the top 10.3 247 Cartel episodes and significant 1 percent have fallen, but the rate for overcharges have been observed the median worker increased in 65 across all regions economies between 1980 and 2007 10.4 248 The European Commission has 9.3 223 Tariffs have been liberalized across imposed large fines on car parts sectors, but pockets of protection cartels since 2013 remain B10.1.1 252 Impact of China’s Belt and Road 9.4 223 There is room for further liberalization Initiative transport projects with and 9.5 224 Most countries impose higher tariffs without input–output linkages on semifinished and finished goods Maps O.1 2 All countries participate in GVCs— 5.1 131 Supply chain of a solar photovoltaic but not in the same way company 1.1 21 All countries participate in GVCs— 6.1 140 4G network coverage, 2018 but not in the same way 6.2 142 Top e-commerce platforms, by traffic 2.1 53 Growth in Internet density and share, 2019 exporter firm density across 6.3 148 A substantial share of exports from provinces in China, 1999 and 2007 developing countries is in goods that 3.1 78 In Vietnam, employment expansion can be produced by robots was linked to GVC firms 7.1 169 Services trade remains restricted in 3.2 88 In Mexico and Vietnam, GVCs are many countries spatially concentrated B5.2.1 127 Global water savings associated with international trade in agricultural products, 1996–2005 Contents | ix Tables 2.1 48 South Asian countries impose higher 6.1 141 Ten largest global companies, by barriers to trade on each other market capitalization, 2011, 2015, and B2.7.1 57 Backward GVC participation and 2019 determinants, by taxonomy group 9.1 222 Policy rationale, externalities, and B2.7.2 58 Backward GVC participation and cooperative solutions determinants, by region and group of B9.3.1 232 Changes in the United Kingdom’s countries bilateral trade with the European 3.1 91 Sample of results from case studies Union under three Brexit scenarios on gender in specific GVCs 9.2 233 Existing trade agreements in Africa B5.3.1 129 Total reported savings generated by are relatively shallow the Sweden Textile Water Initiative in 10.1 246 Regulation of international transfers its five partner countries, 2015–17 of personal information, by privacy 5.1 132 Estimation of value added at stages regime of the supply chain of a solar photovoltaic module x | Contents Foreword Around the world, the process of delivering goods and services to consumers has become spe- cialized to a degree no one could have ever imagined. Businesses focus on what they do best in their home markets and outsource the rest. Samsung makes its mobile phones with parts from 2,500 suppliers across the globe. One country—Vietnam—produces more than a third of those phones, and it has reaped the benefits. The provinces in which the phones are produced, Thai Nguyen and Bac Ninh, have become two of the richest in Vietnam, and poverty there has fallen dramatically as a result. The face of global trade has been transformed in the three decades since the World Bank’s last major World Development Report on the subject. Until 2008, global value chains (GVCs) expanded rapidly. The expansion was revolutionary for many poorer countries, which boosted growth by joining a GVC, thereby eliminating the need to build whole industries from scratch. The experience of the last three decades has proven that it pays to specialize. Yet GVCs are at a crossroads. Their growth has leveled off since 2008, when GVCs peaked at 52 percent of global trade. The reasons are complex. Slowing global growth and investment are one factor. And value chains have matured, making further specialization more challeng- ing. Meanwhile, the push toward international trade liberalization has stalled. The growth of automation and other labor-saving technologies such as 3D printing may encourage countries to reduce production abroad. Unless trade liberalization is reinforced, value chains are unlikely to expand. Under the circumstances, do GVCs still offer developing countries a clear path to progress? That’s the main question explored in the 2020 World Development Report. And the answer is yes: developing countries can achieve better outcomes by pursuing market-oriented reforms spe- cific to their stage of development. This Report offers a detailed perspective on GVCs. It covers not only the degree to which they contribute to economic growth and poverty reduction, but also the extent to which they lead to inequality and environmental degradation. It discusses how new technologies are reshaping trade, finding that automation will help rather than hurt trade. It also raises concerns about the inadequacies in the global trading system that are fueling disagreements among nations. In particular, the Report highlights what can be done by countries that have been largely left out of the GVC revolution. Important steps such as speeding up customs procedures and reducing border delays can yield big benefits for countries making the transition from simply exporting commodities to basic manufacturing. Strengthening the rule of law reinforces trade as well. Also helpful are investments that improve connectivity by modernizing communica- tions and roads, railways, and ports. Liberalizing road, sea, and air transport is also important, and it is often less costly. In the meantime, knowledge and services have become integral to global production, delivering important benefits to developing countries through the supply chain. In Colom- bia, a program led by a multinational firm induced suppliers to upgrade their coffee farms while planting trees and incorporating more efficient and sustainable practices. About 80,000 Foreword | xi farmers and 1,000 villages benefited from the program: the quality of coffee improved, while farmers’ profits increased by 15 percent.  Overall, participation in global value chains can deliver a double dividend. First, firms are more likely to specialize in the tasks in which they are most productive. Second, firms are able to gain from connections with foreign firms, which pass on the best managerial and technolog- ical practices. As a result, countries enjoy faster income growth and falling poverty. All countries stand to benefit from the increased trade and commerce spurred by the growth of GVCs.  David R. Malpass President The World Bank Group xii | Foreword Preface The growth of international trade and the expansion of global value chains (GVCs) over the last 30 years have had remarkable effects on development. Incomes have risen, productivity has gone up—particularly in developing countries—and poverty has fallen. The fragmentation of production and knowledge transfer inherent in GVCs are in no small part responsible for these advances. Hyperspecialization by firms at different stages of value chains enhances efficiency and productivity, and durable firm-to-firm relationships foster technology transfer and access to capital and inputs along value chains. GVCs account for around half of world trade today. At this moment, however, there is reason to worry that this trade-led path to development is under threat. Although trade bounced back after the global financial crisis of 2008, the high growth rates of the 1990s and 2000s have remained elusive. GVC trade—trade in intermediate products—also stalled in 2008, with only modest, intermittent periods of growth since. There are many reasons for this shift, but one is that trade reform has languished and in some cases is even being reversed. Countries can do much on their own to reinvigorate world trade and GVC expansion. With that in mind, this Report sets out a comprehensive domestic agenda for governments: invest- ments in connectivity, improvements in business climate, and unilateral reductions in trade and investment barriers. But there is much that countries need to do together to improve the current system. Coor- dinated trade liberalization is overdue in agriculture and services, the rules applied to foreign investment are uneven, and subsidies and state-owned enterprises are distorting competition. Unfortunately, international cooperation, too, has begun to falter. Many people are disen- chanted with free trade. Some communities have experienced declining wages and unemploy- ment. Businesses are complaining about the limitations of the current multilateral system in dealing with their concerns about lack of access to large markets, the increasing use of “behind- the-border” measures, and “unfair” competition. Governments are inclined to respond by using trade policy as a tool for social protection and to address inadequacies in the current trade rules. This Report argues that reinvigorating the international trade system will require gov- ernments in certain advanced countries to first look inward to address the discontent and inequality associated with openness. More generally, advanced economies need to rethink the priorities of the welfare state to better help workers adjust to structural change. Developing countries as well need to expand social assistance and improve compliance with labor regulations in order to extend the jobs and earnings gains from participation in GVCs to more people across society. They also need to take steps to ensure that their domestic firms benefit from knowledge transfer from lead global firms. Finally, all countries need to ensure that the growth associated with trade does not lead to environmental degradation. Meanwhile, governments need to cooperate with one another beyond the traditional trade issues to ensure that trade and GVCs can deliver for development. Cooperation on corporate taxes will enable governments to better tax capital in a global, digitalized economy, so that they Preface | xiii have the resources to finance infrastructure projects and social policies. Improved cooperation on competition issues is needed to ensure that firms enjoy a level playing field globally. And finally, new models of cooperation are needed for data flows to strike a balance between the privacy of citizens and the needs of business and innovators. The expansion of trade and GVCs is at an inflection point. There is still time to reinvigorate growth, trade, and GVCs. Trade is vital for development, but it needs rules to function smoothly. And those rules require cooperation by governments. This Report offers governments a road map for action. Pinelopi Koujianou Goldberg Chief Economist The World Bank Group xiv | Preface Acknowledgments This year’s World Development Report (WDR) was prepared by a team led by co-directors Caroline Freund and Aaditya Mattoo (World Bank) and Pol Antràs (Harvard University). Daria Taglioni served as task team leader of the project and as a member of the Report’s leadership. Overall guidance was provided by the chief economist of the World Bank, Pinelopi Koujianou Goldberg. The Report is sponsored by the Bank’s Development Economics Vice Presidency. The core team was composed of Erhan Artuc, Paulo Bastos, Davida Connon, François de Soyres, Thomas Farole, Ana Margarida Fernandes, Michael J. Ferrantino, Bernard Hoekman, Claire H. Hollweg, Melise Jaud, Hiau Looi Kee, Bob Rijkers, and Deborah Winkler. Members of the extended team—Jessie Coleman, Jan De Loecker, Leonardo Iacovone, Kåre Johard, Madina Kukenova, Michele Mancini, Alen Mulabdic, Nadia Rocha, Michele Ruta, Marijn Verhoeven, Michael D. Wong, and Douglas Zhihua Zeng—provided invaluable contributions to the Report. Sources of additional input were Emma Aisbett, Emmanuelle Auriol, Gaëlle Balineau, Christopher Barrett, Benoit Blarel, Alessandro Borin, Fabrizio Cafaggi, Jieun Choi, Ileana Cristina Constantinescu, Wim Douw, Roberto Echandi, Jakob Engel, Dominik Englert, Marianne Fay, Vivien Foster, Sebastián Franco-Bedoya, Emiko Fukase, Gary Gereffi, Tania Priscilla Begazo Gomez, Stephane Hallegatte, Armando Heilbron, Dirk Heine, Etienne Raffi Kechichian, Jana Krajčovičovà, Peter Kusek, Somik Lall, Arik Levinson, Yan Liu, Rocco Macchiavello, Maryla Maliszewska, Julien Martin, Denis Medvedev, Josepa Miquel-Florensa, Antonio Nucifora, Carlo Pietrobelli, Obert Pimhidzai, Christine Qiang, Tom Reardon, Kirstin Ingrid Roster, Gianluca Santoni, Abhishek Saurav, Kateryna Schroeder, Victor Steenbergen, Michael A. Toman, and Gonzalo Varela. Research assistance was provided by Vicky Chemutai, Alexandre Gaillard, Chiara Liardi, Julien Maire, Mitali Nikore, Nicolás Gómez Parra, Xiomara Pulido Ramírez, Juan Miguel Jiménez Riveros, Alejandro Forero Rojas, Maria Filipa Seara e Pereira, Guillaume Sublet, Nicolás Santos Villagrán, and the World Bank’s Digital Development program. The team would like to thank the following colleagues for their guidance during preparation of the Report: Rabah Arezki, Asli Demirgüç-Kunt, Shanta Devarajan, Simeon Djankov, Deon Filmer, Mary Hallward-Driemeier, Daniel Lederman, William Maloney, Martin Rama, Halsey Rogers, Hans Timmer, and Albert Zeufack. The Macroeconomics, Trade, and Investment Global Practice of the Equitable Growth, Finance, and Institutions (EFI) Vice Presidency provided the Report team with support. The team also benefited at an early stage from consultations on emerging themes with experts from the Bank of Italy, European Commission, French Development Agency, German Agency for International Cooperation (GIZ), German Federal Ministry for Economic Cooper- ation and Development (BMZ), International Trade Centre, Japan International Cooperation Agency, Organisation for Economic Co-operation and Development, Swedish Chamber of Com- merce, Swedish International Development Cooperation Agency, U.K. Department for Inter- national Development, United Nations Industrial Development Organization, U.S. Agency Acknowledgments | xv for International Development, and World Trade Organization. An event kindly organized by GIZ and BMZ gave the WDR team a unique opportunity to discuss the Report’s themes with a diverse range of experts from government, civil society, and the private sector. Bruce Ross-Larson provided developmental guidance in drafting the Report, which was edited by Sabra Ledent and proofread by Gwenda Larsen. Kurt Niedermeier was the prin- cipal graphic designer, with support from Bill Pragluski and Patrick Ibay. Mikael Reventar, Anushka Thewarapperuma, and Roula Yazigi, together with Chisako Fukuda, offered guid- ance, services, and support on communication and dissemination. Special thanks go to Stephen Pazdan, who coordinated and oversaw production of the Report and to the World Bank’s Formal Publishing Program. The team would also like to thank Mary Fisk, who facil- itated translation of the overview; Patricia Katayama, who oversaw the overall publication process; and Deb Barker, who managed the printing and electronic conversions of the book and its overview booklets. The team would also like to thank Marcelo Buitron, Michelle Chester, María del Camino Hurtado, Rashi Jain, Gabriela Calderon Motta, Alejandra Ramon, and Consuelo Jurado Tan for fulfilling their coordinating roles. Background and related research, along with dissemination, have been generously sup- ported by the KDI School Partnership trust fund, the World Bank’s Knowledge for Change Program (KCP, a multidonor trust fund), Strategic Research Program, Umbrella Facility for Trade, and Multidonor Trust Fund for Trade and Development. During preparation of the Report, government officials, researchers, and representa- tives of civil society organizations (CSOs) attended consultations in Belgium, Chile, China, France, Germany, India, Poland, Sweden, Turkey, the United Kingdom, and the United States. Participants were drawn from many more countries as well. In addition, a diverse group of CSOs participated in two CSO Forum sessions on this Report held during the 2019 World Bank/ International Monetary Fund Spring Meetings and in an e-forum held in March 2019. The team is grateful to these CSOs for their input and to those who took part in these events for their helpful comments and suggestions. Special thanks go to those organizations and individ- uals who provided written comments and engaged directly with the team, including the Con- sumer Unity and Trust Society International, International Trade Union Confederation, ISEAL Alliance, Save the Children, and Women in Informal Employment: Globalizing and Organizing (WIEGO). In addition, the team is grateful for those who submitted comments in response to blogs posted on the topic. Steve Commins provided support during consultations with think tanks and CSOs. Further information on these events is available at http://www.worldbank.org /wdr2020. The team is grateful as well to the many World Bank colleagues who provided written com- ments during the formal Bank-wide review process. Those comments proved to be invaluable guidance at a crucial stage in the Report’s production. Team members would also like to thank their families for their support during the prepara- tion of this Report. Finally, the team apologizes to any individuals or organizations that contributed to this Report but were inadvertently omitted from these acknowledgments. xvi | Acknowledgments Abbreviations AfCFTA African Continental Free Trade Area AGOA African Growth and Opportunity Act ALMP active labor market policy APEC Asia–Pacific Economic Cooperation ASCM Agreement on Subsidies and Countervailing Measures ASEAN Association of Southeast Asian Nations AVE ad valorem equivalent BCRs Binding Corporate Rules BEAT Base Erosion and Anti-abuse Tax BEC Broad Economic Categories BEPS base erosion and profit shifting BIT bilateral investment treaty BPO business processing outsourcing BRI Belt and Road Initiative CBPRs Cross-Border Privacy Rules CCAC Competition and Consumer Affairs Commission (Guyana) CLOUD Act Clarifying Lawful Overseas Use of Data Act CO2 carbon dioxide COMESA Common Market for Eastern and Southern Africa CORFO Chilean Innovation Agency CPEA Cross-border Privacy Enforcement Arrangement CPI consumer price index CPTPP Comprehensive and Progressive Agreement for Trans-Pacific Partnership CVDs countervailing duties DBCFT destination-based cash flow tax DLTs distributed ledger technologies DR–CAFTA Dominican Republic–Central America Free Trade Agreement DST digital services tax DTA domestic tariff area DTRI Digital Trade Restrictiveness Index EAC East African Community EBA Everything but Arms ECOWAS Economic Community of West African States ECTEL Eastern Caribbean Telecommunications Authority EEA European Economic Area EEC European Economic Community EIP eco-industrial park EKC environmental Kuznets curve EPZ export processing zone ESG environmental, social, and governance Abbreviations | xvii EU European Union FAO Food and Agriculture Organization (of the UN) FDI foreign direct investment fintech financial technology FTA free trade agreement FVA foreign value added GATS General Agreement on Trade in Services GATT General Agreement on Tariffs and Trade GDP gross domestic product GDPR General Data Protection Regulation GHG greenhouse gas GLoBE global antibase erosion Gm3 billion cubic meters GVC global value chain HCI Human Capital Index HS Harmonized System ICT information and communication technology IFC International Finance Corporation ILO International Labour Organization IMF International Monetary Fund IMO International Maritime Organization INEGI National Institute of Statistics and Geography (Mexico) IP intellectual property IPA investment promotion agency IPRs intellectual property rights ISCO International Standard Classification of Occupations ISDS investor-state dispute settlement ISIC International Standard Industrial Classification ISO International Organization for Standardization IT information technology ITC International Trade Centre ITKIB Istanbul Textile and Apparel Exporter Associations LDCs least developed countries LPI logistics performance index m3 cubic meters MENA Middle East and North Africa MFA Multifibre Arrangement MFN most favored nation MIDP Motor Industry Development Programme MMT million metric tons MNC multinational corporation MNE multinational enterprise MOU memorandum of understanding MRIO multiregion input–output NAFTA North American Free Trade Agreement NEC not elsewhere classified NEET not in employment, education, or training NOx nitrogen oxides NTB nontariff barrier NTM nontariff measure OECD Organisation for Economic Co-operation and Development xviii | Abbreviations PEA Privacy Enforcement Authority PHE pollution haven effect PHH pollution haven hypothesis PPP purchasing power parity PSDC Penang Skills Development Centre PTA preferential trade agreement PV photovoltaic PWT Penn World Table QIZ qualified industrial zone R&D research and development RTA regional trade agreement RVC regional value chain SACU Southern African Customs Union SAR special administrative region SCC Standard Contractual Clause SDG Sustainable Development Goal SDT special and differential treatment SEV Samsung Electronics Vietnam SEVT Samsung Electronics Vietnam-Thai Nguyen SEZ special economic zone Sida Swedish International Development Cooperation Agency SMEs small and medium enterprises SO2 sulfur dioxide SOE state-owned enterprise SPS sanitary and phytosanitary STE state trading enterprise STWI Sweden Textile Water Initiative System-GMM System Generalized Method of Moments TBT technical barrier to trade TEN-T Trans-European Transport Network TFA textile, footwear, and apparel TFA Trade Facilitation Agreement TFP total factor productivity TiVA Trade in Value Added TRIMs Agreement on Trade-Related Investment Measures TRIPS Agreement on Trade-Related Aspects of Intellectual Property Rights UN United Nations UNCTAD United Nations Conference on Trade and Development UNIDO United Nations Industrial Development Organization USAID U.S. Agency for International Development USCBTTA U.S.–Cambodia Bilateral Textile Trade Agreement USMCA United States–Mexico–Canada Agreement VAT value added tax WAEMU West African Economic and Monetary Union WDI World Development Indicators (database) WEO World Economic Outlook (database) WIOD World Input–Output Database WITS World Integrated Trade Solution (database) WTO World Trade Organization All dollar amounts are U.S. dollars unless otherwise indicated. Abbreviations | xix PART I Overview World Development Report 2020: Trading for Development in the Age of Global Value Chains What is a global value chain (GVC)? A global value chain breaks up the production process across countries. Firms specialize in a specific task and do not produce the whole product. Exports Exports Exports for consumption Raw materials Parts and components Border Border Border Finished goods Services inputs Semifinished goods How do GVCs work? Interactions between firms typically involve durable relationships. Economic fundamentals drive countries’ participation in GVCs. But policies matter—to enhance participation and broaden benefits. Drivers Outcomes p e ra t i o n Policies: Coo So ci • al ty vi an i ct d ne en d I ne E nd an hy q en ness • Con ow h v i ro p ra bs jo t ua w og m l it G ro en y Ge nmental p ts ecializ rsp at En pe i ct rt y Ins e n vi siz io v e nm ro tit Hy on ut e Po u t io Op rk en r ns d Ma t re ot e s: ct GVC ie i o n lic Po s Fir ip sh -t m o- n fir m relatio Overview GVCs can continue to boost growth, create better jobs, and reduce poverty— provided that developing countries undertake deeper reforms and industrial countries pursue open, predictable policies. I nternational trade expanded rapidly after 1990, 3D printing could draw production closer to the powered by the rise of global value chains (GVCs). consumer and reduce the demand for labor at home This expansion enabled an unprecedented con- and abroad. Second, trade conflict among large coun- vergence: poor countries grew faster and began to tries could lead to a retrenchment or a segmentation catch up with richer countries. Poverty fell sharply. of GVCs. These gains were driven by the fragmentation What does all this mean for developing countries of production across countries and the growth of seeking to link to GVCs, acquire new technologies, connections between firms. Parts and components and grow? Is there still a path to development through began crisscrossing the globe as firms looked for effi- GVCs? Those are the central questions explored in ciencies wherever they could find them. Productivity this Report. It examines the degree to which GVCs and incomes rose in countries that became integral have contributed to growth, jobs, and reduced pov- to GVCs—Bangladesh, China, and Vietnam, among erty—but also to inequality and environmental degra- others. The steepest declines in poverty occurred in dation. It spells out how national policies can revive precisely those countries. trade growth and ensure that GVCs are a force for Today, however, it can no longer be taken for development rather than divergence. Finally, it iden- granted that trade will remain a force for prosperity. tifies inadequacies in the international trade system Since the global financial crisis of 2008, the growth of that have fomented disagreements among nations trade has been sluggish, and the expansion of GVCs and provides a road map to resolving them through has slowed. The last decade has seen nothing like the greater international cooperation. transformative events of the 1990s—the integration This Report concludes that GVCs can continue to of China and Eastern Europe into the global economy boost growth, create better jobs, and reduce poverty, and major trade agreements such as the Uruguay provided that developing countries undertake deeper Round and the North American Free Trade Agreement reforms and industrial countries pursue open, pre- (NAFTA). dictable policies. Technological change is likely to be At the same time, two potentially serious threats more of a boon than a curse for trade and GVCs. The have emerged to the successful model of labor- benefits of GVC participation can be widely shared intensive, trade-led growth. First, the arrival of and sustained if all countries enhance social and labor-saving technologies such as automation and environmental protection. Overview | 1 Figure O.1  GVC trade grew rapidly in lower trade barriers induced manufacturers to extend the 1990s but stagnated after the 2008 production processes beyond national borders (figure global financial crisis O.1). GVC growth was concentrated in machinery, electronics, and transportation, and in the regions 55 GVC share of global trade (%) specializing in those sectors: East Asia, North America, and Western Europe. Most countries in these regions 50 participate in complex GVCs, producing advanced manufactures and services, and engage in innovative 45 activities (map O.1). By contrast, many countries in Africa, Latin America, and Central Asia still produce 40 commodities for further processing in other countries. In recent years, however, trade and GVC growth 35 have slowed (figure O.1). One reason is the decline in overall economic growth, and especially investment. 30 Another reason is the slowing pace and even reversal 70 75 80 85 90 95 00 05 10 15 of trade reforms. Furthermore, the fragmentation of 19 19 19 19 19 19 20 20 20 20 Sources: WDR 2020 team, using data from Eora26 database; Borin and production in the most dynamic regions and sectors Mancini (2019); and Johnson and Noguera (2017). See appendix A for a has matured. China is producing more at home.1 In description of the databases used in this Report. the United States, a booming shale sector reduced Note: See figure 1.2 in chapter 1 for details. Unless otherwise specified, GVC participation measures used in this and subsequent figures throughout the oil imports by one-fourth between 2010 and 2015 and Report follow the methodology from Borin and Mancini (2015, 2019). slightly reduced the incentives to outsource manufac- turing production.2 The expansion of GVCs could Recent increases in protection could also affect the stall unless policy predictability evolution of GVCs. Protectionism could induce reshor- ing of existing GVCs or their shifts to new locations. is restored Unless policy predictability is restored, any expansion GVCs have existed for centuries. But they grew swiftly of GVCs is likely to remain on hold. When future from 1990 to 2007 as technological advances—in trans- access to markets is uncertain, firms have an incentive portation, information, and communications—and to delay investment plans until uncertainty is resolved. Map O.1  All countries participate in GVCs—but not in the same way GVC linkages, 2015 Low participation Limited commodities High commodities Limited manufacturing Advanced manufacturing and services Innovative activities Data gaps IBRD 44640 | AUGUST 2019 Source: WDR 2020 team, based on the GVC taxonomy for 2015 (see box 1.3 in chapter 1). Note: The type of a country’s GVC linkages is based on (1) the extent of its GVC participation, (2) its sectoral specialization in trade, and (3) its engagement in innovation. Details are provided in figure 1.6 in chapter 1. 2 | World Development Report 2020 GVCs boost incomes, create Figure O.2  GDP per capita grows most rapidly when better jobs, and reduce poverty countries break into limited manufacturing GVCs 60 Hyperspecialization enhances efficiency, and durable Cumulated change in GDP per capita (%) 57 firm-to-firm relationships promote the diffusion of 50 48 technology and access to capital and inputs along chains. For example, in Ethiopia firms participat- 40 ing in GVCs are more than twice as productive as similar firms that participate in standard trade. 32 30 Firms in other developing countries also show significant gains in productivity from GVC partici- 20 pation. A 1 percent increase in GVC participation is estimated to boost per capita income by more than 1 percent, or much more than the 0.2 percent income 10 gain from standard trade. The biggest growth spurt typically comes when countries transition out of 0 Event 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 exporting commodities and into exporting basic year manufactured products (for example, garments) Years after event using imported inputs (for example, textiles) (figure Limited manufacturing O.2), as has happened in Bangladesh, Cambodia, and Advanced manufacturing and services Vietnam. Innovative activities Eventually, however, these high growth rates can- Sources: WDR 2020 team, using data from the World Bank’s WDI database and the GVC taxonomy for not be sustained without moving to progressively 1990–2015 based on Eora26 database. more sophisticated forms of participation. But the Note: The event study quantifies the cumulated change in real GDP per capita in the 20 years transitions from limited manufacturing to more following a switch from a lower to a higher stage of GVC engagement. See box 3.3 in chapter 3 for the methodology. advanced manufacturing and services, and finally to innovative activities (the GVC taxonomy used in this Report is explained further in box 1.3 in chapter 1), trade. In Mexico and Vietnam, for example, the become increasingly more demanding in terms of regions that saw more intensive GVC participation skills, connectivity, and regulatory institutions. also saw a greater reduction in poverty. GVCs also deliver better jobs, but the relationship with employment is complex. Firms in GVCs tend The gains from GVCs are not to be more productive and capital-intensive than equally shared, and GVCs can other (especially nontrading) firms, and so their pro- duction is less job-intensive. However, the enhanced hurt the environment productivity leads to an expansion in firm output The gains from GVC participation are not distributed and thus to increases in firm employment.3 As a equally across and within countries. Large corpora- result, GVCs are associated with structural transfor- tions that outsource parts and tasks to developing mation in developing countries, drawing people out countries have seen rising markups and profits, sug- of less productive activities and into more produc- gesting that a growing share of cost reductions from tive manufacturing and services activities. Firms in GVC participation are not being passed on to consum- GVCs are unusual in another respect: across a wide ers.6 At the same time, markups for the producers in range of countries, they tend to employ more women developing countries are declining. Such a contrast is than non-GVC firms.4 They contribute therefore to evident, for example, in the markups of garment firms the broader development benefits of higher female in the United States and India, respectively. employment. Within countries, exposure to trade with lower- Because they boost income and employment income countries and technological change contribute growth, participation in GVCs is associated with a to the reallocation of value added from labor to capital. reduction in poverty.5 Trade in general reduces pov- Inequality can also creep upward in the labor market, erty primarily through growth. Because gains in eco- with a growing premium for skilled work and stag- nomic growth from GVCs tend to be larger than from nant wages for unskilled work.7 Women also face chal- trade in final products, poverty reduction from GVCs lenges: GVCs may offer more women jobs, but they also turns out to be greater than that from standard seem to have even lower glass ceilings. Women are Overview | 3 generally found in the lower value-added segments; it Innovation is leading to the emergence of new is hard to find women owners and managers.8 traded goods and services, which contributes to faster GVCs can also have harmful effects on the envi- trade growth. In 2017, 65 percent of trade was in cate- ronment. The main environmental costs of GVCs are gories that did not exist in 1992. associated with the growing, more distant trade in Surprisingly, new production technologies are intermediate goods compared with standard trade. also likely to boost trade. Automation does encourage This leads to higher carbon dioxide (CO2) emissions countries to use less labor-intensive methods and from transportation (relative to standard trade) and reduces the demand for the labor-intensive products to excess waste (especially in electronics and plastics) of developing countries. However, the evidence on from the packaging of goods. The growth generated reshoring is limited,9 and the evidence on automa- by GVCs can also strain natural resources, especially tion10 and 3D printing11 suggests that these technol- if accompanied by production or energy subsidies, ogies have contributed to higher productivity and a which encourage excess production. On a more posi- larger scale of production. As such, they have increased tive note, the concern that firms may choose to locate the demand for imports of inputs from developing the most polluting stages of production in countries countries (figure O.3). where environmental norms are laxer is not borne Similarly, digital platform firms are reducing the out by the data. cost of trade and making it easier for small firms to break out of their local markets and sell both goods and services to the world. But there are signs that the New technologies on balance rising market power of platform firms is affecting the promote trade and GVCs distribution of the gains from trade.12 The emergence of new products, new technologies of production such as automation and 3D printing, National policies can boost GVC and new technologies of distribution such as digital platforms is creating both opportunities and risks. participation But the evidence so far suggests that on balance these In principle, breaking up complex products such as technologies are enhancing trade and GVCs. cars and computers allows countries to specialize in simpler parts and tasks, making it easier for those at an early stage of development to participate in trade. Figure O.3  Automation in industrial countries has But a country’s ability to participate in GVCs is by no boosted imports from developing countries means assured. 10 GVC participation is determined by factor endow- ments, geography, market size, and institutions. These % change in imports of parts from developing countries 8 fundamentals alone need not dictate destiny, however; policies also play an important role. Policies to attract 6 foreign direct investment (FDI) can remedy the scarcity of capital, technology, and management skills.13 Liber- 4 alizing trade at home while negotiating trade liberal- 2 ization abroad can overcome the constraints of a small domestic market, liberating firms and farms from the 0 limits of domestic demand and local inputs. Improving transportation and communications infrastructure str es ric on uc e M n g m pr s Fo fac ng he ts ac ls cs l to ics e ec ry g pr ing d eta d tile Ed ltur iv io in in M ica C duc El ine ni on iliti Ag ucti ti Au last ot at in ur od tur an M an in th an Tex tro u m and introducing competition in these services can m h t o ct U p u fa u an address the disadvantage of a remote location.14 And C m er O r, n pe bb er no participating in deep integration agreements can spur pa Ru er d, th institutional and policy reform, especially when com- oo O W plemented by technical and financial assistance.15 Industries in order of increasing automation Based on an analysis of the drivers of various Source: Artuc, Bastos, and Rijkers 2018. types of GVC participation, this Report identifies Note: The figure depicts the automation-induced increase in industrial countries’ imports of materials from developing countries by broad sector over 1995–2015. The change in imports of parts is measured the policies that promote integration into more in log points; a 0.10 increase in log points is roughly equivalent to a 10 percent increase in imports. advanced GVCs (figure O.4). Importantly, national 4 | World Development Report 2020 Figure O.4  Transitioning to more sophisticated participation in GVCs: Some examples of national policy Commodities to limited Limited manufacturing to advanced Advanced manufacturing and manufacturing manufacturing and services services to innovative activities Fundamentals Policy priorities Foreign direct investment: adopt supportive investment policy and improve the business climate Endowments Finance: improve access to banks Finance: improve access to equity finance Labor costs: avoid rigid regulation and Technical and managerial skills: Advanced skills: educate for exchange rate misalignment educate, train, and open to foreign skills innovation and open to foreign talent Access to inputs: reduce tariffs and NTMs; Standardization: harmonize or mutually accept standards reform services Market size Market access: pursue trade agreements Market access: deepen trade agreements to cover investment and services Trade infrastructure: reform customs; liberalize transport services; invest in Advanced logistics services: invest in multimodal transport infrastructure ports and roads Geography Basic ICT connectivity: liberalize ICT services; invest in ICT infrastructure Advanced ICT services: expand high-speed broadband Governance: promote political stability Governance: improve policy predictability; pursue deep trade agreements Institutions Standards certification: establish Contracts: enhance enforcement Intellectual property rights: conformity assessment regime ensure protection Source: WDR 2020 team. Note: ICT = information and communication technology; NTMs = nontariff measures. policies can and should be tailored to the specific Overvalued exchange rates and restrictive labor circumstances of countries and to specific forms of regulations raise the cost of labor, preventing labor- participation in GVCs. abundant countries from taking advantage of their Attracting FDI is important at all stages of partici- endowments. For example, manufacturing labor pation. It requires openness, investor protection, sta- costs in Bangladesh are in line with its per capita bility, a favorable business climate, and, in some cases, income, but in many African countries, labor costs investment promotion. Some countries, such as those are more than twice as high. in Southeast Asia that have benefited from foreign Connecting to markets through trade liberalization investment in goods, still restrict foreign investment helps countries expand their market size and gain in services. Others try to draw in investment through access to the inputs needed for production. For example, tax exemptions and subsidies, but they risk antagoniz- large unilateral tariff cuts by Peru in the 2000s are asso- ing their trading partners, and the net benefits may ciated with faster productivity growth and expansion not be positive. Nevertheless, countries such as Costa and diversification of GVC exports.16 Trade agreements Rica, Malaysia, and Morocco have attracted transfor- expand market access, and they have been a critical cat- mative GVC investments by large multinational cor- alyst for GVC entry in a wide range of countries, includ- porations through the use of successful investment ing Bangladesh, the Dominican Republic, Honduras, promotion strategies. Lesotho, Madagascar, and Mauritius. Because goods Overview | 5 and services economies are increasingly linked, reform- 70 percent of the earnings of the poor. Ensuring that ing services policies—in telecommunications, finance, smallholders benefit requires additional support, transport, and a range of business services—should be such as through agricultural extension services, part of any strategy for promoting GVC activity. access to risk management instruments (such For many goods traded in GVCs, a day’s delay as insurance), and coordination to exploit scale is equal to imposing a tariff in excess of 1 percent. through producer organizations. Improving customs and border procedures, promoting competition in transport and logistics services, and Improving the business and investment climate enhancing port structure and governance can reduce for GVCs on a national scale can be costly and take trade costs related to time and uncertainty, mitigating time, spurring many countries to set up special eco- the disadvantages associated with a remote location. nomic zones (SEZs) to create islands of excellence. Because GVCs thrive on the flexible formation of But the results so far suggest that relatively few SEZs networks of firms, attention should also be paid to are successful, and only when they address specific contract enforcement to ensure that legal arrange- market and policy failures. Getting the conditions ments within the network are stable and predictable. right, even in a restricted geographical area, requires Protecting intellectual property rights is especially careful planning and implementation to ensure that important for the more innovative and complex value the resources needed—such as labor, land, water, chains. Strengthening national certification and test- electricity, and telecommunications—are readily avail- ing capacity to ensure compliance with international able, that regulatory barriers are minimized, and that standards can also facilitate GVC participation. connectivity is seamless. The few successful zone pro- Many of the traditional approaches to industrial grams in countries such as China, Panama, the United policy, including tax incentives, subsidies, and local Arab Emirates, and now in Ethiopia—as well as the content requirements, are likely to distort production numerous examples of SEZs that have failed to attract patterns in today’s GVC context. Other proactive investors or grow—offer important lessons on how to policies are more promising—especially when they use SEZs for development. address market failures: •  To strengthen domestic capacity to support upgrad- Other policies can help ensure ing in value chains, countries should invest in GVC benefits are shared and human capital.17 The Penang Skills Development sustainable Centre in Malaysia is an example of an industry-led training center that has played an important role in Beyond policies to facilitate participation in GVCs, supporting Malaysia’s upgrading to electronics and complementary policies are needed to share their engineering GVCs. benefits and attenuate any costs. These include labor •  Targeted policies to unblock constraints to GVC market policies to help workers who may be hurt by trade can be effective. For example, in Bangladesh the structural change; mechanisms to ensure compliance introduction of bonded warehouses, combined with with labor regulations; and environmental protection the “back-to-back” letters of credit (ensuring access measures. to working capital), is acknowledged as a catalyst for As GVCs expand, some workers will gain, but the country’s integration into the apparel GVC. others could lose in some locations, sectors, and occu- •  Countries can connect domestic small and medium pations. Adjustment assistance, which is especially enterprises (SMEs) with lead firms in GVCs—by important in middle- and high-income countries, supporting training and capacity building while pro- will help workers adapt to the changing patterns of viding information to lead firms about supply oppor- production and distribution that GVCs bring about. tunities. Examples of successful supplier linkage Adjustment policies can include facilitating labor programs include Chile and Guinea in mining, Kenya mobility and equipping workers to find new jobs.18 and Mozambique in agriculture, and the Czech Because unemployment resulting from structural Republic in the electronics and automotive sectors. change tends to be persistent, wage insurance can •  For countries participating in agriculture value help keep workers employed in lower-paying jobs chains, policies to help integrate smallholders are without experiencing income loss, leading to bet- particularly important. In Africa, 55 percent of jobs ter long-term outcomes. For example, Denmark’s are in agriculture, which is the source of more than successful “flexicurity” model gives employers the 6 | World Development Report 2020 freedom to hire and fire workers with few restric- reform, market access around the globe, and recourse tions, but it supports workers with generous unem- in case of disputes—even against the trade heavy- ployment benefits and active labor market programs. weights. Today, however, the international trade sys- Labor regulations, when well designed and tem is under tremendous pressure. Three decades of enforced, help ensure the safety and health of trade-led catchup growth in developing countries has workers. Private firms can contribute, especially contributed to shifts in economic power across coun- when their consumers are sensitive to labor condi- tries and increased income inequality within coun- tions in the firm’s global operations. There is also tries. The growing symmetry in the economic size an important role for national policy supported by of countries is placing in sharp relief the persistent international cooperation in establishing and mon- asymmetry in their levels of protection. Meanwhile, itoring appropriate labor standards. In Vietnam, the trade system, which adapted to changes in the past, working conditions improved when firms partic- has faltered in recent years, most notably with the fail- ipated in the International Labour Organization- ure of the Doha negotiations. Regional initiatives such International Finance Corporation (ILO-IFC) Better as the European Union and NAFTA have also been hurt Work Programme, along­side complementary govern- by disagreements among member countries. ment action to publicly disclose the names of firms The trade conflict between the United States and that fail to meet key labor standards.19 China is leading to protection and policy uncertainty, Pricing environmental degradation can prevent and it is beginning to disrupt GVCs. If the trade con- GVCs from magnifying misallocations of resources.20 flict worsens and causes a slump in investor confi- Prices of goods should reflect both their economic dence, the effects on global growth and poverty could and socioenvironmental costs. Appropriate pricing be significant—more than 30 million people could of ­environmental damage would also encourage be pushed into poverty (measured as income levels innovation in environmentally friendly goods and below $5.50 a day), and global income could fall by as production processes. Reducing distortions, such as much as $1.4 trillion. That said, even in the status quo, those created by energy and production subsidies, adverse effects are likely to have resulted from the and shifting toward taxing carbon would improve trade practices that provoked the conflict. resource allocation and reduce CO2 emissions.21 In To sustain beneficial trade openness, it is essential addition, environmental regulations, especially for to “walk on two legs.” The first priority is to deepen specific industries and pollutants, could curb the dam- traditional trade cooperation to address remaining age caused by GVC-related production and transport. barriers to trade in goods and services, as well as other measures that distort trade, such as subsidies and the activities of state-owned enterprises. In par- International cooperation allel, cooperation should be widened beyond trade supports beneficial GVC policy to include taxes, regulation, and infrastructure. participation The international trade system is especially valuable Deepen traditional cooperation in a GVC world. GVCs span boundaries, and policy Looking ahead, the first priority should be to deepen action or inaction in one country can affect produc- traditional trade rules and commitments. International ers and consumers in other countries. International cooperation has so far delivered uneven openness in cooperation can help address the spillover effects goods and services. Trade liberalization is overdue in of national policies and achieve better development agriculture and services, and some industrial goods outcomes. Because the costs of protection are magni- remain restricted in certain markets and by nontariff fied when goods and services cross borders multiple measures. Trade preferences have reduced certain times, the gains from coordinated reduction of barri- tariffs faced predominantly by the poorest countries— ers to trade are even larger for GVCs than for standard but not the tariffs these countries impose on their trade. In view of the inextricable link between foreign imports. Special and differential treatment for devel- investment and GVCs, creating an open and secure oping countries has in some cases accommodated climate for investment is vital for GVC participation, sluggish reform, ultimately inhibiting GVC participa- especially by capital-scarce countries. tion and integration into the global economy. Developing countries have benefited enormously In addition, the escalation of tariffs in some of from the rules-based trade system, particularly its the world’s largest markets—which serve to pro- guarantees against trade discrimination, incentives to tect higher value-added production—is inhibiting Overview | 7 processing activities in agroindustry and other labor- importing countries, as is the case in some recent intensive areas such as apparel and leather goods agreements on data flows. in developing countries. Restrictive rules of origin But developing countries must not be left out of in preferential agreements are curtailing sourcing such arrangements because that would undermine options. Subsidies and state-owned firms are dis- their productive engagement in GVCs. International torting competition, and the existing rules do not support can help them to both make regulatory com- guarantee competitive neutrality. For services, inter- mitments in areas of export interest (such as in data- national negotiations have delivered little liberaliza- based services) and extract commitments from their tion beyond that undertaken unilaterally. Important trading partners when they open their markets (such GVC-relevant services, such as air and maritime as for the enforcement of competition policy). transportation (which most need coordinated lib- Finally, coordination failures in infrastructure eralization), have been excluded from negotiations investment affect GVC investment, expansion, and because of the power of vested interests. upgrading, especially in the poorest countries. From Traditional trade negotiations may deliver more a global perspective, countries underinvest in trade- meaningful outcomes if the major developing coun- related infrastructure because they do not take into try traders engage as equal partners and even leaders account the additional benefits to their trade part- instead of seeking special and differential treatment; ners. Countries that share a border can obtain larger if the large industrial countries continue to place their gains when they act simultaneously to expedite trade. faith in rules-based negotiations instead of resorting Guatemala and Honduras, for example, reduced bor- to unilateral protection; and if all countries work der delays from 10 hours to 15 minutes when they together to define a negotiating agenda that reflects joined a customs union and agreed to accept the both development and business priorities. same electronic documentation. The World Trade Organization’s Trade Facilitation Agreement encour- Widen cooperation on taxes, competition, ages countries to coordinate improvements in trade and data flows facilitation, and provides low-income countries with Taxing capital is increasingly difficult in an era of financial assistance for the necessary investments. A global firms, fragmented production, and growth in similar approach may help exploit synergies for other intangible assets such as intellectual property. Coop- investments in transport, energy, and communica- eration should ensure fair access to tax revenues— tions infrastructure. which rich countries need to help displaced industrial workers and poor countries need to build infrastruc- Notes ture. Ultimately, a joint approach to greater use of 1. Constantinescu, Mattoo, and Ruta (2018). destination-based taxation could eliminate firms’ 2. Constantinescu, Mattoo, and Ruta (2018). incentives to shift profits and countries’ incentives to 3. In Vietnam, firms that both import and export employ compete over taxes, but the consequences for tax rev- more workers than firms that export only and firms enue in small developing countries would have to be that do not trade, controlling for sector and province considered. Meanwhile, other measures to combat tax fixed effects as well as state and foreign ownership. In Mexico, firms that have relationships with buyers, as base erosion and income shifting could alleviate asso- well as firms that export and import, also see higher ciated challenges for domestic resource mobilization. employment than firms that only import or only export. Among consumers, concern is growing about data This finding holds even when considering the regional, flows and the international expansion of digital firms, sector, and foreign ownership characteristics of firms. both of which play an important role in GVCs. The Across a country, then, firms that both import and risks range from privacy abuses in data-based services export employ more workers than one-way traders or nontraders. to anticompetitive practices in platform-based ser- 4. Rocha and Winkler (2019). vices. Governments are resorting to data localization 5. The poverty elasticity of growth depends on various fac- laws to limit the cross-border mobility of data and tors, including its incidence (changes in inequality), the to strict rules on the handling of data domestically. initial distribution of land, wealth and income, education Competition laws, too, remain explicitly nationalist in levels among the poor, other forms of past public invest- focus, and cooperation in bilateral or regional trading ment, as well as local institutions, including unions (Ferreira, Leite, and Ravallion 2010; Ravallion and Datt agreements has been limited. The solution may be 2002). Also see Dollar and Kraay (2002) and Ferreira and a new type of bargain: regulatory commitments by Ravallion (2008). exporting firms to protect the interests of consumers 6. Markups can increase because prices are higher, or abroad in return for market access commitments by because costs are lower, or a combination of both when 8 | World Development Report 2020 markets are not perfectly competitive, meaning that Bown, Chad P., and Caroline L. Freund. 2019. “Active Labor firms can affect prices. The effect on firms’ markups Market Policies: Lessons from Other Countries for the depends on whether the reduction in costs, or the gains United States.” PIIE Working Paper 19–2 (January), Peter- from GVC participation, are passed fully on to the con- son Institute for International Economics, Washington, sumer through lower prices. DC. 7. Feenstra and Hanson (1996, 1997); Verhoogen (2008). Buelens, Christian, and Marcel Tirpák. 2017. “Reading the 8. Rocha and Winkler (2019). Footprints: How Foreign Investors Shape Countries’ Par- 9. Oldenski (2015) provides evidence that reshoring is not ticipation in Global Value Chains.” Comparative Economic widespread in the United States. Studies 59 (4): 561–84. 10. Artuc, Bastos, and Rijkers (2018). Chen, Maggie Xiaoyang, and Min Wu. 2018. “The Value of 11. Freund, Mulabdic, and Ruta (2018). Reputation in Trade: Evidence from Alibaba.” Paper pre- 12. See Chen and Wu (2018); Garicano and Kaplan (2001); sented at Workshop on Trade and the Chinese Economy, Höppner and Westerhoff (2018). King Center on Global Development, Stanford University, 13. The positive association between FDI and capital, Stanford, CA, April 12–13. technology, and management skills is driven by GVC Constantinescu, Ileana Cristina Neagu, Aaditya Mattoo, participation in the manufacturing sector only. There is and Michele Ruta. 2018. “The Global Trade Slowdown: no association between FDI inflows and countries’ GVC Cyclical or Structural?” World Bank Economic Review. integration of their agriculture, commodities, or services Published electronically May 23. https://doi.org/10.1093 sectors. This finding could point to a more favorable /wber/lhx027. role for efficiency-seeking or market-seeking FDI that Cramton, Peter, David J. MacKay, Axel Ockenfels, and Steven looks for internationally cost-competitive destinations Stoft, eds. 2017. Global Carbon Pricing: The Path to Climate and potential export platforms. See Buelens and Tirpák Cooperation. Cambridge, MA: MIT Press. (2017) for further evidence that bilateral FDI stocks are Dollar, David, and Aart Kraay. 2002. “Growth Is Good for the positively associated with the bilateral backward GVC Poor.” Journal of Economic Growth 7 (3): 195–225. participation as well as with bilateral gross trade. Farid, Mai, Michael Keen, Michael G. Papaioannou, Ian W. H. 14. APEC and World Bank (2018). Parry, Catherine A. Pattillo, and Anna Ter-Martirosyan. 15. According to Johnson and Noguera (2017), the European 2016. “After Paris: Fiscal, Macroeconomic, and Finan- Union and other preferential trade agreements, espe- cial Implications of Global Climate Change.” IMF Staff cially deep ones, play an important role in decreasing the Discussion Note 16/01 (January), International Monetary ratio of bilateral value added to gross exports, a sign of Fund, Washington, DC. growth in global production fragmentation. Feenstra, Robert C., and Gordon H. Hanson. 1996. “Foreign 16. Pierola, Fernandes, and Farole (2018). Investment, Outsourcing, and Relative Wages.” In The 17. Evidence from the Eora database by Lenzen, Kanemoto, Political Economy of Trade Policy: Papers in Honor of Jagdish Moran, and Geschke (2012), (https://worldmrio.com/) Bhagwati, edited by Robert C. Feenstra, Gene M. Gross- shows a U-shaped relationship between GDP per capita man, and Douglas A. Irwin, 89–128. Cambridge, MA: and forward GVC integration across countries. MIT Press. 18. Bown and Freund (2019). —— — —. 1997. “Foreign Direct Investment and Relative Wages: 19. Hollweg (2019). Evidence from Mexico’s Maquiladoras.” Journal of Interna- 20. Gollier and Tirole (2015); Nordhaus (2015). tional Economics 42 (3–4): 371–93. 21. Cramton et al. (2017); Farid et al. (2016); Weitzman Ferreira, Francisco H. G., Phillippe George Leite, and Martin (2017). Ravallion. 2010. “Poverty Reduction without Economic Growth? Explaining Brazil’s Poverty Dynamics, 1985– References 2004.” Journal of Development Economics 93 (1): 20–36. Ferreira, Francisco H. G., and Martin Ravallion. 2008. “Global APEC (Asia–Pacific Economic Cooperation) and World Bank. Poverty and Inequality: A Review of the Evidence.” Policy 2018. “Promoting Open and Competitive Markets in Road Research Working Paper 4623, World Bank, Washington, Freight and Logistics Services: The World Bank Group’s DC. Markets and Competition Policy Assessment Tool Freund, Caroline L., Alen Mulabdic, and Michele Ruta. 2018. Applied in Peru, the Philippines, and Vietnam.” Unpub- “Is 3D Printing a Threat to Global Trade? The Trade lished report, World Bank, Washington, DC. Effects You Didn’t Hear About.” Unpublished working Artuc, Erhan, Paulo S. R. Bastos, and Bob Rijkers. 2018. paper, World Bank, Washington, DC. “Robots, Tasks, and Trade.” Policy Research Working Garicano, Luis, and Steven N. Kaplan. 2001. “The Effects of Paper 8674, World Bank, Washington, DC. Business-to-Business E-Commerce on Transaction Costs.” Borin, Alessandro, and Michele Mancini. 2015. “Follow the Journal of Industrial Economics 49 (4): 463–85. Value Added: Bilateral Gross Export Accounting.” Temi di Gollier, Christian, and Jean Tirole. 2015. “Negotiating Effec- discussione (Economic Working Paper) 1026, Economic tive Institutions against Climate Change.” Economics of Research and International Relations Area, Bank of Italy. Energy and Environmental Policy 4 (2): 5–27. —— ——. 2019. “Measuring What Matters in Global Value Hollweg, Claire H. 2019. “Firm Compliance and Public Dis­ Chains and Value-Added Trade.” Policy Research Working closure in Vietnam.” Policy Research Working Paper Paper 8804, World Bank, Washington, DC. 9026, World Bank, Washington, DC. Overview | 9 Höppner, Thomas, and Phillipp Westerhoff. 2018. “The EU’s Oldenski, Lindsay. 2015. “Reshoring by U.S. Firms: What Do Competition Investigation into Amazon Marketplace.” the Data Say?” PIIE Policy Brief 15–14 (September), Peter- Kluwer Competition Law Blog, November 30, Wolters son Institute for International Economics, Washington, Kluwer, Alphen aan den Rijn, The Netherlands. http:// DC. competitionlawblog.kluwercompetitionlaw.com/2018/11 Pierola, Martha Denisse, Ana Margarida Fernandes, and /30/the-eus-competition-investigation-into-amazon Thomas Farole. 2018. “The Role of Imports for Exporter -marketplace/. Performance in Peru.” World Economy 41 (2): 550–72. ILO (International Labour Organization) and IFC (Inter- Ravallion, Martin, and Guarav Datt. 2002. “Why Has Eco- national Finance Corporation). 2016. “Progress and nomic Growth Been More Pro-Poor in Some States of Potential: How Better Work Is Improving Garment India than Others?” Journal of Development Economics Workers’ Lives and Boosting Factory Competitiveness.” 68 (2): 381–400. Inter­ national Labour Office, Geneva. https://betterwork Rocha, Nadia, and Deborah Winkler. 2019. “Trade and .org/dev/wp-content/uploads/2016/09/BW-Progress-and Female Labor Participation: Stylized Facts Using a Global -Potential_Web-final.pdf. Dataset.” Background paper, World Bank-World Trade Johnson, Robert Christopher, and Guillermo Noguera. 2012. Organization Trade and Gender Report, World Bank, “Accounting for Intermediates: Production Sharing and Washington, DC. Trade in Value Added.” Journal of International Economics Verhoogen, Eric A. 2008. “Trade, Quality Upgrading, and 86 (2): 224–36. Wage Inequality in the Mexican Manufacturing Sector.” —— ——. 2017. “A Portrait of Trade in Value-Added over Quarterly Journal of Economics 123 (2): 489–530. Four Decades.” Review of Economics and Statistics 99 (5): Weitzman, Martin L. 2017. “How a Minimum Carbon-Price 896–911. Commitment Might Help to Internalize the Global Lenzen, M., K. Kanemoto, D. Moran, and A. Geschke. 2012. Warming Externality.” In Global Carbon Pricing: The Path “Mapping the Structure of the World Economy.” Environ- to Climate Collaboration, edited by Peter Cramton, David mental Science & Technology 46 (15): 8374–81. J. C. MacKay, Axel Ockenfels, and Steven Stoft, 125–48. Nordhaus, William. 2015. “Climate Clubs: Overcoming Cambridge, MA: MIT Press. Free-Riding in International Climate Policy.” American Economic Review 105 (4): 1339–70. 10 | World Development Report 2020 PART II Global value chains: What are they? 1 The new face of trade 2 Drivers of participation 1 The new face of trade Key findings  lobal value chains (GVCs) expanded in the 1990s and 2000s, but that expansion has • G slowed since the financial crisis of 2008. One reason is lower global economic growth and investment. Another is the lack of major liberalization initiatives in recent years.  VCs matter for development. GVC trade exhibits two features that distinguish it from • G traditional trade: hyperspecialization and durable firm-to-firm relationships. These features allow firms to raise productivity and income, rendering GVC trade more powerful than traditional trade in supporting growth and poverty reduction. • A  ll countries participate in GVCs but in different ways. Developed and large emerging countries participate in complex GVCs producing advanced and innovative manufactures and services. By contrast, many countries in Africa, Central Asia, and Latin America still produce commodities for further processing in other countries or engage in limited manufacturing.  he intensification of GVCs was driven by a handful of regions, sectors, and firms. • T GVCs grew in the machinery, electronics, and transportation sectors and in the regions specializing in those sectors: East Asia, North America, and Western Europe. Within countries, a few large trading firms dominate GVC trade, supported by foreign direct investment.  ore-complex value chains have stronger regional linkages, although GVCs have • M expanded both globally and regionally. GVCs in East Asia and Europe are more focused on trade within the region. GVCs in North America depend somewhat more on global partners. Elsewhere, GVC integration has been mostly global and is primarily continuing in that direction. 14 | World Development Report 2020 P roduction of goods and services was increas- of these features, GVCs are becoming more attractive ingly globalized from 1990 to 2008. The process to policy makers in developing countries. was more pronounced in some regions and Given their development potential, the stagnation sectors than in others as firms began to organize their of trade growth and GVC formation since the finan- production in complex global value chains (GVCs). cial crisis is a concern. The slowdown is partly cyclical. They designed products in one country, procured Trade growth is lower because output growth is lower parts and components from several countries, and in the major trading economies, including Europe— assembled the final products in yet another country. which accounts for one-fourth of global output and As a result, international trade and investment flows one-third of world trade—and China. The slowdown is increased considerably, far outpacing the growth also structural. Trade growth has become less respon- of economic output. However, with the 2008 global sive to income growth over the last decade, particu- financial crisis and the great recession that followed, larly in China and the United States, both major actors the growth of GVCs and trade slowed, prompting in GVCs. Part of this development reflects changes in speculation that the phenomenon had run its course. the two economies as China moves up the value chain Some aspects of this wave of globalization are and the U.S. energy sector expands. But it also reflects not new. International trade in raw materials and the absence of major new liberalization initiatives, intermediate inputs has been a prominent fea- such as the Uruguay Round, and of major reforms by ture of world trade flows since time immemorial. the large emerging markets—reforms similar to those For example, Assyrian merchants who settled in by China and Eastern Europe in the 1990s. Kanesh (in modern-day Turkey) in the 19th century This chapter analyzes the changing patterns in BCE imported luxury fabrics and tin from Aššur global trade and investment over the last 30 years and traded copper and wool within Anatolia.1 Past and the importance of GVCs in shaping these shifts. increases in the ratio of trade to the gross domestic Using new data, it characterizes the GVC phenome- product (GDP) have been substantial and sustained. non across regions, countries, and sectors. In so doing, The “First Globalization” during 1870–1914 saw a it provides a better understanding of what is new in major increase in international trade flows, largely the world of GVCs, setting the stage for the Report’s attributed to the steamship. Similarly, today’s wave analysis of how GVCs affect economic development, of globalization has been fueled by falling trade costs inequality, and poverty alleviation. due to technological developments such as contain- This chapter offers three main findings. First, erization and policy reforms, particularly the inte- countries participate in GVCs in different ways. gration of China and Eastern Europe into the world Argentina, Ethiopia, and Indonesia are more engaged economy and major trade agreements such as the in simple manufacturing production chains, whereas North American Free Trade Agreement (NAFTA) and Algeria, Chile, and Nigeria export commodities or the Uruguay Round, which established the World raw materials for further processing. India and Trade Organization (WTO) in 1995. the United States produce services that are being This wave of globalization has, however, some new increasingly traded and embodied in manufactured features. For example, by integrating in GVCs devel- goods. And mostly advanced countries and large oping countries can take advantage of richer states’ emerging economies are producing innovative goods industrial bases rather than having to build up entire and services. industries from scratch. In this way, they accelerate Second, the intensification of GVC trade is con- their industrialization and development. Moreover, centrated in a handful of regions, sectors, and firms. trade within GVCs intensifies the effects of standard GVC linkages have expanded fastest in the three trade integration. Fragmented production makes it trade hubs—East Asia, Europe, and North America— possible for firms in developing countries to enter for- in part because these regions account for a large eign markets at lower costs, benefit from specializa- share of production in the sectors whose production tion in niche tasks, and gain access to larger markets processes have become the most fragmented across for their output. Companies can also access cheaper countries, particularly electronics, machinery, and and better inputs, productivity-enhancing technolo- transport equipment. In each country, GVCs tend gies, and improved management practices developed to be con­ centrated among 15 percent of large firms elsewhere, and thus grow at a faster rate, contributing that both import and export and together account for to the creation of better, higher-paying jobs. Because 80 percent of total trade flows. Related-party trade, The new face of trade | 15 such as that through multinational corporations, is What is a global value chain? especially important. Third, more-complex value chains tend to have The bicycle is the world’s most popular form of trans- especially strong regional linkages, although the port. Invented in Germany in the early 19th century, expansion of GVCs has been both global and regional. bicycles were mass-produced by the Dutch at the end Europe is the most integrated region, with four times of that century, sometimes with frames imported as many regional linkages as global linkages. In East from England. Global production later grew from Asia, linkages are more regional than global, and the about 10 million units in 1950 to more than 130 million regional linkages have intensified substantially since units today. 1990. By contrast, GVCs in North America depend Bicycles are heavily traded. They are assembled somewhat more on global partners than regional using parts and components from all over the partners, and integration has been increasing on both world, especially Asia and Europe (figure 1.1). For fronts. Elsewhere, GVC integration has been mostly example, Bianchi carries out all of its design, proto- global and has been increasing primarily with global typing, and conception work in Italy, and then partners. Importantly, in recent decades the differ- assembles most of its bicycles in Taiwan, China, ences in GVC participation across regions have been using parts and components from China, Italy, far greater than the changes within regions. The same Figure 1.1 Japan, Malaysia, and many other parts of the dynamic applies to sectors. world. Each parts producer has niche expertise— Figure 1.1 Where do bicycles come from? Frame exports Saddle exports China: US$977 million Brake exports China: US$100 million Vietnam: US$147 million Japan: US$200 million Italy: US$85 million Italy: US$66 million Singapore: US$172 million Spain: US$16 million Malaysia: US$152 million Wheel exports Pedal and crank exports China: US$170 million Japan: US$150 million Italy: US$28 million China: US$137 million France: US$26 million Singapore: US$117 million Source: WDR 2020 team, using data from UN Comtrade database. See appendix A for a description of the databases used in this Report. 16 | World Development Report 2020 Shimano of Japan, for example, makes brakes for A quality saddle requires the know-how to produce Bianchi, and the handlebars are made in Taiwan, high-tech gel. China. Because of the extensive bicycle value chain, the Assembling a bicycle from parts and compo- trade in bicycle parts has outstripped the trade in nents made around the world improves efficiency bicycles by 15–25 percent in recent years. In Finland, and results in a cheaper and higher-quality bicycle 33 percent of value added is from outside the country, for the consumer. The bicycle frame requires steel, including 13 percent from the European Union (EU), aluminum, or carbon fiber tubing and welding. 11 percent from Asia, and 5 percent from North Ameri- The wheel must be straightened in both radial ca.2 Boxes 1.1 and 1.2 define GVCs and explain how data and lateral directions to ensure uniform tension. are used to estimate GVC participation more broadly. Box 1.1  Defining global value chains A global value chain (GVC) is the series of stages in the components converging at an assembly plant, or snakelike production of a product or service for sale to consumers. structures, with value created sequentially in a series of Each stage adds value, and at least two stages are in dif- stages.a ferent countries. For example, a bike assembled in Finland Regardless of the shape of GVCs, the possibility of with parts from Italy, Japan, and Malaysia and exported to fragmenting production across borders gives rise to a the Arab Republic of Egypt is a GVC. By this definition, a finer international division of labor and greater gains from country, sector, or firm participates in a GVC if it engages in specialization. GVCs allow resources to flow to their most (at least) one stage in a GVC. productive use, not only across countries and sectors, but also within sectors across stages of production. As a result, Defining spiders and snakes GVCs magnify the growth, employment, and distributional The definition of a GVC does not specify the form the impacts of standard trade. foreign value added in production will take, although In summary, unlike traditional international trade whose it is often associated with either international trade in transactions involve only two countries (an exporting coun- raw materials (such as tin or aluminum), in intermediate try and an importing country), GVC trade crosses borders inputs (such as car parts), or in tasks (such as back-office multiple times. This approach to trade not only leads to the services). Similarly, the definition does not mention the rich set of determinants and consequences of GVC partici- various configurations that a GVC might take, including pation described in this Report, but also creates challenges simple spiderlike structures, with multiple parts and for measuring GVC activity in the world. a. Baldwin and Venables (2013). Box 1.2  Measuring global value chains The main challenge in measuring where value is added in in the importing country, or whether it will be reexported a GVC arises from the fact that customs data, the standard after the importing country adds value to it. source for international trade flows, provide information on where the transacted good or service was produced, but A macro view of GVCs not on how it was produced—that is, which countries con- With the goal of tracing value-added trade flows across tributed value to it. Similarly, customs data record where countries, a body of work has combined information from the transacted good is flowing to, but not how it will be customs offices with national input–output tables to con- used—that is, whether it will be fully consumed (absorbed) struct global input–output tables. The most widely used are (Box continues next page) The new face of trade | 17 Box 1.2  Measuring global value chains (continued) the World Input–Output Database (WIOD), a collaborative in the production of “motor vehicles” in the United States project led by researchers at the University of Groningen; but cannot infer where more specific components such as the Trade in Value Added (TiVA) database compiled by tires, car engines, or windshield wipers originate. Second, the Organisation for Economic Co-operation and Develop- in constructing the tables, researchers are forced to impose ment (OECD); and the Eora global supply chain database, strong assumptions to back out some bilateral intermediate constructed by a team of researchers at the University of input trade flows that cannot be readily read from either Sydney.a On a very broad level, these collaborative projects customs data or national input–output tables.c can be thought of as “scaled up” versions of product-level studies, such as the bicycle study, which showed that A micro view of GVCs 33 percent of value added came from foreign countries.b A more granular approach to measuring the fragmentation Such global input–output tables can be used to devise of production processes across countries, first suggested by alternative ways of measuring the extent to which pro- Yeats (1998), computes the share of trade flows accounted duction processes have globalized in recent years and for by industry categories that can safely be assumed to how countries and sectors participate in GVCs. Building on contain only intermediate inputs (reflected in the words global input–output tables, a natural measure of the impor- “Parts of” at the outset of the product description). Yeats tance of GVC trade in total international trade is the share of found that intermediate input categories accounted for trade that flows through at least two borders (see Borin and about 30 percent of OECD merchandise exports of machin- Mancini [2015, 2019] for details on the methodology). Such ery and transport equipment in 1995, and that this share trade encompasses two broad types of GVC trade: had steadily increased from 26 percent in 1978. Yeats’s classification has continued to be refined in recent years Backward GVC participation, in which a country’s exports •  based on the Broad Economic Categories (BEC) product embody value added previously imported from abroad. classification of the United Nations Conference on Trade For example, if the bicycles exported by Taiwan, China, and Development (UNCTAD).d use imported intermediates, then its GVC participation is More recently, customs data at the firm level have been considered backward because the intermediates used in used to advance measurement of GVC linkages. An import- exports are from the previous stage. ant strength of these data is that transactions between •  Forward GVC participation, in which a country’s exports firms and their foreign partner countries can be observed are not fully absorbed in the importing country and rather than inferred. In addition, firm-level data capture instead are embodied in the importing country’s exports the heterogeneity in GVC linkages across firms that is to third countries. In the bicycle example, if India sends obscured by aggregated industry-level data and thus allow aluminum tubing to Taiwan, China, where it is further used a finer understanding of firms’ input sourcing decisions, in the production of the bicycle later exported, then India’s how import and export participation are linked, and how GVC participation is considered forward because the multinational firms organize their production networks. exporter is at the early stage of production of the bicycle. However, such data do not trace firm-to-firm transactions Despite their widespread use, global input–output tables across countries. This would require linking customs offices have two limitations. First, because they rely on aggregated and firm identifiers across the world.e Thus in the absence input–output data, the resulting sectoral disaggregation of such data, the best option is to continue improving the of GVC flows is coarse. They therefore miss a lot of GVC measurement of GVC linkages at both the macro and micro activity within the broadly defined sectors. For example, levels across a wider range of countries to gain a more com- one can compute the origin of “fabricated metal products” plete empirical measurement of GVCs. a. This chapter and the rest of this Report rely on several global information on which domestic industries buy which imports. However, input–output databases for the analysis. The choice of database is such assumptions are not necessarily valid. Specifically, under the dictated by the level of geographical or sectoral coverage needed for homogeneity assumption all firms in the same industry are assumed to the analysis. Eora offers the largest country coverage for the longest have the same production function and use the same bundle of inputs. continuous time period, but its sectoral coverage is more aggregate Yet at the country-industry level, input use varies with output because and thus less precise than the WIOD and TiVA databases. See Lenzen, firms exporting to different countries and industries participate in Kanemoto, Moran, and Geschke (2012) for a description of EORA; and different value chains and face distinct rules of origin (de Gortari 2019). Borin and Mancini (2019), Johnson (2018), and appendix A for a more d. UN Trade Statistics, Intermediate Goods in Trade Statistics, https:// detailed description of these and other databases used in this Report. unstats.un.org/unsd/tradekb/Knowledgebase/50090/Intermediate b. Kalm et al. (2013); OECD (2013). -Goods-in-Trade-Statistics. c. The homogeneity and proportionality assumptions are conveniently e. Johnson (2018). imposed to resolve the fact that the available data sets have no 18 | World Development Report 2020 The evolution of GVC of their inputs. In addition, firms were able to disperse production across the world because transport costs participation fell significantly (figure 1.3, panel b). Declining air and The overall share of GVC trade in total world trade— sea freight costs boosted the trade in goods, while ser- encompassing both forward and backward linkages— vices benefited from cheaper communication costs. grew significantly in the 1990s and early 2000s, but Successive rounds of trade liberalization have it appears to have stagnated or even declined in the resulted in rapidly falling barriers to trade and invest- last 10 years (figure 1.2). Still, about half of world trade ment for both developed and developing countries. Tar- appears to be related to GVCs. iffs have declined, especially for manufactured goods, What explains the remarkable rise in GVC par- and the gradual, although still insufficient, lowering ticipation in the 1990s and 2000s? And why has this of nontariff barriers has facilitated the international process stalled since the financial crisis? trade of goods and services (figure 1.4). Finally, the The global wave of fragmentation of production in creation of the European single market—together with the 1990s and 2000s was driven by a combination of the integration of China, India, and the Soviet Union factors. The information and communication technol- into the global economy—created huge new product ogy (ICT) revolution brought forth cheaper and more and labor markets, and so firms could sell the same reliable telecommunications, new information man- goods to more people and take advantage of economies agement software, and increasingly powerful per- of scale leading to the further deepening of GVCs. The sonal computers (figure 1.3, panel a). Manufacturing new supply of cheap labor encouraged profit-seeking firms then found it easier to outsource and coordinate companies to either reallocate their production facili- complex activities at a distance and ensure the quality ties or find local suppliers in low-wage countries.3 Since the global financial crisis in 2008, the Figure 1.2  GVC trade grew rapidly in dynamics of GVC expansion have changed. Trade the 1990s but stagnated after the 2008 has bounced back from its deep crisis level, but it has global financial crisis grown only marginally faster than output. Trade in 55 parts and components also stalled after the financial GVC share of global trade (%) crisis and even fell between 2011 and 2014, with a mod- 50 est increase since then. The factors behind the trade and GVC slowdown 45 are both cyclical and structural in nature. On the one hand, trade growth is lower because global output 40 growth is lower in economies that account for large 35 shares of global trade and global output, such as Europe and China. Trade has also grown at a slower 30 pace because the trade-to-income elasticity—defined as the amount of trade generated as output rises—has 70 75 80 85 90 95 00 05 10 15 decreased. This is especially true in large trading coun- 19 19 19 19 19 19 20 20 20 20 Sources: WDR 2020 team, using data from Eora26 database; Borin and tries, including China and the United States. China is Mancini (2015, 2019); and Johnson and Noguera (2017). See appendix A for a description of the databases used in this Report. producing more at home, thereby becoming less reli- Note: Unless otherwise specified, GVC participation measures used in this ant on imported components for its exports. The share and subsequent figures throughout the Report follow the methodology of intermediate imports in exports of Chinese goods from Borin and Mancini (2015, 2019). The Eora26 database is used because it offers the largest country coverage: 190 countries between 1990 and dropped from about 50 percent in the 1990s to a little 2015. GVC participation corresponds to the share of world exports that flow over 30 percent in 2015. In the United States, a boom- through at least two borders. For 1990–2015, the GVC participation measure is computed as the share of GVC exports in total international exports using ing shale sector reduced oil imports by one-fourth the Borin and Mancini methodology. GVC exports include transactions in between 2010 and 2015.4 which a country’s exports embody value added that it previously imported from abroad (backward GVC participation), as well as transactions in which As for any major liberalization initiatives that a country’s exports are not fully absorbed in the importing country and might have set off a new wave of GVC formation, instead are embodied in the importing country’s exports to third countries (forward GVC participation). For 1970–90, the GVC participation measure there have been none. The Doha Round stalled, and no is backcasted using the above data and the time variation of the measure large emerging markets are engaging in the types of (1-VAX). The VAX by Johnson and Noguera (2017) is an alternative measure of the value-added content of trade. Although the level difference between drastic reforms undertaken decades ago in China and (1-VAX) and the GVC participation measure is sizable, the correlation of Eastern Europe. their change over the overlapping years (1990–2009) is 0.97. This method allows reconstructing a long series covering 1970–2015 rather than simply All countries partake in GVCs, but across the world 1990–2015 for which the Eora26 database is available. their participation is uneven (map 1.1). Some countries The new face of trade | 19 Figure 1.3  The ICT revolution spurred the emergence of GVCs a. ICT use, 1960–2017 b. Transport and communication costs, 1920–2015 120 100 100 80 80 60 60 40 40 20 20 0 0 1960 1970 1980 1990 2000 2010 2020 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2020 Individuals using the Internet (% of global population) Airfare (New York–London round trip, 1946 = 100) Fixed broadband subscriptions (per 100 persons) Telephone call (3 minutes, New York–London, 1931 = 100) Mobile cellular subscriptions (per 100 persons) Sea freight rates (port and maritime charges per ton, 1920 = 100) Fixed telephone subscriptions (per 100 persons) Computers (storage cost per megabyte, 1956 = 100) Sources: WDR 2020 team, using data from ITU’s World Telecommunication/ICT Indicators database for panel a and based on Rodrigue, Comtois, and Slack (2017) for panel b. Note: In panel a, data are available for over 200 countries. Mobile cellular subscriptions per 100 persons may be over 100 as some people may have several mobile phones. In panel b, for each indicator the cost is reported as 100 for the first year with data. ICT = information and communication technology. Figure 1.4  From 1948 to 2016, tariffs dropped thanks to multilateral and regional trade agreements 25 300 Weighted average applied tariff (%) 250 20 Cumulative number 200 15 150 10 100 5 50 0 0 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 78 19 0 19 2 19 4 19 6 19 8 19 0 19 2 19 4 19 6 98 20 0 20 2 20 4 20 6 20 8 20 0 20 2 20 4 16 4 5 5 5 5 5 6 6 6 6 6 7 7 7 7 8 8 8 8 8 9 9 9 9 0 0 0 0 0 1 1 1 19 19 20 Applied tariff, developed countries (left axis) Applied tariff, developing countries (left axis) WTO members (right axis) RTAs in force as of 2019 (right axis) Sources: WDR 2020 team, based on Baldwin (2012). Data for regional trade agreements (RTAs) and World Trade Organization (WTO) members are from the WTO’s RTAs database. Tariff data prior to 1988 are from Clemens and Williamson (2004), and those for subsequent years are from the World Bank’s WDI database using country-level weighted applied tariffs for all products. Note: The figure plots tariffs computed as simple averages for developed and developing countries. Prior to 1988, the developed country sample covers 35 countries, including 21 industri- alized countries (Argentina, Australia, Austria-Hungary, Canada, Chile, Cuba, Denmark, France, Germany, Greece, Italy, New Zealand, Norway, Portugal, Russia, Serbia, Spain, Sweden, the United Kingdom, the United States, and Uruguay) and 14 developing countries at the time: Brazil, Burma (now Myanmar), Ceylon (now Sri Lanka), China, Colombia, Egypt, India, Indonesia, Japan, Mexico, Peru, the Philippines, Siam (now Thailand), and Turkey. After 1988, developed countries are defined as high-income countries and developing countries as not high-income countries based on the World Bank’s 2018 country classification. 20 | World Development Report 2020 Map 1.1  All countries participate in GVCs—but not in the same way GVC linkages, 2015 Low participation Limited commodities High commodities Limited manufacturing Advanced manufacturing and services Innovative activities Data gaps IBRD 44640 | AUGUST 2019 Source: WDR 2020 team, based on the GVC taxonomy for 2015 (see box 1.3). Note: The type of a country’s GVC linkages is based on the country’s extent of backward GVC participation, measured as the portion of imports embodied in manufacturing exports as a percentage of a country’s total exports, combined with the country’s sector specialization of domestic value added in exports and engagement in innovation. Countries in the commodities group have a small share of manufacturing exports and limited backward GVC integration. Their share of commodity exports can be low, medium, or high. Countries specialized in limited manufacturing GVCs engage in some manufacturing exports, often alongside commodities exports, and exhibit medium backward GVC integration. Countries specialized in advanced manufacturing and services GVCs have a high share of manufacturing and business services exports and high backward GVC integration. Countries specialized in innovative GVC activities spend a large share of GDP on research and development, receive a large share of GDP from intellectual property, and exhibit high backward GVC integration. export raw materials for further processing; others Figure 1.5  Country transitions between different import inputs for assembly and exports; and still others types of GVC participation, 1990–2015 produce complex goods and services. In addition, some are heavily reliant on GVCs for trade, whereas others Innovative export largely domestic goods for consumption. To activities capture these distinct features of participation, coun- tries are classified into four main types—commodities, Austria, Canada, Finland, Ireland, Israel, Italy, Republic of Korea, limited manufacturing, advanced manufacturing and Singapore, Spain Advanced manufacturing services, and innovative activities—based on the prod- and services Czech Republic ucts they export and their participation in GVCs. The rules for classification are described in box 1.3. China, Estonia, India, Lithuania, This taxonomy reveals clear distinctions among Philippines, Poland, Portugal, Romania, Thailand, Turkey regions. East Asia, Europe, and North America are Limited manufacturing engaged in advanced manufacturing and services GVCs and innovative GVC activities, whereas Africa, Central Asia, and Latin America are mostly in com- modities and limited manufacturing GVCs. Argentina, Armenia, Jordan, Lesotho Bosnia and Herzegovina, GVC participation intensified between 1990 and Cambodia, Costa Rica, Cyprus, El Salvador, 2015, as illustrated by the many countries that tran- Ethiopia, Indonesia, Kenya, Nepal, Serbia, sitioned up into more sophisticated forms of GVC South Africa, Tanzania Commodities participation (figure 1.5). Transitions were especially common in East Asia and Europe, where countries were heavily engaged in the sectors most amenable to GVCs, such as electronics and machinery. Among advanced countries, small open economies tended to 1990 2015 The new face of trade | 21 Box 1.3  Types of GVC participation Countries participate in GVCs in different ways, but there  Limited commodities: Primary goods’ share of total are regularities in the type of GVC integration and how domestic value added in exports is equal to or greater countries upgrade. In 146 countries over the period than 20 percent but less than 40 percent. 1990–2015, the following four types of GVC participation High commodities: Primary goods’ share of total    are particularly notable: (1) commodities; (2) limited man- domestic value added in exports is equal to or greater ufacturing; (3) advanced manufacturing and services; and than 40 percent. (4) innovative activities. These criteria define countries according to their export dependence on manufacturing. Data and measures Countries are classified based on (1) the goods and ser- Innovative activities (based on remaining countries) vices exported, (2) the extent of GVC participation, and • Small countries: IP receipts as a percentage of GDP (3) measures of innovation. A country’s sectoral specializa- are equal to or greater than 0.15 percent, and R&D tion of exports is based on the domestic value added in gross intensity is equal to or greater than 1.5 percent. • Medium-size and large countries: IP receipts as a exports of primary goods, manufacturing, and business ser- percentage of GDP are equal to or greater than 0.1 vices. A country’s extent of GVC participation is measured as percent and R&D intensity is equal to or greater than backward integration of the manufacturing sector as a share 1 percent. of the country’s total exports. Higher backward integration in manufacturing is an important characteristic of countries These criteria split groups into those that spend a relatively entering or specialized in noncommodity GVCs. Two mea- large share of GDP on research and receive a large share of sures are used to capture a country’s innovative activities: GDP from IP. (1) intellectual property (IP) receipts as a percentage of GDP Advanced manufacturing and services (based on and (2) research and development (R&D) intensity, defined remaining countries) as its expenditure of public and private R&D as a percentage Share of manufacturing and business servicesa in total of GDP. domestic value added in exports is equal to or greater than 80 percent, and Definitions of GVC taxonomy groups • Small countries: Backward manufacturing is equal to The rules take into account country size because smaller or greater than 30 percent. countries naturally rely on trade to a relatively greater extent. • Medium-size countries: Backward manufacturing is The following taxonomy groups are defined sequentially: equal to or greater than 20 percent. Commodities • Large countries: Backward manufacturing is equal to or greater than 15 percent. Manufacturing share of total domestic value added in exports is less than 60 percent, and Limited manufacturing (rest of sample) • Small countries: Backward manufacturing is less than Upgrading trajectories 20 percent. Based on these definitions, the following countries transi- • Medium-size countries: Backward manufacturing is tioned from commodities into limited manufacturing GVCs less than 10 percent. over the period 1990–2015: Argentina, Armenia, Bosnia and • Large countries: Backward manufacturing is less than Herzegovina, Cambodia, Costa Rica, Cyprus, El Salvador, 7.5 percent. Ethiopia, Indonesia, Kenya, Nepal, Serbia, South Africa, and These criteria ensure that manufacturing is a small Tanzania. share of exports and that backward linkages in manufac- The following countries moved into advanced manufac- turing are limited. turing and services from limited manufacturing GVCs: China, This group is further subdivided as follows: the Czech Republic, Estonia, India, Lithuania, the Philippines, Low participation: Primary goods’ share of total    Poland, Portugal, Romania, Thailand, and Turkey. domestic value added in exports is less than 20 The Czech Republic moved further up into the innovative percent. activities group in 2012 and remained in this group over the (Box continues next page) 22 | World Development Report 2020 Box 1.3  Types of GVC participation (continued) period covered. Other countries moved into innovative GVC manufacturing and services and then back to limited manu- activities: Austria, Canada, Finland, Ireland, Israel, Italy, the facturing. Five other countries switched from commodities Republic of Korea, Singapore, and Spain. to limited manufacturing and then back to commodities: Two countries, Jordan and Lesotho, downgraded from Botswana, Jamaica, the Democratic People’s Republic of limited manufacturing to commodities. Meanwhile, some Korea, Nicaragua, and Senegal. countries upgraded and then downgraded. Swaziland (now All other countries remained in the same group over the Eswatini) moved from limited manufacturing to advanced period covered. Business services include maintenance and repair; wholesale trade; retail trade; transport; post and telecommunications; and financial intermediation a.  and business activities. Business services, not total services, were used to detect advanced countries with a developed services sector. show the highest participation. Emerging economies Figure 1.6  Average backward and forward GVC such as China, Poland, and South Africa experienced participation across taxonomy groups rapid growth in GVC participation between 1990 and GVC participation 2015 and as such moved up GVC groups. South Africa transitioned from commodities to limited manufac- High turing while China and Poland transitioned from lim- ited manufacturing to advanced manufacturing and services. Other countries remained in the same group over that period. In Brazil, Morocco, and Pakistan, GVC participation grew less rapidly. The high GVC partici- pation for major commodity exporters such as Algeria, Sectoral Saudi Arabia, and República Bolivariana de Venezuela specialization Low reflects extensive forward integration because natural resources are the most upstream sectors. Commodities Limited Advanced Innovative Countries’ sectoral specialization shapes the extent manufacturing manufacturing activities and services of backward and forward participation. Figure 1.6 shows an approximate distribution of backward and Backward participation Forward participation forward GVC integration across the four taxonomy Source: WDR 2020 team. groups. Backward integration is lowest for countries Note: The approximate distribution is based on backward and forward GVC participation averages by specialized in commodities and starts to expand for taxonomy group for the period 2010–15. For the definition of taxonomy groups, see box 1.3. countries in the limited manufacturing group. Coun- tries specializing in advanced manufacturing and ser- How are GVCs distributed across vices are highly reliant on imported inputs for exports. Backward participation is slightly lower for the coun- regions? tries in the innovative group because their activities GVCs have increased globally and regionally, but are less dependent on imported inputs. the differences across regions remain larger than The abundance of natural resources or agriculture differences over time. Some regional GVCs are more in a country is linked to high forward integration focused on trade within the region, while others are because commodities are used in a variety of down- more dependent on global integration (figure 1.7). stream production processes that typically cross sev- Countries’ trade with regional (or regional bloc) eral borders. Participation in limited manufacturing value chains involves only production partners in reduces forward integration because commodities the region, whereas extraregional value chain trade are less important in trade, and the manufacturing involves only partner countries outside the region. output at this stage (such as garments) is less likely to Importantly, the differences between regions in the be used as inputs in destination countries. However, depth of regional integration are stark and vastly moving to advanced manufacturing and services dominate changes over time. Europe is the most GVCs and especially innovative activities increases regionally integrated region, with four times as many forward participation. regional linkages as global linkages. South Asia and The new face of trade | 23 Figure 1.7  GVC activities increased globally and regionally from 1990 to 2015 50 Share of global GVC trade in total GVC trade (%) 45 More global than regional 2010 40 2015 2010 35 1990 2015 2010 30 1990 2015 More regional than global 25 2010 2010 1990 2015 2015 20 1990 2010 1990 2015 15 1990 2000 2015 2010 10 2005 1990 5 0 0 5 10 15 20 25 30 35 40 45 50 Share of regional GVC trade in total GVC trade (%) East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa North America South Asia Sub-Saharan Africa Source: WDR 2020 team, using data from Eora26 database. Note: For each region and intervals of 5–6 years between 1990 and 2015, the figure plots the share of GVC trade involving only production partners in the same region in total GVC trade (regional GVC integration) against the share of GVC trade involving only partner countries outside the region in total GVC trade (global GVC integration). Regional and global GVC participation measures are computed as weighted averages over the countries in each group. The weights are the share of each country in the corresponding region total trade. The economic size of the trading blocs and the number of potential production partners in the region influence these indicators. The 45-degree line marks instances in which the share of regional and global GVC trade in total GVC trade for a given region are equal. In this figure, Mexico is not included in the Latin America and the Caribbean region but in North America, together with Canada and the United States. The economic size of the trading blocs and the number of potential production partners in the region influence these indicators. See the note to figure 1.2 on methodology and data for GVC participation measures. the Middle East and North Africa are the least region- By contrast, the NAFTA GVCs depend somewhat •  ally integrated regions. more on global partners than regional partners, In all regions, the increase in GVC participation and integration has been increasing on both fronts. between 1990 and 2015 resulted from a combination of GVCs expanded more regionally in the 1990s, regional and global trends: reflecting the coming into force of the NAFTA trade agreement in 1994, while the 2000s saw a marked •  In Europe, regional fragmentation of value chains acceleration in global GVC activities in part owing increased through successive rounds of enlarge- to China joining the world economy. ment in which Eastern European countries, includ- In Latin America and the Caribbean, value chains •  ing Bulgaria, Hungary, and Poland, progressively are more globally linked, but they have increased joined older members’ production networks. But both regionally and globally. global fragmentation was equally important, driven In the three remaining regions, GVC integration •  mostly by the larger European economies such as has been mostly global and has been increasing pri- France, Germany, and the United Kingdom, whose marily with global partners, with South Asia’s GVCs linkages with countries in Asia such as China or expanding almost entirely outside the region. India expanded. •  In East Asia, linkages are more regional than global, A look at backward linkages confirms that produc- and GVCs became more internationally fragmented tion networks in East Asia, Europe, and, to a lesser after 1990 because of both regional and global frag- extent, North America are mostly regional (figure 1.8). mentation in the 1990s and 2000s, although regional In an average European country, 65 percent of the integration dominated. imported intermediates embodied in its exports in 24 | World Development Report 2020 Figure 1.8  Global production networks are organized around three main regions, 2018 Share of foreign value added in exports of each region, by source region a. East Asia and Pacific b. Europe and Central Asia c. Middle East and North Africa d. North America (FVA share = 25%) (FVA share = 36%) (FVA share = 14%) (FVA share = 19%) 6% 1% 2% 4% 8% 1% 3% 2% 1% 2% 2% 2% 2% 7% 8% 8% 2% 20% 23% 12% 17% 10% 50% 65% 39% 24% 22% 55% e. Latin America and the Caribbean f. South Asia g. Sub-Saharan Africa (FVA share = 15%) (FVA share = 13%) (FVA share = 14%) 2% 2% 3% 2% 8% 11% 3% 6% 2% 28% 5% 26% 10% 36% 3% 42% 10% 17% 39% 24% 23% East Asia and Pacific Europe and Central Asia Middle East and North Africa North America Latin America and the Caribbean South Asia Sub-Saharan Africa Source: WDR 2020 team, using data from full Eora database (latest year for which data are available is 2018). Note: The full Eora database is used because it offers the largest country coverage. The geographic breakdown across source countries is available for only one GVC participation index, the foreign value-added (FVA) content of exports. For each region, the figure reports the share of imported intermediates embodied in exports in total exports, computed as the ratio of the FVA content of exports in total gross exports (FVA share is in parenthesis). The figure also reports the contribution of each origin partner region to this FVA share. In this figure, Mexico is not included in the Latin America and the Caribbean region but in North America together with Canada and the United States. 2018 originated from other European countries. This sensitive components closer to home. Trade costs also share is about 55 percent for an average East Asian determine the optimal location for individual produc- economy, and almost 40 percent for a member coun- tion stages along GVCs.5 try of NAFTA. The other regions are all more inte- North and Sub-Saharan Africa have managed grated globally than regionally. The share of imported to join GVCs in the apparel, food, and automotive intermediates embodied in exports originating from industries and in some business services. But Africa regional partners is 26 percent in Latin America and remains a small actor in the global economy, account- the Caribbean but as low as 3 percent in South Asia. ing for just 3 percent of global trade in intermediate In Latin America and the Caribbean, the geo- goods. African exports tend to enter at the very begin- graphic distribution of the foreign content of exports ning of GVCs. A high share serves as inputs for other is almost equivalent across East Asia, Europe, and countries’ exports, reflecting the still-predominant North America. South Asia is especially integrated role of agriculture and natural resources in African in production networks in East Asia and Europe, exports. Botswana, the Democratic Republic of Congo, whereas Sub-Saharan Africa is predominantly inte- and Nigeria have become integrated in GVCs through grated in European supply chains followed by those exports of oil and other natural resources. But Ethiopia, in East Asia. These regional patterns reflect geograph- Kenya, and Tanzania have seen faster GVC integra- ical distances and trade costs because intermediate tion, sourcing foreign inputs for their export-oriented inputs are shipped across borders multiple times. businesses. Most of their integration has occurred For example, just-in-time manufacturing techniques in agribusiness and apparel (especially in Ethiopia have pushed firms to locate the production of time- and Kenya), in manufacturing (in Tanzania), and to a The new face of trade | 25 lesser extent, in transport and tourism. Morocco’s The top contributors to GVC intensification were efforts to attract major manufacturers in the auto- Germany, the United States, Japan, Italy, and France, motive industries over the past decade are paying off. which began using more imported inputs in their A new Peugeot facility opened in 2019, following in exports (figure 1.9). By contrast, China’s contribution the footsteps of another French automaker, Renault- to the expansion of GVC worldwide was predomi- Nissan. Overall, GVC participation in some of these nantly through an increase in its share of world trade, Sub-Saharan countries (Ethiopia, Kenya, South Africa, although its GVC intensification remains significant. and Tanzania) grew by 10 percentage points or more, approaching what Poland or Vietnam—now success How are GVCs distributed across stories—experienced in the late 1990s and 2000s. sectors? The sectoral composition of GVC flows is also quite Which countries have accounted diverse. Some countries specialize largely in agri- for most of the GVC expansion? cultural GVCs (such as Madagascar) or in the natu- A few countries in Asia, Europe, and North America ral resource segments of GVCs (such as Chile and have driven GVC expansion over the past 30 years. Norway). These types of GVCs are classified as Between 1990 and 2015, GVC participation worldwide commodity-linked. Developing economies (such as grew by about 7 percentage points, because production Tanzania) specialize in low-tech simple manufactur- processes in some countries and sectors become more ing, and more developed economies (such as China, fragmented—an intensification effect; or because Mexico, and the Slovak Republic) in medium-tech countries and sectors that were already GVC-intensive manufacturing. One set of countries (including India boosted their share of world trade—a scale effect. and Singapore) largely specializes in the services embodied in GVCs. And a small set of very advanced economies (Germany, Japan, and the United States) Figure 1.9  A handful of countries drove global GVC provide innovative goods and services. expansion from 1990 to 2015 1.4 Most GVCs serve a handful of sectors in GVC intensification contribution to overall 1.2 DEU manufacturing and services GVC growth (percentage points) Some industries have used GVCs heavily for decades. 1.0 USA Examples are basic industries that are resource- JPN intensive and make heavy use of imported primary 0.8 inputs—chemicals, refined petroleum, basic metals, 0.6 ITA and rubber and plastics. These sectors were already dis- NLD playing large GVC participation in 1995 because of their 0.4 FRA GBR high foreign value added in exports (figure 1.10). They SWE 0.2 KOR CHN have intensified their use of supply chains over time. BEL ESP By contrast, the fragmentation of value chains in 0 SGP textiles and leather has not changed over the past two TWN –0.2 decades. Most fragmentation of production in these –2 –1 0 1 2 3 sectors occurred in the 1970s and 1980s, thus the slower Scale effect contribution to overall GVC growth pace. The termination of the Multifibre Arrangement (percentage points) in 2004 further concentrated production chains in East Asia and Pacific Europe and Central Asia fewer countries, with China emerging as the largest Latin America and the Caribbean Middle East and North Africa producer and capturing many stages of production. North America South Asia For services, construction and transport-related activ- Sub-Saharan Africa ities are the most fragmented. For transport-related Source: WDR 2020 team, using data from Eora26 database. activities, GVC participation increased substantially Note: The Eora26 database is used because it offers the largest country coverage, covering 190 countries between 1995 and 2011. between 1990 and 2015. The GVC participation measure reflects the share of a country’s exports that For sectors, most of the GVC intensification over flow through at least two borders. It is computed as the share of GVC exports in total international exports. GVC exports include transactions in which a country’s exports embody value added that it the period was driven by high-tech manufacturing previously imported from abroad (backward GVC participation), as well as transactions in which a industries, whose use of imported inputs increased. At country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries (forward GVC participation). For country abbreviations, the other end of the spectrum, very upstream mining see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. and other primary industries accounted for most of the 26 | World Development Report 2020 Figure 1.10  GVC participation by sector, 1995 and 2011 80 Increasing GVC participation Rubber and plastics Coke and 70 refined petroleum Electricity, gas, and water supply Basic metals and Mining and quarrying fabricated metals 60 Electrical and GVC participation, 2011 (%) Air transport Inland optical equipment Chemicals Machinery, NEC transport 50 Financial intermediation Transport equipment Manufacturing, NEC 40 Post and telecommunications Construction Pulp and paper Wholesale trade Textiles 30 Agriculture Hotels and restaurants Food, beverages, and tobacco Real estate Leather and footwear 20 Education Retail trade Decreasing GVC 10 participation 0 10 20 30 40 50 60 GVC participation, 1995 (%) Industry share in global exports, 2011 (%): <1 1–4 5–9 10+ Primary goods Manufacturing Services Source: WDR 2020 team, using data from WIOD 2013 release database. Note: The WIOD 2013 database is used because it offers a finer sectoral classification than Eora26. In addition, the 2013 release (covering 1995–2011) is used instead of the latest 2016 release (covering 2001–14) in order to compare the change in GVC participation in the 2010s with that in the 1990s. The GVC partici- pation measure reflects the share of world exports that flow through at least two borders. For each industry-year, it is computed as the share of GVC exports in total international exports. GVC exports include transactions in which a country’s exports embody value added that it previously imported from abroad (backward GVC participation), as well as transactions in which a country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries (forward GVC participation). The 45-degree line marks instances in which GVC participation for a given sector is the same in 1995 and 2011. NEC = not elsewhere classified. scale effect, consistent with their high share of GVC in exports. India, Kenya, and the Philippines also have integration and growing share of world trade follow- rapidly expanding ICT and business service sectors. ing the large price surge over the period (figure 1.11). Even in China, traditionally viewed as an exporter of manufactures, more than a third of the value added in GVCs are not just in manufacturing—they its exports comes from services. have also expanded rapidly in services For gross exports of services, such as transport, Services are an invisible but vital part of GVCs. The tourism, or business services, the share in trade is fragmentation of goods production has been associ- fairly flat at about 20 percent. The goods trade is ated with outsourcing not just manufacturing tasks increasingly involving services in production, with but also service tasks, with the back office of many the share of services in valued-added trade rising U.S. manufacturers now in India. In addition, trans- from 31 percent to 43 percent between 1980 and 2009, portation, telecommunications, and financial services a result of both forward and backward use of services facilitate and coordinate the geographic dispersion in production (figure 1.12). of production in all sectors. And service production is itself being fragmented across countries, such as GVCs in agriculture and food industries when preliminary architectural designs, tax returns, have also expanded, including those in and magnetic resonance imaging (MRI) readings are Africa performed in one country and finalized and delivered Although GVCs in the agriculture and food sectors to customers in another. In France, Germany, Italy, the have expanded over the past two decades, they United Kingdom, and the United States, services con- remain a small share of GVC trade. In 2014 agricul- tribute more than half the total value added embodied ture exports accounted for 2 percent of world exports The new face of trade | 27 Figure 1.11  A handful of sectors drove global GVC as contracting and logistics expertise. Taken together, expansion from 1995 to 2011 Asia, Latin America, and Sub-Saharan Africa saw their foreign direct investment (FDI) inflows in the agri-food 1.6 sector grow by a factor of three between 2000 and 2010. GVC intensification contribution to overall Electrical and optical equipment 1.4 But such investments are mainly in large and more Transport GVC growth (percentage points) 1.2 equipment developed markets within Latin America (Argentina, Brazil, Chile, and Mexico) and Asia (China, Indonesia, 1.0 and Vietnam), with little flowing into Sub-Saharan Machinery, NEC 0.8 Africa (Ethiopia, Ghana, Tanzania, and Uganda). These Chemicals and chemical products Mining and investments are mostly aimed at the food industry 0.6 quarrying (processing and retail) instead of agriculture.7 0.4 Pulp and In overall participation in agriculture GVCs Other business activities paper Food Coke and between 1990 and 2015, Ethiopia, Ghana, Kenya, and 0.2 Financial intermediation refined petroleum Rwanda in Africa and Vietnam in East Asia stand out. 0 They increased their GVC participation by almost 10 Textiles and textile products Leather and footwear –0.2 percentage points or more. By contrast, the Lao Peo- –1.0 –0.5 0 0.5 1.0 1.5 2.0 ple’s Democratic Republic, Lebanon, and the Republic Scale effect contribution to overall GVC growth of Yemen—and resource-rich economies such as (percentage points) South Sudan—saw their integration in agriculture Agriculture Food GVCs drop by between 5 and 30 percentage points High R&D–intensive industries Low R&D–intensive industries (figure 1.13, panel a). For food GVCs, Sub-Saharan Medium R&D–intensive industries Mining and quarrying African countries including Ethiopia, The Gambia, Other services Telecommunications, financial, and Tanzania also saw significant increase in partic- Trade and transportation and business services ipation, suggesting that those countries have been successfully developing food processing industries Source: WDR 2020 team, using data from WIOD 2013 release database. (figure 1.13, panel b). Value chains in the food industry Note: The WIOD 2013 database is used because it offers a finer sectoral classification than Eora26. In addition, the 2013 release (covering 1995–2011) is used instead of the latest 2016 release (covering are also important in Eastern European countries such 2001–14) in order to compare the change in GVC participation in the 2010s with that in the 1990s. as Bulgaria, Hungary, and Serbia. The GVC participation measure reflects the share of a country’s exports that flow through at least two borders. It is computed as the share of GVC exports in total international exports. GVC exports include Importantly, the participation of most developing transactions in which a country’s exports embody value added that it previously imported from abroad countries in agriculture and food GVCs is largely (backward GVC participation), as well as transactions in which a country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries forward because it is limited to supplying a specific (forward GVC participation). The 35 WIOD 2013 industries are classified in nine industry groups (see product such as coffee by Ethiopia or Uganda, cocoa by World Bank 2019): (1) agriculture, hunting, forestry, and fishing (ISIC Rev. 3 code 01T05); (2) food (ISIC Rev. 3 code 15T16); (3) mining and quarrying (ISIC Rev. 3 code 10T14); (4) high R&D–intensive Côte d’Ivoire or Ghana, oranges by Brazil, and bananas industries (ISIC Rev. 3 codes 24, 29T34, 352, 353, 359); (5) medium R&D–intensive industries (ISIC by Colombia. Rev. 3 code 25T28, 351, 37); (6) low R&D–intensive industries (ISIC Rev. 3 codes 17T23, 36); (7) trade and transportation (ISIC Rev. 3 codes 50T52, 55, 60T63); (8) post and telecommunications, financial, and Agriculture GVCs are also characterized by the business services (ISIC Rev. 3 codes 64, 65T67, 71T74); and (9) real estate activities, utility, construction, prevalence of informality, which has important con- and other services (ISIC Rev. 3 codes 70, 75, 80, 85, 90T93, 95, 40, 41, 45). ISIC = International Standard Industrial Classification; NEC = not elsewhere classified; R&D = research and development. sequences for workers’ poverty and vulnerability. In developing countries, over 94 percent of employ- in contrast to 60 percent for manufactures and ment in agriculture is informal versus 63 percent in around 20 percent for services. When measured in manufacturing. In African countries, these shares value-added terms, this share rises to about 5 percent. rise to 98 percent for agriculture and 77 percent for This finding reflects the fact that in the agri-food manufacturing.8 Although firms in GVCs pay higher sector, unlike in the manufacturing sector, domestic wages to their formal workers, they also rely heav- value chains are dominant and dynamic, with GVCs ily on informal workers who do not earn the same important but secondary. In Asia and Latin America, premiums. In Peru, 79 percent of all men and 84 per- supermarkets and small and medium enterprises in cent of all women working on artichoke farms and the food sector such as chain restaurants, processors, processing plants have jobs that are not secure. Only and modern wholesale and logistics companies have about half of the migrant workers in the export pine- spread rapidly.6 apple sector in Ghana have permanent contracts.9 Another factor in this finding is that GVCs in the Hiring workers indirectly through subcontractors agri-food sector typically involve less cross-border or agents further contributes to vulnerability within movement of goods than capital investments through GVCs as firms transfer their social responsibilities to direct and portfolio means and business practices such a third party. 28 | World Development Report 2020 Figure 1.12  Services are playing a growing role in GVCs a. Goods and services shares in gross exports b. Backward and forward GVC participation and value-added exports, 1980–2009 in services exports, 1995–2014 100 2,500 18 20 21 31 80 39 43 2,000 US$ (billions) 60 1,500 Percent 40 82 80 79 1,000 69 61 57 20 500 0 0 1980 1995 2009 1980 1995 2009 1995 2005 2014 Gross exports Value-added exports Backward GVC participation Services Goods Forward GVC participation Sources: WDR 2020 team, using data from Johnson and Noguera (2017) for value-added exports measure in panel a and WIOD data from the 2013 release for 1995 and the 2016 release for 2005 and 2014 for panel b. Note: Panel a reports the share of goods and services in gross exports and value-added exports, and panel b the GVC exports of services broken down into their backward and forward components. The GVC exports reflect exports that flow through at least two borders and indicate the extent to which sectors participate in GVCs. The GVC exports include transactions in which a country’s exports embody value added that it previously imported from abroad (backward GVC participation), as well as transactions in which a country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries (forward GVC participation). Figure 1.13  GVCs expanded in both the agriculture and food industries from 1990 to 2015 a. Agriculture GVCs b. Agri-food GVCs 60 90 Agriculture GVC participation, 2015 (%) Agri-food GVC participation, 2015 (%) 80 50 RWA 70 VNM GHA ETH KEN 40 60 HUN TZA 50 SRB 30 LAO GMB 40 SSD BGR 20 30 ETH LBN YEM 20 10 10 0 0 0 10 20 30 40 50 60 0 10 20 30 40 50 60 70 80 Agriculture GVC participation, 1990 (%) Agri-food GVC participation, 1990 (%) East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa North America South Asia Sub-Saharan Africa Source: WDR 2020 team, using data from Eora26 database. Note: The Eora26 database is used because it offers the largest country coverage: 190 countries between 1990 and 2015. Plots report only countries with at least 5 percent of their exports in the agriculture or agri-food sector. Agriculture includes forestry, hunting, and fishing. The GVC participation measure reflects the share of a country’s exports that flow through at least two borders. It is computed as the share of GVC exports in total international exports. GVC exports include transactions in which a country’s exports embody value added that it previously imported from abroad (backward GVC participation), as well as transactions in which a country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries (forward GVC participation). The blue 45-degree line marks instances in which GVC participation for a given country are the same in 1990 and 2015. The red 45-degree lines mark a 10 percentage point change in the rate of GVC participation between 1990 and 2015. For country abbreviations, see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. The new face of trade | 29 A few large trading firms account export. Similarly, firms that both import and export dominate GVC participation (figure 1.14). for most GVC trade Because firms are the main actors in GVCs, another In practice, it is firms, not countries or industries, that way to illustrate an individual country’s GVC partici- participate in international trade (box 1.4). In line with pation is to look at its share of firms engaged in two- this simple observation, economic research on inter- way trade—that is, firms that both import and export national trade underwent a dramatic transformation (figure 1.14). For example, 41 percent of trading firms in the last 20 years, placing firm-level international in China, 32 percent in South Africa, and 22 percent strategies at center stage. Fueling this shift was the in Mexico both import and export—and all three have growing availability of longitudinal plant and firm large GVC participation. The concentration of trade data sets that permitted researchers to unveil new in a few importing–exporting firms is extreme. Two- facts challenging the validity of existing models. An way traders account for about 15 percent of all trading important stylized fact from this literature is that in all firms on average in the sample of countries, and yet countries, rich and poor, trade is highly concentrated they capture almost 80 percent of total trade. These in a small share of large firms that both import and Box 1.4  A firm-level approach to GVCs While most conceptual frameworks and empirical mea- Using a firm-level approach, one can also distinguish sures related to GVCs are at the country or country-industry GVCs organized by a lead firm, which incurs the bulk of level, in practice, it is not countries or industries that trade, the fixed costs associated with setting up the network of but rather firms. In line with this observation, research in producers for a given production process, from those that international trade has undergone a dramatic transforma- are more decentralized, with individual producers incurring tion in the past 20 years, placing firm-level international the costs to set up links upstream and downstream.b strategies at the center stage. This intellectual revolution Firm-level data sets containing information on the was fueled by the increased availability of longitudinal import and export transactions of firms can be used to plant and firm-level data sets that allowed researchers to construct measures of GVC participation similar to those unveil new facts that challenged the validity of existing based on the country-industry information in global input– models. At the theoretical level, a seminal paper was that output tables. Specifically, transaction-level customs data of Melitz (2003), which focuses on the exporting decisions sets of the type available from the World Bank’s Exporter of heterogeneous firms within an industry. In Melitz’s Dynamics Database can identify the set of firms in a coun- framework, firms are assumed to produce differentiated try that participate in trade, further distinguishing firms products using technologies featuring increasing returns that export, firms that import, and firms that both export to scale. Product differentiation confers market power on and import. When a given firm in a given country both firms, whereas scale economies are associated with firms imports and exports, it is natural to conclude that this firm facing fixed costs of production and distribution. The deci- participates in GVCs. sion of a firm to export to a given foreign market is shaped To map this definition more precisely to the definition by a comparison of the potential operating profit obtained of backward GVC participation developed in country- in that foreign market with the fixed costs associated with industry studies, one would ideally also resort to product- distributing products in that market. level information to verify that the goods imported by an This firm-level approach to international trade initially exporting firm are indeed intermediate inputs (rather than involved only the exchange of final goods, but an active final goods), so that one can more comfortably conclude that literature has adopted similar ideas to understand the rise this firm is indeed using foreign value added in its production of GVCs. Because of the fixed costs of engaging in global destined for exports. Without linking customs data across sourcing (that is, of importing parts and components), one countries, it is much harder to come up with analo­ gous would expect that the use of imported inputs in production firm-level measures of forward GVC participation. Even when would require importers to attain a minimum efficient scale a firm is identified as an exporter of intermediate inputs of production, thereby excluding smaller and less produc- (instead of final goods), it is almost impossible to establish tive firms in an industry from GVC participation.a whether those inputs are fully absorbed in the importing (Box continues next page) 30 | World Development Report 2020 Box 1.4  A firm-level approach to GVCs (continued) country or whether they are reexported to third markets by across countries. Even when the entries in these tables the importing firms after having added value to them. provide an accurate account of the origin of inputs in a Firm-level measures identify only the extensive margin country’s industrial production, the standard methods used of GVC participation, while industry-level measures based to compute bilateral value-added trade flows from these on global input–output tables also capture the intensity tables assume that the same combination of inputs is used of GVC participation. Computing intensive measures of in production regardless of the destination of sales of a GVC participation at the firm-level data is challenging, country’s and industry’s output. In practice, firms selling however (especially if complementary census information output to different markets use very different combina- is not available), because customs data do not cover firms’ tions of input sources, and this has implications for the domestic purchases of inputs or domestic sales of goods. type of bilateral value-added trade flows one infers from Thus it is difficult to infer the ratio of foreign inputs used in global input–output tables. For example, because Mexican production, and it is even more difficult to disentangle the exports to the United States embody a disproportionate foreign input content of exports from the foreign content of amount of U.S. value added relative to Mexican exports to overall production.c other countries, the share of U.S. value in U.S.-imported Firm-level information on importing and exporting Mexican manufactures is 30 percent instead of the 17 per- can also shed light on whether global input–output tables cent one would infer from standard techniques applied to provide an accurate description of value-added trade flows global input–output tables.d a. See Antràs, Fort, and Tintelnot (2017); Antràs and Helpman (2004); Gopinath and Neiman (2014); Halpern, Koren, and Szeidl (2015). b. See Bernard, Moxnes, and Ulltveit-Moe (2018). c. See Kee and Tang (2016) for an attempt using processing trade in China. d. de Gortari (2019). Apart from qualifying the type of implications that one can draw from aggregated input–output tables, firm-level data can also be used to test the validity of the “proportionality” assumptions that go into construction of those data. “superstar” firms, many of them multinational,10 drive better understood as sunk costs, which naturally cre- country trade performance.11 ate “stickiness” among participants in a GVC. A source of lock-in for GVC relationships is that Sticky buyer–seller relations participants often make relationship-specific investments Modeling global production sharing as simply an (such as purchasing specialized equipment or custom- increase in the extent to which foreign inputs (or for- izing products), and so they would obtain a much lower eign value added) are used in production misses dis- return if GVC linkages were broken. The need to cus- tinctive characteristics of the recent rise of GVCs. That tomize inputs, coupled with quality sensitivity, makes rise entails much more than the intensification of the matching buyers and sellers particularly important. If a firm suddenly faces an increase in the demand for trade in raw materials and homogeneous intermedi- its goods, it cannot easily scale up by buying more for- ate inputs that has been undertaken since the Bronze eign inputs from some centralized market. Typically, Age. It is also much more than import and export only a handful of suppliers worldwide can provide the firms transacting with each other in world markets. additional customized inputs to scale up. The expansion of GVCs entails a finer international Meanwhile, GVCs are more likely to lead to tech- division of labor, but it also involves several additional nology transfer and standards upgrading. Firms in features, four of them especially important: (1) match- GVCs do not engage only in trade in tangible goods ing buyers and sellers, (2) making relationship-specific with other members of their value chains. They often investments, (3) exchanging intangibles, and (4) living benefit from large flows of intangibles, such as technol- with limited contractual security. ogy, intellectual property, and credit. Lead or parent Matching buyers and sellers in GVCs is not fric- firms may also provide good managerial practices, tionless. The fixed costs of exporting and importing saving resources and lifting productivity, or labor reflect in part the costs of finding suitable suppliers and environmental standards. The exchange of these of parts and components or suitable buyers of a sell- intangibles is much more complex than that of simple er’s products. For this reason, these fixed costs are goods or services. The new face of trade | 31 Figure 1.14  Firms that both import and export dominate GVC participation 100 95 MEX Share of export–import firms in total trade (%) ZAF GAB BWA 90 ZMB SRB CHL HRV 85 CIV CHN ROU MKD GTM SLV 80 PER COL BGR TZA MAR 75 URY ALB DOM SEN 70 ECU KEN 65 MUS BGD MDG 60 MWI PRY 55 EGY 50 0 5 10 15 20 25 30 35 40 45 Share of export–import firms in total number of trading firms (%) Share in global exports, 2015 (%): <1 1–9 10+ East Asia and Pacific Europe and Central Asia Latin America and the Caribbean Middle East and North Africa South Asia Sub-Saharan Africa Source: WDR 2020 team, using firm-matched export–import customs data collected for 32 countries by the Trade and International Integration Unit of the World Bank Development Research Group, as part of efforts to build the Exporter Dynamics Database described in Fernandes, Freund, and Pierola (2016). Note: The figure plots the share of two-way trading firms (firms that both import and export in a given year) in the total number of trading firms (firms that import, export, or do both) against their share in a country’s total trade value (imports plus exports). For each country, the average of each measure is computed over 2005–15 for the largest available sample of countries. The dashed lines mark the average across countries for each measure on the x-axis and y-axis. For country abbreviations, see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. The lock-in effects and flows of intangibles within value added across countries through anonymous GVCs are particularly relevant because of the limited spot exchanges of goods and services. Instead, the contractual security that governs transactions within identity of the agents participating in a GVC is cru- these chains. GVCs often engage in transactions that cial, and within GVCs, relationships are more likely to require a strong legal environment to bind produc- exhibit persistence. ers together and avoid technological leakage. And yet GVCs often lack this strong legal environment Transactions within firm boundaries because cross-border exchanges of goods cannot gen- An extreme version of relational contracting arises erally be governed by the same contractual safeguards when parties in a GVC bypass the market mechanism that typically govern similar exchanges within bor- altogether and undertake transactions within the ders. As a result, GVC participants must have repeated boundaries of firms by having the buyer vertically interactions to ensure implicit contract enforce- integrate with the seller or vice versa. Indeed, many ment. As with matching frictions and relationship- value chains are managed and controlled by multina- specificity, this force contributes to the “stickiness” of tional enterprises that organize their production across GVC relationships. different locations. In some cases, goods are closer to In summary, these features of GVCs lead to a new customers and the costs of trade fall (market- novel, relational conceptualization of GVCs that seeking investment). In others, it is a matter of tak- shifts the focus away from the mere allocation of ing advantage of lower costs of factors of production 32 | World Development Report 2020 (efficiency-seeking investment). Both types of invest- Intrafirm trade flows in world trade flows also ment have contributed to the international dispersion exemplify the relational aspects of the growth of of production, but the second has been especially GVCs. For example, U.S. Census data from 2016 show important for GVC growth, which is evident from the that more than 40 percent of U.S. goods trade involves growth of FDI flows and GVCs, especially since the related-party transactions. At the global level, intra- 1990s (figure 1.15). firm trade has been estimated to be about one-third FDI flows into countries in the South and North of world trade flows. In addition to having their (inward FDI) are positively correlated, suggesting that own affiliates abroad, multinational companies rely the expansion of foreign investments in one market on independent suppliers, including small firms in did not come at the expense of the other. For foreign domestic and foreign markets. investment flows out of developed and developing The hierarchy and direction of knowledge flows countries (outward FDI), those from emerging econo- between the multinational (or lead) firm and its mies have grown quickly, if from a very low base.12 Since suppliers vary across types of GVCs, depending the early 2000s, companies in the South have sought on the complexity of products, the ability to codify opportunities to sell products locally, such as when transactions, and the capabilities of supply firms.14 In the Kenyan supermarket chain Tuskys opened stores producer-driven chains, the lead firm controls the in Uganda. In other instances, firms have focused on design and most of the assembly of products by affil- taking advantage of cheaper labor, such as when Chi- iates and captive suppliers, who are prevented from nese firms invested in Madagascar’s agriculture and sharing technology with competitors. Such chains textile sectors. From 2000 to 2015, the outward direct are typical in industries relying heavily on technology investment of firms in Brazil, China, India, the Russian and R&D, such as electronics, automotive, aerospace, Federation, and South Africa surged—from $7 billion to and pharmaceuticals, where production requires the $200 billion, or almost one-third of global FDI.13 assembly of thousands of customized parts into one Figure 1.15  Foreign direct investment accompanied the fragmentation of production from 1970 to 2018 a. Net flows of FDI into developed and b. Net flows of FDI out of developed and developing countries developing countries 7 60 7 60 Share of GVC exports over total exports (%) Share of GVC exports over total exports (%) 6 6 50 50 Net outward FDI flows (% of GDP) Net inward FDI flows (% of GDP) 5 5 40 40 4 4 30 30 3 3 20 20 2 2 10 10 1 1 0 0 0 0 70 70 75 80 85 90 95 00 05 10 15 20 75 80 85 90 95 00 05 10 15 20 19 19 19 19 19 19 19 20 20 20 20 20 19 19 19 19 19 20 20 20 20 20 Low- and middle-income High-income GVC participation (right axis) Source: WDR 2020 team, using data from the World Bank’s WDI database. Note: Panel a reports the net inflows of investment to the reporting economy from foreign investors divided by GDP, and panel b reports the net outflows of investment from the reporting economy to the rest of the world divided by GDP. To avoid composition effects, the definitions of income groups are time-invariant and based on the World Bank’s 2018 country classi- fication. The GVC participation measure reflects the share of countries’ exports that flows through at least two borders. It is computed as the share of GVC exports in total international exports. GVC exports include transactions in which a country’s exports embody value added that it previously imported from abroad (backward GVC participation), as well as transactions in which a country’s exports are not fully absorbed in the importing country and instead are embodied in the importing country’s exports to third countries (forward GVC participation). FDI = foreign direct investment. The new face of trade | 33 high-end product. Large manufacturers such as Apple, Antràs, Pol, Teresa C. Fort, and Felix Tintelnot. 2017. “The General Motors, Samsung, Sony, and Toyota are typi- Margin of Global Sourcing: Theory and Evidence from cal of producer-driven global supply chains. U.S. Firms.” American Economic Review 107 (9): 2514–64. Antràs, Pol, and Elhanan Helpman. 2004. “Global Sourcing.” By contrast, when production is less complex and Journal of Political Economy 112 (3): 552–80. can be modularized or knowledge can be codified, cap- Baldwin, Richard E. 2012. “Global Supply Chains: Why They tive relationships are less likely. In GVCs driven by the Emerged, Why They Matter, and Where They Are Going.” purchasing firms—so-called buyer-driven GVCs—the CEPR Discussion Paper 9103 (August), Centre for Eco- lead company has few factories of its own and sources nomic Policy Research, London. its products almost entirely from a large network Baldwin, Richard E., and Anthony J. Venables. 2013. “Spiders and Snakes: Offshoring and Agglomeration in the Global of independent suppliers, leaving it to concentrate Economy.” Journal of International Economics 90 (2): 245–54. instead on marketing and sales. This type of GVC is Barjamovic, Gojko, Thomas Chaney, Kerem Coşar, and Ali mostly found in the textile and apparel industries, Hortaçsu. 2019. “Trade, Merchants, and the Lost Cities where products such as clothes, housewares, or toys of the Bronze Age.” Quarterly Journal of Economics 134 (3): require relatively little capital and skills. Large retail- 1455–1503. ers such as JCPenney and Walmart and big brands Barrientos, Stephanie Ware, John Kwasi Anarfi, Nicolina such as Nike are examples. Lamhauge, Adriana Castaldo, and Nana Akua Anyidoho. 2009. “Social Protection for Migrant Labour in the From this relational concept of GVCs emerges a Ghanaian Pineapple Sector.” DRC Working Paper T–30 richer analysis of them, one that puts on center stage (September), Development Research Centre on Migra- the major actors (such as multinational firms and lead tion, Globalisation, and Poverty, University of Sussex, firms in GVCs) that shape GVC activity and FDI flows. Brighton, U.K. Such an analysis underscores the role of institutional Bernard, Andrew B., Andreas Moxnes, and Karen Helene factors in shaping the location of global production. Ulltveit-Moe. 2018. “Two-Sided Heterogeneity and Trade.” Review of Economics and Statistics 100 (3): 424–39. By explicitly modeling the mechanisms for dividing Borin, Alessandro, and Michele Mancini. 2015. “Follow the the gains from specialization across firms, this rela- Value Added: Bilateral Gross Export Accounting.” Temi di tional approach also delivers novel lessons about the discussione (Economic Working Paper) 1026, Economic implications of GVC participation for inequality and Research and International Relations Area, Bank of Italy. for development, as the following chapters review. It —— — —. 2019. “Measuring What Matters in Global Value also provides a rich set of predictions about how an Chains and Value-Added Trade.” Policy Research Working increase in automation or digital technologies may Paper 8804, World Bank, Washington, DC. Chan, Man-Kwun. 2013. “Informal Workers in Global Hor- affect the landscape of the international economy and ticulture and Commodities Value Chains: A Review of the different agents in society. Literature.” WIEGO (Global Trade) Working Paper 28 (June), Women in Informal Employment Globalizing and Notes Organizing, Cambridge, MA. Clemens, Michael A., and Jeffrey G. Williamson. 2004. “Why 1. Barjamovic et al. (2019). Did the Tariff-Growth Correlation Change after 1950?” 2. Kalm et al. (2013); OECD (2013). Journal of Economic Growth 9 (1): 5–46. 3. Freeman (2007). Constantinescu, Cristina, Aaditya Mattoo, and Michele Ruta. 4. Constantinescu, Mattoo, and Ruta (2018). 2018. “The Global Trade Slowdown: Cyclical or Struc- 5. Antràs and de Gortari (2017). tural?” World Bank Economic Review. Published electroni- 6. Reardon and Timmer (2012). cally May 23. https://doi.org/10.1093/wber/lhx027. 7. Reardon and Barrett (2019). Cusolito, Ana Paula, Raed Safadi, and Daria Taglioni. 2016. 8. ILO (2018). Inclusive Global Value Chains: Policy Options for Small and 9. Barrientos at al. (2009); Chan (2013); Gammage (2009). Medium Enterprises and Low-Income Countries. Directions in Development: Trade Series. Washington, DC: World 10. Freund and Pierola (2015). Bank and Organisation for Economic Co-operation and 11. Freund and Pierola (2015); Mayer and Ottaviano (2008). Development. 12. See also UNCTAD (2019). de Gortari, Alonso. 2019. “Disentangling Global Value Chains.” 13. Cusolito, Safadi, and Taglioni (2016); Gómez-Mera et al. NBER Working Paper 25868 (May), National Bureau of (2015). Economic Research, Cambridge, MA. 14. Gereffi, Humphrey, and Sturgeon (2005). Fernandes, Ana Margarida, Caroline L. Freund, and Martha Denisse Pierola. 2016. “Exporter Behavior, Country Size and Stage of Development: Evidence from the Exporter References Dynamics Database.” Journal of Development Economics Antràs, Pol, and Alonso de Gortari. 2017. “On the Geography of 119 (March): 121–37. Global Value Chains.” NBER Working Paper 23456 (May), Freeman, Richard B. 2007. “The Great Doubling: The Chal- National Bureau of Economic Research, Cambridge, MA. lenge of the New Global Labor Market.” In Ending Poverty 34 | World Development Report 2020 in America: How to Restore the American Dream, edited by Producer’s World Tour.” In State of the Region Reports: The John Edwards, Marion Crain, and Arne L. Kalleberg, Top of Europe—Plowing Ahead in the Shadows of a Fractured 55–65. Chapel Hill, NC: Center on Poverty, Work, and Global Economy, 10th ed., edited by Christian Ketels, 126–35. Opportunity, University of North Carolina at Chapel Hill; København, Denmark: Baltic Development Forum. New York: New Press. Kee, Hiau Looi, and Heiwai Tang. 2016. “Domestic Value Freund, Caroline L., and Martha Denisse Pierola. 2015. Added in Exports: Theory and Firm Evidence from “Export Superstars.” Review of Economics and Statistics China.” American Economic Review 106 (6): 1402–36. 97 (5): 1023–32. Lenzen, M., K. Kanemoto, D. Moran, and A. Geschke. 2012. Gammage, Sarah C. 2009. “Gender and Pro-Poor Value Chain “Mapping the Structure of the World Economy.” Environ- Analysis: Insights from the GATE Project Methodology mental Science & Technology 46 (15): 8374–81. and Case Studies.” With inputs from Cristina Manfre and Mayer, Thierry, and Gianmarco I. P. Ottaviano. 2008. “The Kristy Cook. May, U.S. Agency for International Develop- Happy Few: The Internationalisation of European Firms.” ment, Washington, DC. Intereconomics 43 (3): 135–48. Gereffi, Gary, John Humphrey, and Timothy J. Sturgeon. Melitz, Marc J. 2003. “The Impact of Trade on Intra- 2005. “The Governance of Global Value Chains.” Review of Industry Reallocations and Aggregate Industry Produc- International Political Economy 12 (1): 78–104. tivity.” Econometrica 71 (6): 1695–1725. Gómez-Mera, Laura, Thomas Kenyon, Yotam Margalit, Josó OECD (Organisation for Economic Co-operation and Devel- Guilherme Reis, and Gonzalo J. Varela. 2015. New Voices opment). 2013. Interconnected Economies: Benefiting from in Investment: A Survey of Investors from Emerging Countries. Global Value Chains. Paris: OECD. World Bank Study Series. Washington, DC: World Bank. Reardon, Thomas, and Christopher B. Barrett. 2019. “The Gopinath, Gita, and Brent Neiman. 2014. “Trade Adjustment Evolution of Agri-Food Value Chains in Developing Coun- and Productivity in Large Crises.” American Economic tries.” Background paper, World Bank, Washington, DC. Review 104 (3): 793–831. Reardon, Thomas, and Charles Peter Timmer. 2012. “The Eco- Halpern, Laszlo, Miklos Koren, and Adam Szeidl. 2015. nomics of the Food System Revolution.” Annual Review of “Imported Inputs and Productivity.” American Economic Resource Economics 4 (August): 225–64. Review 105 (12): 3660–3703. Rodrigue, Jean-Paul, Claude Comtois, and Brian Slack. 2017. ILO (International Labour Organization). 2018. Women and The Geography of Transport Systems, 4th ed. New York: Men in the Informal Economy: A Statistical Picture, 3d ed. Routledge. Geneva: ILO. UNCTAD (United Nations Conference on Trade and Develop- Johnson, Robert Christopher. 2018. “Measuring Global Value ment). 2019. World Investment Report 2019: Special Economic Chains.” Annual Review of Economics 10 (1): 207–36. Zones. New York and Geneva: United Nations. Johnson, Robert Christopher, and Guillermo Noguera. World Bank. 2019. Global Value Chain Development Report 2019: 2017. “A Portrait of Trade in Value Added over Four Technological Innovation, Supply Chain Trade, and Workers Decades.” Review of Economics and Statistics 99 (5): in a Globalized World. Washington, DC: World Bank. 896–911. Yeats, Alexander J. 1998. “Just How Big Is Global Production Kalm, Matias, Mika Pajarinen, Petri Rouvinen, and Timo Sep- Sharing?” Policy Research Working Paper 1871, Washing- pälä. 2013. “The Rise of Baltic Sea Value Chains—A Bicycle ton, DC, World Bank. The new face of trade | 35 2 Drivers of participation Key findings  lobal value chain (GVC) participation is determined by fundamentals such as factor • G endowments, market size, geography, and institutional quality, but these fundamentals need not dictate destiny. Choosing the right policies can shape each one of these fundamentals and thus GVC participation.  actor endowments matter. Low-skilled labor and foreign capital are central to backward • F participation in GVCs at early stages. An abundance of natural resources drives forward GVC integration. Foreign capital, whether efficiency-seeking or resource-seeking, can enhance host country integration in GVCs.  arket size matters. Small countries are more dependent on imported inputs and foreign • M markets. Trade liberalization can expand effective market size and promote participation in GVCs.  eography matters. Overcoming remoteness by improving connectivity can promote • G GVC participation. Trade in parts and components within international production networks is highly sensitive to logistics performance and uncertainty in bilateral international transport times.  nstitutional quality matters. Entering deep preferential trade agreements (PTAs) can • I enhance institutional quality and increase GVC participation. Deep PTAs cover legal and regulatory frameworks, harmonize customs procedures, and set rules on intellectual property rights. 36 | World Development Report 2020 V ietnam’s electronics sector expanded dramat- positively correlated with backward GVC participa- ically in less than a decade. Today, Vietnam is tion. It also promotes domestic upstream sectors, as the second-largest smartphone exporter, pro- happened in the case of apparel in Bangladesh, elec- ducing 40 percent of Samsung’s global mobile phone tronics in Vietnam, and automotives in Morocco. products and employing 35 percent of its global staff. Market size matters. Trade liberalization can Vietnam’s success can be attributed to a combina- expand market size and promote participation in tion of factors. Trade liberalization—driven by World GVCs. Lower tariffs on manufacturing goods fos- Trade Organization (WTO) accession and an agree- ter backward GVC participation in manufacturing. ment with the United States—a favorable investment Manufacturing tariffs fall sharply in the years before climate, and a large pool of low-cost labor determined a country’s transition from commodity to limited Vietnam’s attractiveness as a global value chain (GVC) manufacturing GVCs. Sectors facing lower tariffs in location. The result was large foreign direct invest- destination markets exhibit stronger backward and ment (FDI) inflows, including from Samsung. Viet- forward GVC participation. Market access for low- nam’s geographical proximity to regional suppliers income countries provided by the Everything but of electronics parts and components such as China, Arms initiative of the European Union (EU) or the Japan, the Republic of Korea, and Thailand helped for- African Growth and Opportunity Act (AGOA), a U.S. eign investors gain access to high-quality inputs from trade pact, can stimulate their exports and GVC inte- abroad. And improved connectivity enabled Vietnam gration. In the long run, however, the effects depend to import and export in a timely manner. on rules of origin and their impacts on developing a The story of Vietnam demonstrates that GVC local supplier base. participation is determined first and foremost by fun- Geography matters. Overcoming remoteness by damentals such as factor endowments, market size, improving connectivity can promote GVC participa- geography, and institutions (box 2.1). But these funda- tion. Longer geographical distances to the major GVC mentals need not dictate destiny. Choosing the right hubs—China, Germany, and the United States—have policies can shape each one of these fundamentals and a strong negative impact on both backward and for- thus GVC participation. Attracting FDI can remedy a ward GVC participation in manufacturing. By con- scarcity of capital, technology, and management skills. trast, longer distances increase a country’s likelihood Liberalizing trade at home and negotiating trade lib- of specializing in commodity GVCs. High transport eralization abroad can overcome the constraints of costs impede entering, establishing, and upgrading a small domestic market, freeing firms and farms in GVCs. Inefficient transport and logistics services from dependence on limited local inputs and narrow and weak competition in these services amplify those domestic demand. Improving transport and commu- costs in many manufacturing GVCs. Trade in parts nication infrastructure and introducing competition and components within international production in these services can address the disadvantage of a networks is highly sensitive to logistics performance remote location. Participating in deep trade integra- and uncertainty in bilateral international transport tion agreements that encompass policy areas beyond times. Connectivity also includes effective commu- traditional trade policy, such as investment, competi- nication among the participants in GVCs, which can tion, and intellectual property rights protection, can be improved by access to the Internet. Higher Internet improve domestic institutions by helping countries usage is linked to stronger backward GVC integration. commit to domestic reform and receive technical and Institutional quality matters. Entering deep prefer- financial assistance. ential trade agreements (PTAs) can enhance institu- Factor endowments matter. Low-skilled labor and tional quality and increase GVC participation. Deep foreign capital are central to backward participation PTAs cover legal and regulatory frameworks, harmo- in GVCs. The abundant supply of low-cost labor in nize customs procedures, and set the rules on intel- lower-income countries is often an entry point for lectual property rights. Weak contract enforcement participation in the labor-intensive manufacturing deters traditional trade flows, and GVCs are partic- segments of GVCs. But upgrading skills becomes ularly sensitive to the quality of contractual institu- necessary for integration in more complex GVCs. tions. Sectors relying more on contract enforcement An abundance of natural resources drives forward see faster growth in GVC participation in countries GVC integration. Foreign capital, whether efficiency- with better institutional quality. Greater political seeking or resource-seeking, can enhance host coun- stability reduces the likelihood of specializing in try integration in GVCs. Indeed, it is strongly and commodity GVCs. Drivers of participation | 37 Box 2.1  Vietnam’s integration in the electronics GVC Today, Vietnam is the second-largest smartphone exporter, 2008. Samsung’s presence in Vietnam now includes the producing 40 percent of Samsung’s global mobile phone world’s largest smartphone production facility, a smart- products and employing 35 percent of its global staff. Viet- phone and tablet display assembly facility, an electro- nam’s backward participation in electronics GVCs increased mechanical assembly operation for camera modules, and from 47 percent in 2000 to 67 percent in 2010, and then the Samsung Vietnam Mobile Research and Development declined slightly after 2012 (figure B2.1.1, panel a). Import Center. Samsung has about 160,000 workers in Vietnam, tariffs in the sector dropped from about 8 percent in 2000 and lead firms LG, Canon, and Panasonic, contract manu- to less than 3 percent by 2015 (figure B2.1.1, panel b). facturers Foxconn and Jabil Circuit, and platform leaders Vietnam has been a member of the Association of Intel and Microsoft also operate there. FDI benefited from Southeast Asian Nations (ASEAN) since 1995, and after generous incentives, including tax concessions provided by entering the World Trade Organization in 2007 the coun- the Vietnamese government. try’s number of preferential trade partners increased from Abundant low-skilled, low-cost labor. Vietnam’s large 10 to 16 by 2014. Most free trade agreements were between pool of low-skilled, low-cost labor was an important deter- ASEAN and third countries (Australia, China, India, Japan, minant of its attractiveness as a GVC location. Over half the Republic of Korea, and New Zealand), but some were of the workforce in Vietnam’s population of more than 95 bilateral with Chile, Japan, and the European Union. The million was estimated to be low-skilled in 2006. But the coverage in Vietnam’s trade agreements expanded sub- quality of education in Vietnam is a significant barrier, and stantially from 13 core provisions in 2007 to 86 in 2014. extensive training is still necessary. Samsung’s software Vietnam owes its success in the electronics sector to the engineers are trained at the Samsung Vietnam Mobile following factors. Research and Development Center, with 90 percent of Stable investment climate. Vietnam’s foreign direct them attaining Samsung’s global standards. The improved investment (FDI) stock picked up from around $400 per technological skills of the Vietnamese workforce may have person in the early 2000s to $500 in 2008 and $880 in 2015 actually contributed to the country’s declining share of low- (figure B2.1.1, panel c). FDI inflows to the electronics sector skilled workers—down to less than 40 percent by 2015. included mostly large investments from Korea’s Samsung Proximity. Most of the electronic inputs imported by Group, which launched Samsung Electronics Vietnam in Vietnam are from China; Hong Kong SAR, China; Japan; Figure B2.1.1  Vietnam’s backward GVC integration increased from 2000 to 2015 as tariffs declined and foreign direct investment (FDI) expanded a. Backward integration of b. Weighted import tariffs in c. FDI stock per person electronics and machinery as a electronics and machinery share of gross exports 80 9 1,000 70 8 900 7 800 60 700 6 50 U.S. dollars 600 5 Percent Percent 40 500 4 30 400 3 300 20 2 200 10 1 100 0 0 0 2000 2005 2010 2015 2000 2005 2010 2015 2000 2005 2010 2015 Sources: WDR 2020 team, using data from Eora and World Bank’s WDI and WITS databases. See appendix A for a description of the databases used in this Report. (Box continues next page) 38 | World Development Report 2020 Box 2.1  Vietnam’s integration in the electronics GVC (continued) Korea; Singapore; Taiwan, China; and Thailand. Although Connectivity. Vietnam reduced the average time to the import content of electronics exports reached two- import by two days—to roughly three weeks over 2006–15— thirds of gross exports in recent years (figure B2.1.1, and yet this is still one week longer than in the Philippines or panel a), the reliance on imported inputs declined slightly Thailand, which have been involved in manufacturing GVCs as the role of local suppliers increased. Samsung’s local for much longer. Meanwhile, Vietnam’s Internet usage shot suppliers include not only foreign-owned suppliers that up from 17 percent of the population in 2006 to 43 percent co-located with Samsung in Vietnam, but also 29 domestic in 2015—higher than the 27 percent in the Philippines and suppliers (such as in display making and plastic molding) in 25 percent in Thailand—reflecting an effort to dominate the 2016, up from just four in 2014, all trained by Samsung to information and communication technology GVC, not only meet quality standards. in hardware but also in business services. Sources: Nikkei Asian Review (2018); Sturgeon and Zylberberg (2016); Viet Nam News (2015). Factor endowments matter The abundance of low-skilled labor in countries is positively linked to the extent of their backward inte- GVCs entail a finer international division of labor gration in GVCs, based on evidence from a large sam- than standard trade, with countries specializing in ple of countries in the Eora database (box 2.2).4 This segments of GVCs rather than in industries (chap- pattern is driven by backward GVC participation in ter 1). Traditional trade theory postulates that factor the manufacturing and services sectors. Countries endowments are an important determinant of special- with larger endowments of low-skilled labor in the ization in GVCs, and they also shape the positioning 2000s were also more likely to be among the group of of countries in GVCs. For example, an abundance of countries specializing in either limited manufactur- natural resources in a country is naturally linked to ing or advanced manufacturing and services in 2011. high forward GVC integration because agricultural Among countries engaged in limited manufacturing, products and commodities are used in a variety of Vietnam had by far the highest average percentage downstream production processes that typically cross of low-skilled workers in its labor force (over 42 several borders. Vietnam’s electronics GVC illustrates percent) during 2006–15, followed by Ethiopia (37 how abundance in low-skilled labor is often an entry percent) and El Salvador (31 percent). Using labor point to backward participation.1 costs as an alternative measure of low-skilled labor endowments for the same large sample of countries A large pool of low-skilled workers matters in the Eora database confirms the positive link with for joining manufacturing GVCs, but higher backward integration. According to evidence for 87 skills matter for upgrading countries, lower wages facilitate participation in the When Samsung decided to invest in Vietnam, it was final assembly stages of GVCs, mostly in the apparel attracted to the young, cheap, and abundant work- sector.5 force.2 On average, Vietnamese workers could be hired But labor costs could rise with a country’s contin- at half the cost of their Chinese counterparts and were ued involvement in and upgrading of GVCs, as has seven years younger. This cheap labor lowers costs in happened in China. Improved technological skills Samsung’s factories, giving the smartphone maker contributed to a declining share of low-skilled work- an edge over Apple in the less expensive handsets. ers in Vietnam (see box 2.1). Upgrading workforce Likewise, Bangladesh’s success in apparel exports after skills becomes necessary to export more advanced conclusion of the Multifibre Arrangement’s quota manufacturing goods and services (box 2.2).6 A regime in 2004 is linked to its large pool of low-skilled, firm-level analysis of Bangladesh confirms that low-cost workers. At less than $200 a month, the aver- the higher skill intensity of a workforce and higher age wage of an apparel sector worker in Bangladesh is wages (relative to other firms in the country) are lower than that in China ($270), India ($255), and Viet- positively associated with the likelihood of being a nam ($248).3 GVC firm.7 Drivers of participation | 39 Box 2.2  Modeling results on the drivers of GVC participation From imports of pistons used as intermediates in car manu- GVC participation of seven broad types of determinants facturing in Morocco (foreign content of exports/backward emphasized in the trade literature: (1) factor endowments, participation) to Chilean exports of copper used in refrig- (2) geography, (3) market size, (4) trade policy and erators produced by firms in China and Mexico (domestic foreign direct investment (FDI), (5) quality of institutions, value added in exports used by partner countries for (6) connectivity, and (7) financial and business environ- export production/forward participation), GVC participa- ment factors. tion is multifaceted and diverse across countries. This assessment estimates the impact of country aver- This assessment of the drivers of GVC participation ages of the determinants in the previous decade (e.g., the across countries relies on GVC participation measures 1990s) on country average GVC participation in the fol- from Borin and Mancini (2019) using the Eora database, lowing decade (e.g., the 2000s). It considers the following which covers 190 countries and draws on a combina- dependent variables: (1) the share of backward or forward tion of international input–output tables, domestic GVC participation in gross exports, which captures the production, and trade data (see appendix A for a intensity of GVC trade relative to that of traditional exports; description of the databases used in this Report). The (2) backward or forward GVC participation levels (logs); and econometric model assesses the marginal impacts on (3) gross exports (logs). Comparing the factors that affect Figure B2.2.1  What explains backward and forward GVC participation? a. Backward GVC participation b. Forward GVC participation FDI inflows (log)** Land/GDP (log)*** Political stability Market size index** (manufacturing) (log)** Low-skilled labor/ GDP (log)* Rents from resources/ GDP*** Capital/GDP (log)** Capital/GDP (log)** Rents from resources/ GDP*** Distance to GVC hubs Average tariff (log)*** rate*** Political stability Distance to GVC hubs index** (log)*** FDI inflows (log)* Land/GDP (log)*** Market size Low-skilled labor/ (manufacturing) (log)** GDP (log)*** –0.4 –0.2 0 0.2 0.4 0.6 0.8 –0.4 –0.2 0 0.2 0.4 0.6 0.8 Gross exports (log) Gross exports (log) Backward GVC participation level (log) Forward GVC participation level (log) Backward GVC participation share Forward GVC participation share Sources: WDR 2020 team, using data from CEPII, Eora, ILO, PWT 9.0, UNCTAD, WDI, WGI, and World Bank. See Appendix A for a description of the databases used in this Report. For more detail, see Fernandes, Kee, and Winkler (2019). Note: The graphs show standardized coefficients for each variable on the y-axis. The coefficients are based on a between-effects regression. The depen- dent variables are average exports and backward or forward GVC participation levels and shares. The determinants are measured as averages in the previous decade and include manufacturing import tariffs, FDI inward inflows, distance to major GVC hubs (China, Germany, and the United States), manufacturing value added, political stability index, ratio of low-skilled labor to GDP, ratio of resource rents to GDP, ratio of land to GDP, ratio of capital stock to GDP, nominal exchange rate appreciation, and decade fixed effects. Significance is based on the GVC participation share regressions. Only determinants with statistically significant coefficients are shown. Standardized coefficients refer to how many standard deviations the dependent variable will change per standard deviation increase in the explanatory variable. FDI = foreign direct investment. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. (Box continues next page) 40 | World Development Report 2020 Box 2.2  Modeling results on the drivers of GVC participation (continued) GVC participation shares with their influence on GVC par- GVC participation, whereas a better score in the logistics ticipation levels and on export levels indicates which deter- performance index (LPI) is linked to stronger backward GVC minants matter beyond traditional exports. This assessment participation. Female labor market participation increases also decomposes backward and forward country-level GVC backward GVC participation. And the share of population participation measures into the four broad sectors of agri- speaking English as a second language weakly increases culture, mining, manufacturing, and services to shed light on both forward and backward GVC participation. which sectors are driving the overall cross-country results. To better understand what determines how countries The estimated impacts of the drivers in the baseline model participate in GVCs, measures of backward and forward are shown in figure B2.2.1 (these drivers explain more than GVC participation at the country-sector level are used half the variation in GVC participation shares): in another econometric model that combines country • Low-skilled labor fosters backward GVC participation, endowments (capital, skilled labor, and natural resources), while endowments of natural resources and land fos- institutional quality, and input, output, and market access ter forward GVC participation. tariffs.a The model allows sectors to differ (largely for tech- • Controlling for factor endowments, liberal trade nological reasons) in their intensity of using endowments policy, higher FDI presence, and better institutional and contracts, and it allows results to be given a causal quality are important in determining backward GVC interpretation (figure B2.2.2): participation, while they do not matter (tariffs) or • Sectors using high-skilled labor or capital more inten- they matter in the opposite direction (FDI, political stability) for forward GVC participation. sively exhibit stronger GVC participation and gross • Domestic market size provides a larger pool of local exports in countries relatively more endowed with input suppliers, which lowers backward GVC partici- skilled labor or capital. pation but increases forward GVC participation. • Countries with better institutional quality exhibit stronger GVC participation and exports in their more Decomposing the country-level backward GVC partici- contractually intensive sectors. pation measures by broad sector suggests that the findings • Input tariffs and market access tariffs reduce GVC in figure B2.2.1 are driven largely by backward GVC partic- participation and gross exports.  ipation of the manufacturing sector. The role of other driv- ers of GVC participation shares is also tested. Membership In a separate additional test, sectors using the Internet in preferential trade agreements and the depth of those more intensively exhibit stronger GVC participation and agreements increase backward GVC participation. The gross exports in countries with a higher number of Internet time required to clear imports weakly reduces backward users, controlling for all other determinants. (Box continues next page) Different types of engagement in GVCs require 1990–2015 (such as China, the Czech Republic, Poland, different types of workers. The average annual labor and Turkey) saw their labor costs increase sharply. costs for countries with limited manufacturing GVCs Even countries with limited manufacturing GVCs (such as Costa Rica, Morocco, South Africa, and Sri (such as Cambodia, Indonesia, Nicaragua, and South Lanka) were about $11,000 per worker over 2006–15. Africa) show strong increases in their labor costs in Labor costs reached $16,500 for countries specializing the five years before transitioning (figure 2.2, panel a). in advanced manufacturing and services GVCs (such Sectors using skilled labor more intensively see faster as Mexico, Poland, Thailand, and Turkey). In coun- growth in GVC participation (and in gross exports) tries focusing on innovative GVC activities—such as in countries relatively more endowed with skilled Germany, Japan, the United Kingdom, and the United labor (see box 2.2). The estimated impacts are large: if States—the employee cost was about $55,000 a year Ghana increased its skilled labor share (7.5 percent) to on average, reflecting their higher skill intensity and the cross-country median (20 percent), its backward productivity (figure 2.1). GVC participation and its gross exports would grow Cross-country evidence supports the positive cor- by an estimated 42 percent, and its forward GVC par- relation between skills and integration in innovative ticipation would grow by 39 percent. Further evidence GVCs. Countries that entered the group of advanced for Sub-Saharan Africa shows that skilled labor and manufacturing and services GVCs at some point over higher values of the World Bank’s Human Capital Drivers of participation | 41 Box 2.2  Modeling results on the drivers of GVC participation (continued) Figure B2.2.2  What explains a country-sector’s GVC participation levels and gross exports? a. Endowments and institutions b. Tariffs Natural resources/GDP × natural resources sector*** Input tariff*** participation (log) participation (log) Forward GVC Forward GVC Rule of law index × sector contract intensity*** Market access Capital/labor × tariff*** sector capital intensity*** Skilled labor/labor x Output tariff*** sector skill intensity*** Natural resources/GDP × natural resources sector*** Input tariff** participation (log) participation (log) Backward GVC Backward GVC Rule of law index × sector contract intensity*** Market access Capital/labor × tariff*** sector capital intensity*** Skilled labor/labor x Output tariff sector skill intensity*** Natural resources/GDP × natural resources sector*** Gross exports (log) Gross exports (log) Input tariff*** Rule of law index × sector contract intensity*** Market access Capital/labor × tariff*** sector capital intensity*** Skilled labor/labor x Output tariff sector skill intensity*** 0 0.1 0.2 0.3 0.4 0.5 0 8 6 4 2 0 02 04 .1 .0 .0 .0 .0 0. 0. –0 –0 –0 –0 –0 Sources: WDR 2020 team, using data from Eora, ILO, NBER-CES Manufacturing Industry Database, PWT 9.0, TRAINS, UNIDO, WGI, WITS, Braun (2003), Felbermayr, Teti, and Yalcin (2019), and Nunn (2007). See Appendix A for a description of the databases used in this Report. For more detail, see Fernandes, Kee, and Winkler (2019). Note: The graphs show standardized coefficients for each variable on the y-axis from three separate regressions using forward GVC participation, backward GVC participation, and gross exports as dependent variables. The regressions use a three-year lag of each of the determinants shown in panels a and b and control for country-year fixed effects and sector fixed effects. Standardized coefficients refer to how many standard deviations the dependent variable will change per standard deviation increase in the explanatory variable. Significance level: * = 10 percent, ** = 5 percent, *** = 1 percent. a. This analysis focuses only on differences across countries in the seven subsectors within the overall manufacturing sector in the Eora database. Index8 are positively associated with GVC participa- integration  can mutually reinforce one another. But tion in the region.9 the link between firm GVC participation and female Female labor market participation is linked to corporate leadership  is  negative. Majority female- higher  backward GVC participation (see box 2.2). owned and female-managed  firms are less likely to Evidence from manufacturing firms across 64 devel- participate  in GVCs.  Chapter 3 discusses further the oping countries confirms that the female share of relationship between  GVC participation and  female total employment is higher for firms participating in employment, ownership, and management. GVCs (defined as those that both import intermediate Automation, robotics, and 3D printing could pose inputs and export).10 Verified in all sectors, this pat- a challenge to the GVC participation of countries tern is especially strong in the apparel and electron- whose comparative advantage lies predominantly in ics sectors. A causal link is not warranted, however, abundant low-cost workers. These enterprises require because female labor market participation and GVC higher skills, and they enable customized production 42 | World Development Report 2020 close to the end markets, such as the 3D printing of Figure 2.1  Countries specializing in limited shoes. Producers in lower-income countries typi- manufacturing rely on low labor costs, and countries cally rely more on low-skilled manual labor than do specializing in commodities derive almost a fifth of producers in higher-income countries. But this could GDP from natural resources become more difficult in the context of new technolo- 60 18 gies in GVCs because new technologies are associated Rents from natural resources (% of GDP) with higher-quality standards and high-skilled labor, Annual labor costs per worker 50 15 raising the hurdle for lower-income countries wish- ing to participate in GVCs.11 (Chapter 6 discusses the (US$, thousands) 40 12 potential impacts of new technologies on countries’ prospects for GVC participation.) 30 9 Natural resources are a driving force for 20 6 forward GVC participation Higher relative endowments of land or natural 10 3 resources are both strongly positively correlated with forward GVC participation (see box 2.2). In other 0 0 Commodities Limited Advanced Innovative words, countries with abundant extractive resources, manufacturing manufacturing activities such as copper, iron ore, and other minerals, exhibit and services higher shares of domestic value added embodied in Annual labor costs per worker their partner countries’ exports downstream. Sub- Rents from natural resources (right axis) Saharan countries rich in non-oil natural resources Sources: WDR 2020 team, based on Penn World Table; World Bank’s WDI database; GVC taxonomy for exhibit greater forward linkages to manufacturing the year 2011. GVCs than other countries exhibit.12 Almost a fifth of Note: The left axis shows average annual labor costs and the right axis the average rents from natural GDP originates from natural resources in countries resources as a share of GDP by GVC taxonomy group, with averages over 2006–15. Labor costs were obtained by multiplying a country’s (deflated) GDP by its labor share and dividing by the number specializing in commodities, compared with 3 percent of employees. The average of labor costs for countries specializing in commodities includes several or less for countries operating in limited manufactur- high-income countries (such as Australia, Norway, and Saudi Arabia). See box 1.3 in chapter 1 for a description of the GVC taxonomy used in this Report. ing GVCs (see figure 2.1). FDI acts as a catalyst for GVC integration, capital FDI offers a solution. Cross-country cross- providing foreign capital and technical sector evidence from the Eora database shows that know-how a relative scarcity of capital deters stronger GVC Higher capital endowments stimulate GVC integra- participation in capital-intensive sectors (see box tion and upgrading, but for those countries with scarce 2.2). Countries moving from commodities to limited Figure 2.2  Increases in labor costs and capital stock accompany upgrading in GVCs a. Mean labor costs relative to b. Mean capital stock relative to year of entry to GVC group year of entry to GVC group 1.007 1.02 1.01 Capital stock Labor costs 1.002 1.00 0.997 0.99 0.992 0.98 –5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5 Year of entry Year of entry Limited manufacturing Advanced manufacturing and services Innovative activities Sources: WDR 2020 team, based on Eora; World Bank’s WDI database; GVC taxonomy. Note: The year of entry is normalized to 0 for all countries in a particular GVC group, and the sample used to compute the means shown is based on countries with at least five years of observations before and after entry to the GVC group. Labor costs and capital stock are measured relative to the year of entry. Additional analysis confirms that labor costs and capital stock increase significantly in the five years before and after a switch. Drivers of participation | 43 manufacturing GVCs exhibit a strong increase in cap- is linked to lower forward GVC participation shares ital stock in the five years before the transition (figure driven by GVC integration of agriculture and ser- 2.2, panel b). Because countries can attract FDI to over- vices. Countries attracting FDI in manufacturing may come relative capital scarcity and thus integrate into reduce their exports of raw agricultural goods and GVCs, GVC activity and FDI inflows go hand in hand. intermediate services (such as transportation) embod- When tight control over foreign production processes ied in exports of resource-intensive goods, thereby is necessary (perhaps because of weak contractual lowering their forward GVC participation.19 enforcement or weak protection of intellectual prop- Foreign-owned firms may also promote domestic erty), lead firms might prefer vertical integration of upstream sectors. They increase the demand for local suppliers over an arm’s-length relationship, resulting intermediate inputs and cultivate local suppliers that in intrafirm trade and FDI flows (see chapter 1). may subsequently supply other downstream domes- It is hard to imagine a GVC in which a multi­ tic firms and even export. FDI can ease the entry of national firm is not involved at some stage of the domestic firms into GVCs by, for example, conferring production chain. Vietnam’s success in smartphones technical know-how and transmitting managerial stemmed from investments by Samsung in Vietnam practices. According to the Moroccan minister of to set up Samsung Electronics Vietnam (SEV) in 2008 industry, trade, and new technologies, Moulay Hafid and Samsung Electronics Vietnam-Thai Nguyen El Alami, when Renault-Nissan set up plants in the (SEVT) in 2013 (see box 2.1). Likewise, the Moroccan north of Morocco’s small city of Melloussa, it aimed automotive industry has relied on investments by the to build an “industry ecosystem.” Later, in fact, it French Renault-Nissan Alliance and PSA Group car attracted many other companies specializing in auto companies. Singapore’s Olam, one of the world’s larg- parts production and seeking to supply Renault- est suppliers of cocoa beans, contributed to Ghana’s Nissan. Meanwhile, the government of Morocco is cocoa exports reaching over 23,000 customers world- looking at ways to deepen the country’s backward wide.13 And then there were the earlier success stories linkages. FDI in the apparel sector in Bangladesh led to such as Intel in Costa Rica (until 2014) and Volkswagen new local input suppliers producing zippers, buttons, in South Africa.14 In addition, investors from Taiwan, and fabrics, which also benefited domestic apparel China, in the 1990s and South African investors in the firms and ensured the country’s competitiveness in 2000s were instrumental in developing and expand- global apparel exports (box 2.3).20 Such linkages of sec- ing the apparel value chain in Lesotho, whereas tors and firms through FDI can further deepen coun- Mauritian investors played a similar role for apparel tries’ participation in GVCs.21 Indeed, China has defied in Madagascar.15 In all these cases, foreign-owned the global decline in the share of domestic value added firms were instrumental in jumpstarting the domestic in exports because its large domestic manufacturing economy and integrating production into GVCs. And capacity is supplying the downstream GVC parties yet the reliance on FDI inflows also poses risks: Costa through favorable FDI and trade policies (box 2.4).22 Rica lost many manufacturing jobs to Vietnam in 2014 The link between FDI and GVC participation after Intel abruptly relocated its operations. makes it difficult to disentangle their determi- Although many of these success stories (particu- nants. In their responses to the World Bank’s Global larly in East Asia) are linked to FDI in manufacturing Investment Competitiveness survey, executives at GVCs, much of the growth in FDI over the past two multinational corporations involved in efficiency- decades has come through natural resource–based seeking FDI viewed country endowments as cru- sectors. Such investment differs considerably from cial for their investment decisions. Endowments traditional manufacturing FDI. Investors tend to be included the available talent and skill of labor, the resource-seeking rather than efficiency-seeking or low cost of labor and inputs (including ease of access market-seeking. Investment is also likely to be dis- to imported inputs), and the capacity and skills of persed across a wider set of countries and to emerge local suppliers.23 Favorable exchange rates, good from a widening set of investors (including large physical infrastructure, and low tax rates are also investors from the global South).16 important, as are PTAs, bilateral investment treaties, FDI inflows play a strong role in the extent of back- and investment incentives. (Some of these policy- ward GVC participation shares and levels (see box amenable factors are discussed throughout the chap- 2.2), driven by GVC integration of the manufacturing ter as important drivers of GVC participation. Other sector.17 The lack of foreign-owned firms in manufac- factors are covered in chapter 7.) turing is an important reason for low backward GVC FDI is critical, particularly for countries upgrading participation in Sub-Saharan Africa.18 Meanwhile, FDI their type of participation in GVCs. From 1990 to 2015, 44 | World Development Report 2020 Box 2.3  Sharing suppliers: How foreign firms benefit domestic firms In the development of Bangladesh’s apparel sector, foreign Figure B2.3.1  In Bangladesh, local firms created incentives for local suppliers to improve their suppliers grew as FDI grew from 1985 quality and productivity. Domestic firms that shared local to 2003 suppliers with foreign firms gained access to newer and 12 better local inputs. The spillover effects of shared suppliers helped explain a quarter of the expanded product scope 10 Number of suppliers and a third of the productivity gains of Bangladesh’s 8 domestic firms in the apparel sector from 1999 to 2003. In 6 Bangladesh, foreign apparel firms also fostered the local market supplying intermediate inputs (figure B2.3.1). 4 But the reverse is true when foreign firms leave. In 2 Malaysia, a local supplier sold a special plastic resin to Panasonic for its fax machines and to local manufacturers 0 1985 1990 1995 2000 2005 of box cutters. When Panasonic closed the plant, manufac- turers of box cutters suffered as well. Local intermediate input suppliers FDI firms in garment industry Source: Kee 2015. Source: Kee 2015. Note: FDI = foreign direct investment. Box 2.4  How liberalizing trade and FDI helped China move up in GVCs Global production fragmentation has allowed firms to rely Chinese exports toward industries with high domestic less on domestic inputs for production, as is evident in the value added or in nonprocessing sectors—cannot explain growing backward GVC participation and the declining the upward trend. ratios of value added to gross exports across the world. China is an intriguing exception. How did it defy the global Figure B2.4.1  Domestic value added decline in domestic content in exports, despite its deep in exports from China increased from engagement in GVCs? 2000 to 2007 Firm-level customs transaction data and manufacturing 75 added in Chinese exports (%) firm survey data are used to measure China’s domestic Share of domestic value content in exports (its ratio of domestic value added in exports to gross exports). From 2000 to 2007, the share of 70 domestic content in Chinese exports rose from 65 percent to 70 percent (figure B2.4.1). This upward trend was driven mainly by China’s processing exporters, who substituted 65 domestic for imported intermediate inputs in both volume and variety. After 2000, China’s structural transformation was fueled by trade and foreign direct investment liber- 60 2000 2001 2002 2003 2004 2005 2006 2007 alization that encouraged intermediate input producers in China to expand their product varieties. Exporters in China Measured domestic value added began to buy more domestic intermediate inputs and to 95% confidence interval (upper bound) rely less on imported inputs. Other factors—such as rising 95% confidence interval (lower bound) wages, firm entry and exit, and the changing composition of Source: Kee and Tang 2016. Drivers of participation | 45 Figure 2.3  FDI increases and tariff declines accompany GVC upgrading a. Mean FDI inflows relative to b. Mean manufacturing tariff year of entry to GVC group relative to year of entry to GVC group 1.4 2.00 1.3 1.75 Manufacturing tariff 1.2 FDI inflows 1.1 1.50 1.0 1.25 0.9 1.00 0.8 0.7 0.75 –5 –4 –3 –2 –1 0 1 2 3 4 5 –5 –4 –3 –2 –1 0 1 2 3 4 5 Year of entry Year of entry Limited manufacturing Advanced manufacturing and services Innovative activities Source: WDR 2020 team, based on World Bank’s WDI database and GVC taxonomy. Note: The year of entry is normalized at 0 for all countries in a particular GVC group, and the sample to compute the means is based on countries with at least five years of observations before and after entry to the GVC group. FDI inflows and manufacturing tariffs are measured relative to the year of entry. Additional analysis confirms that FDI inflows increase significantly in the five years before and after a switch, whereas manufacturing tariffs decline significantly over that same period. FDI = foreign direct investment. net FDI inflows picked up substantially for all coun- To minimize cross-hauling of semiprocessed goods in tries in the years before transitioning into a new GVC GVCs, countries often specialize in contiguous stages group (figure 2.3, panel a). The growth of FDI inflows of production. Because larger countries have a larger continues after countries transition into limited man- industrial capacity, they tend to attract a larger set ufacturing GVCs (such as in Argentina, Cambodia, of contiguous stages and reduce the use of imported Indonesia, and South Africa) and to a lesser degree for inputs relative to domestically sourced inputs in their countries transitioning into advanced manufacturing exports (lower backward GVC integration). and services GVCs (such as in China, the Czech Repub- By their sheer size, large countries are likely to be lic, Romania, and Turkey) or into innovative GVC activ- geographically close to the consumers of final goods, ities (such as in Austria, Italy, Korea, and Singapore). so their more “central” location should make them To attract FDI, lower-income countries that face more prone to specialize in downstream stages of pro- substantial infrastructure and regulatory gaps can duction embodying more foreign value added.25 More- establish special economic zones (SEZs) or export over, a large domestic supplier base reduces search processing zones with less burdensome rules for frictions and facilitates the replacement of domestic business and better access to inputs than in the rest of suppliers if there are production disruptions. the country. This approach was central to Bangladesh, Cambodia, Lesotho, and recently Ethiopia successfully Market size and the role of domestic entering the apparel GVC. Such sites account for a suppliers large share of exports and employment in GVCs, but A story from Poland highlights the relationship linkages to the local economy tend to be small.24 How- between market size and GVCs and how industry link- ever, many other countries have struggled to establish ages through the role of domestic suppliers can affect successful zones. Chapter 7 dives deeper into SEZs and outcomes. In 1992 General Motors, one of the world’s their role for GVCs. largest automakers, set up General Motors Poland to import Opel cars for the large Polish domestic market. Two years later, GM Poland commenced production Market size matters activities, and today Poland has become one of the Backward GVC participation in manufacturing as a world’s major auto exporting countries. Through percentage of total exports is lower in large econo- intensive cooperation with Polish auto part suppliers, mies, including China, Japan, and the United States. GM Poland has contributed to the significant growth 46 | World Development Report 2020 in their number and also plays a role in expanding participation is especially acute.27 Higher import their sales to other GM units around the world. tariffs on manufacturing in the 2000s reduced the The effect of market size on GVC participation is propensity of being in the group of countries spe- crucially mediated by links to domestic industries. cializing in advanced manufacturing and services Markets with larger manufacturing sectors are char- GVCs in 2011. Tariffs on intermediate inputs have a acterized by larger forward GVC participation and strong negative impact on both GVC participation smaller backward GVC participation, highlighting and gross exports (see box 2.2). the importance of domestic suppliers for GVC partic- Tariffs on imported intermediates shape countries’ ipation (see box 2.2). A larger manufacturing sector in export bundles, often preventing them from upgrad- the 2000s also increased the likelihood of countries ing to more sophisticated or more profitable products. participating in advanced manufacturing and services For example, Nepal exports tea almost entirely in bulk GVCs or in innovative GVC activities in 2011. to India at about one-tenth of the price for tea sold packaged to Germany or the United Kingdom. To scale Enhancing market size by liberalizing trade up the exports of branded, packaged tea, Nepalese policies entrepreneurs need intermediate inputs such as filter The constraints of a small market and limited local bags. But those are subject to a tariff of 30 percent, inputs can be overcome by liberalizing trade at home plus a 5 percent excise duty, increasing the world price and negotiating liberalization abroad in order to liber- of filter bags for Nepalese exporters by 36.5 percent ate firms and farms from dependence on local inputs and hampering their competitiveness.28 and narrow domestic demand. Regulatory barriers on Exporters can often circumvent high tariffs on both imports and exports, such as tariffs or quotas, imported intermediates by using duty suspension increase trade costs, with consequences for countries’ mechanisms, but these often do not function effi- GVC participation and positioning. Trade barriers ciently. Two examples from South Asia illustrate this increase the cost of imported intermediate inputs and point. Pakistan’s tariffs on intermediates average 8 thus can reduce backward GVC participation. They percent—four times the average in East Asia—and its also translate into higher costs for a country’s exports, regulatory and additional duties (para-tariffs) are high. lowering forward GVC participation. Because tariffs Thus, Pakistani exporters of textiles and apparel— imposed by partner countries increase the costs of the country’s major export sector—rely mostly on exports, reducing tariff barriers can amplify the bene- domestic cotton rather than on imported artificial fits for internationally fragmented production. fibers such as polyester (the leading input to the fast- growing global imports of apparel).29 In principle, Costly imported intermediates are a barrier Pakistani exporters have access to duty suspension to GVC integration schemes for their imported intermediates, such as Successive rounds of trade negotiations and unilat- the Duty and Tax Remission on Exports. In practice, eral trade liberalization efforts have been a driving approvals for remission takes on average 60 days— force for GVC integration over the last three decades. twice the time specified by law—and clearing customs China’s accession to the WTO in 2001 and the accom- after approval takes an extra 5–10 days. For that rea- panying requirement to reduce more than 7,000 son, a mere 3 percent of textile and apparel exporters tariffs ushered in a new era of globalization that use the scheme. In Bangladesh, by contrast, obtaining stimulated GVC participation not only for its home approval for duty suspension on intermediates takes firms but also for those in partner countries in East on average 24 hours, and about 90 percent of textile Asia and beyond. Meanwhile, accession to the world’s and apparel firms use the scheme.30 largest customs union—the EU—was critical in Despite the gradual decline in tariffs over the last bringing the Czech Republic, Hungary, Poland, and decades, especially for manufactured goods, there the Slovak Republic, and later Bulgaria and Romania, are still important differences in the restrictiveness into GVCs.26 of trade policies across countries. Countries special- Lower tariffs on manufacturing goods encourage izing in commodities imposed manufacturing tariffs countries’ backward GVC participation (see box 2.2). averaging 7.5 percent from 2006 to 2015, and those A 1 percentage point decrease in a country’s average with limited manufacturing GVCs imposed tariffs manufacturing tariff is associated with an increase averaging 6.5 percent. Tariffs drop sharply to less than of 0.4 percentage points in that country’s backward 3 percent for countries with advanced manufacturing GVC participation share in gross exports. In Sub- and services GVCs and to less than 2 percent for those Saharan Africa, the negative impact of tariffs on GVC with innovative GVC activities (figure 2.4). Drivers of participation | 47 Figure 2.4  Manufacturing tariffs are high and Table 2.1  South Asian countries preferential trading partners few in countries impose higher barriers to trade connected to commodity GVCs on each other (overall trade 9.0 60 restrictiveness index, 2011) Number of PTA partners Manufacturing tariff (%) 7.5 50 Origin of imports 6.0 40 Importing country South Asia Rest of world 4.5 30 Afghanistan 3.84 4.65 3.0 20 India 4.59 0.50 1.5 10 Nepal 10.59 6.87 0 0 Pakistan 3.00 0.51 Commodities Limited Advanced Innovative Sri Lanka 1.01 0.33 manufacturing manufacturing activities and services Source: Updated estimates by UNCTAD and World Bank (2018), based on their methodology. Manufacturing tariff Number of PTA partners (right axis) Note: The overall trade restrictiveness indexes are computed using applied Sources: WDR 2020 team, based on World Bank’s WDI and Deep Trade Agreements databases and GVC tariffs that take into account bilateral preferences. taxonomy for the year 2011. Note: The left axis shows average manufacturing import tariffs and the right axis the average number of preferential trading partners by GVC taxonomy group, with averages over 2006–15. PTA = preferential trade agreement. export orientation of its major auto producers and its domestic suppliers.33 High local content requirements For countries upgrading their participation in in the country’s industrial policy toward the auto sec- GVCs, manufacturing tariffs fall substantially in the tor—the Inovar-Auto policy (2011–17)—prevented the years prior to such transitions (see figure 2.3, panel sector from participating in GVCs. b). For countries establishing limited manufactur- ing GVCs at some point during 1990–2015—such as Market access can jumpstart GVC Argentina, Cambodia, Indonesia, and South Africa— participation the average manufacturing tariff rates were on aver- Market access, captured by the tariffs in destination age 25 percent higher five years before the transition markets, also plays a role in GVC participation. Sectors compared with the year of the transition. Countries facing on average lower tariffs in destination markets joining the group of advanced manufacturing and exhibit stronger backward and forward GVC partici- services GVCs—such as China, the Czech Republic, pation (see box 2.2). A 1 percentage point decline in the Romania, and Turkey—saw their tariffs drop by half average tariff facing a sector in destination markets from five years before the transition to the time of is associated with an increase in the country-sector’s upgrading and saw a continued decline in the five backward (forward) GVC participation by 6 percent years after upgrading. (7 percent). Low tariffs are necessary but insufficient for high Preferential access is one aspect of special and backward GVC participation because nontariff mea- differential treatment and its objective has been to sures and other barriers at the border also matter. In encourage export-led growth in developing countries. South Asia, nontariff barriers—including para-tariffs But whether preferential access can help developing and other regulatory constraints—increase firms’ pro­ - countries’ exports has sparked disagreement, with duction costs and alter their input mix, thereby affect- skeptics arguing that trade preferences dilute the case ing their long-term export competitiveness. This out- for policy reform at home and lure beneficiaries into come hurts the already low trade and GVC participation sectors in which they lack a comparative advantage.34 in South Asia.31 The overall trade restrictiveness index Preferential access to foreign markets such as that for South Asia countries—capturing the trade policy provided by the Everything but Arms initiative of the distortions that each country imposes on its import European Union and the AGOA of the United States bundle—shows greater protection for imports from can help developing countries’ exports in the short South Asia than from the rest of the world (table 2.1).32 run.35 In the long run, however, the effects are more Brazil’s large automotive sector, which employed nuanced, depending on the prevalent rules of origin more than 500,000 workers in 2016, developed under and their impacts on the development of domestic the shelter of high tariffs and high nontariff measures. suppliers (box 2.5). There is great heterogeneity across But these policies have also been behind the sector’s African countries in the response to AGOA market poor integration into GVCs, reflected in the lack of access preferences. Evidence suggests that for export 48 | World Development Report 2020 Box 2.5  Trade preferences as catalytic aid? Immediately after the European Union granted duty-free The long-term impacts of the African Growth and and quota-free access to Bangladesh under the Everything Opportunity Act (AGOA) on the apparel export performance but Arms (EBA) initiative in 2001, knitwear exports from of African countries were more nuanced. At first, aggregate Bangladesh to the European Union more than doubled, African apparel exports to the United States boomed after from $1.3 billion in 2000 to $3 billion in 2004. During AGOA was enacted, and they then declined after MFA quo- the same period, knitwear exports from Bangladesh to tas ended in 2004 and preference erosion ensued (with the United States also increased by $30 million. Much to competition from Asian giants). They have stagnated in the surprise of many, such generous trade preferences recent years. The aggregate picture, however, is based on resulted not in trade diversion from the rest of the world four different country-level stories (figure B2.5.1). Countries to the preference-granting markets, but in trade creation mostly in Central and West Africa, such as Cameroon, never to the rest of the world. What could explain this finding? took meaningful advantage of AGOA (panel a). Countries Trade preferences can result in a long-term win-win mostly in Southern Africa, such as Eswatini (formerly Swa- scenario for all parties concerned.a The European Union ziland), experienced a boom right after AGOA was enacted, gained from giving trade preferences to Bangladesh under followed by a bust (panel b). Countries such as Lesotho the EBA because its lost tariff revenues were outweighed experienced growth and then stagnation (panel c). And by gains from the lower prices resulting from higher entry countries in East Africa, such as Ethiopia, saw fairly sus- into exporting in Bangladesh. Preferences raised the prof- tained success, albeit starting late in some cases (panel d).b its of potential exporters in Bangladesh, inducing greater As for other countries in these regions, in Madagascar firm entry exports to the European Union. But as firms the contraction in apparel exports to the United States after overcame the fixed costs of production and exporting, the MFA phase-out was driven by a tremendous exit of some began to export to other markets, and exports from firms. In Mauritius, firms did not exit but contracted their Bangladesh to all markets rose. Moreover, Bangladesh exports sharply until a relaxation of the AGOA rules of ori- solidified its position as a major apparel exporter to the gin in 2009 prompted a revival. The sustained dynamism of European Union, even after the conclusion of the Multi­ Kenya and the late growth in Ethiopia were driven largely fibre Arrangement (MFA) quota regime in 2004. The strict by new firms entering the market after 2010 rather than by origin requirements of the European Union’s EBA and its incumbent firms that benefited from large preference mar- potential encouragement of greater local value added gins during the early AGOA period. Thus trade preferences through nurturing stronger domestic suppliers may have do not seem to have nurtured longer-term comparative helped explain these durable benefits. advantage in African countries. (Box continues next page) success, preferential access per se is not sufficient its GVC participation in the electronics sector (see but needs to be complemented by specific domestic box 2.1). Has remoteness prevented countries in Latin policies: lower tariffs, a reduced regulatory burden, America and Sub-Saharan Africa from participating and enhanced connectivity.36 In some cases, as in Ethi- in GVCs? The total distance from Argentina or Chile opia, trade preferences are fundamental to offsetting to the GVC hubs is almost 40,000 kilometers and that a country’s cost disadvantages stemming from lower from Malawi or Mozambique is more than 30,000 labor productivity and higher logistics costs (relative kilometers. These distances contrast with those for to countries such as Vietnam) and so help attract FDI.37 countries specialized in advanced manufacturing and services GVCs and innovative GVC activities, which average 18,000 kilometers. Geography matters The automotive sector relies heavily on fairly Proximity to the hubs in the global trade network— short regional value chains for at least three reasons. China, Germany, Japan, and the United States— Automotive components such as car seats or engines matters for GVC participation. Many value chains can be heavy, bulky, and easily damaged, thereby are not global but regional. Vietnam’s proximity to increasing transportation costs. Just-in-time pro- its regional suppliers of electronic inputs—such as duction and high product variety often require that China, Japan, Korea, and Singapore—clearly helped subcomponents be produced near final assembly. And Drivers of participation | 49 Box 2.5  Trade preferences as catalytic aid? (continued) Figure B2.5.1  Four stories of AGOA apparel exports from Africa a. Missed opportunity: Cameroon b. Boom-bust: Eswatini 1.8 200 Apparel exports (US$, millions) Apparel exports (US$, millions) 1.6 1.4 150 1.2 1.0 100 0.8 0.6 0.4 50 0.2 0 0 97 99 01 03 05 07 09 11 13 15 17 97 99 01 03 05 07 09 11 13 15 17 19 19 20 20 20 20 20 20 20 20 20 19 19 20 20 20 20 20 20 20 20 20 c. Growth and stagnation: Lesotho d. Late success: Ethiopia 500 60 Apparel exports (US$, millions) Apparel exports (US$, millions) 400 50 40 300 30 200 20 100 10 0 0 97 99 01 03 05 07 09 11 13 15 17 97 01 03 05 17 99 07 09 11 13 15 19 19 20 20 20 20 20 20 20 20 20 19 19 20 20 20 20 20 20 20 20 20 AGOA Other duty-free Dutiable U.S. total Source: WDR 2020 team, using data from the World Bank’s Developing Countries’ Trade and Market Access in the European Union and the United States database (U.S. section). Note: Exports are classified by tariff regime eligibility by product-country-year and do not account for preference use. AGOA = African Growth and Opportunity Act. a. Cherkashin et al. (2015). b. Fernandes et al. (2019). final assembly often happens in large end markets by its weakest link, such as customs delays. Supply with local content requirements in return for mar- chain disruptions are especially costly when firms ket access, such as in Brazil, China, India, and South cannot easily resort to alternative suppliers. Trade Africa.38 Morocco took advantage of its geographical delays associated with inefficient connectivity can be proximity to the EU market to become Africa’s largest a large deterrent for relational GVCs requiring coor- producer of passenger vehicles in 2017, surpassing dination and “just-in-time” delivery. Weak contract South Africa.39 enforcement and the need for stronger cooperation Inefficient infrastructure and delays in clearing and repeated interactions among the several agents customs are important sources of high trade costs. participating in the chain may be severely curtailed by The performance of a GVC is often severely impaired a remote location or inadequate air connectivity. 50 | World Development Report 2020 Trade costs can also shape a country’s positioning unpredictable land transport keeps most Sub-Saharan in GVCs. In sequential (or snakelike) GVCs, trade African countries out of the electronics value chain.48 costs compound along the value chain and occur at Estimates suggest that improving trade facilitation a higher incidence in the downstream stages than in halfway to global best practices would stimulate trade the upstream stages. This situation may give remote in the Sub-Saharan Africa region to a far greater extent countries an incentive to specialize in upstream stages than eliminating all import tariffs.49 And although and more central countries an incentive to specialize air transport could help bridge slow land transport in downstream stages.40 Inefficient transport and or long geographical distances, its high cost limits logistics services and weak competition in these ser- low-income country exports to goods with very high vices amplify the trade costs in many manufacturing unit values (such as gold and silver), time-sensitive GVCs with multiple border crossings and can offset goods (fast fashion clothing), and perishable goods other competitive advantages such as low labor costs.41 (cut flowers).50 A day of delay in transit due to a dif- Strong evidence of the negative role played by ferent transport mode choice has a tariff equivalent of longer geographical distances for GVC participation, 0.6–2.1 percent, and the most sensitive trade flows are both backward and forward, can be found using those involving parts and components.51 Meanwhile, the Eora database. This evidence is driven mainly the private provision of cold storage logistics infra- by manufacturing sector GVCs (see box 2.2).42 The structure has enabled the development of the Ethio- longer geographical distances to the GVC hubs in pian floriculture value chain, whereas lack of such China, Germany, and the United States increase a infrastructure is limiting the upgrading potential in country’s likelihood of specializing in commodities, Bangladesh’s aquaculture value chain.52 whereas countries closer to the GVC hubs are more High logistics costs inhibit landlocked countries likely to participate in limited manufacturing GVCs. from participating in GVCs for electronics and fruits Geographical proximity also matters more for trade in and vegetables.53 The average number of days from GVCs than for trade in final goods.43 a warehouse in the origin economy to a warehouse in the destination country in 2006–15 varied greatly Enhanced connectivity can overcome for different types of GVC participation (figure 2.5). geographical barriers and promote Imports by countries specializing in innovative GVC GVC participation activities need less than nine days on average to reach The disadvantage of a remote location can be a warehouse, but one additional week is required for addressed by improving transport and communica- countries specializing in advanced manufacturing tion infrastructure as well as the regulatory frame- and services GVCs, such as the Philippines, Portugal, work—especially competition—governing these ser- and Thailand. By contrast, the average time to import vices. The most remote countries, such as landlocked exceeds one month in countries specializing in com- ones, have policies for important “linking” services modities (not shown in figure 2.5): 42 days to import such as transport and telecommunications that are in Ghana and 92 days to import in Iraq. Infrastruc- perversely restrictive.44 Better connectivity would ture gaps are partly responsible for longer delays in influence the predictability, reliability, and timeliness Africa, while the lack of electronic systems and to a of GVCs.45 lesser extent customs administration and inspections Transport costs remain, according to developing account for more than half of the total delays, accord- country suppliers, the main obstacle to entering, ing to the Doing Business database (figure 2.6). A large establishing, or upgrading in GVCs.46 The geographic portion of long transport times in Sub-Saharan Africa centrality of a country can attract downstream produc- is attributed to cargo dwell times at ports.54 Despite an tion stages in GVCs. But geographic centrality is more already favorable location, Vietnam reduced its aver- related to centrality in the transport network than to age time to import during the period the electronics distance. Perhaps more important for GVC participa- GVC sector expanded, but its connectivity remains tion is economic distance. Countries in Central Asia, worse than that of its regional competitors such as while central in the distance to neighbors, are isolated Thailand (see box 2.1). because of their poor-quality transport networks, An inability to meet requirements for timely pro- their lack of affordable transport services for contain- duction and delivery hurts GVC participation. Trade ers, and the missing links along main infrastructure in parts and components in international production corridors.47 These issues impair their participation in networks is more sensitive to logistics performance the downstream stages of GVCs. Similarly, slow and than trade in final goods and is more likely to suffer Drivers of participation | 51 Figure 2.5  Connectivity is associated with in the face of higher uncertainty in bilateral interna- specialization in more advanced GVCs tional transport times.55 Evidence from the Eora data- base indicates that better scores in the logistics per- 100 formance index are linked to stronger backward GVC 90 participation (see box 2.2). Unpredictability in border Average Internet use (% of population) 80 clearance times for imports lowers survival rates for manufacturing exporters in 48 developing countries.56 70 Moreover, the quality of the national road infrastruc- 60 ture matters for timely delivery to global markets. For Indonesian manufacturing firms, a higher road den- 50 sity in a firm’s province and in neighboring provinces 40 increases the probability of exporting.57 Connectivity is not confined to the physical supply 30 chain of goods; it also includes effective communi- 20 cation between the participants in GVCs. Two ways that improve effective communication are use of the 10 Internet and of the English language. 0 Stronger Internet usage could be linked to stronger 0 5 10 15 20 25 30 35 40 45 GVC integration for at least two reasons. First, a large Average time to import (days) percentage of inputs embodied in exports—about 30 Limited manufacturing percent—are services such as logistics, information Advanced manufacturing and services and communication technology (ICT), and other Innovative activities business services that rely on the Internet. Second, firms in GVCs need to communicate with both their Sources: WDR team, based on World Bank’s WDI and Doing Business databases and GVC taxonomy for the year 2011. suppliers and their customers through Internet-based Note: The bivariate regression line between average time to import and average Internet use is shown in technologies. blue. Figure excludes countries specializing in commodities. Averages are over 2006–15. Countries in which a higher average share of the population is using the Internet exhibit stronger backward GVC integration (see box 2.2). In China, Figure 2.6  Improving customs and introducing expanding Internet access from coastal provinces electronic systems are as important as infrastructure to hinterland provinces increased the density of man- for African trade ufacturing exporters in hinterland provinces, con- trolling for differences across provinces in changing 900 864 skills, capital, and transport infrastructure (map 2.1).58 800 But many countries still have very low Internet Total time reduction (hours) 700 coverage, particularly those specializing in commod- 600 ities. Over 2006–15, only 21 percent of the population 500 of these countries used the Internet, and coverage was 400 even lower than 5 percent in Burkina Faso, Burundi, 300 and Mali. This coverage contrasts sharply with that in 202 200 countries participating in advanced manufacturing 90 86 68 and services GVCs, where half the population on aver- 100 0 age are online. And this share exceeds three-quarters Sub-Saharan Latin America East Asia and South Asia Middle East in countries focusing on innovative GVC activities, Africa and the Pacific and with coverage of over 85 percent in Denmark, Finland, Caribbean North Africa and Sweden (see figure 2.5). Area of reform: English skills have helped India and the Philip- Infrastructure Customs administration pines become attractive offshore destinations for Risk-based inspections Electronic systems business services, including not only call centers but Source: World Bank 2017, 31. also increasingly complex services such as informa- Note: The time reduction captures reforms that were implemented and had a positive impact on the tion technology and finance serving the United King- time for trading across borders indicator from 2016 to 2017. The reforms recorded during this period are aggregated in four wide-ranging categories: electronic systems, customs administration, risk-based dom and the United States. Morocco and Tunisia have inspections, and infrastructure. Regions with no reforms on time are excluded from the figure. become destinations for French firms. 52 | World Development Report 2020 Map 2.1  Growth in Internet density and exporter firm density across provinces in China, 1999 and 2007 a. Number of persons per optical line kilometer, 1999 and 2007 a.1. 1999 a.2. 2007 BEIJING BEIJING Persons per optical line kilometer (millions) 18.0–22.5 13.5–18.0 9.0–13.5 4.5–9.0 0–4.5 b. Number of manufacturing exporting firms per 1,000 inhabitants, 1999 and 2007 b.1. 1999 b.2. 2007 BEIJING BEIJING Manufacturing exporting firms per 1,000 inhabitants 15–100 10–15 5–10 1–5 0–1 Source: Fernandes et al. 2017. IBRD 44646 | AUGUST 2019 A higher portion of people speaking English in important is the quality of institutions, all else being a country is positively correlated with forward equal, for countries’ participation in GVCs? GVC participation (see box 2.2), and proximity Weak contract enforcement is a significant deter- has been shown to be more relevant for GVC trade rent of traditional trade flows, and GVCs are partic- than for trade in final goods.59 Language frictions ularly sensitive to the quality of contractual institu- inhibit knowledge spillovers in GVCs, such as in tions. Because the performance of a GVC depends on Myanmar, where high communication barriers the strength of its weakest link, production delays between domestic managers and Chinese, Japanese, driven by weak contract enforcement could be partic- and Korean managers limit the productivity spill- ularly harmful to GVCs. The presence of relationship- overs from FDI.60 specific investments (such as for the customization of products) and the exchange of large flows of intan- gibles (such as technology, intellectual property, and Institutional quality matters credit) reinforce the potential role of institutional Among the top 25 most politically unstable countries quality as a significant determinant of relational over 2006–15, only the Philippines and Thailand partic- GVC participation.61 GVC linkages relying heavily ipated in advanced manufacturing and services GVCs, on institutional quality also tend to be particularly and only Israel in innovative GVC activities.  How “sticky,” which calls for reputational mechanisms Drivers of participation | 53 of cooperation that partly substitute for the absence integration for their members, and a positive if weak of formal contracting. Under some circumstances, effect is also found for the North American Free Trade vertical integration through FDI may serve as a Agreement (see box 2.2).65 The depth of trade agree- direct (albeit imperfect) substitute for strong contract ments is particularly relevant now that countries are enforcement in the host countries. signing more deep trade agreements exhibiting higher Evidence based on the Eora database reveals that backward GVC participation (chapter 8 discusses deep political stability greatly matters for backward GVC trade agreements in more detail). The African Conti- integration (see box 2.2). Sectors that rely more on nental Free Trade Area, which came into force in 2019, contract enforcement see greater increases in GVC is expected to unleash opportunities for strong GVC participation (and in gross exports) in countries participation in Africa. The channels for PTAs to nur- with better institutional quality, after controlling ture GVC participation include lower tariffs, larger FDI for resource endowments, geography, tariffs, and inflows, shorter distances to GVC hubs, and stronger macroeconomic cycles (see box 2.2). If Mozambique regulatory frameworks that increase political stability. increased its rule of law index to the cross-country But not all PTAs have been conducive to GVC median, its backward GVC participation level would participation. Mercosur has, if anything, impeded its rise by 29 percent, while its forward GVC participation members’ backward GVC participation (see box 2.2).66 level and its exports would grow by 32 percent.62 By Argentina exhibits low backward integration into contrast, countries characterized by lower political GVCs because of its restrictive trade policies, but high stability exhibit higher forward GVC participation (see forward GVC integration because of its rich natural box 2.2). On average across countries, this is driven by resources. If Mercosur were to add deep provisions, participation of the mining sector in GVCs. Indeed, such as commitments to investment and reforms higher average political stability in the 2000s reduced to remove entry barriers and tackle anticompetitive the likelihood of countries specializing in commodi- business practices, Argentina’s GVC integration ties in 2011. Poor institutional quality linked to land would gain substantially.67 Argentina now has only and property rights in Côte d’Ivoire and Ghana has three PTA partners encompassing 57 enforceable hampered growth in their agriprocessing GVCs (pine- deep provisions, compared with 18 PTA partners for apples and cocoa).63 Colombia and 19 for Peru (covering 250 and 263 deep PTAs, especially those with deep provisions, can provisions, respectively). With a Mercosur agreement improve domestic institutions because they help as deep as the agreement among the EU, Colombia, import both reform and technical and financial assis- and Peru in terms of the number of enforceable tance and so result in stronger GVC participation. provisions, Argentina could increase its exports of Over the last decades, most tariff liberalization has parts and components to Mercosur members by 1–9 arisen from the negotiation of bilateral and regional percent. Large potential gains for GVC participation PTAs by developing and developed countries. Tariff from deepening existing PTAs (and from engaging reductions (and certainty about those reductions) are in new deep PTAs) are also possible for the other an important benefit of PTAs, but more countries are Mercosur giant, Brazil.68 But the impacts of PTAs on signing bilateral and regional PTAs that go beyond GVC participation can be subtle because the rules of simple market access. The depth of trade agreements origin under PTAs can influence how GVCs form and is associated with the international fragmentation of expand (box 2.6). production because behind-the-border policies need to be disciplined in trade agreements for GVCs to Transitioning up the GVC operate efficiently. Participation in more advanced GVCs goes hand typology in hand with countries’ engagement with more PTA Over 1990–2015, many countries upgraded their GVC partners (see figure 2.4). The Eora database reveals categories. The Czech Republic moved from limited a supportive role for regional trade blocs and deep manufacturing GVCs in the 1990s to advanced man- trade agreements in promoting countries’ backward ufacturing and services GVCs in the 2000s and to integration in GVCs. Specific trade agreements, such innovative GVC activities after 2010. as those represented by the European Union and the Several determinants identified here as condu- Association of Southeast Asian Nations (ASEAN),64 cive to stronger GVC integration help to explain the are linked to substantially higher backward GVC Czech Republic’s transitions. After the downfall of 54 | World Development Report 2020 the Soviet Union in 1991, the geographical proximity early 1990s at around 5 percent, they had fallen to of the Czech Republic (Czechoslovakia until 1993) to less than 2 percent by 2000. The Czech Republic’s neighboring Austria and Germany and its supply of accession to the European Union in 2004 opened skilled labor at lower labor costs made the country an the doors for PTAs—the European Union being one attractive location for FDI. In the 1990s, its shares of of the deepest PTAs—and the number of PTA part- high-skilled workers (35 percent) and medium-skilled ners jumped from 0 to 45. The 2000s also launched workers (57 percent) were almost identical to Ger­ a new era in which the country emphasized skill many’s, while the average labor costs of a Czech building and innovation. Internet use rose from worker were around $13,800, or less than a third of 35 percent of the Czech population in 2005 to 75 Germany’s $49,000. The country’s appealing labor percent in 2015. The share of high-skilled workers picture led to strong FDI inflows, particularly in auto- further climbed, reaching 40 percent by 2007, while motive and business services, and it was bolstered by R&D expenditure as a percentage of GDP grew the newfound political stability. from 1.1 percent in 2000 to 1.9 percent in 2015, rank- Although average manufacturing import tar- ing the Czech Republic among the countries with iffs were already low in the Czech Republic in the the highest innovation potential in the world.69 Box 2.6  PTAs and GVCs: The role of rules of origin Rules of origin, a central element of preferential trade intermediate good, identifying all final goods that impose agreements (PTAs), state that the eligibility of a final good rule of origin restrictions on its sourcing.b Regressions for preferential tariff treatment requires the production or performed on the impact of these sourcing restrictions sourcing of some of its inputs within the PTA area. PTAs can show that NAFTA’s rules of origin significantly reduced affect firm-level decisions on intermediate input sourcing, the growth rate of Mexican imports of intermediate goods and thus their GVC linkages, through two channels: prefer- from nonmembers relative to the growth rate of imports of ential tariffs and rules of origin. intermediate goods from members. On average, NAFTA’s For preferential tariffs, inputs imported from PTA rules of origin have reduced the growth rate of imports of members face lower (often zero) tariffs than inputs sourced affected goods from nonmembers relative to NAFTA mem- from nonmembers. Rules of origin distinguish goods orig- bers by 30 percentage points. These findings reveal an inating from PTA members from goods originating from effective strengthening of the regional GVC, Factory North nonmembers with the objective of ensuring that goods America.c But they also point to the trade diversion of PTAs imported by one PTA member from another benefiting through the deterrence of imports of intermediate goods from lower PTA tariffs truly originate from the PTA area and from nonmembers. are not simply assembled from components originating in Exemplifying the dramatic changes in sourcing deci- nonmembers. sions—and thus changes in patterns of GVC participation Rules of origin can constrain PTA members by not stemming from changes in rules of origin under a PTA—is allowing them to select the globally most efficient suppliers the Mauritius apparel sector since 2000. Mauritius had of intermediate inputs. In recent surveys, manufacturing been eligible for U.S. nonreciprocal trade preferences under firms in developing countries repeatedly pointed to rules of the African Growth and Opportunity Act (AGOA) since origin as a crucial nontariff barrier.a Rules of origin are diffi- 2001, but it experienced a swing between stringent rules cult to measure because of their legal complexity, but such of origin (2001–09) and liberal rules of origin (2009–15) in measurements did improve for the world’s largest PTA, the its exports of apparel to the U.S. market (figure B2.6.1). A North American Free Trade Agreement (NAFTA). shift across sources of fabric imports followed closely the A novel mapping of all input–output linkages embed- swing in rules of origin, with fabric originating in African ded in NAFTA’s rules of origin is constructed for each countries or the United States until 2009 and then almost final good, identifying all intermediate inputs required entirely from outside Africa and the United States (mostly for its production subject to rules of origin, and for each from Asian countries) from 2010 on.d (Box continues next page) Drivers of participation | 55 Box 2.6  PTAs and GVCs: The role of rules of origin (continued) Figure B2.6.1  Mauritius’s exports of apparel to the United States, by origin of fabric, 2001–15 300 Less stringent rules of origin Apparel exports (US$, millions) 250 200 150 100 50 0 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 Foreign fabric or yarn that is not available African fabric made from U.S. or African yarn in commercial quantity in the United States U.S. fabric, yarn, and thread Foreign fabric (LDC provision) No preference (MFN) Source: Office of Textiles and Apparel (OTEXA), International Trade Administration, U.S. Department of Commerce Note: LDC = least developed country; MFN = most-favored-nation. a. ITC (2015). b. Conconi et al. (2018). c. The term was coined by Baldwin (2013). d. Fernandes et al. (2019). The productivity of the workforce and the availabil- including tariffs, FDI, political stability, customs, ity of high-quality suppliers are major reasons for and logistics. For countries in different regions, the the country’s continuing attractiveness to German relative importance of these determinants differ. For and other multinationals. example, in Sub-Saharan Africa low FDI inflows are The relative importance of different determi- the most important factor deterring backward GVC nants for GVC participation depends on the type participation, while for countries in the Middle East of GVC engagement and on the characteristics of and North Africa (MENA) and in fragile and conflict countries. Bottlenecks specific to different regions situations, low political stability is the severest obsta- and groups of countries hamper their backward GVC cle. Countries in South Asia, Latin America and the participation (box 2.7). To transition across types, all Caribbean, MENA, and the Pacific Islands stand to determinants and policy areas must be improved, benefit the most from tariff liberalization. 56 | World Development Report 2020 Box 2.7  Most important determinants of GVC participation, by taxonomy group and region The determinants of backward GVC participation differ Overall, it is clear that to transition across different across countries, depending on their type of GVC partici- types of GVC participation, several policy areas require pation (table B2.7.1): substantial improvements. The color-coded averages shown in table B2.7.1 suggest that the time to import • An average country in the commodities group is improves substantially from the commodities to the limited characterized by low political stability (–0.6), low manufacturing group, while tariff rates fall drastically from foreign direct investment (FDI) inflows, high man- the limited manufacturing to the advanced manufacturing ufacturing import tariffs (6.6 percent), low customs and services group. The relative importance of lower tariffs efficiency (35 days to import), and low scores in the coincides with backward integration being much higher for logistics performance index (LPI, 2.6). countries specializing in advanced manufacturing and ser- • Countries in the limited manufacturing group see vices than for countries in limited manufacturing (39.8 per- on average improved political stability, 60 percent cent versus 24.1 percent). The innovative activities group higher FDI inflows, 1 percentage point lower average sees improvements on all fronts, most notably in political tariffs (5.6 percent), improved customs efficiency (20 stability and in logistics performance. days to import), as well as improved LPI scores (2.8), Based on the evidence from the cross-country regres- relative to the commodities group. sions (see box 2.2), the most important bottlenecks ham- • Countries in the advanced manufacturing and ser- pering backward GVC participation shares of each World vices group exhibit on average further improved Bank region or group of countries can be summarized political stability, substantially (150 percent) higher as follows, along with the hypothetical impacts of their FDI inflows, substantially lower average tariffs by improvements (table B2.7.2): 3 percentage points (2.6 percent), better customs • Backward GVC integration in South Asia, Sub-Saharan efficiency (13 days to import), as well as a higher Africa, fragile and conflict situations, and the Carib- LPI (3.3), compared with the limited manufacturing bean and Pacific Islands would benefit the most from group. attracting FDI. South Asia and Sub-Saharan Africa • Countries part of the innovative activities group show rank lowest among all regions in terms of FDI inflows. on average improved political stability, 90 percent If South Asia and Sub-Saharan Africa were to improve higher FDI inflows, lower tariffs by 0.9 percentage their average FDI levels to those of the best-performer points (1.7 percent), higher customs efficiency (8.9 Europe and Central Asia (ECA) region, backward GVC days to import), and a better LPI (3.8), relative to the participation for each would increase by an estimated advanced manufacturing and services group. 16 percent.a If fragile and conflict situations improved Table B2.7.1  Backward GVC participation and determinants, by taxonomy group Average backward Average Average Average Logistics GVC participation political Average FDI tariff rate days to performance Taxonomy group share (%) stability index inflow (log) (%) import index Commodities 13.9 –0.6 6.7 6.6 35.4 2.6 Limited manufacturing 24.1 –0.3 7.3 5.6 19.9 2.8 Advanced manufacturing and services 39.8 0.1 8.8 2.6 13.0 3.3 Innovative activities 37.3 0.8 9.7 1.7 8.9 3.8 Source: WDR 2020 team. Note: Averages shown cover the period 2010–15, using the GVC taxonomy for the year 2015. See box 1.3 in chapter 1 for a description of the GVC taxonomy used in this Report. Dark blue relates to the best performance across taxonomy groups, dark red to the worst performance, and lighter shades to intermediate performance. FDI = foreign direct investment. (Box continues next page) Drivers of participation | 57 Box 2.7  Most important determinants of GVC participation, by taxonomy group and region (continued) Table B2.7.2  Backward GVC participation and determinants, by region and group of countries Average backward GVC participation Average political Average FDI Average   share (%) stability index inflows (log) tariff rate (%) East Asia and Pacific 20.0 –0.2 7.3 5.6 Europe and Central Asia 28.9 –0.2 7.4 3.0 Latin America and the Caribbean 18.1 –0.2 7.2 6.3 Middle East and North Africa 14.7 –1.3 7.3 8.8 South Asia 16.1 –1.1 6.1 11.0 Sub-Saharan Africa 17.3 –0.5 6.0 8.6 Fragile and conflict situations 11.6 –1.3 5.4 9.0 Caribbean Islands 17.5 0.1 5.7 9.5 Pacific Islands 15.3 0.1 4.2 8.4 Source: WDR 2020 team. Note: Averages shown cover the period 2010–15. In each region or group of countries, averages are computed based only on World Bank client countries. These groups include only countries that are eligible for lending and are part of the Eora database. Dark blue relates to the best performance across regions or country groups, dark red to the worst performance, and lighter shades or white to intermediate performance. FDI = foreign direct investment. FDI levels to those of the ECA, backward GVC partic- • Backward GVC integration in MENA, South Asia, and ipation could increase by 34 percent on average. For fragile and conflict situations would increase the the Caribbean Islands, GVC participation is estimated most from improved institutional quality. MENA and to grow by 19 percent under that scenario, while for South Asia rank lowest among all regions in terms of the Pacific Islands the increase would be a dramatic political stability. If MENA and South Asia improved 40 percent. their political stability to that of the best-performer • Backward GVC participation in South Asia, the Middle East Asia and Pacific region, backward GVC partici- East and North Africa (MENA), and the Pacific Islands pation in MENA would increase by an estimated 28 would increase the most from import tariff liber- percent and by 20–36 percent in South Asia and in alization. South Asia imposes the highest average fragile and conflict situations. manufacturing import tariff rates across all regions • For Latin America and the Caribbean (LAC), lower (11 percent). If it reduced its tariff rates to those of tariffs could have a high payoff for GVC integration. the best-performer ECA (3 percent), backward GVC If LAC reduced its tariff rates from their average of participation could increase by 20 percent. Under 6.3 percent to the average rate of the best-performer the same scenario, MENA and the Pacific Islands are ECA, 3 percent, backward GVC participation would estimated to experience growth in backward GVC increase by an estimated 7 percent. participation rates of 14–16 percent. a. For any given determinant, the magnitudes reported are obtained as a ratio of (1) the product between the difference in the determinant in the best-performer region and the determinant in the considered region/group and the estimated coefficient on the determinant in cross-country regressions and (2) the average backward GVC participation share in the considered region/group. Estimated coefficients are shown in Fernandes, Kee, and Winkler (2019). 58 | World Development Report 2020 14.  See Freund and Moran (2017) on how governments were Notes successful in using FDI to increase Costa Rica’s and Morocco’s GVC participation. 1. In this chapter, the definition of low-skilled worker or 15.  See Godfrey (2015); Morris and Staritz (2014). low-skilled labor is based on International Standard 16.  Farole and Winkler (2014). Classification of Occupations (ISCO) categories, and it 17.  This positive association is driven by GVC participation covers “elementary occupations,” labeled skill level 1 by in the manufacturing sector only, while there is no the International Labour Organization (ILO). See https:// association between FDI inflows and countries’ GVC www.ilo.org/public/english/bureau/stat/isco/isco08 integration of their agriculture, commodities, or services /index.htm. 2.  Economist (2018). sectors. This could point to a more favorable role of 3. Stitchdiary (2018). efficiency-seeking or market-seeking FDI that looks for 4. See appendix A for a description of the databases used internationally cost-competitive destinations and poten- in this Report. These results appear to contrast with tial export platforms. See Buelens and Tirpák (2017) for those of the McKinsey Global Institute (MGI 2019), further evidence that bilateral FDI stocks are positively which argues that labor-cost arbitrage is a small share associated with the bilateral backward GVC participation of the GVC activity that declined between 2007 and as well as with bilateral gross trade. 2017. The difference in interpretations stems from dif- 18.  Liu and Steenbergen (2019) use the World Bank’s Enter- ferences in definitions and methodology. MGI defines prise Survey data for 139 countries for 2006–18 to show labor-cost arbitrage as exports from countries whose GDP that a lower foreign ownership presence is linked to per capita is one-fifth or less than that of the importing lower backward GVC participation, measured by export- country, and so convergence between developing and ing and importing at the firm level. Based on the same advanced countries will reduce labor-cost arbitrage. source of data, Gould (2018) shows evidence of a strong Importantly, it finds that the overall share of labor- link between foreign participation and integration into cost arbitrage in goods value chains remained roughly global production chains via exporting and importing for constant at 18–19 percent from 2007 to 2017. Only for firms in the East and Central Asia region. labor-intensive goods, such as textiles and apparel, does 19.  However, FDI inflows are important for forward GVC it note a significant decline in labor-cost arbitrage, albeit participation levels according to the Eora cross-country from high levels. Consistent with the analysis presented evidence (see box 2.2). The negative impact of FDI on for- in this Report, it also observes a sharp increase in labor- ward GVC participation shares may also reflect the fact cost arbitrage from 1995 to 2007 and finds labor-cost that some of the countries abundant in natural resources arbitrage is high and rising even in the most recent that exhibit very high values of those shares have low decade in some sectors, such as autos, and in some institutional quality (as shown later in this chapter) and countries, such the United States. attract relatively less FDI. 5.  See Pathikonda and Farole (2017), who extend the tra- 20. Kee (2015). ditional theory of factor content of trade to construct 21. Alfaro-Ureña, Manelici, and Vasquez (2019) also highlight measures capturing the capabilities most relevant in the similar positive improvements for local suppliers that trade of GVC products, as defined by Athukorala (2010) joined multinational supply chains in Costa Rica. and Sturgeon and Memedovic (2011). 22.  Kee and Tang (2016). 6.  Evidence from the Eora database shows a U-shaped 23.  World Bank (2018). relationship between GDP per capita and forward GVC 24.  Taglioni and Winkler (2016). integration across countries. 25.  Antràs and de Gortari (2017). 7. Engman, Farole, and Winkler (2018). 26.  World Bank (2018). 8. The Human Capital Index (HCI) database provides data 27.  See Abudu and Nguimkeu (2019) focusing on Eora data at the country level for each of the components of the for African countries and exploiting variation in coun- HCI as well as for the overall index, disaggregated by tries’ tariff policies over time. gender. The index measures the amount of human cap- 28.  Narain and Varela (2017). ital that a child born today can expect to attain by age 18, 29.  Rocha and Varela (2018). given the risks of poor health and poor education that 30.  The importance of lower tariffs on intermediate inputs prevail in the country where she lives. to foster the use of imported inputs and improve export 9.  See Yameogo and Jammeh (2019), based on Eora cross- performance at the firm level is true both in countries country data for 23 African countries and their compari- poorly integrated into GVCs such as Nepal and Pakistan, son to global evidence for 115 countries. as well as Peru (see Pierola, Fernandes, and Farole 2018) 10. See Rocha and Winkler (2019) for a study using data from and in countries highly integrated into GVCs such as the World Bank’s Enterprise Surveys. China (Bas and Strauss-Kahn 2015). 11. Rodrik (2018). 31. Kathuria (2018). 12. See the evidence in Abreha et al. (2019) based on the Eora 32. The overall trade restrictiveness index measures the uni- database contrasting GVCs of Africa’s manufacturers to form tariff equivalent of a country’s tariff and nontariff GVCs of other developing regions (including in South barriers that would generate the same level of import Asia and East Asia). value for the country in a given year. See UNCTAD and 13. Olam (2016). World Bank (2018) for details on the methodology. Drivers of participation | 59 33.  Sturgeon, Chagas, and Barnes (2017). 56. Vijil, Wagner, and Woldemichael (2019). 34.  Hoekman and Özden (2005); Ornelas (2016). 57.  Rodríguez-Pose et al. (2013). 35.  An example of the loss of that market access illustrates 58.  See Fernandes et al. (2017), who also provide econometric its importance. The suspension of AGOA market access results for a causal impact of Internet access on firm benefits by Madagascar because of its domestic politi- export participation in China. cal unrest in 2009 led to an outflow of Asian FDI and a 59.  Buelens and Tirpák (2017). reduction in exports of apparel to the United States by 60.  Khandelwal et al. (2018). $156 million, or 75 percent, within a year. 61.  Levchenko (2007); Nunn (2007). 36.  See Fernandes, Forero, et al. (2019). 62.  These computations assume an average (mean) sectoral 37.  Interviews with enterprises in Ethiopia and testimonies contractual intensity. of foreign investors discussed in Fernandes, Forero, 63.  See Amanor (2012). et al. (2019) indicate that lead apparel companies in GVCs 64.  ASEAN is a regional intergovernmental organization would not have set up their production plants in Ethiopia comprising 10 countries in Southeast Asia. had AGOA trade preferences not been in place. 65.  Johnson and Noguera (2017) also find that the EU and 38.  Sturgeon and Thun (2018). other preferential trade agreements, especially deep 39.  Morocco World News (2018). agreements, play an important role in decreasing the 40.  Antràs and de Gortari (2017). ratio of bilateral value added to gross exports, a sign of 41.  APEC and World Bank (2018). growth in global production fragmentation. 42.  A study by Kowalski et al. (2015) finds an important role 66.  Mercosur is an economic and political bloc comprising for geographical distance from GVC hubs, based on Argentina, Brazil, Paraguay, Uruguay, and República Trade in Value Added (TiVA) data on GVC participation Bolivariana de Venezuela. from the Organisation for Economic Co-operation and 67.  This is one of the trade liberalization scenarios for Development. Argentina, whose impacts are obtained from a dynamic 43.  Johnson and Noguera (2017) find distance to be a friction computable general equilibrium model, as discussed by for bilateral value added in exports (as well as for bilateral Martínez Licetti et al. (2018). gross exports), whereas Buelens and Tirpák (2017) find 68.  This finding is shown by Hollweg and Rocha (2018), that distance plays a bigger role in GVC trade relative to based on the impact of deep PTAs in a gravity model of trade in final goods. trade on bilateral trade in parts and components. 44.  Arvis, Raballand, and Marteau (2010) emphasize the 69. OECD (n.d.). crucial role of an uncompetitive market structure in the transport sector in explaining the high logistics costs in landlocked countries. Using the World Bank’s Services References Trade Restrictiveness Index, Borchert et al. (2017) show Abreha, Kaleb Girma, Emmanuel K. K. Lartey, Taye Alemu that landlocked countries have more restrictive policies Mengistae, Solomon Owusu, and Albert G. Zeufack. in the transport and communication sectors than coastal 2019. “Africa in Manufacturing Global Value Chains: countries. Cross-Country Patterns in the Dynamics of Linkages.” 45. WEF (2013). Background paper, Office of the Chief Economist, Africa 46.  OECD and WTO (2013). Region, World Bank, Washington, DC. 47.  The evidence is provided by Briceño-Garmendia, Lebrand, Abudu, Derrick, and Pierre Nguimkeu. 2019. “Public Policy and Abate (2018) using a novel measure of country con- and Country Integration to Manufacturing Global Value nectivity that captures the cost, time, and reliability of Chains: The Roles of Trade and Competition Policies, the transport network that enables users to reach rele- Labor Market Regulation, and Tax Incentives.” Back- vant economic destinations, including global GVC hubs. ground paper, Office of the Chief Economist, Africa 48.  Christ and Ferrantino (2011). Region, World Bank, Washington, DC. 49. WEF (2013). Alfaro-Ureña, Alonso, Isabela Manelici, and José P. Vasquez. 50.  Arvis et al. 2011. 2019. “The Effects of Joining Multinational Supply 51. The estimates obtained by Hummels and Schaur (2013) Chains: New Evidence from Firm-to-Firm Linkages.” are based on transport mode choices by U.S. importers. Unpublished working paper, University of California, Similar magnitudes for the cost of a one-day delay in Berkeley, Berkeley, CA. inland transit were found in the World Bank’s Doing Amanor, Kojo Sebastian. 2012. “Global Resource Grabs, Business database by Djankov, Freund, and Pham (2010). Agribusiness Concentration, and the Smallholder: Two 52. See Ponte et al. (2014). West African Case Studies.” Journal of Peasant Studies 39 53. Arvis, Raballand, and Marteau (2010). (3–4): 731–49. 54. Raballand et al. (2012). Ansón, José, Jean-François Arvis, Mauro Boffa, Matthias A gravity model of trade is used to relate bilateral trade 55.  Helble, and Ben Shepherd. 2017. “Time, Uncertainty, and in parts and components or in final goods to the logistics Trade Flows.” ADBI Working Paper 673 (February), Asian performance index by Saslavsky and Shepherd (2014) and Development Bank Institute, Tokyo. to transit times measured by Ansón et al. (2017) using the Antràs, Pol, and Alonso de Gortari. 2017. “On the Geography of database of parcel deliveries compiled by the Universal Global Value Chains.” NBER Working Paper 23456 (May), Postal Union. National Bureau of Economic Research, Cambridge, MA. 60 | World Development Report 2020 APEC (Asia–Pacific Economic Cooperation) and World Bank. Economist. 2018. “Why Samsung of South Korea Is the Biggest 2018. “Promoting Open and Competitive Markets in Road Firm in Vietnam.” April 12. https://www.economist.com Freight and Logistics Services: The World Bank Group’s /asia/2018/04/12/why-samsung-of-south-korea-is-the Markets and Competition Policy Assessment Tool -biggest-firm-in-vietnam. Applied in Peru, The Philippines, and Vietnam.” Unpub- Engman, Michael, Thomas Farole, and Deborah Winkler. lished report, World Bank, Washington, DC. 2018. “Firm Performance and Job Characteristics of Arvis, Jean-François, Robin Carruthers, Graham Smith, and Global Value Chain Participants in Bangladesh: Evidence Christopher Willoughby. 2011. Connecting Landlocked from New Survey Data.” Unpublished working paper, Developing Countries to Markets: Trade Corridors in the 21st World Bank, Washington, DC. Century. Directions in Development: Trade Series. Wash- Farole, Thomas, and Deborah Winkler, eds. 2014. Making ington, DC: World Bank. Foreign Direct Investment Work for Sub-Saharan Africa: Local Arvis, Jean-François, Gaël Raballand, and Jean-François Spillovers and Competitiveness in Global Value Chains. Direc- Marteau. 2010. The Cost of Being Landlocked: Logistics Costs tions in Development: Trade Series. Washington, DC: and Supply Chain Reliability. Directions in Development: World Bank. Trade Series. Washington, DC: World Bank. Felbermayr, Gabriel, Feodora Teti, and Erdal Yalcin. 2019 Athukorala, Prema-chandra. 2010. “Production Networks and (forthcoming). “Rules of Origin and the Profitability of Trade Patterns in East Asia: Regionalization or Globaliza- Trade Deflection.” Journal of International Economics. tion?” Working Paper Series on Regional Economic Inte- Fernandes, Ana Margarida, Alejandro Forero, Hibret Maemir, gration 56 (August), Asian Development Bank, Manila. and Aaditya Mattoo. 2019. “Are Trade Preferences a Baldwin, Richard E. 2013. “Global Supply Chains: Why Panacea? The African Growth and Opportunity Act and They Emerged, Why They Matter, and Where They Are African Exports.” Policy Research Working Paper 8753, Going.” In Global Value Chains in a Changing World, edited World Bank, Washington, DC. by Deborah Kay Elms and Patrick Low, 13–60. Geneva: Fernandes, Ana Margarida, Hiau Looi Kee, and Deborah World Trade Organization. Winkler. 2019. “Factors Affecting Global Value Chain Bas, Maria, and Vanessa Strauss-Kahn. 2015. “Input-Trade Participation across Countries.” Policy Research Working Liberalization, Export Prices, and Quality Upgrading.” Paper, World Bank, Washington, DC. Journal of International Economics 95 (2): 250–62. Fernandes, Ana Margarida, Aaditya Mattoo, Huy Nguyen, Borchert, Ingo, Batshur Gootiiz, Arti Grover Goswami, and and Marc Schiffbauer. 2017. “The Internet and Chinese Aaditya Mattoo. 2017. “Services Trade Protection and Exports in the Pre-Ali Baba Era.” Policy Research Working Economic Isolation.” World Economy 40 (3): 632–52. Paper 8262, World Bank, Washington, DC. Borin, Alessandro, and Michele Mancini. 2019. “Measuring Freund, Caroline L., and Theodore H. Moran. 2017. “Multi- What Matters in Global Value Chains and Value-Added national Investors as Export Superstars: How Emerging- Trade.” Policy Research Working Paper 8804, World Market Governments Can Reshape Comparative Advan- Bank, Washington, DC. tage.” Working Paper 17–1 (January), Peterson Institute Braun, Matias. 2003. “Financial Contractibility and Asset Hardness.” Mimeo, University of California, Los Angeles. for International Economics, Washington, DC. Briceño-Garmendia, Cecilia, Mathilde Lebrand, and Megersa Godfrey, Shane. 2015. “Global, Regional and Domestic Apparel Abate. 2018. “Transport Connectivity: Cost, Time, and Value Chains in Southern Africa: Social Upgrading for Networks in Europe and Central Asia.” Unpublished Some and Downgrading for Others.” Cambridge Journal of working paper, World Bank, Washington, DC. Regions, Economy and Society 8 (3): 491–504. Buelens, Christian, and Marcel Tirpák. 2017. “Reading the Gould, David Michael. 2018. Critical Connections: Promoting Footprints: How Foreign Investors Shape Countries’ Economic Growth and Resilience in Europe and Central Asia. Participation in Global Value Chains.” Comparative Eco- Europe and Central Asia Studies Series. Washington, DC: nomic Studies 59 (4): 561–84. World Bank. Cherkashin, Ivan, Svetlana Demidova, Hiau Looi Kee, and Hoekman, Bernard, and Çağlar Özden. 2005. “Trade Prefer- Kala Krishna. 2015. “Firm Heterogeneity and Costly ences and Differential Treatment of Developing Coun- Trade: A New Estimation Strategy and Policy Experi- tries: A Selective Survey.” Policy Research Working Paper ments.” Journal of International Economics 96 (1): 18–36. 3566, World Bank, Washington, DC. Christ, Nannette, and Michael Joseph Ferrantino. 2011. “Land Hollweg, Claire H., and Nadia Rocha. 2018. “GVC Participa- Transport for Export: The Effects of Cost, Time, and tion and Deep Integration in Brazil.” Policy Research Uncertainty in Sub-Saharan Africa.” World Development 39 Working Paper 8646, World Bank, Washington, DC. (10): 1749–59. Hummels, David L., and Georg Schaur. 2013. “Time as a Trade Conconi, Paola, Manuel García-Santana, Laura Puccio, and Barrier.” American Economic Review 103 (7): 2935–59. Roberto Venturini. 2018. “From Final Goods to Inputs: ITC (International Trade Centre). 2015. “The Invisible Barriers The Protectionist Effect of Rules of Origin.” American to Trade: How Businesses Experience Non-Tariff Mea- Economic Review 108 (8): 2335–65. sures.” Technical Paper MAR-15-326.E, ITC, Geneva. Djankov, Simeon, Caroline L. Freund, and Cong S. Pham. Johnson, Robert Christopher, and Guillermo Noguera. 2017. 2010. “Trading on Time.” Review of Economics and Statistics “A Portrait of Trade in Value-Added over Four Decades.” 92 (1): 166–73. Review of Economics and Statistics 99 (5): 896–911. Drivers of participation | 61 Kathuria, Sanjay, ed. 2018. A Glass Half Full: The Promise of Olam. 2016. Olam: Maximising Value and Purpose: Annual Report Regional Trade in South Asia. South Asia Development 2015. Singapore: Olam. http://olamgroup.com/investor Forum. Washington, DC: World Bank. -relations/annual-report-2016/. Kee, Hiau Looi. 2015. “Local Intermediate Inputs and the Ornelas, Emanuel. 2016. “Special and Differential Treatment Shared Supplier Spillovers of Foreign Direct Investment.” for Developing Countries.” In Handbook of Commercial Pol- Journal of Development Economics 112 (January): 56–71. icy, vol. 1A, edited by Kyle Bagwell and Robert W. Staiger, Kee, Hiau Looi, and Heiwei Tang. 2016. “Domestic Value 369–432. Handbooks in Economics. Amsterdam: Elsevier. Added in Exports: Theory and Firm Evidence from Pathikonda, Vilas, and Thomas Farole. 2017. “The Capabilities China.” American Economic Review 106 (6): 1402–36. Driving Participation in Global Value Chains.” Journal of Khandelwal, Amit Kumar, Louise Guillouet, Rocco Macchi- International Commerce, Economics and Policy 8 (1): 1–26. avello, and Matthieu Teachout. 2018. “Communication Pierola, Martha Denisse, Ana Margarida Fernandes, and Frictions and Knowledge Transfers: Evidence from FDI.” Thomas Farole. 2018. “The Role of Imports for Exporter Paper presented at “Macroeconomics, Trade, and Finance Performance in Peru.” World Economy 41 (2): 550–72. Seminar,” World Bank, Washington, DC, October 18. Ponte, Stefano, Ingrid Kelling, Karen Sau Jespersen, and Kowalski, Przemyslaw, Javier Lopez Gonzalez, Alexandros Froukje Kruijssen. 2014. “The Blue Revolution in Asia: Ragoussis, and Cristian Ugarte. 2015. “Participation of Upgrading and Governance in Aquaculture Value Developing Countries in Global Value Chains: Implica- Chains.” World Development 64 (December): 52–64. tions for Trade and Trade-Related Policies.” OECD Trade Raballand, Gaël, Salim Refas, Monica Beuran, and Gözde Policy Paper 179, Organisation for Economic Co-operation Isik. 2012. Why Does Cargo Spend Weeks in Sub-Saharan and Development, Paris. African Ports? Lessons from Six Countries. Directions in Levchenko, Andrei A. 2007. “Institutional Quality and Inter- Development: Trade Series. Washington, DC: World national Trade.” Review of Economic Studies 74 (3): 791–819. Bank. Liu, Yan, and Viktor Steenbergen. 2019. “The Role of FDI Rocha, Nadia, and Gonzalo J. Varela. 2018. “Unlocking Private in Global Value Chains (GVCs): Implications for Sub- Sector Growth through Increased Trade and Investment Saharan Africa.” Unpublished working paper, World Competitiveness in Pakistan.” Unpublished working Bank, Washington, DC. paper, World Bank, Washington, DC. Martínez Licetti, Martha, Mariana Iootty, Tanya Goodwin, Rocha, Nadia, and Deborah Winkler. 2019. “Trade and and José Signoret. 2018. Strengthening Argentina’s Inte- Female Labor Participation: Stylized Facts Using a gration into the Global Economy: Policy Proposals for Trade, Global Dataset.” Background paper, World Bank-World Investment, and Competition. International Development Trade Organization Trade and Gender Report, World in Focus Series. Washington, DC: World Bank. Bank, Washington, DC. MGI (McKinsey Global Institute). 2019. Globalization in Rodríguez-Pose, Andrés, Vassilis Tselios, Deborah Winkler, Transition: The Future of Trade and Value Chains. New York: and Thomas Farole. 2013. “Geography and the Determi- McKinsey. nants of Firm Exports in Indonesia.” World Development Morocco World News. 2018. “WSJ: Morocco Is Leading 44 (April): 225–40. Africa’s Automotive Industry.” October 1. https://www Rodrik, Dani. 2018. “New Technologies, Global Value .moroccoworldnews.com/2018/10/254381/morocco-africa Chains, and Developing Economies.” NBER Working -automotive-industry/. Paper 25164, National Bureau of Economic Research, Morris, Mike, and Cornelia Staritz. 2014. “Industrialization Cambridge, MA. Trajectories in Madagascar’s Export Apparel Industry: Saslavsky, Daniel, and Ben Shepherd. 2014. “Facilitating Ownership, Embeddedness, Markets, and Upgrading.” International Production Networks: The Role of Trade World Development 56 (April): 243–57. Logistics.” Journal of International Trade and Economic Devel- Narain, Ashish, and Gonzalo J. Varela. 2017. “Trade Policy opment 23 (7): 979–99. Reforms for the Twenty First Century: The Case of Stitchdiary. 2018. “What Makes Bangladesh—A Hub of Nepal.” World Bank, Washington, DC. Garment Manufacturing?” July 18. https://medium.com Nikkei Asian Review. 2018. “Samsung Suppliers in Vietnam /@stitchdiary/what-makes-bangladesh-a-hub-of-garment Branch Out to New Frontiers.” July 3. https://asia.nikkei -manufacturing-ce83aa37edfc. .com/Business/Business-trends/Samsung-suppliers-in Sturgeon, Timothy J., Leonardo Lima Chagas, and Justin -Vietnam-branch-out-to-new-frontiers. Barnes. 2017. Inovar Auto: Evaluating Brazil’s Automotive Nunn, Nathan. 2007. “Relationship-Specificity, Incomplete Industrial Policy to Meet the Challenges of Global Value Chains. Contracts, and the Pattern of Trade.” Quarterly Journal of Washington, DC: World Bank. Economics 122 (2): 569–600. Sturgeon, Timothy J., and Olga Memedovic. 2011. “Mapping OECD (Organisation for Economic Co-operation and Devel- Global Value Chains: Intermediate Goods Trade and opment). No date. “Business Brief: The Czech Republic’s Structural Change in the World Economy.” Development Fourth Industrial Revolution.” http://www.oecd.org Policy and Strategic Research Branch Working Paper /innovation/czech-republic-fourth-industrial-revolution 05/2010, United Nations Industrial Development Organi- .htm. zation, Vienna. OECD (Organisation for Economic Co-operation and Devel- Sturgeon, Timothy J., and Eric Thun. 2018. “China: New Driv- opment) and WTO (World Trade Organization). 2013. Aid ers of Growth: Case Studies of China’s Automotive and for Trade at a Glance 2013: Connecting to Value Chains. Geneva: ICT Hardware Sectors.” Background paper, World Bank, WTO; Paris: OECD. Washington, DC. 62 | World Development Report 2020 Sturgeon, Timothy J., and Ezequiel Zylberberg. 2016. “The Dynamics.” Policy Research Working Paper 8793, World Global Information and Communications Technology Bank, Washington, DC. Industry: Where Vietnam Fits in Global Value Chains.” WEF (World Economic Forum). 2013. “Enabling Trade: Policy Research Working Paper 7916, World Bank, Wash- Valuing Growth Opportunities.” WEF, Geneva. ington, DC. World Bank. 2017. Doing Business 2019: Reforming to Create Jobs. Taglioni, Daria, and Deborah Winkler. 2016. Making Global Washington, DC: World Bank. Value Chains Work for Development. Trade and Develop- ——— —. 2018. Global Investment Competitiveness Report 2017/2018: ment Series. Washington, DC: World Bank. Foreign Investor Perspectives and Policy Implications. Wash- UNCTAD (United Nations Conference on Trade and Devel- ington, DC: World Bank. opment) and World Bank. 2018. “The Unseen Impact of Yameogo, Nadege Desiree, and Kebba Jammeh. 2019. “Deter- Non-tariff Measures: Insights from a New Database.” minants of Participation in Manufacturing GVCs in Report UNCTAD/DITC/TAB/2018/2, UNCTAD, Geneva. Africa: The Role of Skills, Human Capital Endowment, Viet Nam News. 2015. “Samsung R&D Centre Hitting Stride.” and Migration.” Policy Research Working Paper 8938, December 3. https://vietnamnews.vn/economy/279340 World Bank, Washington, DC. /samsung-r-d-centre-hitting-stride.html. Vijil, Mariana, Laurent Wagner, and Martha Tesfaye Woldemichael. 2019. “Import Uncertainty and Export Drivers of participation | 63 PART III What are the effects of GVCs? 3 Consequences for development 4 Macroeconomic implications 5 Impact on the environment 6 Technological change 3 Consequences for development Key findings  yperspecialization and durable firm-to-firm relationships promote efficient production • H and the diffusion of technology, as well as access to capital and inputs along value chains. The result is increased productivity and income growth—more so than what countries achieve through domestic production but also than what they achieve through trade in finished goods. • H  ow countries participate in global value chains (GVCs) matters for the impact on development. Countries experience the biggest growth spurt during the transition out of commodities into basic manufacturing activities.  VCs deliver more productive jobs, primarily through scale effects that result from • G increased productivity and expanded output. Because they boost income and productive employment, participation in GVCs is associated with reduced poverty.  he gains from GVC participation are not distributed equally across and within countries. • T Inequalities arise in the distribution of firm markups across countries; in the distribution of capital and labor, between skilled and unskilled workers as well as between male and female workers; and geographically within countries.  he expansion of GVCs has magnified the challenges facing the international tax system. • T The tax revenue losses from profit shifting and tax competition are substantial, particularly for lower-income countries. 66 | World Development Report 2020 B angladesh is a powerful example of how partic- better fire, building, and worker safety, and they have ipation in global value chains (GVCs) has sup- taken steps to reduce water waste and environmental ported economic growth and structural change. damage.5 In response to demands from international In 1988 Bangladesh’s exports of apparel and footwear buyers, and learning from international best practices, were negligible, accounting for less than 1 percent of Bangladeshi producers are increasingly recognizing the global total. Since then, the business of exporting that they must not only improve their practices, but apparel made from imported textiles has grown on also ensure that improvements can be independently average by nearly 18 percent a year. Bangladesh now verified by third parties. exports 7 percent of the world’s apparel and footwear— Is Bangladesh an isolated experience? This chapter third only to China (which increasingly sources from examines whether GVC participation promotes devel- Bangladesh) and Vietnam.1 The sector accounts for 89 opment beyond what countries can achieve through percent of the country’s exports and 14 percent of GDP, standard trade, or whether it makes the development and it employs 3.6 million workers, 55 percent of them path harder. It considers cross-country evidence, but women.2 Diversification is also under way. The plas- also dives deeper into firm-level evidence from a few tics sector has benefited from complementarities with countries—especially Ethiopia, Mexico, and Viet- the ready-made garment sector because garments nam—to demonstrate the complexities of GVC par- are enclosed in plastic packaging. Leather goods and ticipation. The evidence indicates that the challenges, footwear are growing rapidly (second-largest export opportunities, successes, and failures of Bangladesh category). Meanwhile, agriculture’s share of GDP fell reflect how other countries are forging their develop- from 70 percent in 1988 to 38 percent in 2018, and the ment path in a GVC world. However, their outcomes share of people in extreme poverty from 44 percent to are also shaped by national choices about policies, 15 percent in 2016.3 institutions, and other factors. Navigating globalization has been challenging. GVCs support productivity gains and income Low wages drive Bangladesh’s export success, and in growth because of their two defining characteristics: the past 30 years there has been little upgrading to long-term firm-to-firm relationships and hyperspe- better-paid tasks. Demands for higher wages in the cialization in specific tasks. In cross-country studies, factories recently spilled into social unrest in the streets a 10 percent increase in the level of GVC participation in the form of strikes and protests.4 Tragic incidents, is estimated to increase average productivity by close such as the April 2013 collapse of the Rana Plaza build- to 1.6 percent and per capita GDP by 11–14 percent— ing in Dhaka and the garment factory it housed, where or much more than the 2 percent income gain from 1,134 lives were lost, highlighted the poor safety condi- increasing trade in products fully produced in one tions in some parts of the value chain, particularly in country by a comparable amount. the more peripheral but numerous contractor factories. In GVCs, domestic firms become interdependent Moreover, unplanned growth of the sector has strained with foreign firms that share know-how and technol- scarce land resources as well as water resources—the ogy with their buyers and suppliers. Because of hyper- sector consumes nearly twice as much water as the specialization, exporting no longer requires mastering entire population of the capital, Dhaka, and ground­ the entire production process of a good; countries can water levels are dropping at more than 2 meters a year. specialize in only a few tasks in the production pro- The relational nature of GVCs may help gradually cess. For these two important reasons, firms in devel- to mitigate these problems. Large, formal exporters in oping countries that participate in GVCs tend to be GVCs tend to pay well and offer safe conditions, unlike more productive, and all forms of GVC participation the less visible subcontractors further up the value are associated with higher income growth than stan- chain. But because those suppliers are associated with dard trade. The biggest growth spurt, however, comes global brands, poor working conditions, safety and when countries such as Bangladesh, Cambodia, and environmental concerns, and worker dissatisfaction Vietnam break out of commodities or agriculture into have captured the attention of global consumers and basic manufacturing. Empirical evidence suggests civil society, who are urging improvement. With the that within three years of joining a manufacturing support of donors and in coordination with local public GVC, a country is more than 20 percent richer on a per institutions, some international buyers have ramped capita basis. up monitoring of indirect suppliers and undertaken a Alongside the productivity and income gains, GVCs series of initiatives to improve the governance of the deliver more and better jobs. Production is more cap- value chain, together with social and environmental ital-intensive, perhaps because machines allow pro- practices. Among others, they have begun to enforce duction on a large scale and can deliver the precision Consequences for development | 67 required for compatible parts. Because of the greater Finally, GVCs do not cause tax avoidance and tax reliance on machinery, GVC exports require fewer competition, but their evolution has magnified the units of human work per unit of production com- challenges facing the international tax system. The pared with non-GVC exports. But the overall effects growth of intangibles in global business and the digital on employment in the relevant firms and sectors have delivery of services are further exacerbating a preexist- been positive because of the large boost to exports. ing problem. Moreover, in GVCs that involve affiliates The new activities that GVCs bring to countries pull of the same firm, fragmentation of production also workers out of less productive tasks and into more leads to greater intrafirm trade and more opportuni- productive manufacturing jobs. Between 2000 and ties for tax avoidance by manipulating where profits 2014, for example, the labor force of Ethiopian firms are recognized for tax purposes. The tax revenue losses that became importers and exporters—a measure of from profit shifting are substantial, and they are par- GVC participation—grew by 39 percent relative to ticularly large for developing countries. In 2013 non- when they were nontraders, despite the fact that they OECD (Organisation for Economic Co-operation and also utilized 145 percent more capital per worker than Development) countries missed out on $200 billion in nontrading firms. tax revenue as a result of this practice. GVC firms also tend to employ more women than Policy intervention is important to address the other firms, improving their livelihoods and those challenges, attenuate the costs, and share the benefits of their families. In Bangladesh, for example, young of GVC participation. Although GVCs have been able women in villages exposed to the GVC-dominated gar- to drive pro-poor growth over the past 30 years, with ment sector delay marriage and childbirth, and young the steepest declines in poverty occurring in precisely girls gain an additional 1.5 years of schooling. those countries that became integral to GVCs, only By boosting income and employment growth, GVC additional efforts can pull the remaining 2 billion peo- participation also reduces poverty. Because economic ple out of poverty without exceeding environmental growth and employment gains from GVCs are larger limits. The policy chapters of this Report discuss these than from conventional trade, poverty reduction from considerations in detail. GVCs can also be expected to be larger than that pro- duced by such trade. GVCs, however, create some challenges. First, the Economic growth gains from GVC participation may be distributed Trade openness and GVC integration are contrib- unequally within and across countries. Large corpo- uting to better economic performance (figure 3.1).7 rations that outsource parts and tasks to developing The rise of GVCs has generated even greater income countries have seen an increase in markups, suggest- gains than a commensurate expansion of traditional ing that cost reductions are not being passed on to con- trade.8 These gains stem from the productivity effects sumers.6 At the same time, markups for the producers of GVCs. Figure 3.2 depicts the positive association of these inputs in developing countries are declining. between growth in manufacturing productivity and So, too, is the share of income accruing to labor in both growth in GVC participation. Backward participa- developed and developing countries. Technological tion in GVCs is particularly important—a 10 percent change and higher markups reallocate value added increase in the level of GVC participation increases in from labor to capital within countries. Inequality can turn average productivity by close to 1.6 percent.9 also arise within the labor market, with growing pre- Because GVCs are a firm-level phenomenon, the miums for skills. Women are generally employed in greater productivity gains are attributable to firms lower-value-added segments, and women owners and becoming more productive. In the cashew value managers are largely missing in GVCs. Inequality has chain in Mozambique, for example, processors for a geographic dimension too, with GVCs concentrated international brands introduced new semiautomatic in urban agglomerations and in border regions for equipment that increased capacity, reduced costs, and countries neighboring GVC partners. boosted productivity.10 Firm-level empirical evidence Second, in some countries and sectors, firms could supports the association of GVC participation with be stuck in dead-end tasks with few opportunities to higher productivity observed in cross-country data innovate, upgrade, and diversify. The skill mix of the and anecdotally. Firm-level data can identify the set domestic workforce, the organization and governance of firms in a country that participate in trade, further of some value chains, and the nature of certain tech- distinguishing between firms that export, firms that nologies may not favor the process of learning and import, and firms that both export and import. When innovation typical of relational GVCs. a given firm in a country both imports and exports, 68 | World Development Report 2020 Figure 3.1  GVC participation is associated with growth in exports and incomes a. Total export growth b. Income growth 40 40 30 30 Growth in GDP per capita (%) 20 20 Growth in total exports (%) 10 10 0 0 –10 –10 –20 –20 –30 –30 –40 –40 –50 –50 –1.0 –0.5 0 0.5 1.0 –1.0 –0.5 0 0.5 1.0 Growth in foreign value added in exports (%) Growth in foreign value added in exports (%) Source: WDR 2020 team, using data from World Bank’s WDI database. See appendix A for a description of the databases used in this Report. Note: Each dot is a country-year observation. In both panels, the x-axis is the average annual growth in foreign value added in exports between 1990 and 2015. In panel a, the y-axis is the average annual growth in total exports between 1990 and 2015. Total export growth includes exports of goods and services. In panel b, the y-axis is the average annual growth in per capita GDP in purchasing power parity terms between 1990 and 2015. R-squared is 0.73 for total export growth and 0.25 for GDP per capita growth. the likely conclusion is that this firm participates Figure 3.2  GVC participation is associated with in GVCs. In Ethiopia and across a large sample of growth in productivity countries, GVC firms in manufacturing show higher 0.4 productivity (labor productivity, controlling for capital intensity) than one-way traders or nontraders (figure Labor productivity (log difference) 0.3 3.3). Firms that both import and export are 76 percent 0.2 more productive than nontrading firms, compared with a 42 percent difference for export-only firms and 0.1 a 20 percent difference for import-only firms.11 In Viet- 0 nam, this relationship holds across firms in all sectors: manufacturing, services, and agriculture alike. –0.1 Intuitively, there are two complementary explana- –0.2 tions for higher growth and productivity. First, GVCs allow countries to benefit from the efficiency gained –0.3 from a much finer international division of labor. –0.4 GVCs exploit the fact that countries have different –0.6 –0.4 –0.2 0 0.2 0.4 0.6 0.8 comparative advantages not only in different sectors, GVC participation (log difference) but also in different stages of production within sec- Source: Constantinescu, Mattoo, and Ruta 2019. tors. By breaking up complex products, GVCs allow Note: Each dot represents a country-year combination for 1995–2009. GVC participation is measured countries to specialize in specific parts or tasks of as the sum of the foreign value added embodied in a country’s gross exports (backward linkages) and production, escaping domestic supply and demand the country’s domestic value added embodied in other countries’ gross exports (forward linkages). Labor productivity is computed as the real value added divided by the number of persons employed in constraints. China’s “Button Town,” where hundreds manufacturing (excluding petrochemicals). R -squared is 0.22. of factories produce more than 60 percent of all but- tons on Earth, is an extreme example.12 competition for domestic producers. In GVC trade, Second, growth and productivity gains stem from openness also increases imports of intermediate better access to a greater variety of higher-quality inputs, and domestic firms using those inputs observe or less costly intermediate inputs.13 In traditional positive effects on their productivity. Because of these trade, where products cross borders only as finished mechanisms, export growth can be expected to raise products, greater openness to imports entails greater domestic income and employment even when exports Consequences for development | 69 Figure 3.3  Firms that both export and import are Interdependent firms may share know-how and more productive technology with suppliers because such sharing boosts their own productivity and sales, leading to 120 trading firms and nontrading firms (%) faster catch-up growth across countries. Unlike in Productivity difference between 100 traditional trade in which firms in different countries compete, GVCs are networks of firms with common 80 goals. Those goals include minimizing the costs of production or maximizing the profits of the entire pro- 60 duction chain of which they are part.17 Downstream firms typically benefit when their suppliers become 40 more productive and vice versa. A direct implication 20 of this simple observation is that firms from countries specializing in innovation-intensive GVC tasks might 0 find it beneficial to share process and product inno- Ethiopia Vietnam Developing countries vations with their GVC coparticipants specializing Firm type: in simple or advanced manufacturing and services Export and import Export only Import only GVC tasks. Furthermore, the stickiness—or long-term Sources: Ethiopia: Choi, Fukase, and Zeufack (2019), based on 2000–2014 manufacturing census (firms nature—of relational GVCs makes firms particularly with 10 or more employees). Vietnam: WDR 2020 team, based on 2014 Enterprise Surveys (firms with more than five employees). Developing countries: WDR 2020 team, based on World Bank’s Enterprise prone to benefit from learning-by-importing and Surveys (sample of 81 developing countries). learning-by-exporting through repeated interactions Note: The figure reports the percentage difference in productivity between nontrading firms and with highly productive firms at the global frontier of (1) firms that both export and import or (2) firms that only export or (3) firms that only import. The results are obtained by regressing firm labor productivity (log sales per worker) on dummy variables knowledge. marking the type of firm (export and import, export only, or import only), controlling for log capital per In Kenya, South Africa, and Uganda, for example, worker and fixed effects. The Ethiopia estimation controls for sector, year, and region fixed effects, as well as for whether the firm is state owned. The Vietnam estimation controls for sector and region fixed improved processes in horticulture were induced by effects as well as for whether the firm is state- or foreign-owned. The developing countries estimation demand for higher quality and sourcing requirements controls for country-sector, subnational region, and year fixed effects. All coefficient estimates are statistically significant. The percent differences reported in the graph are obtained as 100 multiplied by by global and regional supermarket chains, allowing the exponential of the coefficient estimates minus 1. in turn diversification and higher yields of fresh fruit and vegetable exports.18 In Kenya, incomes increased have lower domestic content (discussed shortly).14 after contract farmers adopted the quality standards Reinforcing this productivity enhancement is the fact demanded by their international buyers, and these that exporting to the global market allows for greater firms supported better traceability of the product economies of scale.15 along the entire supplier network.19 These observations are consistent with empirical Trade between firms engaging in GVCs has char- findings. Increasing direct and indirect exports and acteristics very similar to those of intrafirm trade imports of goods, services, parts, and components pro- because external international sourcing requires the duced through GVCs has been associated with much same high levels of coordination, intense bilateral larger per capita income growth than other forms of information flows, and harmonization and integration trade openness (box 3.1). of many business services as intrafirm internationally fragmented production.20 In the coffee value chain in Relational GVCs are a vehicle for Costa Rica, trade transactions conducted within inte- technology transfer grated firms (intrafirm) and those conducted within It is well accepted that real income grows when epi- long-term relationships with other firms (interfirm) sodes of trade liberalization boost the diffusion of new are similar to one another but starkly different from technology.16 Those positive effects are even greater trade transactions conducted between anonymous in relational GVC trade. As observed in chapter 1, firms.21 in contrast to “standard” trade carried out in anony- Additional empirical evidence supports the mous markets, GVCs typically involve longer-term hypothesis that firms in GVCs work toward common firm-to-firm relationships. This relational nature of goals. A 2018 survey of 1,476 apparel, textile, and infor- GVCs makes them a particularly powerful vehicle mation and communication technology (ICT) firms for technology transfer along the value chain. Firms in Ethiopia and Vietnam found that the probability have a shared interest in specializing in specific tasks, of a buyer providing its suppliers with some form of exchanging technology, and learning from each other. assistance is greater in strongly relational GVCs—that 70 | World Development Report 2020 Box 3.1  Dynamic estimations of the relationship between GVC participation and per capita income growth Growth regressions have been estimated for a panel of 100 Figure B3.1.1  GVC trade is associated countries across income groups for the period 1990–2015. with larger per capita income than A standard Solow growth model was augmented with non-GVC trade measures of GVC participation. Specifically, the log GDP per 1.2 capita was regressed on its lagged value, a vector of the standard determinants of growth, and measures of backward covariate on the x-axis on per capita GDP (%) and forward GVC participation. To reflect the dynamic nature Long-run effects of a 1% increase in the 1.0 of growth, the equation was estimated in a dynamic panel setting, through a System Generalized Method of Moments 0.8 (System-GMM). A 1 percent increase in GVC participation is associated with a more than 1 percent increase in per capita income 0.6 in the long run. The point estimates of the relationship are reported in figure B3.1.1. The estimation is robust to various statistical tests, 0.4 including reverse causality, diagnostic tests for weak instruments, and those for the strength of the chosen 0.2 instruments. The difference in coefficients for backward and forward GVC integration suggests that the development impact for 0 Backward Forward Non-GVC Gross a commodity producer integrated in GVCs only through GVC GVC exports savings forward linkages is much lower than that for a country (% of output) (% of GDP) producing intermediate inputs, which benefits from both Sources: WDR 2020 team, using data from World Bank’s WDI database. forward and backward linkages. is, firms selling exclusively to a single buyer are 38 growth, and wage increases: a 1 percent increase in percent more likely to receive assistance than firms training is associated with 0.6 percent increase in with a diversified client base. Firms without strong value added per hour and a 0.3 percent increase in the relationships are 29 percent less likely to receive assis- hourly wage.22 A case study of the impact of a Japanese tance from a client (figure 3.4). The survey also asked multinational company on skilled labor in Malaysia about know-how assistance specifically: firms selling shows that the integration of the subsidiary’s produc- exclusively to a single buyer are 34 percent more likely tion network into its GVC spurred greater needs for to receive know-how than firms with a diversified cli- skill development, particularly in management and ent base, while firms without strong relationships are engineering services.23 The development implications 31 percent less likely. Lead firms may be more willing of GVC firm efforts in the on-the-job training in sup- to share knowledge and know-how that benefit the plier companies are of primary importance: employer- supplier firm if they believe those benefits will not be sponsored training is the most important source of passed on to other buyers. The survey also shows that further education in OECD countries, and it is more suppliers’ main support from their foreign partner is effective than both government-financed active labor in capacity building, which may help firms overcome programs and training self-financed by employees.24 skill constraints. Buyer support can take other, sometimes surpris- Through firm-to-firm relationships, GVC firms can ing, forms. For example, Samsung, which in 2018 also play an important role in on-the-job learning, and employed 160,000 people in Vietnam to produce its employer-sponsored training within GVCs can be an Galaxy smartphones, is trying to build a stronger local effective mechanism for skill development, economic supplier base—not only through its own initiative, but Consequences for development | 71 Figure 3.4  GVC firms with relationships supply MNCs, domestic firms experience strong and receive more assistance persistent improvements in performance, including gains in total factor productivity (TFP) of 6–9 percent 50 % more likely or less likely than average four years later. Moreover, the sales of domestic firms firm to receive assistance from buyer 38 40 to buyers other than the first MNC buyer grow by 20 30 percent through both a larger number of buyers and 20 larger sales per buyer. 10 The relational nature of GVCs does not automati- cally result in technology transfer, however. Lead firms 0 can use relational dependence to prevent technologies –10 from spilling over from their supplier network to –20 potential competitors. As a result, new capabilities –30 may be especially difficult to gain when lead firms in –29 GVCs tightly control their technology. –40 Firms that sell to a single buyer In the car industry, where production is complex, Firms that are weakly connected lead firms maintain control over the supply chain, and the technology is not easily diffused. Brands system- Source: WDR 2020 team, using data from a 2018 survey of 1,476 apparel, textile, and information and communication technology firms in Ethiopia atically coordinate production from start to finish, and Vietnam. and incentives for suppliers to innovate, upgrade, and Note: Survey question: “Is any type of assistance—financial, technology, diversify into new market opportunities are relatively know-how, or material assistance—provided by the largest client?” The survey further asks the respondent to characterize the largest client. Single weak. buyer is a variable that takes the value of 1 for firms whose total sales Recent research from the mining industry has sim- (100 percent) are to a single client. Finally, the survey asks respondents to identify their “GVC connectedness.” A weakly connected firm is a firm with ilarly shown that the hierarchical form of governance no direct linkages to GVCs. The variable takes on a value of 1 when a firm is typically prevailing in the mining sector has often not connected to the industry leader as either buyer or supplier, and it does not participate in exporting activity or in trade with foreign entities directly served as an obstacle to learning and innovation.26 or indirectly through intermediaries. The regressions control for country, Though the industry is evolving, rarely do mining sector, and size fixed effects. All coefficient estimates are statistically significant. companies forge long-term formal links with local suppliers or collaborate with them on innovation also by pushing its suppliers from other countries to projects. When new technological challenges arise help in the effort and instructing them to train local that offer new technological opportunities for the firms in customizing production to Samsung’s needs. mining industry in developing countries, they rely on Sometimes, lead firm involvement benefits the wider solutions from their headquarters abroad or interna- educational system of the host country. For example, tional suppliers to the disadvantage of their new local Synopsys, one of the world’s leading companies in suppliers (box 3.2). chip design and testing, established a presence in The extent to which a GVC relationship supports Armenia. Today, Synopsys is one of the largest infor- the growth potential of GVC participants from devel- mation technology (IT) employers in the country, oping countries is therefore likely to be determined with 800 employees—mostly engineers—in Yerevan. by a multitude of factors. The sensitivity and value With the goal of preparing qualified microelectronics of the intellectual property embedded in a lead firm’s specialists, it initiated bachelor’s, master’s, and PhD relationship with its suppliers, technical dependence, programs at both its own educational centers and five codification of transactions, the complexity of both Armenian universities. the product and the value chain, and the technical and In the agri-food sector, long-term relational managerial competence of suppliers all converge to contracts can also be beneficial by helping improve determine suppliers’ upgrading opportunities.27 connectivity, provide better access to technology and capital inputs that increase quality and yield for local How countries participate in GVCs matters producers, achieve higher and more stable prices for Because of the forces just described, how countries farmers, lead to new managerial practices, and achieve participate in GVCs matters. Backward participation a better reputation. Recent research has investigated and forward participation drive the positive associa- the effects of becoming a supplier to multinational tion between GVC participation and growth in per corporations (MNCs) using administrative data track- capita GDP. Inputs that are high in services content—a ing all firm-to-firm transactions in Costa Rica.25 Esti- proxy for knowledge-intensive products—and exports mates from event studies reveal that after starting to that are high in domestic manufacturing content 72 | World Development Report 2020 have the strongest associations with per capita GDP Countries such as Bangladesh, Cambodia, and Viet- growth. Meanwhile, trade in unprocessed agricul- nam leveraged GVCs to move out of commodities into tural goods and commodities has no systematic and basic manufacturing activities and experienced large statistically significant relationship with growth in growth spurts during this transition. Firms in GVCs per capita GDP. contribute to their country’s economic transformation Box 3.2  Mining GVCs: New opportunities and old obstacles for local suppliers from developing countries Mining activities are no longer always organized as huge, The hierarchical form of governance typically prevailing in vertically integrated (multinational) corporations. The the mining sector has often proved to be a true obstacle.c shift toward focusing on core activities while outsourcing Information is highly asymmetric; power between the lead and subcontracting many others is surfacing in this sector mining companies or buyers and their (local) suppliers is and allowing for the emergence of relational GVCs. Lead unbalanced; and many other market imperfections and fail- companies in mining GVCs must contain costs, and so their ures affect transactions along the value chain. As a result, activities have become more knowledge-intensive. They the demand for locally and sometimes even internationally are increasingly searching for local innovative solutions provided suppliers is not easily fulfilled. from local firms to problems such as falling ore grades, Can public policies help? The World Class Supplier falling productivity, rising production costs, exposure to Program in Chile attempts to do so by matching demand local labor and environmental disputes, and the challenges and supply with an open innovation approach, but it has of extreme geographical conditions such as in Bolivia, Chile, had mixed results thus far.d Public intervention can help and Peru, where mines are operated at high altitudes, address other obstacles, particularly when these require a in narrow veins, and in very dry climates. long-term commitment or do not happen because of coor- Mining companies are relying on local suppliers not dination failures. An example of a long-term commitment only for simple intermediate products, but also increas- is developing the skills required by the mining industry, ingly for knowledge-intensive ones. According to recent while an example of the coordination required is bringing research, scientific advances and new forms of innovation together the many different stakeholders. In the mining have opened new technological opportunities for the industry, the latter is an important obstacle because many mining industry in developing countries.a These include actors beyond the mining industry must concur to create revolutionary advances in information and communication the enabling environment needed for firms to thrive. These technologies, computer vision systems, satellites and other actors range from local communities in the mining regions remote sensing applications, advances in molecular and to water and energy interests, education and training synthetic biology for bioleaching (extracting heavy metals institutions, and regulatory institutions—notably, those from minerals with living organisms), and bioremediation dealing with the environment.e Most important, time is of of pollutants for copper and gold. It is precisely these and the essence for this sector. Technology is hardly modifiable similar advances that open opportunities for new suppliers once in use, and the opportunities for local firms to meet to access and add value to mining value chains.b mining firms’ demands and become suppliers can be gen- That said, the organization and governance of the value erated only in the early stages of extraction process design chain do not appear to favor learning and innovation by and implementation. Once exploitation is under way, mining suppliers, as sometimes happens in other sectors. opportunities for developing country producers may shrink. Source: Prepared by Carlo Pietrobelli, Roma Tre University and UNU-MERIT, drawing on Pietrobelli and Olivari (2018). a. Pietrobelli, Marin, and Olivari (2018). b. For example, in Chile the company Micomo has developed highly innovative monitoring technologies that assist the extraction process through fiber optics. Power Train has entered the market with new remote-control systems for trucks operating at high temperatures and with wireless monitoring systems that predict where crucial equipment will wear and have to be replaced, thereby preventing stoppages. In Brazil, Geoambiente has developed sophisticated geological maps, sensors, and radar images that help in the exploration phases, predicting the contents of minerals or areas prone to ero- sion in order to monitor environmental impacts. This company is now Google’s largest partner in Brazil. The use of new materials is also revolutionizing the industry. For example, Verti in Brazil has developed dust suppressors that run on excess glycerin from biodiesel plants. Meanwhile, Innovaxxion in Chile has applied new approaches to mechanical, robotic, and electrical engineering to substantially reduce the waste generated in copper mining. c. Pietrobelli, Marin, and Olivari (2018). d. Navarro (2018). e. Katz and Pietrobelli (2018). Consequences for development | 73 by becoming suppliers of materials and components capita GDP growth was largest for countries as they to a global buyer. Previously only marginally and moved away from being commodity or agriculture intermittently involved in exporting or importing, suppliers and relatively closed to foreign inputs and these firms now source foreign goods and services began to build international linkages in simple manu- to process and reexport as part of a global buyer’s facturing GVC tasks—that is, “limited” manufacturing value chain. During this initial phase of manufactur- GVCs (figure 3.5 and box 3.3). In the first year after ing engagement, domestic per capita income grows entering limited manufacturing GVCs, countries’ steeply, reflecting firms’ learning of new processes GDP per capita is 6 percent higher than in the year of and capabilities, access to large-scale international entry. In the first year after entering advanced manu- demand, and inflow of know-how and technology facturing and services GVCs their GDP per capita is from GVC partners.28 2 percent higher. And in the first year after entering Productive firms drive the transition from limited innovative tasks of GVCs, they are 3 percent higher. to advanced GVC participation in manufacturing However, there are diminishing—and even negative— and services by growing in sophistication and size. returns in staying indefinitely in this phase of devel- They adopt a more complex production structure and opment. Higher rates of growth can be sustained by improve managerial practices. They hire more work- transitioning into advanced manufacturing and ser- ers in nonproduction functions, including in supply vices, and then into innovative activities. The Czech chain management, product development, ICT, and Republic, which upgraded from limited to advanced professional services. They become more capital- manufacturing and services in 2000 and then to and data-intensive, and also tend to expand middle- innovation in 2012 (see chapter 2) is now the most management functions to handle the bigger scale productive economy in Eastern Europe and the OECD of operations and the growing complexity. In this country with the lowest share of population having a enhanced phase, relation-specific feedback loops with disposable income below the poverty line (measured GVC partners become more relevant. Success requires as 60 percent of median household income). The econ- not only continued access to markets, capital, and omy is thriving. Growth is balanced. Internal demand opportunities, but also learning more cutting-edge and household consumption are strong, supported by technologies and skills.29 both per capita income growth and private investment. Consistent with these observations, regression Finally, the unemployment rate has steadily declined results reveal that from 1990 to 2015 cumulative per since the country’s accession into the European Union (EU) in 2004, and it is now below 3 percent, one of the lowest rates in the OECD. Figure 3.5  GDP per capita grows most rapidly when What does this all mean for countries’ industrial- countries break into limited manufacturing GVCs ization options? It is well understood that GVCs can 70 facilitate industrialization by reducing the range of Cumulated change in GDP per capita (%) 60 “capabilities” required to produce and export indus- 57 trial goods. For example, in the auto industry coun- 50 48 tries can participate through GVCs even when they 40 do not have any domestic car makers or any domestic 32 provider of car engines. 30 But more sophisticated tasks in value chains 20 require skills and capabilities that many developing countries lack. As a general rule, learning to handle 10 simple products and production processes is likely to 0 be easier than acquiring the capabilities to transition Event 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 from simple production tasks to specializing in intan- year Years after event gible capital and breaking into new industries. The Limited manufacturing wrong skill mix could end up providing few opportuni- Advanced manufacturing and services ties to innovate, upgrade, and diversify after new GVC Innovative activities ties with international partners are created. Suppliers may find it difficult to upgrade beyond a certain task Sources: WDR 2020 team, using data from the World Bank’s WDI database and the GVC taxonomy for 1990–2015 based on Eora26 database. complexity because doing so may require an ability Note: The event study quantifies the cumulated change in real GDP per capita in the 20 years following a to handle growing firm size and more sophisticated switch from a lower to a higher stage of GVC engagement. See box 3.3 for the methodology. management, sourcing, and learning strategies.30 74 | World Development Report 2020 Box 3.3  Assessing outcomes of GVC participation using event studies Event studies are used in this chapter and in chapter 5 to (in logarithms); inequality as measured by the Gini coef- quantify the changes in outcomes in the 20 years following ficient; $5.50 per day poverty share; and CO2 emissions a switch from a lower to a higher stage of GVC engage- (kilograms of CO2 per $1 of GDP at 2011 values, purchasing ment. Based on data for 146 countries over the period power parity–adjusted). 1990–2015, four types of GVC engagement were identified: The explanatory variable, tswitch +n , is a vector of dummy (1) commodities, (2) limited manufacturing, (3) advanced variables taking a value of 1 in the nth year after a tran- manufacturing and services, and (4) innovative activities sition to a more advanced GVC engagement stage and 0 (see box 1.3 in chapter 1 for a detailed description). otherwise; t and i are time and country fixed effects The event study involves computing average within- to control for conditions in different calendar years and country deviations in a given outcome in each year follow- in different countries, respectively; and eit is the error ing the year of a transition for all countries that stay at least term. The analysis quantifies the effect of transitions four years in a particular GVC engagement stage, had one into limited manufacturing GVC participation (“limited”), transition toward a more advanced GVC engagement stage, into advanced manufacturing and services GVC partici- and had no transitions back to a lower stage. pation (“advanced”), and into innovation GVC participa- The econometric specification is expressed as tion (“innovation”). The estimated coefficients on each 20 dummy variable are multiplied by 100 to give the percent ln( outcome va riable it ) = α 0 + ∑ (δ tswitch + n ) + δ t + δ i + eit change in the outcome variable relative to the outcome n= 1 level at the time of the transition. Figures 3.5, 3.9, and 3.13 where the outcome variables are real income per capita and figure 5.2 in chapter 5 plot those coefficients. (in logarithms); employment, aggregated and by skill level As discussed earlier, in some cases the organization strategies: strong connectivity to international tech- and governance of the value chain, the nature of tech- nology ecosystems, and investments in design and nology, and large bargaining power imbalances may marketing capabilities. These strategies allowed firms trap suppliers from developing countries in dead-end to develop innovative and cost-efficient products tasks instead of favoring the processes of learning and compatible with global markets by using cutting-edge innovation typical of relational GVCs. technologies and capabilities in marketing and design The rise of GVCs may thus lead countries engaged to respond rapidly to changes in market demand and in highly hierarchical or captive GVCs, or those that lag consumer taste. A few successful companies started behind in skills and human capital, connectivity, and developing their own research and development institutional quality (chapter 2), to become locked in (R&D) capabilities and high-technology expertise, but in relatively low value-added segments of production they did so as part of the global ecosystem of technol- with little scope for upgrading. Bangladesh’s and Cam- ogy, not through just indigenous innovation. bodia’s experiences in the apparel sector are examples Because of deepening global integration, Whit- of the difficulties developing country firms face in taker et al. (2010) suggest that the viable growth upgrading from basic assembly functions to more path for developing countries is now “compressed sophisticated segments of the value chain, which development”—that is, leveraging globally engaged require a very different skill set (box 3.4). They may, production systems rather than nationally integrated then, find it simpler to “industrialize” in the age of production systems. GVCs introduce international GVCs, but the returns to doing so by replicating the interdependencies that are unlike those faced by ear- strategies of earlier developers may not be as high as lier developers (chapter 4). Accordingly, the efficacy of they were in the past. Moreover, the gradual increase industrialization and development strategies depends in automation may compound these effects (chapter 6). on how well policy makers understand these new China’s experience suggests, however, that indus- conditions and learn, seize opportunities, adapt, and trialization may still be possible, but it requires develop innovative solutions in concert with a wide new approaches to development. Chinese firms range of actors, domestic and foreign. These issues are that upgraded in the smartphone market used two discussed further in the chapters on policies. Consequences for development | 75 Box 3.4  Skills and upgrading in Cambodia’s apparel value chain The foreign direct investment that Cambodia’s apparel apparel manufacturers, and their foreign branch plant sector has attracted over the past two decades has been locations. important for jobs and growth. Foreign investors set up Opportunities for functional upgrading of these multi- manufacturing locations in Cambodia 20 years ago to take national corporations (MNCs) is also limited because the advantage of lower production costs stemming from a mix apparel industry is buyer-driven. The company or brand of lower minimum wages and trade preferences. These responsible for setting the final price and selling the prod- multinational manufacturing firms have head offices in Hong uct is not the same company that owns manufacturing Kong SAR, China; Taiwan, China; or the Republic of Korea. facilities. Apparel manufacturers (whether at the head- They also have manufacturing facilities in other Asian coun- quarters or branch locations) do not control retail, market- tries. Despite the presence of these firms, Cambodia has not ing, branding, or creative new product development, which moved up the apparel GVC and is still performing many of are the most lucrative and knowledge-intensive activities in the same assembly activities largely carried out by the same the sector. Thus branch plants of foreign operations there- original foreign investors. More than 95 percent of its apparel fore have little opportunity for functional upgrading. exporters are branch plants of foreign-owned firms. And yet there are still opportunities for upgrading in three All the activities associated with functional upgrading areas. The first is in the preproduction and production stages take place at the headquarters location, leaving little or no currently performed in Cambodia by foreigners. The second room for branch manufacturing sites to take on more activ- is in the sourcing of inputs and arranging the logistics of ities. These activities include textile sourcing and sales/ shipments, currently carried out abroad at the headquarters buyer acquisition and technical product development. of foreign MNCs with manufacturing locations in Cambodia, This experience is not unique to Cambodia. It is, in fact, but that could be transferred to Cambodia. The third is in cre- difficult for countries to upgrade in this industry because ative design and branding, which could be done by private of relationships between global lead firms, multinational domestic firms that are locally headquartered. Source: Based on Frederick (2018). Finally, integration in agricultural GVCs can also and income. For example, many farmers reported that support economic transformation in the sector wher- their income and output increased by half or more as a ever lead firms are able to encourage the upgrading result of contractual arrangements.31 of farmers through long-term relationships. Formal or informal contractual arrangements that regulate the provision of production inputs, such as fertilizer, Employment technology, extension services, and market informa- Apart from higher overall productivity, firms in devel- tion, have positively affected the upgrading of farm- oping countries that participate in GVCs tend to be ers in Ghana, Kenya, and Zambia who are growing more capital-intensive. Machines can be equipped to maize, cassava, or sorghum. Having a contract with a deliver the precision needed for the compatibility of buyer is significantly and positively associated with parts. They can also deliver the higher-quality out- upgrading to higher-value intermediate processes put demanded by foreign consumers and help firms and moving to higher-value-added products. Farmers achieve higher productivity and greater scale. It may under contract seem to have better access to inputs therefore make sense for firms to adopt more capital- and technologies through the out-grower company intensive methods, even those in poor countries with or other external sources. In a random sample of 1,200 relatively large labor forces. The costs of accessing farmers in Ghana, Kenya, and Zambia, over 50 percent capital may also be lower for GVC firms because of of surveyed contract farmers attributed their use of the relational dimension of participation—they have fertilizer to their contractual arrangement. Exten- easier access to finance, foreign machinery, and train- sion services, seeds and pesticides, and tractors were ing for their operations. In Vietnam, firms that both other cited forms of support. Moreover, the majority import and export use more capital inputs per worker of the farmers under contract said the scheme had a than firms that export only or firms that sell exclu- positive to very positive impact on their production sively to the domestic market.32 Firms in Ethiopia that 76 | World Development Report 2020 export and import are also more capital-intensive than Figure 3.6  In Ethiopia, GVC firms are relatively more one-way traders or nontraders. This observation holds capital-intensive but their employment is increasing across a sample of developing countries.33 fastest Can GVCs deliver higher productivity and greater a. Capital intensity b. Employment capital intensity, as well as more and better-paying 160 160 switching from not trading to trading (%) Difference between trading firms and jobs? Or is economic growth through GVCs at the 140 140 expense of job growth? GVCs are becoming more Before-after difference for firms important for exports (chapter 1), but at the same time 120 120 nontrading firms (%) exports are becoming less job-intensive.34 In some 100 100 countries, exports are contributing a smaller share of 80 80 total jobs,35 leading some observers to conclude that the employment consequences of GVCs have been dis- 60 60 appointing.36 According to these observers, rather than 40 40 contributing to more and better-paying jobs in devel- 20 20 oping countries, capital-intensive production by GVC firms may lead to stagnant or lower overall employ- 0 0 Capital intensity Employment ment, and the path to development by moving workers from agriculture to manufacturing may be suppressed. Firm type: Because GVCs boost exports, their overall effects Export and import Export only Import only on employment in developing countries have been Sources: Choi, Fukase, and Zeufact (2019), using data from Ethiopia 2000–2014 manufacturing census positive. Even though production is becoming more (firms with 10 or more employees). capital-intensive and less job-intensive, the positive Note: For the period 2000–2014 panel a reports the percentage difference in capital intensity between nontrading firms and trading firms. The results are obtained by regressing firm capital intensity (log productivity effects at the firm level are (unexpectedly) capital per worker) on dummy variables if a firm exports and imports (GVC firm), exports only, or good for scale and employment. Through scale effects, imports only, controlling for whether the firm is state-owned, as well as sector, year, and region fixed effects. Panel b reports the percentage difference in employment before and after the switch for firms higher productivity is expanding aggregate output and that switched from nontrading to trading status. The results are obtained by regressing firm employ- employment. GVC firms tend to employ more workers ment (log number of workers) on dummy variables if a firm exports and imports (GVC firm), exports only, or imports only, controlling for whether the firm is state owned, as well as year and firm fixed than other firms.37 When the higher productivity of effects. All coefficient estimates are statistically significant. For the capital intensity and employment these firms leads to sufficient scale—through more regressions, the coefficients for export-only and GVC firms are not statistically different. The percent differences reported in the graphs are obtained as 100 multiplied by the exponential of the coefficient competition and market restructuring, demonstration estimates minus 1. effects, demand effects, technology spillovers, and investment in infrastructure—the overall effect on share in total employment, albeit slightly.42 In fact, the jobs is positive. In Ethiopia, firms that both export and provinces that became more GVC-intensive also expe- import are more capital-intensive and increased their rienced faster growth in the employment share of the labor force faster than other firms between 2000 and population (map 3.1). No province experienced net job 2014 (figure 3.6). These firms utilized 145 percent more losses. Net job creation nationally exceeded 12 million, capital per worker than nontrading firms between and the share of employment in the population (ages 15 2000 and 2014, compared with a 102 percent difference and over) increased from 70 percent to 76 percent.43 It for export-only firms and a 19 percent difference for is likely these experiences would extend to other low- import-only firms.38 Ethiopian firms that became two- income countries that have been able to integrate into way traders saw their labor force grow by 39 percent basic manufacturing, such as textiles or agribusiness. (relative to when they were nontraders), while the In Mexico, employment expansion is more strongly growth for firms becoming exporters was 29 percent linked to GVCs than one-way trade (figure 3.7). and for firms becoming importers was 6 percent. Between 1993 and 2013, municipalities in Mexico with Employment in manufacturing expanded from 2000 a larger share of employees in manufacturing firms to 2014, and GVC firms accounted for an increasing that both export and import experienced stronger share of manufacturing employment.39 In Mozam- growth in their total employment and increased their bique, despite adopting more mechanical technologies share in the country's total employment. in the cashew value chain, as discussed earlier, employ- The new activities that GVCs bring to countries ment also increased alongside output in the sector.40 can also induce shifts in type of employment. In Viet- Vietnam is another powerful example. Between nam, the number of self-employed, wage, and salaried 2004 and 2014, total jobs in firms that both import and workers, as well as employers, all increased between export expanded faster than in firms that import only 2004 and 2014. But wage and salaried jobs nearly or export only.41 As a result, GVC firms increased their doubled, outpacing other employment types, and the Consequences for development | 77 Map 3.1  In Vietnam, employment expansion was linked to GVC firms a. Change in employment in GVC firms b. Change in employment-to-population per capita, 2004–14 ratio, 2004–14 Percent Percent 8–10 14–18 6–8 11–14 4–6 7–11 2–4 4–7 0.5–2 0.1–4 IBRD 44649 | AUGUST 2019 Sources: WDR 2020 team, using data from GSO (2012) and General Statistics Office of Vietnam's Enterprise Surveys. Note: GVC firms are firms that both export and import. Employment is measured as the total number of employees reported by registered firms, summed across firms with more than five employees within each province. The employment-to-population ratio is measured as employment relative to population in the province. Figure 3.7  In Mexico, employment expansion is more share of total employment increased 11 percentage strongly linked to GVC expansion than non-GVC trade points, from 25 to 36 percent. Formal employment (jobs covered by social security) in the manufactur- 0.20 ing sector also grew as GVC firms assumed greater Change associated with increased presence of trading firms within 0.18 importance in formal manufacturing employment in 0.16 Mexican municipalities Vietnam.44 However, as discussed shortly, informal or 0.14 noncontract work can also be important in agriculture 0.12 and manufacturing value chains. 0.10 The overall result is that GVCs are associated 0.08 with structural transformation, with exports pull- 0.06 ing people out of less productive activities and into 0.04 more productive manufacturing jobs. In Vietnam, 0.02 manufacturing absorbed nearly 2.5 million workers 0 between 2005 and 2014, increasing its share of the Employment (log) Employment share country's total employment from 12 to 14 percent.45 Firm type: This is not unique to Vietnam. The 2016 World Bank Export and import Export only Import only report Stitches to Riches? reveals that, based on data on Sources: WDR 2020 team, based on INEGI (2014) and CONEVAL and World Bank (2013). the apparel sector in South Asia between 2000 and Note: Standardized coefficient estimates are reported for the period 1993–2013 from a regression of 2010, when a country experienced a 1 percent increase log of municipality employment or municipality employment share in total employment on the number of employees per capita in manufacturing firms that export and import, export only, and import only, in apparel output (a proxy for apparel exports), there controlling for total population of the municipality, distance of the municipality to the U.S. border, was a 0.3–0.4 percent increase in employment. This and state and year fixed effects. All coefficient estimates are statistically significant. Standardized coefficients refer to how many standard deviations the dependent variable will change per standard rise in employment increased overall welfare as work- deviation increase in the explanatory variable. ers moved out of agriculture or the informal sector 78 | World Development Report 2020 toward better-paying, higher-value-added jobs.46 Simi- Figure 3.8  Worldwide, GVC firms hire more women larly, Lesotho’s integration in the global apparel sector than non-GVC firms accounted for 10 percent of the country’s workforce 60 and half of manufacturing employment in 2009, help- ing to transform an agrarian economy.47 In Haiti, the UKR Female labor share, nonparticipant (%) LVA MNG apparel sector employed 37,000 workers in 2014.48 50 BGR BIH LKA MDA RUS GVCs support employment of not just men, but HRV MMR NIC also women. Female employment grew faster than 40 COL KGZ SRB MAR TJK GTM JAM BLR male employment in Vietnamese provinces where CHN NAM UZB IDN GEO ROU ECU SLV ETH GVC participation expanded the most.49 Notably in MEX PER AZE ARM 30 CRI TTO PAN MKD the apparel and electronics sectors, where assembly of BOL POL TUN ZAR CHL KAZ VNM many small parts must be done manually, firms report PRY LBN UGA 20 ZWE ARG BGD preferences for female employees because of the high NGA KEN TZA levels of dexterity required. In Ethiopia, women con- GHA ZMB EGY stitute 75 percent of the workforce in the apparel sec- 10 IND SEN JOR WBG NPL tor,50 65 percent in Haiti,51 and 77 percent in Sri Lanka.52 YEM IRQ PAK Across the world, firms that both export and import tend to employ more women than firms that do not 0 10 20 30 40 50 60 70 80 90 participate in GVCs (figure 3.8). Foreign-owned firms Female labor share, GVC participant (%) as well as firms that export or import also have higher Source: Rocha and Winkler (2019), using data from World Bank’s Enterprise Surveys. female labor shares on average than firms that do not, Note: Each dot represents a country-year observation. The x-axis plots the employment-weighted share but the relationship is stronger for GVC participants. of female workers of total workers in firms that both export and import (GVC participant). The y-axis These jobs have positive effects on other aspects of plots the employment-weighted share of female workers of total workers in firms that do not export and import (nonparticipant). For country abbreviations, see International Organization for Standardization women’s livelihoods. In Bangladesh, for example, (ISO), https://www.iso.org/obp/ui/#search. young women in villages exposed to the garment sector delay marriage and childbirth, and young girls gain an additional 1.5 years of schooling (box 3.5).53 wages in 2000–2014 than did those that exported only The gender dimension of GVCs though is not without or imported only, controlling for sector, location, and challenges. year effects. In Mexico, wages are also significantly Not only do GVC firms employ more people, but higher in firms that both import and export than they also pay better. In Ethiopia, manufacturing firms in firms that do not. Firms that have relationships that both import and export paid significantly higher with buyers or suppliers also pay higher wages than Box 3.5  GVC participation can lead to indirect welfare improvements for women How does getting a job change one’s life beyond the income negative effects of early marriage and childbirth. Girls in itself? Bangladesh is an interesting case study because the villages close to garment factories had on average signifi- country’s ready-made garment industry employs 3.6 mil- cantly higher educational attainment—they appeared more lion people, 53 percent of whom are women.a Meanwhile, likely to stay in school than those with no factory nearby. the country has seen remarkable progress in health and This effect was particularly strong for younger girls ages education. How might these factors be related? One study 5–9. The most plausible explanation appears to be that the used an innovative approach, looking at 1,395 households chances of getting a job increase the returns to staying in in 60 villages to identify how the arrival of ready-made school and improving literacy and numeracy. In addition, garment jobs may have affected various welfare-related parents, through higher income from these jobs, can better indicators.b Exposure to the sector was associated with afford to send their children to school. a drop in both marriages and childbirths for girls ages The study compared these demand-led welfare effects 12 to 18—an important finding because of the long-term with a more supply-side intervention in the form of a (Box continues next page) Consequences for development | 79 Box 3.5  GVC participation can lead to indirect welfare improvements for women (continued) large-scale conditional cash transfer program to encourage those in other villages. There were also indirect positive girls’ school enrollment. The demand-led welfare effects effects from BPO employment on girls’ school enrollment, were much larger than the effects of conditional cash trans- nutrition, health, delayed marriage, and childbirth.c fers. In other words, expanding light manufacturing provides Evidence of improved welfare for women working in not only benefits in the form of jobs but also, more indirectly, GVCs can be found elsewhere as well. One study looked benefits for education, health, and workers’ children. That at the subjective well-being of women employed in Sen- said, there was a small negative effect on school enrollment egal’s export-oriented horticulture industry.d Employment of girls ages 17–18. For them, the opportunity cost of getting improved subjective well-being for the poorest women, a garment factory job may outweigh the returns to staying generally through improved living standards, but not as in school. As discussed in box 3.6, the relationship between much for women whose incomes were well above the pov- human capital formation and participation in GVCs is hetero- erty threshold. For low-income women employed in Ethio- geneous across countries’ contexts. pia’s cut flower industry, savings in relation to their incomes Together, these results suggest that the type of job mat- are higher than for those employed in other sectors, and ters, and that as countries move into more value-added and the subjective valuation of their jobs is also higher.e skill-intensive activities, the returns to education for girls Finally, by analyzing workers’ experiences in the Kenyan will improve, and dropout rates are likely to fall. Evidence cut flower industry through interviews, the authors of one from India seems to confirm this point. An investigation article found a clear link between employment and wom- of the more skill-intensive business processing outsourc- en’s empowerment—such as in greater independence, new ing (BPO) industry in the country showed that women in opportunities, and decision making within the household.f villages linked to the industry had higher aspirations and The strength of the effect, however, depends on the quality invested more in computer or English courses than did of the job. a. Moazzem and Radia (2018). b. Heath and Mobarak (2015). c. Jensen (2012). d. Van den Broeck and Maertens (2017). e. Suzuki, Mano, and Abebe (2018). f. Said-Allsopp and Tallontire (2015). firms without relationships in Mexico.54 In China, which reduces poverty.  However, GVCs may have GVC engagement improved firms’ wages (more so in additional channels through which trade affects pov- capital-intensive and foreign-invested firms) both by erty. Labor-saving productivity growth through the improving productivity within firms and by reallocat- hyperspecialization of GVCs may directly displace ing labor to more productive firms.55 Again, across a jobs. However, adoption of techniques and technolo- sample of developing countries, firms that both export gies that save on labor can spur job creation through and import pay higher wages than import-only and three indirect channels that are more challenging to export-only firms and nontraders.56 conceptualize and measure. First, productivity gains How countries participate in GVCs also matters for in supplier industries can yield steep increases in the wage growth. From 1990 to 2015, wage growth was the demand for labor because of input–output linkages. largest for countries that broke out of commodities or Second, productivity growth can boost final demand. agriculture into basic manufacturing (“limited manu- And, third, such growth may lead to compositional facturing” in figure 3.9). shifts in the structure of the economy and could sup- port jobs by spurring the growth of sectors with high labor shares. Poverty and shared prosperity In a cross section of countries, growth in GVC By supporting employment and income growth, participation is indeed associated with a decline in GVCs also support poverty reduction and shared the number of people living on less than $5.50 a day prosperity. The classical trade literature suggests that (in 2011 international prices)—see figure 3.10. Open- trade creates growth, better jobs, and higher incomes, ness affects poverty primarily through growth, the 80 | World Development Report 2020 main driver of the remarkable reduction in global Figure 3.9  The boost to wages is largest in countries poverty since 1990.57 Where economic growth from after they first enter limited manufacturing GVCs GVCs is larger than from conventional trade, poverty 80 Cumulated change in wages (%) reduction from GVCs will also likely be larger. In Mexico, municipalities with a larger share of 60 employees in internationalized firms experienced a 58 55 greater reduction in poverty between 1993 and 2013 for the poorest as well as vulnerable households. A 40 greater presence of import and export firms is posi- tively associated with the poorest households’ ability 20 to obtain a basic food basket. Municipalities with 11 greater GVC participation also experienced a lower incidence of capabilities poverty and asset poverty— 0 Event 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 that is, their access to enough financial resources to year provide for other needs, including health, education, Years after event and transport, improved.58 They also experienced a Limited manufacturing Limited manufacturing (ns) decline in the marginalization index, which captures Advanced manufacturing Advanced manufacturing deprivation and inaccessibility to basic goods and and services and services (ns) services for welfare. The relationship among pov- Innovative activities Innovative activities (ns) erty, marginalization, and international integration Sources: WDR 2020 team, using data from the World Bank’s WDI database and the GVC taxonomy for is generally stronger for firms that both export and 1990–2015 based on Eora26 database. import than for those that export only or import Note: The event study quantifies the cumulated change in wages in the 20 years following a switch from a lower to a higher stage of GVC engagement. Dotted lines indicate statistically nonsignificant (ns) only (figure 3.11). All this said, even though GVCs can coefficients. See box 3.3 for the methodology. create opportunities for poor households, they have also been found to create risks for the accumulation Figure 3.10  GVC participation is of human capital throughout the life cycle, such as in associated with poverty reduction Mexico (box 3.6). In Vietnam, provinces with more internationalized 1.0 firms also experienced greater reductions in poverty 0.8 between 2004 and 2014 (figure 3.12). This decline likely 0.6 Change in poverty rate (%) worked through the employment and ultimately the 0.4 income channels, as just discussed. Provinces with 0.2 more internationalized firms similarly experienced 0 higher growth in the incomes of the bottom 40 per- cent of the population between 2004 and 2014. The –0.2 impacts were not restricted to those provinces with –0.4 more GVC participation, and poverty also fell in neigh- –0.6 boring provinces in Vietnam.59 –0.8 The positive effects of GVC participation on –1.0 income growth are likely to extend to everyone in –1.0 –0.5 0 0.5 1.0 society—if the welfare state works. GVC integration in Change in foreign value certain regions of a country can give people the incen- added in exports (%) tive to migrate within their country, which can be a powerful mechanism for reducing poverty. Higher Sources: WDR 2020 team, using data from Eora and World Bank’s WDI database. incomes will also generate more demand for a greater Note: Each dot is a country-year observation. The x-axis is the average number and diversity of goods and services, imported annual growth in foreign value added in exports between 1990 and 2015. The and domestic. This demand will lead to diversification y-axis is the average annual growth in the poverty rate between 1990 and 2015. The poverty rate is measured as a percentage of the population living of the economy, which will increase opportunities on less than $5.50 a day (in 2011 international prices). for a broader and more diverse set of agents. GVCs are also likely to make a larger variety of goods more Agriculture value chains can be a particularly affordable, such as cell phones, thereby allowing the powerful factor in poverty reduction by integrating poor to participate more widely in society. rural households and smallholder farmers into supply chains. In Madagascar and Senegal, more high-value Consequences for development | 81 Figure 3.11  In municipalities in Mexico, the expanded exports and the modernization of export supply chains presence of GVC firms is more strongly associated of green beans and tomatoes had important positive with poverty reduction than the presence of firms that welfare effects. Most notable were higher incomes for export only or import only these farmers, particularly those in the lower quartile of the income distribution. The result was a reduction 0 in the absolute poverty levels.60 Change associated with increased –0.01 There is no apparent relationship between GVC presence of trading firms within participation and growth in income inequality in Viet- Mexican municipalities –0.02 nam or Mexico, as measured by the Gini coefficient –0.03 using household data at the provincial or municipal –0.04 level.61 Despite this finding, there can be important distributional implications of GVC participation –0.05 across and within countries. –0.06 The lack of a systematic relationship between GVC –0.07 participation and growth in income inequality for developing countries is at first sight confirmed by the –0.08 cross-country event study described in this chapter (see Food poverty rate Asset poverty rate Capabilities poverty rate box 3.3). Greater income inequality within countries, as Firm type: measured by the Gini coefficient, is observed only in Export and import Export only Import only the group of countries that switch to the innovation Sources: WDR 2020 team, using data from INEGI (2014) and CONEVAL and World Bank (2013). stage of GVC engagement, and it becomes statistically Note: Standardized coefficient estimates are reported for the period 1993–2013 from a regression of food significant only after about a decade (figure 3.13). poverty, asset poverty, and capabilities poverty rates at the municipal level on the number of employees per capita in manufacturing firms that export and import, export only, and import only, controlling for total population of the municipality, distance of the municipality to the U.S. border, and state and year Distribution of gains fixed effects. Ratios are defined as the number of food, asset, or capabilities poor over total population in the municipality. All coefficient estimates are statistically significant. Standardized coefficients refer Paralleling the gains that GVCs have delivered for to how many standard deviations the dependent variable will change per standard deviation increase in the explanatory variable. For definitions of food poverty, asset poverty, and capabilities poverty, see countries, a large majority of people in both high- and note 55 at the end of this chapter. lower-income countries view two elements of GVCs Box 3.6  Does GVC participation lead to human capital accumulation? By boosting productivity and enabling structural trans- But the rates of human capital formation differ signifi- formation, participation in GVCs has been associated with cantly among countries that increased their participation in rising incomes and less poverty. But the extent to which GVCs. Although Mexico experienced an increase in open- countries reap long-term development gains from GVC ness after the launch of the North American Free Trade participation hinges critically on its consequences for the Agreement (NAFTA), income growth and human capital human capital of workers and their children. formation remained disappointing, despite rising public Many developing countries are giving priority to raising spending on education. human capital formation while deepening GVC participa- What explains these different experiences? Recent tion and pursuing export-led industrialization. The experi- empirical evidence suggests that the skill intensity of ence of East Asia—such as Korea in the 1980s and 1990s newly created manufacturing jobs may play a critical and more recently China and Vietnam—suggests that these role. Sub­national evidence from Mexico reveals that the two goals are compatible and may reinforce one other. GVC school dropout rate rose with the local expansion of participation fosters industrialization and urbanization, export manufacturing industries: for every 25 jobs created, boosting parental income and productivity. It also raises one student dropped out of school at grade 9 instead of tax collection and creates room for larger private and public continuing through grade 12.a These effects are driven by investments in education. Human capital formation further the export-manufacturing jobs that require fewer skills supports GVC participation and industrial development. and therefore raised the opportunity cost of schooling for (Box continues next page) 82 | World Development Report 2020 Box 3.6  Does GVC participation lead to human capital accumulation? (continued) students at the margin. Subnational evidence from China capital goods raised the demand for skills and led to greater reveals that high-skill export shocks raise both high school educational attainment.d and college enrollments, whereas low-skill export shocks These findings point to a mutually reinforcing relation- depress both.b The amplified differences in skill abundance ship between the skill intensity of tasks and skill acquisition. across regions reinforce the initial patterns of industry spe- On balance, participation in GVCs may still support human cialization. Broader cross-country evidence for 102 coun- capital formation via income growth and the weaker finan- tries over 45 years points in the same direction: growth in cial constraints facing parents and governments. But these less skill-intensive exports depresses average educational positive effects may be offset by reduced skill formation in attainment, whereas growth in skill-intensive exports raises areas in which participation in GVCs leads to an expansion schooling.c At the same time, in China rising imports of of low-skill-intensive sectors and tedious tasks. a. Atkin (2106). b. Li (2018). c. Blanchard and Olney (2017). d. Li (2019). positively: free trade and international business ties. Figure 3.12  In Vietnam, poverty reduction was greater However, the number of skeptics in all countries grew in locations with a higher presence of GVC firms between 2002 and 2014 (figure 3.14). Although the dis- 60 content is greater in high-income countries, the num- Change in expenditure poverty rate, ber of those perceiving themselves to be losers from 40 global integration is also nonnegligible in developing 20 countries. GVCs may have fueled some of this public discon- 0 2004–14 (%) tent. Rather than being distributed equally across and –20 within countries, the gains have been concentrated, accruing to specific firms, workers, and locations. –40 People can feel left out, even if they are not worse off. –60 Markups and firms –80 The public sentiment on trade and international –100 business ties captures the fact that since the 1980s 0 10 20 30 40 50 60 70 80 there has been a widespread rise in firms’ profits. In Share of GVC firms in total employment, 2014 (%) 134 countries, the average global markup increased by 46 percent between 1980 and 2016, with the largest Sources: WDR 2020 team, using data from GSO (2012) and General Statistics Office of Vietnam’s Enterprise Surveys. increases accruing to the largest firms in Europe and Note: GVC firms are firms that both export and import. Employment is measured as the total number North America and across a broad range of economic of employees reported by registered firms, summed across firms with more than five employees within sectors.62 each province. The expenditure poverty rate is measured as the poverty headcount. The presence of firms that export only had no additional relationship with poverty reduction. The growth of GVC activity appears to be a con- tributor to the rise in markups for several reasons. First, GVCs lower the costs of inputs for companies, Firms that import and export are not constrained by through importing, and increase their productivity, domestic inputs and domestic demand, which helps through the scale of expansion afforded by exporting. them grow and realize economies of scale. This factor Second, in the presence of economies of scale is especially important in the mass production man- GVCs disproportionately favor large firms that can ufacturing that dominates the limited manufacturing afford the fixed costs of exporting and importing. GVC group. The size distribution of firms is likely to Consequences for development | 83 Figure 3.13  Rising income inequality is a greater Third, markups increase only if these cost reduc- problem for countries breaking into the innovation tions are not fully passed on to consumers through stages of GVC engagement lower prices.63 Participating in GVCs justifies some markup increase to cover the greater fixed costs of 10 more complex sourcing or exporting. But the markup 6 5 growth in GVC-intensive sectors is also likely to have Cumulated change in Gini coefficient (%) 0 increased the profit rate of these companies. It is well established empirically that large firms pass through –5 a smaller share of a price shock to consumers. Con- –10 –11 sistent with this, these large firms are also likely to –13 only partly pass on lower costs due to offshoring to –15 consumers. The California company Everlane, which –20 is committed to transparent pricing, reports the cost Event 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 year breakdown of all its products as well as the average Years after event price of its items in the market. According to the Limited manufacturing Limited manufacturing (ns) company’s website, a pair of jeans that customarily Advanced manufacturing Advanced manufacturing sells for $170 is produced for $34, which includes cost, and services and services (ns) insurance, and freight. Innovative activities Innovative activities (ns) Indeed, U.S. industries are increasingly concen- Sources: WDR 2020 team, using data from the World Bank’s WDI database and the GVC taxonomy for trated, with a small number of productive firms 1990–2015 based on Eora26 database. accounting for large shares of the market and large Note: The event study quantifies the cumulated change in the Gini coefficient in the 20 years following a profits.64 This rise of “superstar” firms in the United switch from a lower to a higher stage of GVC engagement. Dotted lines indicate statistically nonsignifi- cant (ns) coefficients. See box 3.3 for the methodology. States and other advanced economies may be asso- ciated in part with the rise of GVCs and in part with technological change and innovations. In other Figure 3.14  A majority worldwide views trade and words, GVCs have boosted superstar firms that earn international business ties positively, but skepticism superstar profits and may dominate the market. In grew from 2002 to 2014 Ethiopia, for example, measures of markups are also 40 highly correlated with industry concentration in TUR manufacturing.65 “somewhat bad“ or “very bad,“ 2014 There is evidence that firms in developed coun- % of respondents that answered ITA 30 JOR tries that outsource parts and tasks to suppliers in USA FRA JPN developing countries have seen higher profits. In the MEX GHA textile sector, for example, markups of Japanese firms 20 ZAF have increased since 1990 in line with backward GVC ARG participation (figure 3.15, panel a). This positive asso- RUS KOR POL KEN ciation holds for other developed countries and other DEU PER 10 PAK UGA GBR PHL sectors that have also transferred large parts of their TZA LBN production to developing countries.66 CHN 0 Within developing countries, there is also evidence 0 10 20 30 40 50 of incomplete pass-throughs of cost reductions to % of respondents that answered consumers through lower prices, resulting in higher “somewhat bad“ or “very bad,“ 2002 profits. After India’s trade liberalization in the 1990s, Source: Pew Research Center 2014. when input tariffs on intermediate inputs fell, both Note: Each dot is a country-year observation. The figure shows the share of respondents that answered costs and prices dropped, but markups went up by in 2002 and in 2014 “somewhat bad” or “very bad” to the question “What do you think about the grow- about 13 percent when the economy opened to trade.67 ing trade and business ties between [survey country] and other countries—do you think it is a very good thing, somewhat good, somewhat bad, or a very bad thing for our country?” For country abbreviations, Consumers still benefited through lower prices (as see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. well as higher quality and greater variety), but they were worse off than if firms had fully passed on those be significantly more skewed in a world of GVCs than cost reductions. in a world without them, which is consistent with GVC activity—and the relational nature of GVCs in evidence that firms participating in GVCs tend to be particular—similarly appear to be a likely contributor larger than other firms. to the international dispersion of the markups that 84 | World Development Report 2020 Figure 3.15  Increasing GVC participation is associated with rising markups in developed countries but falling markups in developing countries a. Japanese textile sector b. Indian textile sector 0.09 1.35 0.13 1.30 Backward GVC participation Forward GVC participation Aggregate markup Aggregate markup 0.12 0.07 1.20 0.11 0.05 1.25 0.10 1.10 0.03 0.09 0.01 1.15 0.08 1.00 90 92 94 96 98 00 02 04 06 08 10 12 90 92 94 96 98 00 02 04 06 08 10 12 14 16 19 19 19 19 19 20 20 20 20 20 20 20 19 19 19 19 19 20 20 20 20 20 20 20 20 20 GVC participation Aggregate markup (right axis) Sources: WDR 2020 team, using data from Eora and Worldscope. Note: Graphs plot data between 1991 and 2011 for panel a and between 1990 and 2015 for panel b. The left y-axis in panel a measures the share of foreign value added in gross exports of the Japanese textile sector (backward GVC participation). The left y-axis in panel b measures the share of domestic value added in India embodied in importing countries’ exports to third countries (forward GVC participation). The right y-axis in both panels measures the share-weighted average markup of listed companies in the textile sector. Markups are calculated following De Loecker and Eeckhout (2018). Similar results hold across countries and sectors.  GVCs generate.68 The implications of GVCs for the Figure 3.16  In Ethiopia, firms entering emergence of superstar firms huge in scale, high in GVCs experience greater declines in market power, and large in profit rates are exacerbated markups, 2000–2014 by the disproportionate bargaining power that these switching from not trading to trading 0 large lead firms may have over their suppliers. Before-after difference for firms Although buyer firms in developed countries are –0.01 seeing higher profits, supplier firms in developing countries are getting squeezed. Across 10 developing –0.02 countries, the relationship between markups and forward participation is negative for developing –0.03 countries in the textile and apparel sector (see figure 3.15, panel b, for India).69 Some developing countries, –0.04 including China, enjoy a positive correlation. This finding is consistent with a growing number of firms –0.05 from emerging economies graduating from supplier Export and Export only Import only to lead firms in GVCs. import Other country-level evidence suggests markups Firm type have increased mostly in advanced economies but Sources: Choi, Fukase, and Zeufack (2019), using data from Ethiopia: not in emerging markets.70 In Ethiopia, firms that 2000–2014 manufacturing census (firms with 10 or more employees). buy inputs abroad to sell in the external market have Note: Standardized coefficient estimates are reported for the period 2000–2014 from a regression of the log of markup at the firm level on lower markups than other types of firms (one-way dummy variables for firm type (export only, import only, export and import), traders or nontraders).71 And the more intensely a firm controlling for state ownership, labor (log), capital (log), firm fixed effects, and year fixed effects. No data shown for "Import only" because only statisti- is integrated into a GVC (measured as the share of the cally significant coefficient estimates are reported. Standardized coefficients export value added and imported inputs in total sales), refer to how many standard deviations the dependent variable will change per standard deviation increase in the explanatory variable. the lower is its markup. As Ethiopian firms become integrated into GVCs, they also experience reductions in their markups, which are strongest for two-way traders (figure 3.16). In Poland, increased GVC partici- In South Africa, markups charged by manufacturing pation—including the use of imported components in exporters are on average significantly lower than production as well as the rising presence of domestic those charged by nonexporters. Firms with a rela- firms in foreign markets—is associated with the tively small proportion of exports (up to 10 percent) observed decline in markups between 2002 and 2016.72 charge markups that are about 1.2 percent lower than Consequences for development | 85 nonexporters, while firms with a medium (11–25 per- share,74 but the rise in GVC activity appears to be a cent) and large (more than 25 percent) share of exports contributor. By increasing the profit rate of compa- charge markups that are 1.8 percent and 2.3 percent nies, GVCs also generate a force that results in a lower lower than those of nonexporters, respectively.73 The share of an economy’s income being paid to labor. risk that firms from developing countries experience In the United States, superstar firms that are more limited profits after becoming suppliers for global productive and earn higher profits also have lower firms mirrors the rise in profits in developed countries. labor shares, and their increasing concentration has In short, GVCs primarily reward large interna- contributed to the declining labor share within indus- tional firms by reducing their production costs. How- tries.75 It may be that producers are not passing on ever, these gains are only partly passed on to consum- their cost savings to both workers and consumers. ers or shared with suppliers. Because suppliers are Similarly, the movement of relatively labor- predominantly in developing countries, the gains may intensive tasks from developed to developing coun- be distributed unequally, even across countries in the tries could explain why the composition of production value chain. becomes more capital-intensive with GVC participa- tion in developed countries. In developing countries, Markups and labor’s share of profits this could also reduce the labor share insofar as it The rise in the market power of firms is contributing accompanies production that has become relatively to the changing distribution of capital and labor in more capital-intensive than before.76 countries. The share of income accruing to workers— In 63 developed and developing economies, GVC or how much of a country’s GDP accrues to labor integration as well as other domestic within-industry through wages as opposed to physical capital and forces, such as technology or markups, contributed profits—is the other side of the markup phenomenon: significantly to the reallocation of value added from profits are rising, but labor’s share of income is falling labor to capital within countries between 1995 and (figure 3.17, panel a). 2011. The labor share declined by 2.2 percentage points, There are, of course, many possible explanations with GVCs contributing 0.6 percentage point (figure for the observed global decline in the so-called labor 3.17, panel b). Similarly, global integration, particularly Figure 3.17  GVCs have contributed to the declining labor share within countries a. Labor share by country income group, b. Factors contributing to labor share decline 1995–2011 58 48 Labor share (over GDP), high-income (%) 57 47 Total 56 46 lower-middle-income (%) Labor share (over GDP), 55 45 54 44 World demand 53 43 Domestic 52 42 within-industry 51 41 factors 50 40 49 39 GVCs 48 38 –2.5 –2.0 –1.5 –1.0 –0.5 0 95 97 99 01 03 05 07 09 11 19 19 19 20 20 20 20 20 20 Percentage points High-income Lower-middle-income (right axis) Source: WDR 2020 team, using data from OECD’s TiVA database. Note: In panel a, the green line plots the labor share in 29 advanced economies, and the blue line plots the labor share in 34 developing economies. In panel b, the decomposition explores the contribution of world demand, domestic within-industry factors, and GVCs to the total percentage point decline in the average labor share of 63 developed and developing economies between 1995 and 2011. V is the diagonal matrix of the share of value added in gross output; B is the Leontief inverse; and Y is the diagonal matrix of final goods and services produced in a country and sold worldwide. The results are obtained from three counterfactual exercises to decompose the relative contribution of each component by asking what the contribution to the observed overall changes in labor share would be if only domestic within-industry factors (V), GVCs (B), or world demand (Y) are allowed to change over time. The decomposition follows the methodology of Reshef and Santoni (2019). 86 | World Development Report 2020 the expansion of GVCs, has been identified as the firms in GVCs tend to adopt more capital-intensive primary trigger of the rise of overall capital intensity techniques than comparable domestic firms.83 Physi- in production in emerging markets and developing cal capital deepening and upgrading contribute to the economies.77 Alongside globalization, explanations increase in the relative demand for skilled workers have also focused on economies of scale, innovation, because of the capital–skill complementarity—phys- and new technologies.78 ical capital (and especially capital equipment) is less substitutable with skilled labor than with unskilled Skills and wage inequality labor.84 Consistent with this finding, in countries par- Inequality can also arise within the labor market, ticipating in GVCs and in the more capital-intensive with a growing wage premium for the skilled. The parts of the value chain firms demand more-skilled Stolper-Samuelson theorem, one of the key tenets of workers.85 The result is that as workers tend to move traditional international trade, indicates that rising toward less routine and more interactive tasks, GVCs trade integration is likely to increase wage inequal- produce more jobs for skilled workers.86 ity (skilled versus unskilled workers) in relatively Firm-level analysis confirms a positive and signif- advanced countries with abundant skilled labor. But icant relationship between GVCs and skilled employ- rising integration would be expected to reduce wage ment—that is, between the number of skilled workers inequality in lower-income countries in which skilled and firms with international links that export or are labor is scarce. In a world of fragmentation, however, foreign-owned.87 In 27 transition economies, import- the theorem’s validity is undermined. And, indeed, it ing inputs increases the demand for skilled labor.88 In is widely accepted both theoretically and empirically fact, global sourcing explains more than a quarter of that greater fragmentation of production increases the unconditional difference between importers and wage inequality in countries at all income levels for at nonimporters in the employment share of high-skilled least three reasons.79 First, when production is moved workers. In Madagascar, upgrading by diaspora- and across countries, the workers in those economies find Mauritian-owned firms in the apparel sector corre- themselves employed in new production processes sponded with in-firm training and skills upgrading.89 and tasks. In higher-income countries, these pro- In Africa more broadly, with Chinese import pene- cesses and tasks may be considered low-skilled and tration firms increase their share of skilled workers labor-intensive, but in lower-income countries they by almost 4 percent, which is associated with a shift are considered skilled labor-intensive when compared in production from low-skill to high-skill-intensive with the outside opportunities of workers.80 Thus off- products.90 shoring increases the demand for skilled workers in low- and middle-income economies and puts upward Geographical disparities pressure on wage inequality. Inequality arising from GVCs also has a geographical A second reason for increased wage inequality in dimension. GVC integration is strongly associated low- and middle-income economies is that GVCs are with greater concentration in cities,91 as well as border often more skill-sensitive than traditional trade flows, regions for countries neighboring GVC partners. This in part because they often produce goods destined for finding is consistent with evidence from Mexico and quality-sensitive consumers in rich countries,81 and in Vietnam showing that economic integration across part because of the high complementarities among national borders is associated with greater spatial con- the various stages of production carried out in differ- centration within national borders (map 3.2). ent countries.82 Because some regions grow faster than others, The disproportionate importance of the matching regional inequalities in developing countries can between buyers and sellers in GVCs may also drive increase when labor is not perfectly mobile. In Viet- up wage inequality. Because the identity of these pro- nam, the only areas with double-digit job growth were ducers matters, especially when sensitivity to quality concentrated around Hanoi and Ho Chi Minh City. is high, relational GVCs may set off “a war for talent,” By contrast, in developed countries some regions are with the price of particularly attractive producers or being hollowed out by GVCs. In the United States, the the wage of particularly skilled individuals bid up outsourcing of manufacturing tasks and the expo- disproportionately relative to that in a world without sure of industries to foreign competition have led to relational GVCs. the emergence of a “rust belt.”92 Such a phenomenon A third reason for the increase in wage inequality can result in localized and persistent income losses in countries in which skilled labor is scarce is that for years for people in negatively affected regions or Consequences for development | 87 Map 3.2  In Mexico and Vietnam, GVCs are spatially concentrated b. Vietnam, 2014 a. Mexico, 2015 Log of employment of GVC firms per capita, by province –3.3 to –1.2 –5.3 to –3.3 % of firms that participate –7.3 to –5.3 in GVCs, by state –9.3 to –7.3 3.7 to 5.5 –11.3 to –9.3 2.9 to 3.7 No data 1.4 to 2.9 0 500 km 0.7 to 1.4 0 300 km IBRD 44666 | SEPTEMBER 2019 Sources: Mexico: WDR 2020 team, using data from ENAPROCE 2015. Vietnam: WDR 2020 team, using data from GSO (2012) and General Statistics Office of Vietnam’s Enterprise Surveys. Note: In Mexico, state-level GVC participation is measured as the percent of firms that participate in GVCs. In Vietnam, province-level GVC participation is measured as log of employment of GVC firms per capita. sectors when people cannot move easily.93 Both experi- Subcontracted home-based workers (so-called ences highlight the need for internal mobility of labor homeworkers) make up significant shares of employ- to distribute the gains from trade. Place-based policies ment in other supply chains. Among other things, they that could reduce the negative consequences of the weave textiles, package products, process rice, and economic forces that disproportionately benefit some make food products. An estimated 5 million home- areas are discussed in the final chapters of this Report. workers are part of India’s garment and textile sup- ply chains alone. Most homeworkers are informally Unequal work conditions employed without employer contributions to their Small-scale farmers and home-based workers form social protection, and the vast majority are women. the base of some value chains, often on unequal terms. Their average earnings are not only lower than those A review of 49 studies related to the commodities and of factory workers but also erratic, and subcontracted horticulture value chains concluded that “informal- homeworkers also pay for many of the nonwage costs ity is the norm rather than the exception: informal of production, such as workplace, equipment, utilities, workers make up the majority of the workforce, even and transport. Integrating homeworkers into supply in formal enterprises.”94 In a random sample of 1,200 chains on fairer terms will require better regulation farmers in Ghana, Kenya, and Zambia growing maize, from above and better integration from below (box 3.7). cassava, or sorghum, between 82 percent of farmers in GVC participation can increase casual employ- Zambia and 97 percent in Kenya had no contract. For ment. A case study in Ghana and Côte d’Ivoire on those with a contract, informal contracts dominated participation in the pineapple and cocoa value chains the landscape. In Kenya, 86 percent of contracts were found that, although participation benefits success- informal.95 ful farmers through improved growing processes, higher yields, and higher incomes, it is also associated 88 | World Development Report 2020 Box 3.7  Home-based work in GVCs By organizing in collectives, homeworkers can link to global • Capital—physical and social networks supply chains in efficient ways and on fair terms. To do this, More appropriate and enabling laws and regulations •  collective enterprises of homeworkers—cooperatives or regarding cooperatives and commercial transactions other collective forms—would have to seek the following because the existing laws in most countries are not types of support: appropriate for cooperatives and the commercial transactions of those at the base of the economic Management and business training, including how •  pyramid. to forecast market demand and how to manage businesses Additional spillovers from forming collective enterprises Professional managers and knowledge of how to •  include greater bargaining power in market transactions recruit and retain managers and an enhanced ability to challenge the social norms that Professional advice and assistance on how to link to •  constrain women’s time, mobility, and access to resources global supply chains, how to upgrade products and (such as the social norms governing inheritance and prop- production systems, and how to reduce dependence erty rights) and the economic policies that ignore or under- on intermediaries value their economic activities and contributions. with an increase in casual labor hiring, as well as anti-sweatshop campaigns in the 1990s brought atten- displacement of farmers from land because of their tion to poor working conditions in the textile, foot- low bargaining positions and lack of knowledge on wear, and apparel (TFA) sector.101 As a result of activist their rights to land ownership.96 Earlier research has pressure, multinational enterprises (MNEs) signed documented the growing use of casual and seasonal codes of conduct pledging to raise wages and improve contract labor both on farms and in packhouses in working conditions in factories producing their South Africa (fruit exports) and Kenya (fresh vegeta- products. The result was large real wage increases in ble exports).97 the targeted enterprises, by as much as 30 percent in GVCs may also be associated with poor worker large foreign-owned and exporting TFA plants relative conditions. Work practices often fall short of inter- to other TFA plants. In fact, wages were no worse in national standards in supplier countries, ranging MNEs than in domestic plants to begin with. Within from violations of core labor standards to unsafe the TFA sector, real annual wages in domestic plants working conditions, low wages, excessive working were lower than those in foreign-owned or exporting hours, and precarious contracts.98 This problem is plants. particularly associated with labor-intensive GVCs, Relationships within value chains can also catalyze where outsourcing to developing country locations improved working conditions. CocoaAction, promoted is often motivated by low-wage labor.99 This situation by nine main global producers of chocolate and cocoa, has led many observers to question the social value was set up to regenerate the cocoa plantations in West of the GVC business model, pointing to incidents at Africa. It also sought to help smallholder cocoa farmers contract manufacturers such as the 2013 collapse of who often subsist on incomes below the poverty line the Rana Plaza garment factory in Bangladesh. In and who face deficits in literacy, low school attendance the copper-cobalt belt of the Democratic Republic of rates, child labor, and gender inequality. In launching Congo, for example, children often work in the min- CocoaAction, the leading chocolate and cocoa compa- ing sector.100 And yet putting a halt to sourcing from nies recognized that their individual commitments these artisanal mines as a way to counter child labor could not solve the complex and systemic challenges could have unintended negative effects for household and that more sustainable production of cocoa would income, where poverty and then social norms are the also be good for their profits. Similar efforts were main reasons for children working in mining. made in Ethiopia, Mexico, and Vietnam. There are signs that GVCs can transmit sensitiv- However, this may not be enough. While private ities about working conditions in host countries and firms can play an important role, there is also a clear induce remedial actions. In Indonesia, for example, role for policy action supported by international Consequences for development | 89 cooperation to determine the appropriate standards processing workforce, but they are poorly represented and ensure their enforcement. These policies are in enterprise management. The trends are similar in addressed in the final chapters of this Report. aquaculture in Nigeria and Vietnam,103 cocoa and cof- fee in Papua New Guinea,104 and horticulture in Azer- baijan105 and Afghanistan.106 In the cashew value chain The gender gap in Mozambique, lack of gender equality limits the Although firms in GVCs tend to employ more women access of women farmers to agricultural inputs, credit than other firms, women are generally in lower- services, and markets. Despite the fact that more than value-added segments of the value chain, mostly in half of the industry’s workforce are women, almost no labor-intensive production jobs and in occupations women hold leadership positions within factories.107 that require lower skills and pay less.102 The positive In call centers in the Arab Republic of Egypt, women relationship between GVC participation and the make up the majority of call agents, whereas men female labor share is much higher for production dominate jobs in higher-value segments and manage- workers than for administrators or sales workers ment.108 In Kenya, women are overrepresented in the in manufacturing firms (figure 3.18, panel a). Many accommodation and excursion segments of tourism, countries have few women-owned or women-run but they tend to work as low- to mid-skilled employ- GVC firms. Firms that import and export are signifi- ees, unless engaged as entrepreneurs.109 cantly  less likely to be majority female–owned than Why are so few GVC firms owned or run by other firms and are significantly less likely to have a women? Women’s placement in value chains stems top female manager. Thus GVCs do not appear to be in part from the same reasons that hold back women breaking the glass ceiling (figure 3.18, panel b). in the non-GVC economy. These include disadvan- The asymmetry between production, on the one tages in endowments, such as assets, education, hand, and management and ownership, on the other, skills, experience, networks, and social capital, as is particularly visible in agriculture, but it is on view well as gender-biased regulations or discriminatory in other sectors as well (table 3.1). In southern Africa’s social norms. According to the World Bank’s Women, fish-aquaculture sector, women contribute mostly to Business, and the Law database, 20 countries have yet primary production and make up 90 percent of the to grant men and married women equal ownership Figure 3.18  Women are more likely to be production workers and less likely to own or manage GVC firms a. Female production workers versus b. Female owners and managers nonproduction workers 5.0 0 female-owned or -managed between GVC firms and non-GVC firms (%) Difference in probability of being 4.5 –0.02 female share of total workers, GVC firms vs. non-GVC firms Percentage point premium in 4.0 –0.04 3.5 –0.06 3.0 –0.08 2.5 –0.10 2.0 –0.12 1.5 1.0 –0.14 0.5 –0.16 0 –0.18 Production workers Administrators or Majority ownership Top management sales workers Source: Rocha and Winkler (2019), using data from the World Bank’s Enterprise Surveys. Note: Exporters are firms with an export share (direct or indirect) of at least 10 percent of total sales. Importers are firms with an imported input share of at least 10 percent of total inputs. GVC participants are firms classified as both exporter and importer. Panel a plots the coefficient of estimations of the female labor share (production workers and nonproduction workers) on a dummy variable if the firm is a GVC participant, controlling for capital intensity, sales, and total factor productivity (TFP), as well as country-sector, subnational region, and year fixed effects. Panel b plots the coefficient of estimations of whether a firm is majority female-owned or has a female top manager on a dummy variable if the firm is a GVC participant, controlling for country-sector, subnational region, and year fixed effects. All coefficient estimates are statistically significant. 90 | World Development Report 2020 Table 3.1  Sample of results from case studies on gender in specific GVCs Author and year of publication Sector and country(ies) Results Veliu et al. (2009) Aquaculture, Nigeria Women represent a significant share of employment, especially in and northeast processing and packaging, but they are poorly represented in enterprise Vietnam management. World Bank and IFC (2014) Cocoa, coffee, and Women provide substantial labor in both coffee and cocoa cultivation fresh produce value and predominate in the fresh produce sectors, especially in tasks relevant chains, Papua New for the quality of exports such as postharvesting. Guinea IFC (2018) Horticulture, A higher share of women are employed in horticulture than in other Azerbaijan sectors. For products that depend on manual harvesting, women constitute more than 50 percent of harvesters. World Bank (2011) Horticulture, Women provide the majority of labor in the lower levels of the value Afghanistan chains for horticulture—harvesting and postharvesting—although this is often unpaid household work. Ahmed (2013) Call centers, Arab Women make up the majority of call agents, whereas men dominate jobs Republic of Egypt in higher-value segments and management. Christian (2013) Tourism, Kenya Women are overrepresented in the accommodation and excursion segments of the tourism sector, although they tend to work as low- to mid- skilled employees, unless they are engaged as entrepreneurs. Barrientos (2014) Apparel, globally In 2014 on average 60–80 percent of production workers in the top 27 apparel-exporting countries were women. rights to property, and 41 countries do not grant sons Figure 3.19  Gender equality in business regulations and daughters equal rights to inherit assets from ensures that women are more fairly rewarded their parents. Even when the legal system does not 6 discriminate against female ownership of assets, social norms inhibiting land ownership by women 5 Increase in women’s wages for an additional year of experience (%) are a recurring theme across low- and middle- 4 income countries. In Afghanistan’s rural areas, social and cultural norms severely limit women’s access 3 to services, including credit, training, extension, inputs, and trading and marketing networks.110 In 2 Honduras, efforts by female entrepreneurs to enter 1 value chains and upgrade into higher-value activ- ities appear to be complicated by limited access to 0 important inputs such as land, finance, and market –1 information.111 In call centers in Egypt, limited access to education, training, promotion, and networks –2 0 20 40 60 80 100 made it difficult for women to take advantage of the rising demand for higher technical skills Gender equality in business regulation score generated by product upgrading.112 These gender- Source: World Bank 2019b. intensified constraints can restrict a country’s abil- Note: Each dot represents a country observation. The x-axis plots the country score for gender equality ity to remain competitive and upgrade to higher- in business regulation. The y-axis plots the expected percentage increase in wages for each additional year of experience for women. The World Bank’s Women, Business, and the Law database (2019) value segments of the chain—a topic discussed in a documents the gender legal disparities for 189 economies. forthcoming report by the World Bank and World Trade Organization on trade and gender, “How Can resources can be an effective first step. The larger the 21st Century Trade Help to Close the Gender Gap?” number of legal restrictions women face, the lower Removing legal restrictions that make it harder is the payoff from work experience (figure 3.19). Sim- for women to start businesses and access productive ply mandating a nondiscrimination clause in hiring Consequences for development | 91 increases women’s employment in formal firms by lowering corporate income tax rates and granting 8.6 percent.113 tax incentives such as tax holidays and preferential tax zones. Such measures can help countries achieve development objectives by promoting job growth and Taxation technology transfer. But they can also be inefficient Raising tax revenue is a challenge in today’s globalized if such benefits do not outweigh the cost of lower tax and digitalized economy. GVCs have magnified the revenues.115 In a race to the bottom, corporate income challenges facing the international tax system. The tax rates have declined by almost half since 1990 current system of international taxation relies prin- (figure 3.20).116 cipally on identifying the physical place where value Revenues from corporate income taxes are further is created by firms. The mobility of certain factors of eroded by international tax avoidance, which takes production, combined with the fragmentation of pro- advantage of loopholes and weaknesses in the inter- duction processes across countries, make firms even national tax architecture. In GVCs that involve affili- more sensitive to the differences in taxation from ates of the same common corporate structure, firms country to country. In GVCs that involve affiliates of can locate activities that generate high profits with the same firm, fragmentation of production also leads relatively little input, or “substance,” in jurisdictions to greater intrafirm trade and more opportunities for where those profits are taxed at low rates. Such prac- tax avoidance by manipulating where value is recog- tices are legal, but they run counter to the principle of nized for tax purposes. Exacerbating the problems are taxing activities where value is created. Firms can also the growth of intangibles in global business and the manipulate transfer prices between their affiliates to digital delivery of services.114 shift profits to lower-tax jurisdictions. Countries are under pressure to engage in tax In principle, transactions between affiliates of a competition by lowering the burden of corporate multinational corporation are “priced” according to income tax to retain domestic and attract foreign the arm’s-length principle, which means that they investment. Meanwhile, lower communication and are in line with comparable transactions between transport costs are facilitating the relocation of firms unrelated enterprises under comparable circum- and the fragmentation of production across coun- stances.  These rules for affiliated-party transactions tries. Indeed, firms can locate production chains and are intended to ensure that profits of MNCs are regis- procurement across the globe, choosing countries tered in countries where value is created. In practice, that make the most sense from a business perspec- however, the arm’s-length principle is hard to apply, tive. That includes taking advantage of differences leaving scope for manipulating transfer prices to shift between national tax systems to shift production profits (but not substantial activities) to low-taxed to lower-tax jurisdictions. Countries compete by entities without violating transfer pricing rules.117 Figure 3.20  Corporate income tax rates have declined by almost 50 percent since 1990 a. Corporate income tax rates, G-7 countries, b. Average corporate income tax rates, by country 2012 and 2018 income group, 1990–2018 40 50 35 45 30 25 40 Percent Percent 20 35 15 30 10 5 25 0 20 n s ce y ly da m 16 90 92 96 98 02 06 08 10 12 94 00 18 04 14 e an pa Ita do at an a 19 19 19 19 20 20 20 20 20 20 19 20 20 20 20 m Ja an St ng Fr er C d Ki G te i d Un te i Un Low-income Middle-income High-income 2012 2018 or announced OECD (Europe) OECD (not Europe) Source: IMF 2019. Note: Data include average subnational rates. OECD = Organisation for Economic Co-operation and Development. 92 | World Development Report 2020 Figure 3.21  As a share of GDP, non-OECD countries lose the most from profit shifting a. Revenue loss in U.S. dollars b. Revenue loss as a share of GDP 450 1.4 400 1.2 350 1.0 300 US$ (billions) 0.8 Percent 250 200 0.6 150 0.4 100 0.2 50 0 0 Non-OECD OECD Non-OECD OECD Source: Crivelli, de Mooij, and Keen 2016. Note: OECD = Organisation for Economic Co-operation and Development. Other avenues for international tax avoidance include 2. Moazzem and Radia (2018); Solotaroff et al. (2019). debt transactions between affiliated parties in low-tax 3. World Bank, Women, Business, and the Law (database). jurisdictions (lender) and high-tax jurisdictions (bor- See appendix A for a description of the databases used in this Report. rower), locating intangible assets in low-tax jurisdic- 4. Guardian (2019). tions, and treaty shopping.118 5. International buyers have joined together to work in a Tax revenue losses from profit shifting are sub- coordinated way through the Fire and Building Safety stantial: an estimated 30 percent of global cross-border Accord (mostly European companies and unions) and corporate investment stocks are routed through off- through the Alliance for Bangladesh Worker Safety shore hubs, and the associated tax losses for develop- (a group of mostly North American buyers). These groups ing countries are estimated at about $100 billion.119 In have committed to inspecting their supplier factories and developing plans for training and remediation. In March 2013 non-OECD countries missed out on $200 billion 2013, the government, business organizations, and trade in tax revenue as a result of profit shifting, a relatively unions signed the National Action Plan on Fire Safety, larger loss than in OECD countries (figure 3.21).120 which calls for action to improve legislation, expand The growth of intangibles in GVCs and the digital labor inspection capacity, and implement systematic delivery of services pose special challenges. Intangi- inspections of all factories. The Accord, the Alliance, and ble assets such as data, patents, and trademarks are the National Action Plan have agreed to use a common inherently more mobile than the traditional physical standard for certification to ensure that building struc- tural integrity and fire safety are adequate. The World factors of production. Such assets are hard to value, Bank Group has also been working with the private sec- and their share in overall capital goods is rising in tor on improving water usage through the Partnership the digital economy. In the United States, the share of for Clean Textiles and labor standards through ILO-IFC intangible assets in the nonresidential capital stock Better Work (ILO and IFC 2016). In April 2018, after the doubled between 1966 and 2016.121 Firms can choose five-year anniversary of the Rana Plaza disaster, a crowd- to move only certain parts of the production process sourcing effort to map all garment factories in Bangla- abroad, thereby minimizing any associated risk and desh and make the mapping publicly available was initiated by the private sector, with collaboration among maximizing the potential gains.122 Thus small changes Sourcemap, the C&A Foundation, and BRAC University. in tax policy can prompt large locational shifts by GVC 6. Markup is a measure of market power. It is the ratio firms, increasing pressure on countries to compete for of the price to the marginal cost of production after all economic activity through their national tax systems. tangible and intangible factors of production have been remunerated. 7. (See also UNCTAD 2013.) Notes 8. Quantitative methods that trace the internationally frag- 1.  Harmonized System (HS) categories 61, 62, and 64, mented nature of GVCs through global input–output using mirror data for 2017. The HS trade statistics cod- links typically predict larger gains from trading across ing system is an internationally standardized system of borders than models without those international links names and numbers to classify traded products. (Antrás and de Gortari 2017; Caliendo and Parro 2013). Consequences for development | 93 9. Based on regression results from Constantinescu, determine outcomes (Costinot, Oldenski, and Rauch Mattoo, and Ruta (2019). 2011). 10. Costa (2019). 31. Dihel et al. (2018). 11. Output per worker controlling for capital as well as for- 32. General Statistics Office (GSO), Vietnam, Enterprise eign ownership status, sector, and regional differences. Surveys, 2014 (database). The percent differences are obtained by multiplying 33. World Bank, Enterprise Surveys (database). The surveys 100 by the exponential of the coefficient in figure 3.3 are administered to a developing country sample of 81 minus 1. countries. 12. Doane (2015). 34. Calì et al. (2016). In a sample of 39 countries, the num- 13. Amiti and Konings (2007); De Loecker et al. (2016); ber of jobs supported by $1 million in gross exports Goldberg et al. (2010). declined from 38 in 2001 to 16 in 2011. The number of 14. For example, Antràs, Fort, and Tintelnot (2017) show that manufacturing jobs supported by $1 million in gross U.S. firms that began importing from China after that exports declined from 20 in 2001 to 12 in 2011. Similarly, country’s accession to the World Trade Organization the number of jobs per unit of domestic value added in also increased their sourcing from domestic suppliers exports declined between 2000 and 2014 in seven devel- in the United States. oping countries, where technical change in GVCs has 15. Amiti and Konings (2007); Constantinescu, Mattoo, and been biased against the use of labor (Pahl and Timmer Ruta (2017); Goldberg et al. (2010); Halpern, Koren, and 2019). Szeidl (2015); Taglioni and Winkler (2016). 35. Calì et al. (2016). In a sample of 39 countries with data 16. Sampson (2015). for 2001 and 2011, 26 countries experienced a decline in 17. The benefits for producers in developing countries that export jobs’ share of total jobs. On average, 28 percent of relational GVCs produce are sizable. In Colombia, a jobs were supported by exports in 2011, compared with program led by a multinational firm induced suppliers 31 percent in 2001. to upgrade their coffee farms while planting trees and 36. Rodrik (2018). incorporating more efficient and sustainable practices. 37. In Vietnam, firms that import and export employ About 80,000 farmers and 1,000 villages benefited from more workers than firms that export only and firms the program: the quality of coffee improved, while that do not trade, controlling for sector and province farmers’ profits increased by 15 percent (Macchiavello fixed effects as well as state- and foreign-ownership. In and Miquel-Florensa 2019). Mexico, firms that have relationships with buyers, as 18. Barrientos et al. (2016). well as firms that export and import, also see higher 19. Krishnan (2017). employment than firms that only import or only 20. Gereffi, Humphrey, and Sturgeon (2005). export. This holds even when considering regional, 21. Macchiavello and Miquel-Florensa 2017). sector, and foreign ownership characteristics of firms. 22. Dearden, Reed, and Van Reenen (2006). Across developing countries, firms that import and 23. Iberahim (2013). export employ more workers than one-way traders or 24. Hansson (2008). nontraders. 25. Alfaro-Ureña, Manelici, and Vasquez (2019). 38. The percent differences are obtained as 100 multiplied 26. Pietrobelli, Marin, and Olivari (2018). by the exponential of the coefficient in figure 3.6 27. Gereffi, Humphrey, and Sturgeon (2005). minus 1. 28. De Loecker (2007, 2013); Iacovone and Crespi (2010). 39. Manufacturing firms with 10 or more workers. The 29. Kugler and Verhoogen (2012). share of GVC firms in total employment increased from 30. More abundant use of capital and skills is important for 19 percent to 23 percent between 2000 and 2014. upgrading (Bustos 2011). Handling greater product com- 40. Costa (2019). plexity requires reinforcing intermediate management 41. In the nonagricultural enterprise sector. Employment layers relative to plant workers (Caliendo et al. 2015). in GVC firms increased by 130 percent between 2004 Sourcing strategies are also essential for upgrading. As and 2014, compared with 115 percent for import-only part of this process, the organization of procurement and 47 percent for export-only firms (with six or more practices and sourcing becomes an integral part of a employees). Total employment in nontrading firms firm’s strategy and an increasingly important part of increased slightly faster, by 136 percent, although this its competitive advantage (Antràs, Fort, and Tintelnot difference is likely not statistically significant. 2017). It not only matters which international linkages a 42. From 30 percent to 31 percent. firm creates, but also which domestic supplier linkages 43. World Bank, World Development Indicators (database). it creates (Eslava, Fieler, and Xu 2015). Connections with 44. Total manufacturing employment increased from 2.7 demanding and refined customers supports learning million to 5.7 million between 2004 and 2014. Formal (Fieler, Eslava, and Xu 2014). But these customers have manufacturing jobs (covered by social security) in high standards of quality and delivery that may be more formal enterprises increased from 2.4 million to 4.5 difficult to learn than simpler tasks, even with deep million. GVC firms accounted for 14 percent of formal relationships within the GVC. A firm’s ability to quickly jobs in 2004 and 17 percent in 2014. absorb increasing amounts of more complex technol- 45. According to the General Statistics Office (GSO) of ogy, know-how, and an ability to produce at high quality Vietnam, the number of employees age 15 and older in 94 | World Development Report 2020 manufacturing was 5,031,200 in 2005 (first year of avail- some increases in income inequality emerged as well. able data) and 7,414,700 in 2014 (see annual employed In Turkey, the average worker has experienced some population and annual employed population 15 years benefits, although many of the benefits have accrued of age and above, with breakdown by kind of economic to high-skilled workers, thereby revealing the greatest activity, items, and year). increase in income inequality. 46. Lopez-Acevedo and Robertson (2016). 62. De Loecker and Eeckhout (2018). 47. Kumar (2017). 63. Markups can increase because prices are higher or 48. Faucheux et al. (2014). because costs are lower, or it may be a combination 49. Within provinces, a 1 percentage point increase in the of both when markets are not perfectly competitive, share of firms that participate in GVCs is associated meaning that firms can affect prices. The effect of GVC with a 3.2 percent increase in female employment and participation on firms’ markups depends on whether the a 2.1 percent increase in male employment. reductions in costs, or the gains from GVC participation, 50. Staritz, Plank, and Morris (2016). are fully passed on to the consumer through lower prices. 51. Faucheux et al. (2014). 64. Autor et al. (2017). 52. World Bank (2018). 65. Including the Herfindahl index and the number of 53. Heath and Mobarak (2015). There is a 28 percent firms within an industry. decrease in the likelihood of getting married and a 29 66. Based on regression analysis that considers country- percent decrease in the likelihood of childbirth. and industry-specific characteristics. It is possible that 54. There is no additional wage premium for relational these producers focus on tasks that have the highest firms over export-only or import-only firms. value added because of demand (such as for a particu- 55. Using firm-level and customs transaction–level data lar design, concept, or service—that is, where market covering the period 2000–2006 with the methods of power is the result of innovation or merit), and they propensity score matching, difference in differences, outsource those tasks that have lower value added (such and generalized propensity score. See Lu et al. (2019). as producing homogeneous parts). Ideally, one would 56. Shepherd and Stone (2012) also find that firms with the disentangle the channels and the different effects, but strongest international linkages—export, import, and this is not possible with the data. foreign-owned—pay higher wages. 67. De Loecker et al. (2016). 57. The poverty elasticity of growth depends on various 68. De Loecker and Warzynski (2009). factors, including its incidence (changes in inequality); 69. Excluding China and controlling for country fixed the initial distribution of land; wealth and income; edu- effects. The negative correlation also holds without cation levels among the poor; other forms of past public controlling for country fixed effects for samples that investment; and local institutions, including unions exclude and include China. (Ferreira, Leite, and Ravallion 2010; Ravallion and Datt 70. Diez, Fan, and Villegas-Sánchez (2019). 2002). Also see Dollar and Kraay (2002) and Ferreira and 71. Choi, Fukase, and Zeufack (2019). Controlling for state Ravallion (2008). ownership, firm size, and capital intensity. 58. Food poverty is defined as the inability, even if all avail- 72. Gradzewicz and Mućk (2019). able income is used in the home, to buy only the goods 73. Dauda, Nyman, and Cassim (forthcoming). of a basic food basket. Capabilities poverty is defined as 74. Karabarbounis and Neiman (2013). the insufficiency of disposable income to acquire the 75. Barkai (2016) finds that in the United States profits have value of the food basket and carry out the expenses risen as a share of GDP and that the pure capital share of necessary in health and education, even dedicating the income (defined as the value of the capital stock times total income of the home to nothing more than these the required rate of return on capital over GDP) has purposes. Asset poverty is defined as the insufficiency fallen alongside the labor share. of income available to acquire the food basket, as well 76. Autor et al. (2017) also point to outsourcing as a possible as to make the necessary expenses in health, clothing, explanation for the declining labor share in the United housing, transportation, and education, even if the States. entire household income is used exclusively for the 77. Reshef and Santoni (2019) investigate the same phe- acquisition of these goods and services. nomenon in a sample of 26 EU countries over the period 59. This is consistent with the observation that migrant 1995–2014. They suggest that the labor-reducing effect workers are more likely to work in the formal sector in of capital intensity may be a short-run phenomenon. Vietnam (McCaig and Pavcnik 2015). The authors document a recovery from 2007 onward, 60. Maertens, Minten, and Swinnen (2012). explained by within-industry changes, notably for 61. Foreign direct investment (FDI) inflows—an important skilled labor associated with the complementarities determinant of GVC participation—has similarly been between capital intensity and skilled labor. Domestic associated with poverty reduction, but also income within-industry factors also explain a recovery in the inequality in Ethiopia, Vietnam, and Turkey (World labor share from 2007 on in the larger sample of 63 Bank 2019a). In Ethiopia, the overall effects of FDI are developing and developed economies. largely positive, with large effects on poverty reduction 78. Oxera Consulting (2019). and limited effects on income inequality. In Vietnam, 79. Goldberg and Pavcnik (2007). FDI has significantly benefited shared prosperity, but 80. Feenstra and Hanson (1996, 1997). Consequences for development | 95 81. Verhoogen (2008). 116. A s corporate profits have gone up, the average revenue 82.  Antràs, Garicano, and Rossi-Hansberg (2006); Kremer from corporate income tax has remained stable over and Maskin (2006). the same period. But as other sources of income (such 83.  Bernard, Moxnes, and Ulltveit-Moe (2018). as wage income) as a share of GDP have gone down, 84.  Griliches (1969); Krusell et al. (2000). these factors have indirectly reduced the scope for gov- 85.  Becker, Eckholm, and Muendler (2013); Bloom and Van ernments to secure adequate tax revenue. Reenen (2011); Dearden, Reed, and Van Reenen (2006); 117. Cooper et al. (2016). Hansson (2008). 118. For a more complete listing, see Beer, de Mooij, and Liu 86.  Hijzen et al. (2013); Javorcik (2014); Markusen and (2018). Trofimenko (2009); te Velde and Morrissey (2003). 119. UNCTAD (2015). Other studies link GVC participation to increased 120. Crivelli, de Mooij, and Keen (2016). wage disparity between skilled and unskilled labor in 121. Auerbach (2017). developed countries, although little literature exists 122. de Mooij and Ederveen (2008). for developing countries. Using matched employer- employee data for Denmark for 1995–2006, Hummels et al. (2014) characterize the link between offshoring References and wages across skill levels and find that offshoring Ahmed, Ghada. 2013. “Global Value Chains, Economic Upgrad- increases (lowers) the high-skilled (low-skilled) wage. ing, and Gender in the Call Center Industry.” In Global Similarly, using matched employer-employee data for Value Chains, Economic Upgrading, and Gender: Case Studies Italy, Borghi and Crinò (2013) confirm that offshoring of the Horticulture, Tourism, and Call Center Industries, edited contributes to widening the wage gap between skilled by Cornelia Staritz and José Guilherme Reis, 73–105. and less skilled employees. Report 83233 (January). Washington, DC: World Bank. 87.  Shepherd and Stone (2012). Alfaro-Ureña, Alonso, Isabela Manelici, and José P. Vasquez. 88. Crinò (2012). 2019. “The Effects of Joining Multinational Supply 89.  Morris and Staritz (2014). Chains: New Evidence from Firm-to-Firm Linkages.” 90.  Darko, Occhiali, and Vanino (2018). Paper presented at the National Bureau of Economic 91.  World Bank staff estimates. Analysis of OECD’s TiVA Research conference, “Firms, Networks, and Trade,” Cam- database for 61 countries shows that a one-unit stan- bridge, MA, March 15. dard deviation increase in domestic value added in Amanor, Kojo Sebastian. 2012. “Global Resource Grabs, Agri- exports of intermediate products is associated with a 0.1 business Concentration, and the Smallholder: Two West decline in the dispersion of the urban size distribution African Case Studies.” Journal of Peasant Studies 39 (3–4): within countries. 731–49. 92.  Autor, Dorn, and Hanson (2016). Amiti, Mary, and Jozef Konings. 2007. “Trade Liberalization, 93.  de Vries (2017); Farole, Hollweg, and Winkler (2018). Intermediate Inputs, and Productivity: Evidence from 94. Chan (2013). Indonesia.” American Economic Review 97 (5): 1611–38. 95.  Dihel et al. (2018). Antràs, Pol, and Alonso de Gortari. 2017. “On the Geography 96. Amanor (2012). of Global Value Chains.” NBER Working Paper 23456 97. Dolan and Sutherland (2003); Kritzinger et al. (2004). (May), National Bureau of Economic Research, Cam- 98. ILO (2016). bridge, MA. 99.  ILO and IFC (2016). Antràs, Pol, Teresa C. Fort, and Felix Tintelnot. 2017. “The 100. Faber, Krause, and Sánchez de la Sierra (2017). Margins of Global Sourcing: Theory and Evidence from 101. Harrison and Scorse (2010). U.S. Firms.” American Economic Review 107 (9): 2514–64. 102. Staritz and Reis (2013). Antràs, Pol, Luis Garicano, and Esteban Rossi-Hansberg. 103. Veliu et al. (2009). 2006. “Offshoring in a Knowledge Economy.” Quarterly 104. World Bank and IFC (2014). Journal of Economics 121 (1): 31–77. 105. Muñoz-Boudet (2018).  Atkin, David. 2016. “Endogenous Skill Acquisition and Export 106. World Bank (2011).  Manufacturing in Mexico.” American Economic Review 106 107. Costa (2019). (8): 2046–85. 108. Staritz and Reis (2013).  Auerbach, Alan J. 2017. “Demystifying the Destination- 109. Christian (2013). Based Cash Flow Tax.” NBER Working Paper 23881 (Sep- 110. World Bank (2011). tember), National Bureau of Economic Research, Cam- 111. Staritz and Reis (2013). bridge, MA. 112. Ahmed (2013). Autor, David H., David Dorn, and Gordon H. Hanson. 2016. 113. World Bank, Women, Business, and the Law (database). “The China Shock: Learning from Labor-Market Adjust- 114. This section focuses on direct taxation. However, ment to Large Changes in Trade.” NBER Working Paper GVCs also pose challenges for indirect taxes, such as the 21906 (January), National Bureau of Economic Research, value added tax (VAT), although these are more tracta- Cambridge, MA. ble (see Clavey et al., forthcoming). Autor, David H., David Dorn, Lawrence F. Katz, Christina 115. See IMF, OECD, United Nations, and World Bank (2015). Patterson, and John Van Reenen. 2017. “Concentrating 96 | World Development Report 2020 on the Fall of the Labor Share.” American Economic Review: Organisation in Portuguese Firms.” CEPR Discussion Papers and Proceedings 107 (5): 180–85. Paper 10993, Centre for Economic Policy Research, Barkai, Simcha. 2016. “Declining Labor and Capital Shares.” London. Working Paper 2, Stigler Center for the Study of the Econ- Caliendo, Lorenzo, and Fernando Parro. 2013. “Estimates of omy and the State, Booth School of Business, University the Trade and Welfare Effects of NAFTA.” Review of Eco­ of Chicago. nomic Studies 82 (1): 1–44. Barrientos, Stephanie. 2014. “Gender and Global Value Chan, Man-Kwun. 2013. “Informal Workers in Global Hor- Chains: Challenges of Economic and Social Upgrading ticulture and Commodities Value Chains: A Review of in Agri-Food.” EUI Working Paper RSCAS 2014/96, Robert Literature.” WIEGO (Global Trade) Working Paper 28 Schuman Centre for Advanced Studies, European Univer- (June), Women in Informal Employment Globalizing and sity Institute, San Domenico di Fiesole, Italy. Organizing, Cambridge, MA. Barrientos, Stephanie, Peter Knorringa, Barbara Evers, Choi, Jieun, Emiko Fukase, and Albert Zeufack. 2019. “Global Margareet Visser, and Maggie Opondo. 2016. “Shifting Value Chain (GVC) Participation, Competition, and Regional Dynamics of Global Value Chains: Implications Markup: Firm-Level Evidence from Ethiopia.” Back- for Economic and Social Upgrading in African Horticul- ground paper, World Bank, Washington, DC. ture.” Environment and Planning A: Economy and Space 48 (7): Christian, Michelle. 2013. “Global Value Chains, Economic 1266–83. Upgrading, and Gender in the Tourism Industry.” In Becker, Sascha O., Karolina Eckholm, and Marc-Andreas Global Value Chains, Economic Upgrading, and Gender: Case Muendler. 2013. “Offshoring and the Onshore Composi- Studies of the Horticulture, Tourism, and Call Center Industries, tion of Tasks and Skills.” Journal of International Economics edited by Cornelia Staritz and José Guilherme Reis, 43–71. 90 (1): 91–106. Report 83233 (January). Washington, DC: World Bank. Beer, Sebastian, Ruud Aloysius de Mooij, and Li Liu. 2018. Clavey, Colin, Jonathan Leigh Pemberton, Jan Loeprick, and “International Corporate Tax Avoidance: A Review of the Marijn Verhoeven. Forthcoming. “International Tax Channels, Magnitudes, and Blind Spots.” IMF Working Reform, Digitization, and Developing Economies.” World Paper WP/18/168 (July 23), International Monetary Fund, Bank, Washington, DC. Washington, DC. CONEVAL (Mexico, Consejo Nacional de Evaluación de la Bernard, Andrew B., Andreas Moxnes, and Karen Helene Política de Desarrollo Social [National Council for the Ulltveit-Moe. 2018. “Two-Sided Heterogeneity and Trade.” Evaluation of Social Development Policy]) and World Review of Economics and Statistics 100 (3): 424–39. Bank. 2013. “Desigualdad, pobreza, y politica social en Blanchard, Emily, and William W. Olney. 2017. “Globalization Mexico: Una perspectiva del largo plazo.” CONEVAL and and Human Capital Investment: Export Composition World Bank, Mexico City. Drives Educational Attainment.” Journal of International Constantinescu, Cristina, Aaditya Mattoo, and Michele Ruta. Economics 106: 165–83. 2017. “Does Vertical Specialization Increase Productiv- Bloom, Nicholas, and John Van Reenen. 2011. “Human ity?” Policy Research Working Paper 7978, World Bank, Resource Management and Productivity.” In Handbook of Washington, DC. Labor Economics, vol. 4B, edited by David Card and Orley —— — —. 2019. “Does Vertical Specialization Increase Produc- Ashenfelter, 1697–1767. San Diego: Elsevier. tivity?” World Economy. Published electronically April 10. Borghi, Elisa, and Rosario Crinò. 2013. “Service Offshoring https://doi.org/10.1111/twec.12801. and Wages: Worker-Level Evidence from Italy.” LIUC Papers in Economics 264 (April), Serie Economia e Cooper, Joel, Randall Fox, Jan Loeprick, and Komal Mohindra. Impresa 70, Carlo Cattaneo University, Castellanza, Italy. 2016. Transfer Pricing and Developing Economies: A Handbook Borin, Alessandro, and Michele Mancini. 2015. “Follow for Policy Makers and Practitioners. Directions in Develop- the Value Added: Bilateral Gross Export Accounting.” ment: Public Sector Governance Series. Washington, DC: Temi di discussione (Economic Working Paper) 1026, World Bank. Economic Research and International Relations Area, Costa, Carlos. 2019. “The Cashew Value Chain in Mozam- Bank of Italy. bique.” With contributions by Christopher Delgado. Jobs —— — —. 2019. “Measuring What Matters in Global Value Working Paper 32, World Bank, Washington, DC. Chains and Value-Added Trade.” Policy Research Working Costinot, Arnaud, Lindsay Oldenski, and James Rauch. 2011. Paper 8804, World Bank, Washington, DC. “Adaptation and the Boundary of Multinational Firms.” Bustos, Paula. 2011. “Trade Liberalization, Exports, and Review of Economics and Statistics 93 (1): 298–308. Technology Upgrading: Evidence on the Impact of MER- Crinò, Rosario. 2012. “Service Offshoring and the Skill Com- COSUR on Argentinian Firms.” American Economic Review position of Labor Demand.” Oxford Bulletin of Economics 101 (1): 304–40. and Statistics 74 (1): 20–57. Calì, Massimiliano, Joseph Francois, Claire H. Hollweg, Crivelli, Ernesto, Ruud Aloysius de Mooij, and Michael Miriam Manchin, Doris Anita Oberdabernig, Hugo Alex- Keen. 2016. “Base Erosion, Profit Shifting, and Develop- ander Rojas-Romagosa, Stela Rubinova 2016. “The Labor ing Countries.” FinanzArchiv: Public Finance Analysis 72 (3): Content of Exports Database.” Policy Research Working 268–301. Paper 7615, World Bank, Washington, DC. Darko, Christian K., Giovanni Occhiali, and Enrico Vanino. Caliendo, Lorenzo, Giordano Mion, Luca David Opromolla, 2018. “The Chinese Are Here: Firm Level Analysis of and Esteban Rossi-Hansberg. 2015. “Productivity and Import Competition and Performance in Sub-Saharan Consequences for development | 97 Africa.” FEEM Working Paper 014.2018 (April), Note di for Effective Global Action, University of California, Lavoro Series, Fondazione Eni Enrico Mattei, Milan. Berkeley. Dauda, Seidu, Sara Nyman, and Aalia Cassim. Forthcoming. Farole, Thomas, Claire H. Hollweg, and Deborah E. Winkler. “Product Market Competition, Productivity, and Jobs: 2018. “Trade in Global Value Chains: An Assessment of The Case of South Africa.” Policy Research Working Labor Market Implications.” Jobs Working Paper 18 (July Paper, World Bank, Washington, DC. 16), World Bank, Washington, DC. Dearden, Lorraine, Howard Reed, and John Van Reenen. Faucheux, Benoit, J. R. del Rosario, Gomera Economistas 2006. “The Impact of Training on Productivity and Asociados, and Ayitika. 2014. “Analyse des Chaînes Logis- Wages: Evidence from British Panel Data.” Oxford Bulletin tiques en Haïti, Fiche Filière: Textile.” World Bank, Wash­ of Economics and Statistics 68 (4): 397–421. ington, DC; Catram Consultants, Paris. https://www De Loecker, Jan K. 2007. “Do Exports Generate Higher Pro- .catram-consultants.com/portfolio-item/logistiques ductivity? Evidence from Slovenia.” Journal of International -haiti/#tab-id-2. Economics 73 (1): 69–98. Feenstra, Robert C., and Gordon H. Hanson. 1996. “Foreign ————. 2013. “Detecting Learning by Exporting.” American Investment, Outsourcing, and Relative Wages.” In The Economic Journal: Microeconomics 5 (3): 1–21. Political Economy of Trade Policy: Papers in Honor of Jagdish De Loecker, Jan K., and Jan Eeckhout. 2018. “Global Market Bhagwati, edited by Robert C. Feenstra, Gene M. Gross- Power.” NBER Working Paper 24768 (June), National man, and Douglas A. Irwin, 89–128. Cambridge, MA: MIT Bureau of Economic Research, Cambridge, MA. Press. De Loecker, Jan K., Pinelopi Koujianou Goldberg, Amit Kumar —— — —. 1997. “Foreign Direct Investment and Relative Wages: Khandelwal, and Nina Pavcnik. 2016. “Prices, Markups, Evidence from Mexico’s Maquiladoras.” Journal of Inter­ and Trade Reform.” Econometrica 84 (2): 445–510. national Economics 42 (3–4): 371–93. De Loecker, Jan K., and Frederic Warzynski. 2009. “Markups Ferreira, Francisco H. G., Phillippe George Leite, and Martin and Firm-Level Export Status.” NBER Working Paper Ravallion. 2010. “Poverty Reduction without Economic 15198 (July), National Bureau of Economic Research, Growth? Explaining Brazil’s Poverty Dynamics, 1985– Cambridge, MA. 2004.” Journal of Development Economics 93 (1): 20–36. de Mooij, Ruud Aloysius, and Sjef Ederveen. 2008. “Corporate Ferreira, Francisco H. G., and Martin Ravallion. 2008. “Global Tax Elasticities: A Reader’s Guide to Empirical Findings.” Poverty and Inequality: A Review of the Evidence.” Policy Oxford Review of Economic Policy 24 (4): 680–97. Research Working Paper 4623, World Bank, Washing- ton, DC. de Vries, Gaaitzen J. 2017. “Drivers of Job Polarization: A Fieler, Ana Cecília, Marcela Eslava, and Daniel Yi Xu. 2014. Global Supply Chain Perspective.” With Laurie Reijn­ “Trade, Skills, and Quality Upgrading: A Theory with ders, WTO Trade Dialogues Lecture Series, presented at Evidence from Colombia.” NBER Working Paper 19992, Groningen Growth and Development Centre, Faculty National Bureau of Economic Research, Cambridge, MA. of Economics and Business, University of Groningen, Frederick, Stacey. 2018. “Apparel Skills Mapping and Func- Groningen, the Netherlands, March 2. tional Upgrading in Cambodia: Jobs Diagnostic.” World Diez, Federico J., Jiayue Fan, and Carolina Villegas-Sánchez. Bank, Washington, DC. 2019. “Global Declining Competition.” IMF Working Gereffi, Gary, John Humphrey, and Timothy Sturgeon. 2005. Paper WP/19/82, International Monetary Fund, Wash- “The Governance of Global Value Chains.” Review of Inter­ ington, DC. national Political Economy 12 (1): 78–104. Dihel, Nora, Arti Grover Goswami, Claire H. Hollweg, and Global Reporting Initiative and United Nations Global Com- Anja Slaney. 2018. “How Does Participation in Value pact. 2017. “Business Reporting on the SDGs: An Analysis Chains Matter to African Farmers?” Policy Research of the Goals and Targets.” https://www.globalreporting Working Paper 8506, World Bank, Washington, DC. .org/resourcelibrary/GRI_UNGC_Business-Reporting-on Doane, Seth. 2015. “A Look at China’s ‘Button Town.’  ” CBS -SDGs_Analysis-of-Goals-and-Targets.pdf. Evening News, October 8. https://www.cbsnews.com/news Goldberg, Pinelopi Koujianou, Amit Kumar Khandelwal, /welcome-to-button-town-china/. Nina Pavcnik, and Petia Topalova. 2010. “Imported Inter- Dolan, Catherine, and Kirsty Sutherland. 2003. “Gender and mediate Inputs and Domestic Product Growth: Evidence Employment in the Kenya Horticulture Value Chain.” from India.” Quarterly Journal of Economics 125 (4): 1727–67. Globalisation and Poverty Discussion Paper No. 8, Glo- Goldberg, Pinelopi Koujianou, and Nina Pavcnik. 2007. “Dis- balisation and Poverty Research Programme, Institute of tributional Effects of Globalization in Developing Coun- Development Studies, Brighton, U.K. tries.” Journal of Economic Literature 45 (1): 39–82. Dollar, David, and Aart Kraay. 2002. “Growth Is Good for the Gradzewicz, Michał, and Jakub Mućk. 2019. “Globalization Poor.” Journal of Economic Growth 7 (3): 195–225. and the Fall in Markups.” Warsaw School of Economics, Eslava, Marcela, Ana Cecília Fieler, and Daniel Yi Xu. 2015. Warsaw. https://www.nbp.pl/badania/seminaria/20ii2019 “(Indirect) Input Linkages.” American Economic Review: .pdf. Papers and Proceedings 105 (5): 662–66. Griliches, Zvi. 1969. “Capital-Skill Complementarity.” Review Faber, Benjamin, Benjamin Krause, and Raúl Sánchez de la of Economics and Statistics 51 (4): 465–68. Sierra. 2017. “Artisanal Mining, Livelihoods, and Child GSO (General Statistics Office, Vietnam). 2012. Result of the Labor in the Cobalt Supply Chain of the Democratic Viet Nam Household Living Standards Survey 2012. Hanoi: Republic of Congo.” Policy Report (May 6), Center GSO. 98 | World Development Report 2020 Guardian. 2019. “Bangladesh Strikes: Thousands of Garment IMF, OECD, UN, and World Bank. Washington, DC: World Workers Clash with Police over Poor Pay.” January 13. Bank. https://www.theguardian.com/world/2019/jan/14 INEGI (Mexico, Instituto Nacional de Estadistica y Geografia /bangladesh-strikes-thousands-of-garment-workers [National Institute of Statistics and Geography]). 2014. -clash-with-police-over-poor-pay. “Economic Censuses.” INEGI, Aguascalientes, Mexico. Halpern, Laszlo, Miklos Koren, and Adam Szeidl. 2015. https://en.www.inegi.org.mx/programas/ce/2014/. “Imported Inputs and Productivity.” American Economic Javorcik, Beata Smarzynska. 2014. “Does FDI Bring Good Jobs Review 105 (12): 3660–3703. to Host Countries?” Policy Research Working Paper 6936, Hansson, Bo. 2008. “Job-Related Training and Benefits for World Bank, Washington, DC. Individuals: A Review of Evidence and Explanations.” Jensen, Robert. 2012. “Do Labor Market Opportunities Affect OECD Education Working Paper 19, Organisation for Young Women’s Work and Family Decisions? Experimen- Economic Co-operation and Development, Paris. tal Evidence from India.” Quarterly Journal of Economics 127 Harrison, Ann, and Jason Scorse. 2010. “Multinationals and (2): 753–92. Anti-sweatshop Activism.” American Economic Review 100 Karabarbounis, Loukas, and Brent Neiman. 2013. “The Global (1): 247–73. Decline of the Labor Share.” NBER Working Paper 19136 Heath, Rachel, and A. Mushfiq Mobarak. 2015. “Manufactur- (June), National Bureau of Economic Research, Cam- ing Growth and the Lives of Bangladeshi Women.” Journal bridge, MA. of Development Economics 115 (July): 1–15. Katz, Jorge, and Carlo Pietrobelli. 2018. “Natural Resource- Hijzen, Alexander, Pedro S. Martins, Thorsten Schank, and Based Growth, Global Value Chains, and Domestic Richard Upward. 2013. “Foreign-Owned Firms around the Capabilities in the Mining Industry.” Resources Policy 58 World: A Comparative Analysis of Wages and Employ- (October): 11–20. ment at the Micro-Level.” European Economic Review 60 Kremer, Michael R., and Eric Maskin. 2006. “Globalization (May): 170–88. and Inequality.” Weatherhead Center for International Hummels, David L., Rasmus Jørgensen, Jakob Munch, and Affairs, Harvard University, Cambridge, MA. Chong Xiang. 2014. “The Wage Effects of Offshoring: Krishnan, Aarti. 2017. “The Origin and Expansion of Regional Evidence from Danish Matched Worker-Firm Data.” Value Chains: The Case of Kenyan Horticulture.” Global American Economic Review 104 (6): 1597–1629. Networks 18 (2): 238–63. Iacovone, Leonardo, and Gustavo Crespi. 2010. “Catching Up Kritzinger, Andrienetta, Stephanie Barrientos, and Hester with the Technological Frontier: Micro-Level Evidence on Rossouw. 2004. “Global Production and Flexible Employ- Growth and Convergence.” Industrial and Corporate Change ment in South African Horticulture: Experiences of 19 (6): 2073–96. Contract Workers in Fruit Exports.” Sociologia Ruralis 44 Iberahim, Hadijah. 2013. “Vertical Integration into Global (1): 17–39. Value Chains and Its Effects on Skill Development.” In Krusell, Per, Lee E. Ohanian, José-Víctor Ríos-Rull, and IEEE Business Engineering and Industrial Applications Collo­ Giovanni L. Violante. 2000. “Capital-Skill Complementar- quium, 28–33. Piscataway, NJ: Institute of Electrical and ity and Inequality: A Macroeconomic Analysis.” Economet­ Electronics Engineers. IFC (International Finance Corporation). 2018. “Horticulture: rica 68 (5): 1029–53. Value Chain Analysis, Azerbaijan.” South Caucasus Gen- Kugler, Maurice, and Eric A. Verhoogen. 2012. “Prices, Plant der Assessment Technical Assistance (P160432) Report Size, and Product Quality.” Review of Economic Studies 79 (1): (April), IFC, Washington, DC. 307–39. ILO (International Labour Organization). 2016. Sectoral Studies Kumar, Ruchira. 2017. “Global Value Chains: A Way to Create on Decent Work in Global Supply Chains: Comparative Analy­ More, Better, and Inclusive Jobs.” Jobs and Development sis of Opportunities and Challenges for Social and Economic (blog), July 12. https://blogs.worldbank.org/jobs/global Upgrading. Geneva: International Labour Office. https:// -value-chains-way-create-more-better-and-inclusive-jobs. www.ilo.org/public/libdoc/ilo/2016/490742.pdf. Li, Bingjing. 2018. “Export Expansion, Skill Acquisition and ILO (International Labour Organization) and IFC (Interna- Industry Specialization: Evidence from China.” Journal of tional Finance Corporation). 2016. “Progress and Poten- International Economics 114 (September): 346–61. tial: How Better Work Is Improving Garment Workers’ — — ——. 2019. “Skill-Biased Imports, Human Capital Accumu- Lives and Boosting Factory Competitiveness.” Interna- lation, and the Allocation of Talent.” https://pdfs.semantic tional Labour Office, Geneva. https://betterwork.org/dev scholar.org/1497/7c5d507017f14afaeec804011564af95514b /wp-content/uploads/2016/09/BW-Progress-and .pdf. -Potential_Web-final.pdf. Lopez-Acevedo, Gladys, and Raymond Robertson, eds. 2016. IMF (International Monetary Fund). 2019. “Corporate Taxa- Stitches to Riches? Apparel Employment, Trade, and Economic tion in the Global Economy.” Policy Paper 19/007 (March Development in South Asia. Directions in Development: 10), IMF, Washington, DC. Poverty Series. Washington, DC: World Bank. IMF (International Monetary Fund), OECD (Organisation Lu, Yue, Yunlong Lu, Rui Xie, and Xiao Yu. 2019. “Does for Economic Co-operation and Development), United Global Value Chain Engagement Improve Firms’ Wages: Nations, and World Bank. 2015. Options for Low-Income The Empirical Evidence from China.” World Economy. Countries’ Effective and Efficient Use of Tax Incentives for Invest­ Published electronically April 16. https://doi.org/10.1111 ment: A Report to the G-20 Development Working Group by the /twec.12805. Consequences for development | 99 Macchiavello, Rocco, and Josepa Miquel-Florensa. 2017. “Ver- Reshef, Ariell, and Gianluca Santoni. 2019. “Are Your Labor tical Integration and Relational Contracts: Evidence from Shares Set in Beijing? The View through the Lens of the Costa Rica Coffee Chain.” CEPR Discussion Paper No. Global Value Chains.” Working paper (May), Paris School DP11874, Centre for Economic Policy Research, London. of Economics, Paris. SSRN: https://ssrn.com/abstract=2924734. Rocha, Nadia, and Deborah E. Winkler. 2019. “Trade and ————. 2019. “Buyer-Driven Upgrading in GVCs: The Sus- Female Labor Participation: Stylized Facts Using a Global tainable Quality Program in Colombia,” CEPR Discussion Dataset.” Background paper, How Can 21st Century Trade Papers 13935, Centre for Economic Policy Research, Help to Close the Gender Gap?, World Bank and World Trade London. Organization. Maertens, Miet, Bart Minten, and Johan F. M. Swinnen. Rodrik, Dani. 2018. “New Technologies, Global Value Chains, 2012. “Modern Food Supply Chains and Development: and Developing Economies.” NBER Working Paper 25164, Evidence from Horticulture Export Sectors in Sub- National Bureau of Economic Research, Cambridge, MA. Saharan Africa.” Development Policy Review 30 (4): 437–97. Said-Allsopp, Muhaimina, and Anne Tallontire. 2015. “Path- Markusen, James, and Natalia Trofimenko. 2009. “Teach- ways to Empowerment? Dynamics of Women’s Participa- ing Locals New Tricks: Foreign Experts as a Channel of tion in Global Value Chains.” Journal of Cleaner Production Knowledge Transfers.” Journal of Development Economics 88 107 (November): 114–21. (1): 121–31. Sampson, Thomas. 2015. “Dynamic Selection: An Idea Flows McCaig, Brian, and Nina Pavcnik. 2015. “Informal Employ- Theory of Entry, Trade, and Growth.” Quarterly Journal of ment and a Growing Globalizing Low-Income Country.” Economics 131 (1): 315–80. NBER Working Paper 20891 (January), National Bureau of Shepherd, Ben, and Susan Stone. 2012. “Global Production Economic Research, Cambridge, MA. Networks and Employment: A Developing Country Per- Moazzem, Khondaker Golam, and Marzuka Ahmad Radia. spective.” OECD Trade Policy Paper 154 (May 14), Organ- 2018. “ ‘Data Universe’ of Bangladesh’s RMG Enterprises: isation for Economic Co-operation and Development, Key Features and Limitations.” Centre for Policy Dia- Paris. logue, Dhaka. Solotaroff, Jennifer L., Aphichke Kotikula, Tara Lonnberg, Morris, Mike, and Cornelia Staritz. 2014. “Industrialization Snigdha Ali, Rohini P. Pande, and Ferdous Jahan. 2019. Trajectories in Madagascar’s Export Apparel Industry: Voices to Choices: Bangladesh’s Journey in Women’s Economic Ownership, Embeddedness, Markets, and Upgrading.” Empowerment. Washington, DC: World Bank. World Development 56 (April): 243–57. Staritz, Cornelia, Leonhard Plank, and Mike Morris. 2016. Muñoz-Boudet, Ana María. 2018. “Value Chain Selection: “Global Value Chains, Industrial Policy, and Sustainable Azerbaijan.” South Caucasus Gender Assessment Tech- Development: Ethiopia’s Apparel Export Sector.” ICTSD nical Assistance (P160432) Report (May), International Country Case Study (November), International Centre for Finance Corporation, Washington, DC. Trade and Sustainable Development, Geneva. Navarro, Lucas. 2018. “The World Class Supplier Program for Staritz, Cornelia, and José Guilherme Reis, eds. 2013. Global Mining in Chile: Assessment and Perspectives.” Resources Value Chains, Economic Upgrading, and Gender: Case Studies Policy 58 (October): 49–61. of the Horticulture, Tourism, and Call Center Industries. Report Oxera Consulting. 2019. “Increased Market Power: A Global 83233 (January). Washington, DC: World Bank. Problem That Needs Solving?” Oxera Agenda (January), Suzuki, Aya, Yukichi Mano, and Girum Abebe. 2018. “Earn- Oxera Consulting, Oxford, U.K. https://www.oxera.com ings, Savings, and Job Satisfaction in a Labor-Intensive /wp-content/uploads/2019/01/Increased-market-power Export Sector: Evidence from the Cut Flower Industry in .pdf. Ethiopia.” World Development 110 (October): 176–91. Pahl, Stefan, and Marcel Peter Timmer. 2019. “Jobs in Global Taglioni, Daria, and Deborah E. Winkler. 2016. Making Global Value Chains: A New Measurement Framework and Value Chains Work for Development. Trade and Develop- Analysis for Fourteen Countries in Africa and Asia.” Back- ment Series. Washington, DC: World Bank. ground paper, World Bank, Washington, DC. te Velde, Dirk Willem, and Oliver Morrissey. 2003. “Do Work- Pew Research Center. 2014. “Faith and Skepticism about ers in Africa Get a Wage Premium if Employed in Firms Trade, Foreign Investment.” Pew Report (September 16), Owned by Foreigners?” Journal of African Economies 12 (1): Pew Research Center, Washington, DC. https://www 41–73. .pewresearch.org/global/2014/09/16/faith-and-skepticism UNCTAD (United Nations Conference on Trade and Devel- -about-trade-foreign-investment/. opment). 2013. World Investment Report 2013: Global Value Pietrobelli, Carlo, Anabel Marin, and Jocelyn Olivari. 2018. Chains: Investment and Trade for Development. Geneva: “Innovation in Mining Value Chains: New Evidence United Nations. from Latin America.” Resources Policy 58 (October): 1–10. — —— —. 2015. World Investment Report 2015: Reforming Interna­ Pietrobelli, Carlo, and Jocelyn Olivari, eds. 2018. “Special Issue tional Investment Governance. Geneva: United Nations. on Mining Value Chains, Innovation, and Learning.” Van den Broeck, Goedele, and Miet Maertens. 2017. “Does Resources Policy 58 (October). Off-Farm Wage Employment Make Women in Rural Ravallion, Martin, and Guarav Datt. 2002. “Why Has Eco- Senegal Happy?” Feminist Economics 23 (4): 250–75. nomic Growth Been More Pro-Poor in Some States of Veliu, Atdhe, Nebiyeluel Gessese, Catherine Ragasa, and India than Others?” Journal of Development Economics 68 Christine Okali. 2009. “Gender Analysis of Aquacul- (2): 381–400. ture Value Chains in Northeast Vietnam and Nigeria.” 100 | World Development Report 2020 Agriculture and Rural Development Discussion Paper 44, ——— —. 2018. “Apparel GVC Analysis: Sri Lanka.” World Bank, World Bank, Washington, DC. Washington, DC. https://www.theciip.org/sites/ciip/files Verhoogen, Eric A. 2008. “Trade, Quality Upgrading, and /Apparel%20GVC%20analysis.pdf. Wage Inequality in the Mexican Manufacturing Sector.” ——— —. 2019a. “Global Investment Competitiveness Report Quarterly Journal of Economics 123 (2): 489–530. 2019/2020.” World Bank, Washington, DC. Whittaker, D. Hugh, Tianbiao Zhu, Timothy Sturgeon, Mon ——— —. 2019b. World Development Report 2019: The Changing Han Tsai, and Toshie Okita. 2010. “Compressed Develop- Nature of Work. Washington, DC: World Bank. ment.” Studies in Comparative International Development 45 World Bank and IFC (International Finance Corporation). 2014. (4): 439–67. “The Fruit of Her Labor: Promoting Gender-Equitable World Bank. 2011. “Understanding Gender in Agricultural Agribusiness in Papua New Guinea.” Report ACS10004 Value Chains: The Cases of Grapes/Raisins, Almonds, (June 30), World Bank and IFC, Washington, DC. and Saffron in Afghanistan.” Report 62323-AF (May), World Bank and World Trade Organization. Forthcoming. Agriculture and Rural Development Unit, Sustainable “How Can 21st Century Trade Help to Close the Gender Development Department, South Asia Region, World Gap?” Washington, DC, and Geneva. Bank, Washington, DC. Consequences for development | 101 4 Macroeconomic implications Key findings • G  lobal value chains (GVCs) are associated with greater synchronization of economic activity across countries. When production in one country relies on inputs from another country, then economic activity in the two countries is linked. • G  VCs create strong links in price formation, implying that inflation in one country is more likely to spill over to its direct and indirect trading partners. In this sense, GVC participation is associated with the rising synchrony in inflation across countries. • I n GVC countries, episodes of export growth are linked to episodes of import growth. This finding implies that the consequences of currency movements for export volumes are likely to be dampened. • G  VCs amplify the costs of protectionism for trade and growth. The back-and-forth movement in tasks and parts across borders means that trade barriers are incurred multiple times. Protectionism is therefore costlier for growth and welfare.  rade agreements have the potential to reshape the geography of production. The • T prevalence of rules of origin as well as the productivity gains associated with a reduction in the price of imported inputs imply that trade agreements have systemic consequences for the allocation of production across countries in GVCs. 102 | World Development Report 2020 G lobal value chains (GVCs) strengthen the Posing new challenges for governments and pol- economic connections between countries. icy makers, these realities require closer cooperation Instead of individually selling final goods between countries. National policies are now trans- and competing for the same customers, countries are mitted to other countries, and the GVC feedback loop increasingly related through rigid production linkages can reduce their effectiveness. Because of the high that bind them to a common fate. This international interdependence of production structures, decisions interdependence means that policies and economic by governments and central banks are more likely conditions in one country affect its trading partners to have a systemic impact, and their effectiveness and propagate to the rest of the world. As a result, the depends on policies in various parts of the world. benefits of international coordination (and the costs Moreover, regional agreements can have global rip- of not coordinating) have increased. Four are investi- ples, and economic issues are more global, calling for gated in this chapter. coordinated solutions. In view of their rigid ties, GVCs First, production linkages are associated with would benefit from multilateral institutions helping greater synchrony of economic activity across coun- to coordinate policy worldwide, including through tries. When production in one country relies on inputs the formulation of product standards, investment and from its trading partners, the economic conditions in intellectual property protections, or the timing of fis- other countries affect its domestic activity and its abil- cal adjustments. ity to thrive. Although international trade in finished products cannot be associated with any change in the synchronization of GDP across countries, trade in Synchronizing economic activity intermediate inputs can be. When the production of a good in Vietnam requires Second, input–output linkages create strong links inputs from Indonesia and is then used for production in price formation, implying that inflation in one in China, it is only natural that supply and demand country is more likely to spill over to its direct and shocks in one country will be felt by its suppliers and indirect trading partners. In this sense, GVC partici- customers. One example of such an arrangement is pation is associated with the rising synchrony of both Nike, one of the world’s most valuable sport brands. real economic activity and inflation across countries. It has segmented its footwear production across these At the same time, the actions of national central banks, three South Asian countries. through production linkages between domestic and Over the past 30 years, the co-movement of foreign firms, can have important consequences in economic activity has surged globally. In the 1980s, other countries as well. economic cycles in different countries were largely Third, because of the interconnections in pro- independent of one another, especially in middle- and duction, episodes of export growth are linked with low-income countries, with correlations of less than similar growth in imports. Thus the consequences of 0.1. But economic activity has since become much currency movements for export volumes are likely to more correlated—the most for high-income countries be reduced and become harder to predict. Export vol- followed by middle-income and low-income countries umes do not react to the exchange rate with the direct (figure 4.1). partner; they react to the exchange rate in the country The recent worldwide increase in economic syn- of final consumption. When a government changes chrony stems in part from the rise of GVCs. Produc- the value of its currency, it affects the trade flows of tion is increasingly organized according to a “world other countries throughout the production chain. factory” view, which drastically changes how shocks Fourth, the rise of GVCs influences the impact of are transmitted across borders. To understand the regional trade agreements and how policy makers importance of those changes, one could start by tak- should think about the possible diversion of trade ing a look at the world before the recent increase in flows. When firm-to-firm relationships are rigid, international production linkages.1 the benefits of accessing new markets can be shared If two countries were open to trade and produce throughout the production chains with countries not a similar final good—say, clothes—their firms com- part of the trade agreement. Conversely, the disrup- peted in the same markets for the same customers. tion created by trade wars and dismantled agreements As a result, a country’s increase in productivity could may be transmitted to other trading partners and may enhance consumer welfare everywhere, but it could not be easily avoided by reorganizing buyer–seller also mean tougher competition for its competitors. relationships. In this sense, good news in one country could be bad Macroeconomic implications | 103 Figure 4.1  In all income groups, countries’ economic the design, production, and after-sales services of activity has become more synchronized since the many goods are spread over many countries. This new mid-1990s reality changes the extent to which economic fluctu- ations are transmitted across countries. If a country’s 0.6 productivity rises, the consequences are good for trad- 0.5 ing partners buying its goods as inputs as well as for 0.4 the country’s own suppliers: they can share the com- Correlation petitive gains throughout all production stages, and 0.3 they are less likely to cannibalize each other’s market 0.2 shares.2 The positive historical association between total 0.1 trade and business cycle co-movement was driven 0 by trade in intermediate inputs (figure 4.2). Although GVCs are not the only factor explaining the surge in 81 83 85 87 89 91 93 95 97 99 01 03 05 07 09 11 13 19 19 19 19 19 19 19 19 19 19 20 20 20 20 20 20 20 GDP correlation across countries, evidence about High-income Middle-income Low-income their role is growing. From both a microdata and firm Source: WDR 2020 team, using data from the World Bank’s WDI database. See appendix A for a perspective3 and a more macroaggregate perspective,4 description of the databases used in this Report. studies have shown that the recent increase in input– Note: Each year represents the midpoint of a 10-year moving window. Each line represents the average output linkages enhanced the co-movements in eco- of all country-pair GDP correlations, taken over all country-pairs containing at least one country in the income group considered (high, middle, and low). nomic activity. The economic fates of countries participating in GVCs are tied to one another. Even if at the micro­ Figure 4.2  Greater synchrony of economic activity is economic level individual firms in different countries associated with GVCs continue to compete, the aggregate health of an econ- 5 omy now depends on the health of other economies supplying inputs or buying outputs. Based on a panel 4 of 150 countries for the last 50 years, one study finds Change in GDP synchrony that moving from the 25th to the 75th percentile of 3 trade in intermediate inputs is associated with an 2 increase in the GDP correlation of 28 percentage points.5 1 Synchrony of economic activity across countries is a key indicator for many macroeconomic policies. 0 For example, the extent to which the West African Economic and Monetary Union (WAEMU) can be –1 considered an optimal currency area largely depends –0.4 –0.2 0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 on the synchrony of business cycles among all mem- Change in production connectivity ber countries. And beyond currency considerations, Sources: WDR 2020 team, using data from the World Bank’s WDI and WITS databases. the synchrony of economic activity among countries Note: Each dot represents a pair of regions—for example, East Asia and Pacific and Sub-Saharan Africa signals interdependence, so both good news and bad are one region-pair, and Latin America and the Caribbean and South Asia are another. The x-axis measures the change in production connectivity defined as the total trade in intermediates as a share of news are transmitted from one country to the next. GDP of both regions, taken between the 2002–11 and 1982–1991 time windows. The y-axis measures the Large firms dominate the global economy. For 32 proportional change in the GDP correlation over time. Region-pair values are computed by taking the simple average of all country-pairs between two regions. developing countries, the five largest exporters in a country account on average for a third of its exports and nearly half of its export growth.6 Although the news for its trading partners. For example, higher effi- importance of large firms in driving economic growth ciency in the garment industry in Bangladesh could be is not new, their impact reached a more global scale linked to a contraction in production in India. with the expansion of GVCs. With production more In a world dominated by GVCs, however, countries fragmented across countries, any local decision that increasingly trade intermediate inputs and are tied by improves a global firm’s ability to thrive will have a rigid production linkages. From the iPhone to Nutella, positive impact on many countries. 104 | World Development Report 2020 Propagating shocks GVCs are also linked to greater synchrony of financial cycles and stock market returns. Looking The strength of propagating shocks across firms at the consequences of natural disasters, firms expe- and countries is a function of the “specificity” of the rience a larger drop in stock market returns when input–output relationship, which is not always well disasters hit their specific suppliers than their non- represented by simply looking at cost shares across specific ones.7 A specific supplier is a supplier that countries. When an input is needed for production, produces an input tailored just for its customers. losing access to it can be disastrous, even though When such a relationship exists, both buyer and the input may not represent a large share of total supplier may face less flexibility in changing their production costs (box 4.1). The interdependence of business partners when needed, as it takes time to firms and countries thus increases to the extent that find another firm willing and able to produce or buy GVCs involve custom products that cannot be easily specific inputs. replaced. Box 4.1  The Japanese earthquake and the costs of supply chain disruptions Businesses tend to focus on the possibility that inputs will and sensors produced in Japan had few or no substitutes. increase in price or be delivered late. But disruptions by At the time, it was estimated that about a third of Apple’s extreme events are a rising threat. In 2018 extreme weather flash drives came from Toshiba, Japan, and the rest from caused $81 billion in global losses; in 2017, $300 billion.a the Republic of Korea.c Natural disasters can have unanticipated cascading impacts Quantifying the global impact of such a disaster is not along GVCs, shocking distribution and supply networks easy. For the transport equipment industry, the disruption worldwide. cost an estimated $139 billion (in value added), with Japan In March 2011, the Tōhoku earthquake, measuring 9.0 on suffering about 40 percent of the impact and the rest falling the Richter scale, hit Japan’s northeast coast. Several tsu- mainly on the United States (25 percent), China (8 percent), namis followed, devastating coastal areas, flooding and dis- the European Union (8 percent), and Canada (7 percent).d abling local nuclear power stations, and creating a national The substitutability of inputs is a critical determinant nuclear crisis. The triple disaster was catastrophic for GVCs, of supply chain shocks. In one study of U.S. affiliates of particularly the automotive, computer, and consumer elec- Japanese firms, the degree of the shock depended not on tronics producers that rely heavily on Japanese suppliers of the level of Japanese ownership, but on the U.S. affiliate’s specialized parts and components. As Japanese production ability to replace in the short run imported intermediates of automotive equipment drew to a halt, senior executives from Japan with alternative inputs.e In the month following at Toyota, Honda, Opel, Nissan, and General Motors froze the crisis, U.S. manufacturing output fell about 1 percent production lines in several factories worldwide, leading to and remained significantly below previous levels for the losses of $70 million a day.b next six months. Famously, automakers temporarily stopped orders These findings are particularly relevant to buyers and for cars in colors that required a specialty pigment called suppliers holding low inventories and relying heavily on Xirallic, which gives cars a glittery shine. Xirallic is produced just-in-time production to keep inventory costs low. Risk only in Japan, and its production was badly affected during management strategies to diversify suppliers and reduce the nuclear crisis. firm sensitivity to inventory shortages and delays in logis- In electronics, the problems were similar. The many tics will become more important as environment-related specialized connectors, speakers, microphones, batteries, disasters increase. a. Swiss Re (2019). b. Automotive News (2011). c. Lohr (2011). d. Arto, Andreoni, and Rueda Cantuche (2015). e. Boehm, Flaaen, and Pandalai-Nayar (2019). Macroeconomic implications | 105 Synchronizing inflation the “slack” in an economy) in estimating the inflation pressure in the economy, GVCs have been shown to International input–output linkages also create strong significantly increase the global factors at the expense links in price formation, implying that inflation in one of purely domestic ones. country is more likely to spill over to its direct and Furthermore, an increase in imports and exports of indirect trading partners (figure 4.3). Such linkages intermediate inputs is associated with a decline in the account for an estimated half of the global compo- relative weight of the domestic output gap in favor of nent of producer price index inflation.8 Although global economic conditions in the formation of infla- imported inflation has been a factor in the consumer tion (figure 4.5). Because imported intermediates can price index, its extension to producer prices has policy be used to produce goods that are either reexported implications for central banks. further or consumed in the domestic economy, such When designing their monetary policy and target- a result points to a synchronization of inflation across ing a given inflation rate, authorities need to account all sectors. This finding is in line with Ha, Kose, and for the economic conditions and strategies of their Ohnsorge (2019), who show that inflation synchroni- direct and indirect trading partners. In this sense, GVC zation has been significant across all inflation mea- participation is associated with the rising synchrony sures since 2001, whereas it was previously prominent of not only real economic activity but also inflation only for inflation measures that included mostly trad- across countries. able goods. Backward GVC participation is associated with Finally, GVCs are not only associated with the an increase in the globalization of inflation. For co-movement of inflation patterns but also may be each country, the change in the correlation between linked to the global reduction in inflation. The emer- domestic and world inflation over the past decade is gence of independent central banks and better moni- associated with an increase in the amount of imported toring in many countries has played an important role, inputs used in production. but a recent study by the Organisation for Economic The fragmentation of production across countries Co-operation and Development (OECD) also suggests also plays a role in the synchrony of inflation expecta- that GVCs have contributed to lower inflation via down- tions, which feeds back into current inflation (figure ward pressures on labor through heightened competi- 4.4). For example, although economists have long rec- tion across countries to attract tasks, in particular when ognized domestic and global output gaps (measures of low-wage countries are integrated in supply chains.9 Figure 4.3  The synchrony of inflation increased between 1988 and 2010 a. CPI inflation b. GDP deflator inflation 0.7 0.5 0.6 0.4 0.5 Correlation Correlation 0.3 0.4 0.3 0.2 0.2 0.1 0.1 0 0 88 90 92 94 96 98 00 02 04 06 08 10 88 90 92 94 96 98 00 02 04 06 08 10 19 19 19 19 19 19 20 20 20 20 20 20 19 19 19 19 19 19 20 20 20 20 20 20 Developed countries Developing countries Source: WDR 2020 team, using data from the World Bank’s WDI database. Note: For each country, the correlation between domestic and world inflation was determined using two measures of price levels: the consumer price index (CPI) in panel a and the GDP deflator in panel b. An average was then taken across two income groups in order to plot the evolution of this average correlation. Each year represents the midpoint of a 15-year rolling window. 106 | World Development Report 2020 Reducing the effect of Figure 4.4  GVCs are associated with greater inflation devaluations synchrony in some countries 7 Economics textbooks presume a relationship between movements in a country’s exchange rate and its export 6 Change in inflation synchrony volumes. When the currency depreciates, export vol- 5 using GDP deflator umes are expected to increase by some amount, and 4 that amount is called the exchange rate elasticity of 3 exports. Yet some recent significant exchange rate 2 movements, such as those in the United Kingdom 1 in 2007–09 and in Japan in 2012–14, were not associ- ated with large movements in trade volumes.10 This 0 perceived unresponsiveness of exports to exchange –1 rate fluctuations has raised the question of whether –2 the exchange rate elasticity of export volumes has –0.2 0 0.2 0.4 0.6 0.8 changed or even dropped to zero. Change in backward GVC participation For all country income groups, changes in a Sources: WDR 2020 team, using data from the World Bank’s WDI database. country’s exports and imports have become more cor- Note: Each dot represents a pair of regions—for example, East Asia and Pacific and Sub-Saharan related over time (figure 4.6). Upper-middle-income Africa are one region-pair and Latin America and the Caribbean and South Asia are another. The x-axis countries especially, many of whom are engaged in measures the change, taken between the 2000–2009 and 1990–1999 time windows, in production connectivity defined as the total trade in intermediates as a share of GDP of both regions. The y-axis GVCs in Europe, have seen a sharp rise. measures the proportional change in inflation correlation for the same time windows, where inflation is Over the last decades, short-term growth in exports measured by the changes in the GDP deflator. has been accompanied by import growth. Contrary to what the standard quantitative trade models of importing predict, a country’s aggregate imported Figure 4.5  Trade in intermediate inputs increased the input share increases after large depreciations.11 This weight of global factors in inflation formation from can be explained by the fact that exporting firms are 1983 to 2006 often also importers, and export opportunities are 1.5 2001 accompanied by a need to import. 2002 1.0 1994 2000 The latest research suggests that all production 1999 Relative global factor 1990 2003 2005 1998 linkages can have an impact on export elasticities and 0.5 1989 1997 2004 2006 that GVCs can have complex effects on devaluation. By 1993 1991 1996 1995 loosening the effectiveness of devaluations, the expan- 0 1992 1987 sion of GVCs complicates the task of policy makers 1988 –0.5 1986 1984 and creates the need for international coordination 1985 (box 4.2). –1.0 Greater participation in GVCs is expected to gen- 1983 erate larger bilateral balances, but it is not necessarily –1.5 14 16 18 20 22 24 26 28 associated with a larger overall trade balance. Indeed, current accounts at the country level are mostly Share of exports and imports of intermediate goods and services in GDP (%) determined by savings, investments, and cross-border finance and are little affected by changes in trade Source: Auer, Borio, and Filardo 2017. policy or by the links between imports and exports. Note: The relative global factor is the difference between domestic and global output gaps in the formation of inflation. The upward-sloping line shows the positive relationship between the global Policy makers in countries participating in GVCs weight in domestic inflation (y-axis) and participation in GVCs (x-axis). should track not only the currency composition of inputs for production, but also the currency in the country of final absorption (figure 4.7).12 In doing so, reduces the change in export price in response to they should keep in mind the following points: exchange rate movements, thereby lessening the associated change in export volumes. •  An increase in an export’s share of foreign value A greater share of exports that return as imports to •  added from a country with a different currency a country sharing the same currency weakens the Macroeconomic implications | 107 Figure 4.6  Export boosts tend to coincide with An increase in the share of exports used in the •  import boosts—more now than 30 years ago destination country to produce further reexports that are ultimately consumed in a third country 1.0 increases the responsiveness of trade flows to the Correlation between import and export growth 0.9 direct trading partner’s nominal effective exchange 0.8 rate, creating significant interdependence across 0.7 countries. This mechanism underlines the inter- national interconnections that characterize today’s 0.6 production processes. 0.5 With the international fragmentation of produc- 0.4 tion across countries, export performance in one 0.3 country can be driven by the demand addressed by 0.2 firms located in other countries. In this sense, the 0.1 consequences of devaluing a country’s currency value propagate upstream in the supply chain and trigger 0 export growth from its suppliers. –0.1 Interestingly, greater participation in international 1985 1990 1995 2000 2005 2010 2015 production decreases the exchange rate elasticity of Low-income Lower-middle-income exports, and a currency devaluation could also reduce a Upper-middle-income High-income sector’s exports to a specific destination. This happens Sources: WDR 2020 team, using data on annual trade flows from the World Bank’s WITS database and whenever a sector has both a high share of foreign income group classification from the World Bank (2017 version). value added in exports and a high share of exports Note: The year-on-year growth rate of total imports and exports was computed for each year. Then the correlation between these growth rates was computed for a rolling window of 10 years. Each year reimported and consumed in a country with the same represents the midpoint of the 10-year rolling window. currency. Current GVC participation around the world exchange rate responsiveness of exports. Simply already accounts for a significant decline in the effi- put, if the final demand driving exports is located ciency of devaluation in boosting exports (figure 4.8). at home or in a country with the same currency, a Sectors in the top decile of the backward GVC partici- devaluation can do little to boost those trade flows. pation have an export elasticity that is only two-thirds Box 4.2  Blunting the effects of devaluation on Turkey’s exports Changes in imports and exports are driven by many ele- large current account deficits over the past 15 years. In ments, and the value of currency is only one of them. Other 2015 the country was well integrated in GVCs, with its important determinants are economic and financial condi- share of foreign value added in exports reaching 30 per- tions and the uncertainty in both direct and indirect trading cent, almost 10 percentage points above the world aver- partners, as well as possible changes in tariffs and nontariff age. Between 2015 and 2018, the real effective exchange barriers and the design of industrial policy in both domestic rate depreciated by 25 percent, and such a large move- and foreign economies. ment translated into only a modest 5 percent in export Without accounting for all these other factors, it is growth (much slower than the world’s export growth of difficult to draw strong conclusions about the way GVCs 8 percent during the same period) and 11 percent in import are changing the link between devaluations and export growth. volumes. With those caveats in mind, the recent devalua- This relatively small adjustment is especially striking tions in Turkey illustrate the mechanisms described in this because recent World Bank studies have shown that his- chapter. torically Turkey’s current account balance has been less Turkey has moved rapidly from a current account that persistent than is typically found in the cross-country lit- was relatively in balance up to 2000 to sustaining relatively erature, suggesting that it adjusts more rapidly to shocks.a Knight, Nedeljkovic, and Portugal-Perez 2019. a.  108 | World Development Report 2020 Figure 4.8 Figure 4.7 With GVCs, devaluations can have complex consequences Exports less sensitive to currency devaluations Exports sensitive to the exchange rate of trade partner Two-way trade Trade flow in which the origin Trade flow in which the origin between countries with and final destination countries and final destination countries di erent currencies share the same currency have di erent currencies Vietnam Mali Vietnam (dong) (franc) (dong) Intermediate Final Intermediate Intermediate China The China Gambia Senegal Thailand (yuan) (franc) (baht) (yuan) (dalasi) Final Final Exports to China contain Exports to The Gambia are reexported and absorbed in a country with Exports to Thailand are reexported and absorbed in a third value added from China. the same currency. (Mali and Senegal are both part of the WAEMU.) country—in this case, China—that has its own currency. Source: WDR 2020 team. Note: This figure summarizes the different channels through which GVCs can influence the elasticity of exports to devaluations. WAEMU = West African Economic and Monetary Union. Figure 4.8 GVCs dampen the reaction of export volumes to currency movements a. With GVCs, exports are less reactive b. With GVCs, exports are sensitive to devaluations to other countries’ currency value % change in trade volume following a % change in export volume when a direct 1% change in currency value, partner changes its currency value by 1%, for different GVC participation for different GVC participation No GVC No GVC High backward GVC participation High forward High forward third country return domestic 0 0.05 0.10 0.15 0.20 0 0.05 0.10 0.15 0.20 Elasticity Elasticity Source: WDR 2020 team. Note: In panel a, the bars plot the value of the exchange rate elasticity for the 10 percent highest GVC participation indexes. “High backward GVC participation” refers to the foreign value added embedded in exports. “High forward return domestic” refers to the domestic value added embedded in exports and reexported back in the domestic economy. In panel b, the bars plot the value of the elasticity of export volume to the change in the nominal effective exchange rate of a direct partner. The values displayed in this figure use the estimation coefficients from de Soyres et al. (2018) on the elasticity of exports to exchange rate, as well as the interaction of this elasticity with a variable marking the intensity of GVC participation. of the elasticity corresponding to sectors with no transactions are invoiced in U.S. dollars, some coun- participation in GVCs. Moreover, the rise of produc- tries are more sensitive to the U.S. dollar exchange rate tion interconnections is associated with a significant than the bilateral exchange rate.13 Moreover, foreign sensitivity of export volume to foreign devaluations. investment enterprises in China are absorbing cur- The relationship between exchange rate move- rency movements to partially stabilize their prices in ments and export growth is also affected by the choice local currency terms. By contrast, the prices charged of invoicing currency as well as possible changes in by private, locally owned Chinese firms exhibit much markups. For example, because many international more sensitivity to currency movements.14 Macroeconomic implications | 109 Most current measures of trade imbalances are is smaller when measured in trade in value added than based on gross trade data, reflecting the difference when measured in gross trade. The reason? China between the value of total exports and total imports. buys many of the inputs for its exports from other But for GVCs, gross exports and gross imports are poor countries. But U.S. bilateral trade deficits with many measures of the domestic value added exported and of those other countries are larger (or U.S. trade sur- of the foreign value added consumed (box 4.3). Thus pluses with them are smaller) when measured in trade GVCs bias the distribution of trade deficits across in value added. The reason? Many U.S. imports from trading partners, which might mislead trade policy. China incorporate the value of inputs originating in For example, the U.S. bilateral trade deficit with China these countries. Box 4.3  Trade imbalances in using value-added data Most current measures of trade imbalances are based on measured in gross trade because China buys many of the gross trade data and simply reflect the difference between the inputs for its exports from other countries. However, U.S. value of total exports and total imports. For GVCs, however, bilateral trade deficits with some of those other countries gross exports and gross imports are not accurate measures are larger when measured in value-added terms because of the domestic value added exported and of the foreign many U.S. imports from China contain inputs originating in value added consumed. Thus over the past few years several these countries. researchers have highlighted the importance of building a Based on the Trade in Value Added (TiVA) database more accurate picture of bilateral trade flows and the need to from the Organisation for Economic Co-operation and account for the evolution of bilateral value-added balances.a Development (OECD) for 2015, figure B4.3.1 shows the 10 For example, the U.S. bilateral trade deficit with China is country-pairs with the largest differences between bilat- smaller when measured in trade in value added than when eral trade balances using gross exports and value-added Figure B4.3.1  Computing bilateral trade balance in gross exports or in value-added exports matters 60 Gross balance − value-added balance (53%) (50%) 50 (49%) 40 (US$, billions) (49%) (13%) 30 (69%) 20 (112%) (102%) (61%) (108%) 10 0 s na o s am g . s es s ep e e e te ic ur at at at at hi a ex tn R bo St St St St St C ie a, M em d d d d d –V , re a– an te te te ite ite ux Ko na i ni ni n w Un Un Un hi –L U –U ai hi a– o– C g– a– e– om C –T na n ic or ad ur hi na gd hi ex C ap bo an hi C n M m ng Ki C C xe Si d Lu tei Un Source: WDR 2020 team, using data from OECD’s TiVA database. Note: The figure shows the 10 pairs of countries with the largest differences between their gross trade balance and their value-added trade balance. At the top of each bar, the difference in the gross and value-added balance is also expressed as a share of the gross trade balance. The indicator used to compute the gross trade balance is “EXGR: gross exports” in the TiVA database. The value-added trade balance is given by “BALVAFD: Value added embodied in final demand, balance” in the TiVA database. It is computed as the difference between domestic value added in foreign final demand, FFD_DVA, and foreign value added in domestic final demand, DFD_FVA. (Box continues next page) 110 | World Development Report 2020 Box 4.3  Trade imbalances in using value-added data (continued) exports, respectively. The difference, also expressed as United States are imported from the United States itself. a share of gross balance, shows how important it is to By contrast, only 18 percent of the imported parts used by account for the value-added balance. Mexican firms exporting to Germany come from the United One method often used to compute value-added trade States. This finding implies that just looking at sectors to flow is based on manipulation of input–output tables to understand trade flows is not enough: one needs to deepen calculate the share of value added coming from any country the analysis at both the sector and destination levels. embedded in any gross flow. Such a method, however, relies Finally, even though greater participation in GVCs is on strong proportionality assumptions. For example, when expected to generate larger bilateral balances, such an looking at the automotive industry in Mexico, this method outcome is not necessarily associated with a larger overall assumes that the share of inputs from the United States is trade balance. According to the International Monetary the same regardless of the destination of the trade flows—in Fund’s 2019 World Economic Outlook, there is a strong other words, it assumes the same production process for all positive relationship between a country’s participation in destinations for a given industry in a given country. GVCs and the size of its absolute bilateral balances, whereas New findings from de Gortari (2019) reveal that this the relationship is much weaker when it comes to the size of assumption does not hold in the data. Using data from the overall trade balance. Moreover, it has been shown that the automotive industry, de Gortari finds a strong link targeting bilateral trade deficits does not, in general, reduce between the destination of exports and the origin of the a country’s overall current account deficit. Indeed, macro- imported inputs: about 74 percent of all the foreign parts economic policies as well as financial conditions tend to be used by vehicle assemblers in Mexico that export to the the key forces explaining countries’ overall trade balances.b a. Johnson and Noguera (2012a, 2012b, 2017); Koopman, Wang, and Wei (2014). b. IMF (2019). See also Ahn et al. (2019) for more on this subject. Moreover, as noted by Amiti, Freund, and Bodine- accrue only to foreign exporters are shared—and often Smith (2017), production linkages across countries divided differently—on both sides of the border.16 lead to bilateral imbalances across countries, in the GVCs also change another standard paradigm of same way that large companies routinely run deficits trade policy: the diversion of trade from a more effi- with their suppliers: the company purchases inputs cient producer outside of a trade agreement to a less but sells little to these smaller firms. For example, Ger- efficient producer inside of it. Traditionally, signing a many, despite running a large aggregate trade surplus, trade agreement has been associated with an increase runs bilateral trade deficits with the Czech Republic, in trade flows within the agreement zone as well as a Hungary, and the Slovak Republic, the main low-cost decrease in trade flows between the agreement zone suppliers in the European production chain.15 Indeed, and the rest of the world.17 However, because of pro- a bilateral deficit has little meaning for the aggregate duction linkages within GVCs, this standard view has trade balance. The same is true for the U.S.–Mexico been challenged. relationship, where new measures from de Gortari A look at all regional trade agreements over the (2019) highlight the high integration in the automotive past 60 years reveals that agreements are associated industry (box 4.3). with strong, positive trade creation: on signing the agreement, exports between member countries grow significantly, with estimates ranging from less than Mitigating trade diversion and 10 percent to more than 80 percent, depending on increasing trade the agreement and the countries. But there is also Production fragmentation knits together the eco- trade diversion: exports from nonmember to member nomic interests of firms (and workers) up and down countries can decrease, while exports from member to the supply chain. Before the proliferation of GVCs, nonmember countries tend to increase slightly. trade liberalization often benefited local consumers The reduction in imports within the agreement at the expense of local producers. But with these new zone is, among other things, related to rules of origin linkages, the producer gains from trade that used to on final goods. Those rules are defined to prevent Macroeconomic implications | 111 nonmember countries from transshipping products on each other, covering more than half of their bilat- through low-tariff agreement members to avoid high eral trade (approximately 70 percent of U.S. exports tariffs. In effect, the rules act as an input tariff in the to China and almost half of U.S. imports from China). sense that they distort sourcing decisions and divert The United States also imposed tariffs on other coun- trade in intermediate goods to higher-cost agreement tries covering solar panels, washing machines, steel, members. Those mechanisms are quantitatively quite and aluminum, sparking retaliation from the affected relevant: on average, Mexican imports of intermedi- trading partners. At the same time, negotiations con- ate inputs from third countries relative to its partners tinued over the terms and timing of the United King- in the North American Free Trade Agreement (NAFTA) dom’s departure from the European Union (EU). would have been 45 percent higher without rules of In the age of GVCs, this new wave of protectionism origin.18 is likely to have significant costs: GVCs thus fundamentally change how local trade agreements affect global trade flows. With production • The hyperspecialization in tasks and parts across fragmented across countries, rigid linkages have the borders means that trade costs are incurred multi- potential to mitigate the diversion usually associated ple times. with regional trade agreements. For example, if a • Protective measures against any country have member country relies significantly on intermediate knock-on effects on all its trading partners in the inputs from other member countries, signing a trade value chain. agreement actually strongly increases its exports to • GVCs also amplify the costs of trade policy uncer- nonmember countries. The explanation comes from tainty because firms are more reluctant to make fur- the supply side: firms gaining preferred access to their ther investments in new or existing relationships supplier within the free trade zone have a lower mar- with foreign suppliers. ginal cost and can expand their market share in other • Significant tariffs on inputs can force firms to incur countries. In other words, countries forming a trade large costs to reshape their existing supply chains, agreement import less from and export more to the thereby causing potentially long-lasting disruptions rest of the world. Such an effect can lead to efficiency in global investment and production. gains not only within the regional free trade zone but also in other parts of the world.19 According to a recent estimate, the tariffs already Despite the rules of origin, when the share of implemented would lead to a decline in U.S. imports of intermediate goods increases between a nonmember intermediate goods from China over the longer term country and a member country, the trade diversion of by over 40 percent, much more than the declines in exports from the nonmember country to the member consumption and investment goods.20 Furthermore, country is largely mitigated. Indeed, firms in member if the trade conflict worsens and leads to a slump in countries gaining access to larger markets within investor confidence, effects on global growth and a free trade zone can transmit this positive shock to poverty could be significant—up to 30.7 million people their own suppliers outside the agreement zone. could be pushed into poverty measured as an income Moreover, an increase in the share of intermedi- level of less than $5.50 a day, and global income could ate inputs between two countries is associated with fall as much as $1.4 trillion in a worst-case scenario.21 higher trade creation upon signing a trade agreement Low- and middle-income countries other than China and lower trade diversion when one of the two coun- would bear roughly half of the global income loss.22 tries enters a separate trade agreement with other partners. This finding has consequences for trade GVCs amplify the costs of protection for negotiations. If signing a trade agreement creates trade and growth positive spillovers to nonmember countries, the whole A large body of empirical research has shown that design of trade negotiations could be adapted to allow an increase in trade costs significantly reduces trade for more cross-country coordination, including coun- flows. GVCs are affected to an even greater extent. The tries that are not directly part of the trade agreement. hyperspecialization in tasks and parts across borders means that trade barriers are incurred multiple times. Recent evidence reveals that protection and disinte- The return of protectionism gration reduce both backward and forward linkages.23 Protectionism saw a resurgence over the last two As shown in chapter 3, GVC trade has a bigger effect years, fueled by tensions between the United States on growth and employment than standard trade. Pro- and China. In 2018 the two countries imposed tariffs tectionism is therefore costlier for growth and welfare. 112 | World Development Report 2020 Protection not only affects whether and how much the country that imposes protection. For example, countries participate in GVCs; it also affects how they U.S. tariffs on the car industry would penalize U.S. participate. In sequential (or snakelike) GVCs, trade companies reliant on Chinese parts, which are often costs compound along the value chain and have a difficult to source at home. Brexit will likely hit the bigger effect on the downstream stages than on the U.K. producers relatively harder than those in the upstream stages. This effect leads remote countries to EU-27 because the United Kingdom is losing a larger specialize in upstream stages and more central coun- market for suppliers.28 tries to specialize in the more downstream stages.24 An Evidence reveals that in the months after tariffs implication is that the effect of trade costs would be were imposed by the United States in 2018, they were more significant for backward GVC participation than paid in full by U.S. importers, generating aggregate for forward participation and therefore would have welfare losses.29 The resulting price hikes also affected stronger negative impacts on growth. Consistent with supply chains. Imports of products subject to tariffs this view, recent studies estimate that the negative declined sharply, in part because importers turned impacts of Brexit on trade and employment will be to domestic products, but in part because companies considerably larger than commonly expected because shifted their sourcing to more expensive nontargeted of backward linkages.25 sources such as Mexico and Vietnam.30 The effects of protection on consumer prices GVCs fuel the transmission of protection and welfare are likely to be even stronger if tariffs In a world of GVCs, bilateral trade barriers may spill are applied globally, leaving firms unable to shift to over to products and countries not directly targeted other suppliers. Recent evidence reveals that when by those barriers. As noted, protective measures global tariffs on washing machines were applied in against any country have knock-on effects on all its early 2018, their prices climbed about 12 percent for trading partners in the value chain.26 For example, U.S. consumers—foreign manufacturers could no China’s exports to the United States have significant longer shift production to other countries.31 Because value added from developed countries such as Japan, protection and disintegration create incentives for the Republic of Korea, or the United States and from firms to restructure their supply networks, the con- developing countries including Indonesia or Malay- sequences of even a temporary increase in protection sia (figure 4.9). U.S. tariffs on Chinese final goods could persist. therefore affect the intermediate producers in those If U.S.–China trade tensions are not resolved, they economies. Similarly, Chinese tariffs on U.S. goods could disrupt GVCs. In particular, tariffs imposed by affect producers in Canada and Mexico. The sup- the United States on intermediate goods are likely to ply chain diffusion channels  determine how the local lead to a reallocation of sourcing of inputs across value effects of a shock propagate upstream and down- chains between the United States and China, possibly stream to trade partners in the same supply chain. causing adjustment costs in the sectors and locations affected by trade diversion. Protection may cause lasting disruptions in Recent evidence also reveals how the impact of supply chains U.S.–China tariffs changes with time, the magnitude Bilateral measures of protection create incentives for of protection, and the nature of products. Economet- firms to reorganize their supply chain. The effects ric analysis of the value and quantities of imports in of protection on GVC participation may differ when the United States in 2018 and the first quarter of 2019 GVCs are relational in nature. Because of protection- finds that the tariffs have led to significant declines ism, some of the links in the chain may be unable in the affected imports by the United States from to provide parts, components, or services in time or China.32 This decline is relative not only to imports of under prespecified terms. These supply chain disrup- affected products prior to tariff implementation, but tions are particularly costly when firms cannot easily also to imports of unaffected products, whether from resort to alternative suppliers. China or third countries. The analysis also shows that The lock-in effects associated with costly search higher tariff rates lead to larger declines and that and relationship-specific investments also have impli- declines become bigger over time as the policy change cations for the role of market size in attracting GVC is perceived to be ongoing and agents adjust to the activity. With relational GVCs, a large market may new situation (figure 4.10). reduce search frictions.27 Trade barriers imposed on The U.S.–China tariffs have also affected products large markets such as China or the European Union traded via GVCs. For intermediates likely to be asso- may therefore be particularly disruptive for firms in ciated with GVCs, such as parts and components and Macroeconomic implications | 113 Figure 4.9  The multilateral dimension of the U.S.–China trade war a. Chinese exports to United States: Share of value added by Chinese trade partner, 2015 2.5 2.0 Value added (%) 1.5 1.0 0.5 0 n . s y n lia ce sia ia m ly na a l re nd nd a s m zi p te nd an pa tio di ad Ita do iu ys po Re an a a hi la la a ne In rla lg Br ra m str Ja St an la C ng er ai a Fr a, do Be er de a ng Au he Th itz C d R, Ki M re G ite In Fe Si et SA Sw Ko d Un N ite n ng ia Un ss Ko Ru g on H b. U.S. exports to China: Share of value added by U.S. trade partner, 2015 3.0 2.5 Value added (%) 2.0 1.5 1.0 0.5 0 a na RB n o go s y m ce p. n ia a ia ly il nd a sia nd az an pa tio ic ad di bi Ita do al er Re an hi la ba ay ex a, In ra rla Br ra m str Ja ig an C ng er Fr a, To el M al iA er de N Au he itz C Ki zu M re G ud Fe d et Sw Ko ne d an N Sa ite n Ve ia ad Un ss id Ru in Tr Source: WDR 2020 team, using data from Eora database. Note: All countries including self are considered as sources of value added in U.S. and Chinese bilateral exports. That is, we include U.S. and Chinese domestic value added into the respective total value added in exports. The figures, however, plot only the share of the top 20 foreign partner countries in the total value added embedded in the U.S. and Chinese bilateral exports. Exports of goods and services are considered. processed industrial supplies, the decline in import even larger. It is estimated that U.S. imports of inter- values and quantities are smaller than those for other mediate goods from China are likely to decline in the products. This finding is consistent with the existence longer term by over 41 percent, or much more so than of long-term relationships in GVCs. Moreover, the the declines in consumption goods by 9 percent and finding that declines are larger and statistically sig- investment goods by 26 percent.33 nificant for products targeted by higher tariff rates holds for GVC products as well. Although more data Policy uncertainty is costlier under GVCs are needed, this result points to the first signs of GVC GVCs also amplify the costs of sudden increases in disruptions associated with the trade tensions. trade policy uncertainty because firms may wait to Analysis using a computable general equilibrium invest in relationships with foreign suppliers until the model suggests that the longer-term effects may be uncertainty is resolved.34 Firms experiencing more 114 | World Development Report 2020 Figure 4.10  Impact of U.S. tariffs on markup fluctuations. See Ahmed, Appendino, and Ruta (2017) and de Soyres et al. (2018) for the role of down- imports from China (average decline) stream GVC participation. See Mattoo, Mishra, and Posttariff Subramanian (2017) for related evidence on the role of implementation –47.4 third-country effects. Decline in value of U.S. imports from China (%) Four months after 13. Boz, Gopinath, and Plagborg-Møller (2017). implementation –63.2 14. Corsetti et al. (2018). 15. See Amiti, Freund, and Bodine-Smith (2017) for more 25% tariff –74.2 details. 16. Blanchard (2010, 2017). –29.9 17. Mattoo, Mulabdic, and Ruta (2017). GVC intermediates 18. Conconi et al. (2018). Posttariff 19. de Soyres, Maire, and Sublet (2019). Decline in quantity of U.S. implementation –35.7 20. Corong et al. (2019). imports from China (%) Four months after 21. Constantinescu et al. (2019). implementation –49.2 22. Constantinescu et al. (2019). 23. Laget et al. (2018). 25% tariff –55.8 24. Antràs and de Gortari (2017). 25. See Vandenbussche, Connell, and Simons (2017). 26. Bellora and Fontagné (2019). GVC intermediates –21.9 27. Grossman and Helpman (2005). –80 –70 –60 –50 –40 –30 –20 –10 0 28. Sampson 2017; Vandenbussche, Connell, and Simons (2017). Percent 29. Amiti, Redding, and Weinsein (2019); Fajgelbaum et al. Source: WDR 2020 team, using data from U.S. Census (as of April 2019). (2019). Note: Intermediates are defined as categories 42 and 53 in the Broad 30. Amiti, Redding, and Weinstein (2019). Economic Categories (BEC). 31. Flaaen, Hortaçsu, and Tintelnot (2019). 32. Constantinescu et al. (2019). 33. Corong et al. (2019). permanent uncertainty in specific countries are likely 34. Graziano, Handley, and Limão (2018) find that Brexit to shift sourcing to more expensive nontargeted mar- uncertainty induced a net exit of traded products and a reduction in U.K.–EU bilateral trade flows, especially kets and diversify their set of suppliers. Generalized in industries with high sunk costs. Crowley, Exton and and long-lasting trade policy uncertainty is likely to Han (2018) estimate that in 2016 over 5,200 U.K. firms have even stronger negative impacts on GVC trade declined to export new products to the European Union, and investment. The increase in policy uncertainty in and almost 4,000 U.K. firms halted product exports to the 2018 likely contributed to the recent trade slowdown. European Union. Entry (exit) in 2016 would have been The negative association between economic policy 5.1 percent higher (4.3 percent lower) if firms exporting uncertainty and trade growth emerges from a broader from the United Kingdom to the European Union had not sample spanning 18 countries over 30 years.35 faced greater trade policy uncertainty after June 2016. 35. Constantinescu, Mattoo, and Ruta (2017). Notes 1. Burstein, Kurz, and Tesar (2008). References 2. Johnson (2014). Ahmed, Swarnali, Maximiliano Appendino, and Michele 3. Boehm, Flaaen, and Pandalai-Nayar (2019); di Giovanni, Ruta. 2017. “Global Value Chains and the Exchange Rate Levchenko, and Méjean (2018); Liao and Santacreu (2015). Elasticity of Exports.” B.E. Journal of Macroeconomics 17 (1): 4. de Soyres and Gaillard (2019a, 2019b). 1–24. 5. de Soyres and Gaillard (2019a). Ahn, JaeBin, Emine Boz, Maurice Obstfeld, and Petia 6. Freund and Pierola (2015). Topalova. 2019. “Bilateral Trade Balances and Global 7. Barrot and Sauvagnat (2016). Value Chains.” Paper presented at 2019 International 8. Auer, Borio, and Filardo (2017); Auer, Levchenko, and Economics and Finance Society–East Asian Economic Sauré (2019). Review Joint Seminar, “Structural Changes in the Global 9. See Andrews, Gal, and Witheridge (2018) and de Soyres Economy: Global Value Chains and Financial Risks,” and Franco (2019). Seoul National University Asia Center, June 5. 10. IMF (2015). Amiti, Mary, Caroline L. Freund, and Tyler Bodine-Smith. 11. Blaum (2018). 2017. “Why Renegotiating NAFTA Could Disrupt Sup- 12. See Amiti, Itskhoki, and Konings (2014) for a discussion ply Chains.” Liberty Street Economics (blog), April 18, of upstream GVC participation as well as its link with Federal Reserve Bank of New York. https://libertystreet Macroeconomic implications | 115 economics.newyorkfed.org/2017/04/why-renegotiating Tōhoku Earthquake.” Review of Economics and Statistics 101 -nafta-could-disrupt-supply-chains.html. (1): 60–75. Amiti, Mary, Oleg Itskhoki, and Jozef Konings. 2014. “Import- Borin, Alessandro, and Michele Mancini. 2019. “Measuring ers, Exporters, and Exchange Rate Disconnect.” American What Matters in Global Value Chains and Value-Added Economic Review 104 (7): 1942–78. Trade.” Policy Research Working Paper 8804, World Amiti, Mary, Stephen J. Redding, and David Weinstein. 2019. Bank, Washington, DC. “The Impact of the 2018 Trade War on U.S. Prices and Boz, Emine, Gita Gopinath, and Mikkel Plagborg-Møller. Welfare.” NBER Working Paper 25672 (March), National 2017. “Global Trade and the Dollar.” NBER Working Bureau of Economic Research, Cambridge, MA. Paper 23988 (November), National Bureau of Economic Andrews, Dan, Peter Gal, and William Witheridge. 2018. “A Research, Cambridge, MA. Genie in a Bottle? Globalisation, Competition, and Infla- Burstein, Ariel, Christopher Kurz, and Linda Tesar. 2008. tion.” OECD Economics Department Working Paper 1462, “Trade, Production Sharing, and the International Trans- Document ECO/WKP(2018)10 (March 20), Organisation mission of Business Cycles.” Journal of Monetary Economics for Economic Co-operation and Development, Paris. 55 (4): 775–95. Antràs, Pol, and Alonso de Gortari. 2017. “On the Geography Conconi, Paola, Manuel García-Santana, Laura Puccio, and of Global Value Chains.” NBER Working Paper 23456 Roberto Venturini. 2018. “From Final Goods to Inputs: (May), National Bureau of Economic Research, Cam- The Protectionist Effect of Rules of Origin.” American bridge, MA. Economic Review 108 (8): 2335–65. Arto, Iñaki, Valeria Andreoni, and José Manuel Rueda Can- Constantinescu, Ileana Cristina, Aaditya Mattoo, and tuche. 2015. “Global Impacts of the Automotive Supply Michele Ruta. 2017. “Trade Developments in 2016: Pol- Chain Disruption Following the Japanese Earthquake of icy Uncertainty Weighs on World Trade.” Global Trade 2011.” Economic Systems Research 27 (3): 306–23. Watch (February 21), World Bank, Washington, DC. Auer, Raphael, Claudio Borio, and Andrew Filardo. 2017. Constantinescu, Ileana Cristina, Aaditya Mattoo, Michele “The Globalisation of Inflation: The Growing Impor- Ruta, Maryla Maliszewska, and Israel Osorio- tance of Global Value Chains.” BIS Working Paper 602 Rodarte. 2019. “Global Trade Watch 2018: Trade Amid (January 9), Bank for International Settlements, Basel, Tensions.” Global Trade Watch (May 29), World Bank, Switzerland. Washington, DC. Auer, Raphael, Andrei A. Levchenko, and Philip Sauré. 2019. Corong, Erwin, Maryla Maliszewska, Maria Pereira, and “International Inflation Spillovers through Input Link- Dominique van der Mensbrugghe. 2019. “Global and ages.” Review of Economics and Statistics 101 (3): 507–21. Regional Impacts of Trade Tensions on Global Value Automotive News. 2011. “Opel, Renault Production Hit by Chains.” World Bank, Washington, DC. Shortage of Japanese Parts.” March 18. https://www Corsetti, Giancarlo, Meredith Allison Crowley, Lu Han, and .autonews.com/article/20110318/COPY01/303189860/opel Huasheng Song. 2018. “Markets and Markups: A New -renault-production-hit-by-shortage-of-japanese-parts. Empirical Framework and Evidence on Exporters from Barrot, Jean-Noël, and Julien Sauvagnat. 2016. “Input Spec- China.” CFM Discussion Paper 1803, Centre for Macro- ificity and the Propagation of Idiosyncratic Shocks in economics, London School of Economics and Political Production Networks.” Quarterly Journal of Economics 131 (3): 1543–92. Science, London. Bellora, Cecilia, and Lionel Fontagné. 2019. “Shooting Oneself Crowley, Meredith Allison, Oliver Exton, and Lu Han. 2018. in the Foot? Trade War and Global Value Chains.” Paper “Renegotiation of Trade Agreements and Firm Export- presented at Global Trade Analysis Project’s 22nd Annual ing Decisions: Evidence from the Impact of Brexit on Conference on Global Economic Analysis, “Challenges UK Exports.” Cambridge-INET Working Paper 2018/10 to Global, Social, and Economic Growth,” University of (December), Institute for New Economic Thinking, Uni- Warsaw, Poland, June 19–21. versity of Cambridge, Cambridge, U.K. Blanchard, Emily J. 2010. “Reevaluating the Role of Trade de Gortari, Alonso. 2019. “Disentangling Global Value Agreements: Does Investment Globalization Make the Chains.” NBER Working Paper 25868 (May), National WTO Obsolete?” Journal of International Economics 82 (1): Bureau of Economic Research, Cambridge, MA. 63–72. de Soyres, François, and Sebastian Franco. 2019. “Inflation ————. 2017. “Renegotiating NAFTA: The Role of Global Synchronization through GVCs.” Unpublished working Supply Chains.” In Economics and Policy in the Age of Trump, paper, World Bank, Washington, DC. edited by Chad P. Brown, 175–84. London: Centre for de Soyres, François, Erik Frohm, Vanessa Gunnella, and Economic Policy Research. Elena Pavlova. 2018. “Bought, Sold, and Bought Again: Blaum, Joaquin. 2018. “Global Firms in Large Devaluations.” The Impact of Complex Value Chains on Export Elastic- 2018 Meeting Paper 593, Society for Economic Dynamics, ities.” Policy Research Working Paper 8535, World Bank, Department of Economics, Stonybrook University, Stony- Washington, DC. brook, NY. de Soyres, François, and Alexandre Gaillard. 2019a. “Trade, Boehm, Christoph Emanuel, Aaron B. Flaaen, and Nitya Global Value Chains, and GDP Comovement: An Empir- Pandalai-Nayar. 2019. “Input Linkages and the Trans- ical Investigation.” Unpublished working paper, World mission of Shocks: Firm-Level Evidence from the 2011 Bank, Washington, DC. 116 | World Development Report 2020 — —— —. 2019b. “Value Added and Productivity Linkages across —— — —. 2017. “A Portrait of Trade in Value-Added over Four Countries.” Unpublished working paper, World Bank, Decades.” Review of Economics and Statistics 99 (5): 896–911. Washington, DC. Knight, David Stephen, Milan Nedeljkovic, and Alberto de Soyres, François, Julien Maire, and Guillaume Sublet. Portugal-Perez. 2019. “Turkey: An Empirical Assessment 2019. “An Empirical Investigation of Trade Diversion and of the Determinants of the Current Account Balance.” Global Value Chains.” Unpublished working paper, World Policy Research Working Paper 8982, World Bank, Wash- Bank, Washington, DC. ington, DC. di Giovanni, Julian, Andrei A. Levchenko, and Isabelle Jeanne Koopman, Robert B., Zhi Wang, and Shang-Jin Wei. 2014. Méjean. 2018. “The Micro Origins of International Busi- “Tracing Value-Added and Double Counting in Gross ness-Cycle Comovement.” American Economic Review 108 Exports.” American Economic Review 104 (2): 459–94. (1): 82–108. Laget, Edith, Alberto Osnago, Nadia Rocha, and Michele Fajgelbaum, Pablo D., Pinelopi K. Goldberg, Patrick J. Ken- Ruta. 2018. “Deep Agreements and Global Value Chains.” nedy, and Amit Kumar Khandelwal. 2019. “The Return Policy Research Working Paper 8491, World Bank, Wash- to Protectionism.” NBER Working Paper 25638, National ington, DC. Bureau of Economic Research, Cambridge, MA. Liao, Wei, and Ana Maria Santacreu. 2015. “The Trade Flaaen, Aaron B., Ali Hortaçsu, and Felix Tintelnot. 2019. “The Comovement Puzzle and the Margins of International Production, Relocation, and Price Effects of U.S. Trade Trade.” Journal of International Economics 96 (2): 266–88. Policy: The Case of Washing Machines.” BFI Working Lohr, Steve. 2011. “Stress Test for the Global Supply Chain.” New Paper 2019–61 (April), Becker Friedman Institute for York Times, March 19. https://www.nytimes.com/2011/03 Research in Economics, University of Chicago. /20/business/20supply.html. Freund, Caroline L., and Martha Denisse Pierola. 2015. Mattoo, Aaditya, Prachi Mishra, and Arvind Subramanian. “Export Superstars.” Review of Economics and Statistics 97 2017. “Beggar-Thy-Neighbor Effects of Exchange Rates: (5): 1023–32. A Study of the Renminbi.” American Economic Journal: Graziano, Alejandro, Kyle Handley, and Nuno Limão. 2018. Economic Policy 9 (4): 344–66. “Brexit Uncertainty and Trade Disintegration.” NBER Mattoo, Aaditya, Alen Mulabdic, and Michele Ruta. 2017. Working Paper 25334 (December), National Bureau of “Trade Creation and Trade Diversion in Deep Agree- Economic Research, Cambridge, MA. ments.” Policy Research Working Paper 8206, World Grossman, Gene M., and Elhanan Helpman. 2005. “Out- Bank, Washington, DC. sourcing in a Global Economy.” Review of Economic Studies Sampson, Thomas. 2017. “Brexit: The Economics of Interna- 72 (1): 135–59. tional Disintegration.” Journal of Economic Perspectives 31 Ha, Jongrim, Ayhan Kose, and Franziska Lieselotte Ohnsorge. (4): 163–84. 2019. “Global Inflation Synchronization.” Policy Research Swiss Re. 2019. “Swiss Re Estimates Its Fourth Quarter 2018 Working Paper 8768, World Bank, Washington, DC. Claims Burden from Large Natural Catastrophes at USD IMF (International Monetary Fund). 2015. World Economic 1.0 Billion; Large Man-Made Losses Caused Additional Outlook, October 2015: Adjusting to Lower Commodity Prices. USD 0.3 Billion of Claims.” News release, January 15, World Economic and Financial Surveys Series. Washing- Swiss Re Insurance Co., Zurich. https://www.swissre.com ton, DC: IMF. /media/news-releases/nr-20190115-q4-2018-claims.html. — —— —. 2019. World Economic Outlook, April 2019: Growth Vandenbussche, Hylke, William Connell, and Wouter Slowdown, Precarious Recovery. Washington, DC: IMF. Simons. 2017. “Global Value Chains, Trade Shocks, and Johnson, Robert Christopher. 2014. “Trade in Intermediate Jobs: An Application to Brexit.” Discussion Paper DPS17/13 Inputs and Business Cycle Comovement.” American (September), Center for Economic Studies, Department of Economic Journal: Macroeconomics 6 (4): 39–83. Economics, University of Leuven, Belgium. Johnson, Robert Christopher, and Guillermo Noguera. 2012a. World Bank. 2017. “New Country Classifications by Income “Accounting for Intermediates: Production Sharing and Level: 2017–2018.” Data Blog, July 1, World Bank, Wash- Trade in Value Added.” Journal of International Economics 86 ington, DC. https://blogs.worldbank.org/opendata/new (2): 224–36. -country-classifications-income-level-2017-2018. — —— —. 2012b. “Fragmentation and Trade in Value Added over Four Decades.” NBER Working Paper 18186 (June), National Bureau of Economic Research, Cambridge, MA. Macroeconomic implications | 117 5 Impact on the environment Key findings  lobal value chains (GVCs) are a mixed blessing for the environment. Scale effects— • G which refer to the rapid growth of GVC economic activity—are bad for the environment, whereas composition effects—which refer to how tasks are distributed across the globe— have ambiguous effects. Technique effects—which refer to the environmental cost per unit of production—are positive for the environment. • G  VCs are associated with more shipping and more waste in the aggregate than standard trade. Both have environmental costs. • O  ne important concern has been that industries might migrate to jurisdictions where environmental regulations are lax, but that concern is not borne out by the data. Rather, by locating production where it is most efficient, GVCs can lower the net resource intensity of global agricultural production.  he relational aspect of GVCs can attenuate environmental concerns. Knowledge • T flows between firms can enable the spread of more environmentally friendly production techniques throughout a GVC. The large scale of lead firms in GVCs can accelerate environmental innovation and push for higher standards.  VCs also facilitate the production of new environmentally friendly goods. Products • G such as solar panels, electric cars, and wind turbines are produced at lower costs in GVCs and help reduce the environmental costs of consumption. 118 | World Development Report 2020 T he $4,995 Pedego Conveyor electric bike is pro- • Technique effect. GVCs can also promote improve- duced in Vietnam with parts from all over the ments in production techniques. The knowledge world (figure 5.1).1 Gears, pedals, brakes, and flows among networks of firms can enable the other components are shipped from China, Europe, development or quicker application of more envi- Indonesia, Japan, and other economies to Vietnam ronmentally friendly techniques. With their large for assembly, and then the bike itself is shipped to the scale, the lead firms in GVCs are able to sustain United States for final sale. Roughly 60 percent of the high rates of innovation. Market concentration bike’s value is from outside Vietnam. can lower the difficulty in managing common Because parts are crisscrossing the globe, produc- pool resources such as fisheries and forests. The ing the Conveyor through a global value chain (GVC) relational aspect of GVCs is also important in has greater environmental costs than standard trade. this context because lead firms are increasingly Even more worrisome, some of the most environmen- transferring environmentally friendly technol- tally damaging parts, such as the batteries and tires, ogies to their suppliers and pushing for higher may end up being produced in countries with the standards. weakest regulations, leading to more environmental degradation. Policies can influence the net impact of GVCs. But GVCs are also engines of innovation that help Subsidies on fuel, for example, can exacerbate the drive the creation and diffusion of less-damaging overproduction of fuel-intensive exports. But sub- products and processes. GVCs make new environmen- sidies for environmental goods can promote their tally friendly products like this electric bike possible. production and further innovation. GVCs in new Big international brands can use GVCs to encourage environmental goods, from solar panels to LED light the global adoption of clean and efficient technologies bulbs, many subsidized over the years, expanded and processes aimed at enhancing both profitability rapidly, thereby facilitating the diffusion of low- and sustainability. carbon technology. Variations in regulation can also Environmental consequences arise from features lead to net global increases in environmental damage of GVCs, including the hyperspecialization of tasks, if polluting tasks migrate to countries with lax reg- geographic dispersion of production, economies of ulations—part of the composition effect called the scale, and the market power of lead firms. The total pollution haven hypothesis (PHH). However, a large environmental impact of GVCs is considered here body of literature does not find evidence in support of along three dimensions: this hypothesis. Comparative advantage for many of the most polluting industries rests primarily on fac- • Scale effect. If GVCs spur the growth of economic tors such as capital and resource abundance, and so activity, and if composition, consumer preferences, these industries tend not to migrate to the least reg- and production techniques remain the same in the ulated countries. However, low- and middle-income sense that pollution per unit of output is constant, countries are often reluctant to raise environmental then growth leads to environmental deterioration. standards because in a world of liberalized trade and GVCs also have some consequences that extend investment they fear losing the interest of foreign beyond those of standard trade. In particular, GVCs investors.2 Policies for preserving the environment in are associated with more waste and more shipping a world of GVCs are discussed in chapter 8. in the aggregate, both of which have environmental costs. • Composition effect. GVCs, by promoting trading in Scale effects of trade and growth tasks, prompt certain types of economic activity to As GVCs grow and economic activity expands, emis- relocate internationally, thereby transforming pat- sions increase—a simple scale effect. The effect would terns of production and trade. Shifts in production be greater if production increased more in higher- toward countries with abundant natural resources polluting industries—a composition effect. Absent allow the preservation of scarce resources, helping technological innovation, the scale effect of GVC trade to sustain global resources such as land and water. tends to be negative for the environment because, However, the redistribution of “dirty” and “clean” although production-related pollution and carbon tasks among countries may create environmental dioxide (CO2) emissions fall with a country’s income, benefits for some countries and environmental consumption-related environmental emissions and costs for others. degradation tend to increase. Impact on the environment | 119 Figure 5.1 The complexity of producing the Pedego Conveyor electric commuter bike in Vietnam with parts from all over the world 1.7% 27% 1.2% 1% 19% 2% Tires Motor Seat Sensors Batteries Display Indonesia Germany Italy Czech Republic Taiwan, China Germany (Schwalbe) (Brose) (Selle Royal) (Gearsensor) (Hitech Energy) (Brose) 4.9% 4.5% 0.2% 0.5% 21.2% 1.8% Gears Belt drive Kickstand Pedals and Spokes and Brakes Japan United Kingdom Italy crank arms wheels China (Shimano) (Gates) (Ursus) Taiwan, China China (Tektro) (VP Components) (Various) Source: Frothingham 2018. Note: Diagram shows the percent of total value added from each component. Countries that recently transitioned into limited significant (figure 5.2). Indeed, in some countries manufacturing-linked GVCs tend to experience faster manufacturing has expanded without rising emis- growth of production-related CO2 emissions relative sions. Meanwhile, countries that recently transitioned to the previous period, although some countries have into advanced manufacturing and services GVCs, as also seen their emissions growth decline—which well as into innovation hubs, typically experience a is why the effect of transitioning is not statistically decline in average production-related CO2 emissions. 120 | World Development Report 2020 In these countries, which tend to be at a higher stage Figure 5.2  Production-related CO2 emissions drop of development, consumers may demand more reg- in countries that recently transitioned into advanced ulations, and the technology of production becomes GVCs and innovation hubs more environmentally friendly. 30 Cumulated growth in production-related CO2 emissions (%) These contrasting results are consistent with the literature. On the one hand, the environmental 20 19 Kuznets curve (EKC),3 an inverted-U, reveals that economic growth increases the presence of local pol- 10 lution and production-related CO2 emissions when country incomes are low. Beyond a certain turning 0 point, it is instead associated with improvements in environmental indicators, and rising country incomes appear to lead to an increase in demand for environ- –10 mental quality.4 On the other hand, there is a clearly positive correlation between higher GVC activity –20 –21 and a number of indicators of global environmental damage. Because of the urgency of the global environ- –30 mental challenge, relying on countries growing first and cleaning up later may be misguided, and such an –40 –39 approach may fail to deliver the reductions in emis- sions needed to avoid a climactic catastrophe. –50 One way in which GVCs can encourage manu- Event 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 facturing while also protecting the environment is year Years after event by inducing GVC firms to opt for industrial parks that have higher standards and encourage environ- Limited manufacturing (ns) Advanced manufacturing and services mentally friendly production techniques. More than Advanced manufacturing and services (ns) 300 industrial parks now consider themselves to be Innovative activities eco-industrial parks (EIPs)—a number that is expected Innovative activities (ns) to rise. In many countries, governments have become Sources: WDR 2020 team, using data from the World Bank’s WDI database and the GVC taxonomy for more conscious of green approaches to manufactur- 1990–2015 based on Eora26 database. See appendix A for a description of the databases used in this ing, and lead firms, concerned about reputation, are Report. eager to improve the sustainability of production (see Note: The event study quantifies cumulated CO2 emissions in the 20 years following a switch from a lower to a higher stage of GVC engagement. Carbon emissions are normally expressed in kilograms per box 8.5 in chapter 8). 2011 dollars, adjusted for purchasing power parity (PPP). Dotted lines indicate statistically nonsignifi- cant (ns) coefficients. See box 3.3 for a discussion of the methodology. Transportation One concern about GVCs is their more intensive use and more than 70 percent by value is transported by of transportation than other types of trade. Parts sea.7 The capacity of the merchant shipping industry and components are shipped to a country only to has surged since 1990, and so have emissions from be shipped out after assembly. This back-and-forth shipping. In 2016, CO2 emissions from international transport of goods across long distances generates shipping were about 2.0 percent of global CO2 emis- CO2 emissions through the combustion of fossil-based sions (figure 5.3, panel a). This is not a small number: fuels, thereby directly contributing to climate change. if a country had the same percentage of emissions, it CO2 emissions from international freight trans- would be the seventh-largest emitter, ranking between portation account for about 7 percent of total CO2 Germany and the Republic of Korea. emissions globally.5 By 2050, CO2 emissions related Under business-as-usual conditions, these emis- to international freight are estimated to quadruple, sions are projected to increase by 50–250 percent by which threatens the temperature goals of the Paris 20508—that is, if the maritime sector continues to Agreement.6 In the past, industries more heavily into expand at an annual rate of more than 3 percent, as it offshoring produced the greatest increases in carbon has over the past 40 years.9 Although emissions from emissions related to international trade. other sectors have begun to decline or are expected to GVCs are most closely linked to maritime trans- peak soon, none of the business-as-usual scenarios for port. More than 80 percent of world trade by volume shipping foresee a decline in emissions before 2050. Impact on the environment | 121 Figure 5.3  International shipping emissions are increasing a. International shipping emissions and b. Pathways for international shipping share of global CO2 emissions, 1970–2016 CO2 emissions, 2008–78 800 2.5 1,600 Share of global CO2 emissions (%) 700 1,400 2.0 Level of emissions (MMT) 600 1,200 CO2 emissions (MMT) 500 1,000 1.5 400 800 300 1.0 600 200 400 0.5 If possible 100 200 0 0 0 08 28 70 10 18 68 78 78 82 86 90 98 02 06 14 38 48 58 74 94 20 20 19 20 20 20 20 19 19 19 19 19 20 20 20 20 20 20 19 19 International shipping emissions Business as usual Shipping’s share of global CO2 emissions (right axis) 50% reduction by 2050 (85% reduction in carbon intensity) 100% reduction by 2050 Sources: Panel a: Muntean et al. 2018; panel b: UCL Energy Institute, London (https://www.ucl.ac.uk/bartlett/energy/research/themes/transport/shipping). Note: MMT = million metric tons. As a result, the United Nations Conference on Trade transportation, fuels in international transportation and Development predicts that the maritime sector’s are not subject to excise taxes. Charging for maritime share of global CO2 emissions could account for 10–17 fuels based on their true social cost could support fully percent by 2050.10 Technological advances and ambi- exploiting the potential of existing energy efficiency tious climate policy will have to counter this trend. As and developing alternative fuels. The challenge is that transport technology has improved, growth in emis- ships are highly mobile: they travel mostly in interna- sions since 1990 is already less (1.85 times) than the tional waters and can easily be registered anywhere. near tripling of capacity over the same period (figure Thus pricing emissions appropriately would work best 5.3, panel a). with a global solution such as taxing maritime fuels at Aware of these rapidly rising emissions challeng- a single international carbon rate.13 And yet because a ing the world’s remaining carbon budget, the Interna- global solution is not in place yet, and notwithstand- tional Maritime Organization (IMO) has committed ing potential market distortions, some governments to at least halving CO2 emissions by 2050, aiming to are exploring unilateral measures. The European Par- eliminate CO2 emissions from shipping as quickly liament, for example, is considering regional carbon as possible (figure 5.3, panel b).11 Although technical pricing on maritime fuels in the absence of a global and operational efficiency measures could reduce agreement.14 Other options include taxing ships based emissions by 30–55 percent by 2050, according to the on the type of vessel or taxing based on bills of lading Intergovernmental Panel on Climate Change,12 tech- that show the distance the imported cargo traveled. nological innovations will be required to achieve full These and other policy considerations are discussed decarbonization of the sector as envisaged by the IMO. in chapter 8. This energy transition in shipping toward Maritime shipping also poses major pollution zero-emissions fuels could be facilitated by effective challenges in other areas. However, some interna- policy support in the form of carbon taxes, emissions tional solutions have begun to emerge and lead to trading, low-carbon fuel standards, and a gradual ban improvements: of fossil fuels, among other measures. From an envi- ronmental perspective, maritime activities are cur- • Air pollution. Shipping accounts for roughly 15 per- rently undercharged. For example, unlike in domestic cent of global emissions of sulfur dioxide (SO2) and 122 | World Development Report 2020 nitrogen oxides (NOx). Ship engines burn the dirtiest the environment. Rail is the lowest emitter of CO2 fuel possible (heavy fuel oil, a residual product of the (3 percent of the total), whereas road freight is over refinery processes of gas, diesel, kerosene, among 50 percent of the total. other fuels). A recent study by the International Council on Clean Transportation attributed 60,000 GVCs and waste premature deaths a year to shipping emissions.15 GVCs can influence the amount and type of waste The IMO therefore recently decided to reduce the generated during the production and transport of mandatory sulfur limit from 3.5 percent to 0.5 per- goods from source to consumer. They have contrib- cent as of 2020 for maritime fuels. uted to a large share of the waste in the electronics • Maritime litter. Although most plastic waste that ends and other GVC-intensive sectors, but they are also up in the ocean comes from land-based sources and well positioned to be part of the solution. is transported through rivers into the sea, about 20 E-waste is the fastest-growing waste stream in the percent originates directly from ships and other sea- world, accounting for more than 70 percent of the based sources, including aquaculture, fishing, and toxic waste in U.S. landfills (figure 5.4).17 GVCs have dumping of waste and other matter from deep-sea enabled rapid declines in the cost of electrical and elec- platforms. Next to environmental misconduct, a big tronic devices,18 benefiting large numbers of people problem is that port reception facilities—waste dis- who otherwise could not afford even low-cost items. posal facilities provided for ships by authorities—are GVCs also drive the rate of technological innovation often nonexistent, or they are inadequately equipped, that leads to high replacement rates worldwide. complicated to use, or too expensive. Shipbreaking But GVCs have the potential to close the loop and (that is, scrapping vessels) is also a problem.16 turn e-waste into valuable resources. The United • Invasive species. To float in a balanced way, ships Nations University conservatively estimated the value often have to take on board ballast water. This water of recoverable materials in last year’s e-waste to be is then discharged at another location when the $55 billion, or more than the 2016 gross domestic prod- weight and volume requirements change. Invasive uct of most countries.19 Some countries such as Japan species are transported around the globe in this have e-waste management laws that make manufac- water and released at locations where they may not turers and retailers responsible for taking back used have any natural predators and can pose a threat to home appliances, recycling them, and publishing the sensitive ecosystems. costs of recycling. • Water pollution. Other pollution-related problems are E-waste flows should be viewed as sources of linked to oil spills, sewage disposal (from ship oper- inputs for next-generation products.20 The World ations), and bilge water (a cocktail of oil and chem- Economic Forum’s call for a circular electronics value icals leaking from the engines and machinery and chain represents a model of sustainability that is dif- water that accumulates in the lowest part of vessels ficult to envisage without GVCs.21 Inputs from retired and must be pumped out from time to time). electronics should be removed and recycled by the very companies that produce them. Road and rail transport are two additional sources The global trade in plastic waste grew in lockstep of the impacts of GVCs on the environment because with the expansion of GVCs through the 1990s and of their predominance in domestic value chains. The 2000s. In 1990 worldwide imports of plastic waste efficiency and performance of the trucking industry were worth less than $1 billion, and by 2010 they had can have a significant impact on the carbon footprint peaked at around $10 billion. In the last decade, they of GVCs. The adoption of more fuel-efficient vehicles have begun to level off and even decline.22 Meanwhile, reduces associated emissions, and the reduction of plastic and microplastic waste have proven to be a empty backhauls improves overall efficiency, results major challenge for solid waste management and have in less waste, and contributes to lower prices. For become a global crisis for the environment, especially example, when the Lao People’s Democratic Repub- the oceans. In 2018 the Center for Biological Diversity lic abolished restrictions on backhauling by foreign estimated that swirling convergences of plastic make trucking companies, road transport prices declined by up about 40 percent of the world’s ocean surfaces and 20 percent. Substitution between road and rail modali- that at current rates they could outweigh all the fish in ties and the associated development of more seamless the sea by 2050.23 containerized logistics are another important area Gross trade data from UN Comtrade are not well that will determine the overall impact of GVCs on suited to portraying what is happening to plastic Impact on the environment | 123 Figure 5.4 The world produced 50 million metric tons of e-waste in 2018 ycling for rec l e c ted t co l 4% No % Thrown into 80 ed , t raded , or recycled und e r inf e household waste mp rio r (du co wn n dit no io u nk ns ) te fa — ed 20% 17% t en Large Temperature um equipment exchange oc td equipment Do an no cu d rec me sal 38% 15% nte ycled po Small Screens Dis d, c equipment ollec 76% ted, In 2018 approximately 50 million metric tons of 9% Small IT e-waste was produced. 1% Lamps In addition, an estimated 100 million old devices or more are stored in homes. Source: Adapted from Ryder and Zhao (2019). Note: IT = information technology. waste worldwide. Input–output data are in principle Changes in the composition of better able to track plastic waste, but in both statisti- production cal sources the information is too aggregated to track international flows. The two most common polymers, Falling trade costs, tighter environmental PET and PP, lack specific codes in UN Comtrade regulations, and pollution havens because trade codes for these waste materials are not Because trade costs are falling while environmental yet harmonized across countries, and the available regulations are tightening in many countries, pol- multiregion input–output data do not include a cate- luting manufacturers may respond to new environ- gory for waste. Thus calculating plastic waste urgently mental regulations by relocating to countries with requires better statistical measurement. less strict standards. Moreover, because GVCs foster Today’s recycling technologies cannot handle the hyperspecialization, with tasks moved to the most rapidly growing quantities of global waste. For many productive location, lead firms from countries with years, China was accepting a large share of the world’s tight environmental regulations may locate “dirty” plastic waste, but eventually the environmental costs production in countries where environmental norms of recycling “dirty” plastics became formidable, and are lax—that is, in so-called pollution havens. Relocat- China raised the import standards in 2017, all but cut- ing conventional local pollutants thus improves the ting off acceptance of plastic waste (box 5.1). With most air and water quality in places with strict regulations plastic waste now ending up in landfills or incinera- at the expense of environmental quality in pollution tors, reducing waste and developing better technology havens. for packaging goods and recycling are environmental In theory, concerns about pollution havens are priorities in many countries. These countries are pro- well founded.24 Pollution is a production input, just moting a shift away from plastics in bags and water like labor and capital. One could think of pollution bottles, encouraging reuse, and using more econom- as the disposal services of the environment, where ical and environmentally friendly packaging of parts, the unregulated price is zero. Countries export the components, and goods traveling the world. goods in which they have a comparative production 124 | World Development Report 2020 Box 5.1  The ban on plastics by China disrupted the waste GVC One way GVCs extended a product’s life was through recy- scrap, and when their contents could not be recycled it cling of paper and plastic waste. In recent decades, goods was burned or dumped. As a result, Indonesia, Malaysia, shipped from China to the United States were consumed, the Philippines, Thailand, and Vietnam have announced and paper and plastic containers, along with domestic plas- they would ban and send contaminated waste back to the tic and paper waste, were sent back to China for recycling. countries of origin, with the threat of abandoning the waste At the end of 2017, China stopped accepting large in countries’ territorial waters if the waste is not accepted. amounts of imported waste for recycling because a large Reducing the paper and plastics in packaging and using share was “dirty” and causing environmental damage. The cleaner technology for recycling has become a priority for prices of plastic scrap and low-grade paper then collapsed, environmentally concerned countries. In May 2019, 187 disrupting the global recycling industry. In the first half of countries—not including the United States—agreed to 2017, China and Hong Kong SAR, China, absorbed 60 per- amend the Basel Convention on the Control of Transbound- cent of the plastic waste exported by G-7 countries. A year ary Movements of Hazardous Wastes and Their Disposalb to later, they imported less than 10 percent.a better regulate the global trade of plastic waste and make it In their place, Malaysia, Thailand, and Vietnam, among more transparent. Among the commitments, private com- other East Asia and Pacific countries, experienced signifi- panies will have to secure the consent of receiving countries cant increases in contaminated and plastic waste imports. before they can trade contaminated and most mixes of However, many containers were misrepresented as plastic plastic waste. a. Hook and Reed 2018. b. The Basel Convention on the Control of Transboundary Movements of Hazardous Wastes and Their Disposal was adopted on March 22, 1989, by the Conference of Plenipotentiaries in Basel, Switzerland, in response to a public outcry following the discovery in the 1980s in Africa and other parts of the developing world of deposits of toxic wastes imported from abroad. advantage—that is, their costs of producing those 10 percent increase in pollution abatement costs in goods are lower relative to their costs of producing the United States leads to a 0.6 percent increase in net other goods. Countries with lax pollution regulations imports from Mexico and a 5 percent increase in net have a comparative advantage in goods whose produc- imports from Canada.27 tion is pollution-intensive, and they will export those The association of falling trade costs and tighter goods—becoming pollution havens. environmental regulations could drive polluters to Evidence of the pollution haven effect (PHE) has, flee to developing countries. But this has not hap- however, been very limited so far. Polluting indus- pened. Take, for example, what happened to the types tries—paper, metals, cement, and refineries—tend to be of goods produced in the United States compared with costly to relocate, and production is tied to local factor U.S. imports as trade costs declined and U.S. environ- or product markets. Paper plants locate near the trees, mental standards became stricter (see chapter 1).28 and cement plants near their customers. It is therefore Emissions from U.S. domestic manufacturing fell by not obvious that countries with lax regulations will 60 percent from 1990 to 2008, stemming from changes have a comparative—or even an absolute—advantage in environmental policy.29 Meanwhile, the structure of in polluting goods. Environmental regulations are imports shifted toward cleaner goods. Contrary to the a small part of costs. Consistent with this, empirical conventional wisdom about industrialized countries evidence shows that strict environmental regulation “offshoring” production of polluting goods, imports of polluting industries has not led to large relocations to the United States have been shifting away from to countries with less-strict standards.25 In some cases, pollution-intensive goods even faster than U.S. domes- polluting industries and strict regulations are in fact tic production (figure 5.5). As trade costs fall, the U.S. positively correlated.26 Of all the recent papers finding increasingly imports goods in which it has a compar- a PHE, few attempt to untangle the causal negative ative disadvantage, which happen to be those that are effect of pollution regulations on polluting industries. relatively less pollution-intensive. Trends in Europe Those that do untangle that effect find a statistically are similar, with imports becoming progressively significant but quantitatively modest effect for the less pollution-intensive, especially from low-income most polluting industries. One study showed that a countries. Impact on the environment | 125 Figure 5.5  U.S. output has increasingly shifted away water embodied in cereals and oils. Arid countries from polluting goods, but imports have done so even that do not have a comparative advantage in water- faster intensive culture no longer need to grow these prod- ucts domestically. They can import them for consump- 10 tion or further processing with considerable savings in SO2 emissions (pounds per million US$) water usage (box 5.2). Trade in “virtual water,” the water 8 embodied in agricultural production, is estimated to have saved 4 percent of the global water footprint. National policies can make the environment worse 6 by subsidizing activities that lead to environmental problems. Subsidizing fisheries can lead to overfish- 4 ing, which has been recognized as a major global issue since at least the 1990s.30 When agriculture is subsi- dized, deforestation, soil erosion, and chemical runoff 2 into bodies of water are greater than they would be otherwise, and natural biodiversity will decline.31 0 Even in the absence of subsidies, GVCs and trade 1972 1976 1980 1984 1988 1992 1996 2000 2004 create some concerns about hyperspecialization and U.S. imports U.S. domestic manufacturing degradation of land for agricultural use, a major driver Source: WDR 2020 team, using data from Levinson (2010). of forest loss. Four products—soy, cattle, palm oil, and Note: The figure shows the pounds of sulfur dioxide (SO2) air pollution per million U.S. dollars of value wood products—alone are responsible for 40 percent produced by the U.S. manufacturing sector between 1972 and 2001 and those of imported value. Those totals are calculated using the World Bank’s Industrial Pollution Projection System, which is simply a list of global deforestation, at an average rate of 3.8 million of emission intensities for each of more than 400 manufacturing industries in 1987 (Hettige et al. 1995). hectares a year.32 But many more commodities—such Averaging across industries, weighted by their values shipped in each year, gives the average pollution intensity of the entire U.S. manufacturing sector each year. The blue line in the figure plots that average, as cocoa, coffee, spices, vanilla, bananas, cut flowers, holding pollution intensities fixed as of 1987. The green line in the figure reports the same calculation for orange juice, and natural rubber—are experiencing imports. These averages drop over time because of changes in the composition of the manufacturing sector. U.S. output has increasingly shifted away from goods that generate the most pollution per dollar a growing global demand that threatens the envi- of output toward cleaner goods. Graphs for nitrogen oxides, volatile organic compounds, and carbon ronment in the hotspots where these goods grow. monoxide look similar, with the pollution intensity of U.S. domestic manufacturing falling less quickly than that for imports. Some fear that this demand may translate into the depredation of resources from developing countries— Although the PHE has been overplayed to date, it especially because incomplete markets mean that the may become more relevant as some countries adopt biodiversity contained in forests is not valued suffi- more ambitious climate policies to reduce emissions ciently. Through more efficient production and lower rapidly. prices, trade and GVCs increase the global quantity demanded of certain agricultural resources and com- Environmental effects of agriculture and modities. The result can be deforestation, biodiversity commodity GVCs loss, and other environmental problems in countries Much of the literature on trade and the environment where resources are concentrated. and the nascent literature on GVCs and the environment However, GVCs also present an opportunity to use focus on carbon emissions and, to some extent, other value chain connections with concerned consumers forms of pollution. However, land use changes such as to address these issues through voluntary standards deforestation and overfishing are equally important and regulated changes. Meanwhile, large-scale oper- from a purely environmental and human health per- ations and upstream connections allow lead firms to spective. These are conceptually distinct issues, with efficiently provide information and services that give very different impacts from trade and GVCs. small-scale producers an opportunity to demonstrably In agriculture, GVCs can help save scarce resources meet standards that they otherwise could not. The by ensuring that raw materials are sourced closest appropriate regulations and policies will, however, to natural resources. But they can also lead to over- have to be put in place for achieving large scale impact. use because of specialization and a growing global The challenges and the possible solutions in a GVC demand. The pernicious effects are magnified when world are well illustrated by the cocoa and chocolate resource use is subsidized. industry. Cocoa—the primary ingredient in the world’s GVCs allow countries to preserve scarce resources chocolate—has been identified as a major driver of by importing raw agricultural products from countries deforestation in West Africa. For many years, the soar- with more abundant resources. A good example is the ing global demand and expanding cocoa production 126 | World Development Report 2020 Box 5.2  Virtual water Are countries that have scarcer water reserves importing Because water-efficient countries can export water- water-intensive goods? intensive goods, especially agricultural products, to less The global water footprint in 1996–2005 was estimated efficient countries, trade has helped reduce the amount at 9,087 billion cubic meters per year: 74 percent green (the of water used in aggregate production. The global rainwater stored in the soil used to produce agricultural water savings related to trade in agricultural products in goods), 11 percent blue (the freshwater used to produce 1996–2005 was an estimated 369 billion cubic meters goods and services), and 15 percent gray (polluted water per year (58.7 percent green, 26.6 percent blue, and 14.7 from production). Agricultural production contributes percent gray), which is equivalent to 4 percent of the 92 percent to this total footprint, and about one-fifth of the global water footprint related to agricultural production global water footprint is attributed to production for export.   (map B5.2.1). Map B5.2.1  Global water savings associated with international trade in agricultural products, 1996–2005 25 Gm³/yr 5% 3% 3% 85% 24% 12% 6% 37% 5.6 71% 6% 60% 5.4 Gm³/yr 17% 1% Gm³/yr 77% 21% 5.7 73% 54% 45% 12 13% 2% 10% 1% Gm³/yr Gm³/yr 14% 16 85% 40% 50% Gm³/yr 85% 5.1 9% 8% 7.8 9.3 Gm³/yr Gm³/yr Gm³/yr 83% 5.0 Gm³/yr 7.4 Gm³/yr 30% 38% 32% Rainwater stored in the soil used to produce 5% 6% agricultural goods 5.0 Freshwater used to produce 89% Gm³/yr goods and services Polluted water from production IBRD 44678 | SEPTEMBER 2019 Source: Mekonnen and Hoekstra 2011. Note: Virtual water balance per country and direction of gross virtual water flows related to trade in agricultural and industrial products over the period 1996–2005. The thicker the arrow, the bigger the virtual water flow. Only the biggest water savings are shown—more than 5 billion cubic meters per year (Gm3/yr). have degraded forests. Suitable land is shrinking Certification schemes are one possible means of because of climate change, and trees are aging and addressing the environmental and socioeconomic need to be replanted, particularly in Côte d’Ivoire and issues in the industry. This opportunity for moving Ghana. However, the 5–6 million smallholder farms to more sustainable methods of cocoa production is that produce almost the entire global supply of cocoa supported by the downstream industry. Processing lack good agricultural practices to address these chal- is dominated by a few large traders, grinders, and lenges. They also face difficulties in obtaining farming chocolate producers.33 Six companies alone process supplies and financing any improvements they may and trade 89 percent of the annual global cocoa pro- want to make. Ongoing deforestation to increase duction, and five chocolate producers buy 39 percent cocoa production is not sustainable. of it. Because a few large companies dominate and Impact on the environment | 127 compete at the downstream stages of production, they attention to how their supply chains function in are well placed to cooperate in fighting environmen- terms of social and environmental standards. Typi- tal degradation, a huge threat to their productivity, cally, the lead firms in GVCs are well known, and so particularly as climate change is making cocoa har- their behavior can be easily monitored. Some firms vest yields extremely unpredictable. And yet despite actively promote standards along the value chain, the strong incentives to work together to improve the including by assessing the monetary value of better social and environmental footprint of the upstream social and environmental standards in their balance operations, the private sector commitments are not sheets. Consumers are demanding more sustainable translating into improved sustainability of the supply products, and so producing such products can have chain in the absence of regulatory change. To improve positive economic returns from either cost savings, sustainability of the cocoa value chains, domestic reg- risk mitigation, or product recognition.34 ulators and international development partners need Recent studies provide empirical evidence that to work together with the private sector. stricter regulation can enhance business perfor- mance.35 At the country-industry level, higher compli- Relational GVCs and production ance with social and environmental standards is cor- related with economic upgrading.36 An example of how techniques higher standards help save the water and energy of Environmental concerns associated with globaliza- supplying firms is described in box 5.3. As in other suc- tion may be alleviated in the age of GVCs. Because cessful cases, the example described in box 5.3 involves lead firms have a brand name to protect, they pay a joint effort of private and public stakeholders. Box 5.3  Toward sustainable fashion In 2018 the greenhouse gas emissions from textile pro- Figure B5.3.1  Swedish lead firms in duction totaled 1.2 billion metric tons of CO2 equivalent, apparel and textiles produce a lot of or more than that from international flights and maritime value added with little CO2, and their shipping combined.a Textile production (including cotton suppliers produce a lot of CO2 with farming) uses about 93 billion cubic meters of water a little value added year, and 20 percent of industrial water pollution globally 100 is attributable to dyeing and treating textiles. If the sector continues on its current trajectory, resource consumption 31 80 will triple between 2015 and 2050, while the industry share 57 of the carbon budget associated with a 2°C pathway could 60 Percent increase to 26 percent. 26 Most emissions associated with the Swedish textile 40 and apparel sector are produced by its suppliers outside 30 Sweden, suggesting that cross-country and cross-industry 20 43 collaboration is needed to reduce emissions (figure B5.3.1). 13 A partnership between Swedish textile producers and the 0 CO2 emissions Value added Swedish International Development Cooperation Agency (Sida) reveals how higher standards can help save the Indirect suppliers water and energy of supplying firms, with environmental Direct suppliers and economic gains. Own operations The Sweden Textile Water Initiative (STWI), launched in Sources: WDR 2020 team, using data from OECD’s TiVA database; WIOD; 2014 and supported by Sida, was a public-private develop- Exiobase. ment partnership with 24 textile and leather companies. Its Note: Estimates were obtained through a multiregional input–output model extended with satellite accounts for carbon emissions. The direct goal was to help establish a network of private companies and indirect suppliers of the Swedish textile and apparel sector include committed to improving the efficiency of water use by the upstream industries from both Sweden and foreign countries. (Box continues next page) 128 | World Development Report 2020 Box 5.3  Toward sustainable fashion (continued) suppliers and subsuppliers associated with their brands generally sustainable because of the cost savings in water in Bangladesh, China, Ethiopia, India, and Turkey. Sida and chemicals over time, and companies’ awareness and provided the financing; clothing brands contributed by capacity increased. These numbers confirm that development engaging their factories; and the Stockholm International interventions can play a catalytic role in improving the sus- Water Institute oversaw implementation. This collaboration tainability of GVCs by raising awareness and providing tech- generated significant cost savings and time savings in terms nical assistance. But cost sharing with companies is important of rolling out the initiative.b Although Sida exited in 2018, to ensure ownership and engagement. the network continues to expand globally and to pursue The initiative had a limited impact on national water its mandate of supporting sustainability champions with governance practices in each country. The STWI’s upcoming business intelligence, networking, and advice on resource Mill Improvement Alliance hopes to extend the program to efficiency. a larger number of factories to achieve broader sector- and In the first three years, STWI supported 276 factories in the economywide impacts. But governments also will have five initial countries, training more than 1,300 managers and to join the effort, particularly in updating their water 37,000 staff. The savings amounted to almost 11 million cubic governance frameworks. Private actors in initiatives such meters of water and almost 80 million kilowatt-hours of elec- as the STWI can submit recommendations for regulatory tricity (table B5.3.1). Despite some variation in savings among change—and possibly counter pushback that might other- countries and across factories, the factory investments were wise come from affected companies. Table B5.3.1  Total reported savings generated by the Sweden Textile Water Initiative in its five partner countries, 2015–17 Savings Bangladesh China Ethiopia India Turkey Total Water (m ) 3 2,680,005 6,316,597 99,323 339,659 1,085,973 10,521,557 Electricity (kWh) 18,364,890 45,526,706 21,780 6,074,612 9,599,713 79,587,701 Thermal use (metric tons) 1,708,103 4,695,729 115,881 0 0 6,519,714 Chemical use (kg) 1,187,505 18,611,056 5,185 281,635 2,497,178 22,582,559 Waste water (m3) 16,319 2,435,680 0 0 229,860 2,681,859 Natural gas (m3) 20,798,126 1,407,313 0 24,514 5,130,815 27,360,768 Fossil fuel (metric tons) 702,334 0 444 1,904 625 705,309 Coal (kg) 0 1,002 0 6,319,396 3,823,737 10,144,135 GHG emissions (metric tons) 45,365 353,277 0 41,274 24,850 464,766 Source: Swedish International Development Cooperation Agency (Sida). Note: GHG = greenhouse gas; kg = kilograms; kWh = kilowatt-hours; m3 = cubic meters. a. Ellen MacArthur Foundation (2017). b. Andersson et al. (2018). The relational nature of GVCs can also promote the sustainability index. Levi’s has a comparable arrange- transfer of clean technology and know-how. Firms that ment with its suppliers through the International have a brand to defend naturally tend to align practices Finance Corporation’s Global Trade Supplier Finance within the corporation. The clothing firm Puma, in col- Program. Investment firms are also pushing for more laboration with the International Finance Corporation, sustainable practices among the major brands. They the bank BNP Paribas, and the fintech firm GT Nexus are paying more attention to environmental, social, launched a program in 2016 that offers better receivable and governance (ESG) performance and pushing the financing terms to suppliers who score high on Puma’s major brands to adopt higher standards. Impact on the environment | 129 Box 5.4  Demanding environmental standards in GVC upstream firms Saitex International (Vietnam) and Zakład Pierzarski cutting energy usage by 5.3 million kilowatt-hours a year— Konrad Ożgo  (Poland) are GVC suppliers whose compar- and CO2 emissions by nearly 80 percent. It also plants ative advantage includes their ability to meet demanding trees to offset its emissions. Furthermore, it minimizes voluntary environmental standards. the waste from production. All denim creates a toxic Saitex produces denim jeans in a LEED (Leadership in by-product called sludge, but at Saitex the sludge is Energy and Environmental Design)-certified facility for the extracted and shipped to a nearby brick factory. Mixed California company Everlane, whose “radical transparency” with concrete, the toxic material can no longer leech into is the core of its marketing strategy. According to Everlane’s the environment. The resulting bricks are used to build website,a Saitex recycles 98 percent of its water, relies on affordable homes. alternative energy sources, and repurposes by-products Zakład Pierzarski Konrad Ożgo, which preprocesses to create premium jeans minimizing the waste. Standard white goose down for the outdoor clothing firm Patagonia, denim manufacturers use “belly” washing machines, which has a fully traceable supply chain to comply with its brand consume as much as 1,500 liters of water to produce one philosophy. Internal audits and third-party verification pair of jeans. Saitex instead consumes only 0.4 liter of ensure that the birds are neither live plucked nor force- water per pair of jeans thanks to state-of-the-art recycling. fed and that they are raised in humane conditions. The On-site rainwater collection pools allow Saitex to adoption of this costly technology allows this supplier to minimize the impact of the consumption it does have, have a long-lasting relationship with the buyer, Patagonia, and its sophisticated five-step filtration process separates which in this way can trace its supply back to the more water from toxic contaminants and then sends the clean than 100 individual smallholder farms—including parent water back into the system. Saitex is also committed to farms, hatcheries, and raising farms—whose output passes using renewable energy resources such as solar power and through the preprocessor. a. https://www.everlane.com/factories/denim-saitex. The long-term nature of firm-to-firm relationships Green goods and contracts in relational GVCs can be a force for con- vincing companies in their supply chain to adopt new One of the biggest contributions of GVCs to the costly technology (box 5.4). This point is important environment may be the many new and innovative because many of the environmental impacts are borne environmental products they make possible. Trade upstream, by the suppliers, even if most of the value and GVCs have a positive impact on the environment is created downstream, as in the Swedish example in by promoting innovation and by making these clean box 5.3. technologies and environmental goods more afford- The positive role of relational GVCs does have able. This section describes some of the most import- its limits, however. First, the technology transfer ant green goods value chains.37 tends to benefit direct suppliers the most, and to a much lesser extent second- and lower-tier suppliers, Solar energy which in some cases are invisible to the GVC lead The solar value chain relies on innovation and com- firm. Second, the positive local effects of relational plex production systems. Countries may be part of GVCs may not translate into an overall gain for the the value chain through producing silicon, manufac- environment globally. When a lead firm relocates turing solar cells, or assembling modules, inverters, production to a developing country, and it produces mounting systems, combiner boxes, and other compo- there with carbon intensity that is lower than the nents.38 Older companies appear to be more vertically prevailing carbon intensity of the host country, that integrated, whereas newer entrants tend to source is not in itself a reduction in pollution and emis- from multiple locations for assembly on-site. sions. The carbon intensity can still increase overall Solar photovoltaic (PV) products are generally relative to a counterfactual where the firm did not tradable. Map 5.1 illustrates the supply chain of a relocate. PV company. Solar cell production is concentrated 130 | World Development Report 2020 Map 5.1 Supply chain of a solar photovoltaic company Netherlands Shipping and United Kingdom distribution System installed 25+ years recycling Germany Polysilicon Manufacturing Switzerland equipment Manufacturing equipment, local United States marketing Manufacturing China equipment R&D; product design; wafer, cell, and module manufacturing IBRD 44662 | SEPTEMBER 2019 Source: European Commission 2016. Note: Solar cell production is primarily concentrated in China and elsewhere in Asia and is dependent on the production of components from several countries. Europe and the United States lead upstream service provision, including shipping, distribution, installation, and recycling. R&D = research and development. primarily in China and elsewhere in Asia and is depen- only about 50 percent of the value of components dent on the production of components from several is from domestic sources. Several European coun- countries. Europe and the United States lead upstream tries, such as Denmark and Germany, used to be the service provision, including shipping, distribution, main manufacturing hubs, but the sector is growing installation, and recycling. increasingly diverse geographically, with more than Large parts of the supply chain have generally been 50 percent of suppliers from China, India, and other located in countries or regions with strong demand, Asian countries, as well as Brazil.40 such as the European Union. Low labor costs, natural In the electric vehicle industry, global sales of new resources, and government policies have driven some vehicles passed a million units for the first time in production to China. Meanwhile, policies to encour- 2017. On current trajectories, this figure could quadru- age deployment have expanded in other countries.39 ple by 2020, to about 5 percent of the total global light Value created along the solar value chain starts vehicle market.41 with polysilicon and ends with the PV module (table China is the largest global market for electric vehi- 5.1). Downstream activities generally account for a cles, and it is dominated by independent domestic firms. large share of value added, especially for services China’s electric vehicle industry showcases how trade such as installation, system design, and research and liberalization and greater access to foreign suppliers, development. combined with government intervention and strong competition in the traditional automotive market, allow Other examples of green goods independent domestic companies to enter the niche The wind energy supply chain, though not as global- market of electric vehicles and become both innovative ized as solar, has grown increasingly complex and and cost-competitive. In the years after China joined fragmented. A single wind turbine has more than the World Trade Organization, the import volumes of 8,000 parts. And major components include rotor parts for electric motors and generators picked up, as blades, towers, and nacelles. In the U.S. supply chain, exports of electronic motors also increased. Impact on the environment | 131 Table 5.1  Estimation of value added at stages of the supply chain of a solar photovoltaic (PV) module US$ per kilowatt Sales receipts Cost of intermediate products Stage of production (turnover or gross output) and services Value added Polysilicon 150  50 100 Silicon wafer 330 150 180 Solar cell 460 330 130 Final product (PV module) 660 460 200 Total 1,600 990 610 Source: Jha 2016. Notes 4.  2 See Copeland and Taylor (2004) for a formulation of the pollution haven hypothesis. 1. Roosevelt (2018). 25.  Cherniwchan, Copeland, and Taylor (2017); Dechezleprêtre 2.  This phenomenon, known as the pollution haven effect and Sato (2017); Ederington, Levinson, and Minier (2005). (PHE), is different from the pollution haven hypothesis 26. Demsetz (1967). (PHH). The empirical literature provides much more sup- 27.  Levinson and Taylor (2008). port for the PHE than for the PHH (Copeland and Taylor 28.  Shapiro and Walker (2018). 2004; He 2006; Kellenberg 2009; Levinson and Taylor 29.  Shapiro and Walker (2018). 2008). The PHE and the associated problem of “carbon 30.  Jackson et al. (2001). leakage” hold even when the PHH does not. Although 31.  van der Werf and Petit (2002). 32.  Kroeger et al. (2017). environmental policy is not a predominant determinant 33.  Kroeger et al. (2017). of comparative advantage, it does matter at the margin, 34.  Impact Valuation Roundtable (2017). particularly for countries whose competitiveness is 35.  Lanoie et al. (2011). based on producing at low cost. 36. Kummritz, Taglioni, and Winkler (2017). 3.  The environmental Kuznets curve (EKC) describes a 37. Not all green goods have a positive environmental foot- relationship between per capita income in a location and print. In some cases, such as in the mining of rare miner- environmental outcomes. See Grossman and Krueger als, this is not the case. (1991, 1995). 38. Jha (2016). 4.  Copeland and Taylor (2004); Stern (2017). 39. Jha (2016). 5. ITF (2015). 40. Jha (2016). 6. ITF (2015). Hertzke et al. (2018). 41.  7.  UNCTAD (2017, 2018). 8. IMO (2015). 9. UNCTAD (2017). References 10.  ETC (2018); European Parliament (2015). Andersson, Jens, Reza Iftekhar Patwary, Weronika Rehnby, 11.  The initial strategy, launched in April 2018, envisages and Emilie Pellby. 2018. “Evaluation of STWI Projects reducing total greenhouse gas (GHG) emissions from 2014–2018: Final Report.” November 19, NIRAS Sweden, international shipping, which should reduce the total Stockholm. annual GHG emissions by at least 50 percent by 2050 Anenberg, Susan, Joshua Miller, Daven Henze, and Ray compared with 2008, while pursuing efforts toward Minjaresicct. 2019. “A Global Snapshot of the Air phasing them out entirely by that date (IMO 2018b). Pollution–Related Health Impacts of Transportation 12.  Sims et al. (2014). Sector Emissions in 2010 and 2015.” International Council 13.  Keen, Parry, and Strand (2012). on Clean Transportation, Washington, DC. 14.  Dominioni, Heine, and Martínez Romera (2018). Brooks, Amy L., Shunli Wang, and Jenna R. Jambeck. 2018. 15.  Anenberg et al. (2019). “The Chinese Import Ban and Its Impact on Global 16.  For more information, see IMO (2018a). Plastic Waste Trade.” Science Advances 4 (6). doi: 10.1126 17. Holgate (2018). /sciadv.aat0131. 18. UNU (2018). Center for Biological Diversity. No date. “Ocean Plastics Pol- 19. UNU (2018). lution: A Global Tragedy for Our Oceans and Sea Life.” 20. Lepawsky (2015). Center for Biological Diversity, Tucson, AZ. https://www 21. Ryder and Zhao (2019). .biologicaldiversity.org/campaigns/ocean_plastics/. 22. Brooks, Wang, and Jambeck (2018). Cherniwchan, Jevan, Brian R. Copeland, and M. Scott Tay- 23.  Center for Biological Diversity (n.d.). lor. 2017. “Trade and the Environment: New Methods, 132 | World Development Report 2020 Measurements, and Results.” Annual Review of Economics /industries/automotive-and-assembly/our-insights/the 9 (August): 59–85. -global-electric-vehicle-market-is-amped-up-and-on-the Copeland, Brian R., and M. Scott Taylor. 2004. “Trade, -rise?. Growth, and the Environment.” Journal of Economic Liter- Hettige, Hemamala S., Paul Jonathan Martin, Manjula Singh, ature 42 (1): 7–71. and David R. Wheeler. 1995. “The Industrial Pollution Dechezleprêtre, Antoine, and Misato Sato. 2017. “The Impacts Projection System.” Policy Research Working Paper 1431, of Environmental Regulations on Competitiveness.” World Bank, Washington, DC. Review of Environmental Economics and Policy 11 (2): 183–206. Holgate, Peter. 2018. “How Do We Tackle the Fastest Demsetz, Harold. 1967. “Toward a Theory of Property Rights.” Growing Waste Stream on the Planet?” Global Agenda: American Economic Review 57 (2): 347–59. Circular Economy, February 9, World Economic Forum, Dominioni, Goran, Dirk Heine, and Beatriz Martínez Geneva. https://www.weforum.org/agenda/2018/02/how Romera. 2018. “Regional Carbon Pricing for International -do-we-tackle-the-fastest-growing-waste-stream-on-the Maritime Transport: Challenges and Opportunities for -planet/. Global Geographical Coverage.” Policy Research Working Hook, Leslie, and John Reed. 2018. “Why the World’s Paper 8319, World Bank, Washington, DC. Recycling System Stopped Working.” Financial Times, Ederington, Josh, Arik Levinson, and Jenny Minier. 2005. October 25. https://www.ft.com/content/360e2524-d71a “Footloose and Pollution-Free.” Review of Economics and -11e8-a854-33d6f82e62f8. Statistics 87 (1): 92–99. IMO (International Maritime Organization). 2015. Third IMO Ellen MacArthur Foundation. 2017. A New Textiles Economy: Greenhouse Gas Study, 2014. London: IMO. Redesigning Fashion’s Future. Cowes, Isle of Wight, U.K.: — —— —. 2018a. “Marine Litter.” In Focus, IMO, London. http:// Ellen MacArthur Foundation. https://www.ellenmacarthur www.imo.org/en/MediaCentre/HotTopics/marinelitter foundation.org/publications/a-new-textiles-economy /Pages/default.aspx. -redesigning-fashions-future. — —— —. 2018b. “UN Body Adopts Climate Change Strategy ETC (Energy Transitions Commission). 2018. Mission Pos- for Shipping.” Press Briefing 06, April 13, IMO, London. sible: Reaching Net-Zero Carbon Emissions from Harder-to- http://www.imo.org/en/MediaCentre/PressBriefings Abate Sectors by Mid-Century. London: ETC. http://www /Pages/06GHGinitialstrategy.aspx. .energy-transitions.org/sites/default/files/ETC_Mission Impact Valuation Roundtable. 2017. “White Paper: Operation- Possible_FullReport.pdf. alizing Impact Valuation: Experiences and Recommen- European Commission. 2016. Trade Sustainability Impact dations by Participants of the Impact Valuation Round- Assessment on the Environmental Goods Agreement. Final table.” Geneva, World Business Council for Sustainable report. Brussels: Directorate-General for Trade, European Development. Commission. ITF (International Transport Forum). 2015. “The Carbon European Parliament. 2015. “Emissions Reduction Targets Footprint of Global Trade: Tackling Emissions from for International Aviation and Shipping.” Report IP/A/ International Freight Transport.” Policy Brief, November ENVI/2015-11 (November), Policy Department A: Eco- 30, ITF, Paris. https://www.itf-oecd.org/carbon-footprint nomic and Scientific Policy, Directorate General for -global-trade. Internal Policies, European Parliament, Brussels. Jackson, Jeremy B. C., Michael X. Kirby, Wolfgang H. Berger, Frothingham, Steve. 2018. “Pedego Shifts Some E-Bike Pro- Karen A. Bjorndal, Louis W. Botsford, Bruce J. Bourque, duction to Vietnam in Response to Tariffs. Bike Retailer Roger H. Bradbury, et al. 2001. “Historical Overfishing and Industry News, September 12. https://www.bicycle and the Recent Collapse of Coastal Ecosystems.” Science retailer.com/international/2018/09/12/pedego-shifts-some 293 (5530): 629–37. -e-bike-production-vietnam-response-tariffs#.XVrgOp Jha, Veena. 2016. “Building Supply Chain Efficiency in Solar NKiCd. and Wind Energy: Trade and Other Policy Consider- Grossman, Gene M., and Alan B. Krueger. 1991. “Environmen- ations.” Issue Paper (May), International Centre for Trade tal Impacts of a North American Free Trade Agreement.” and Sustainable Development, Geneva. NBER Working Paper 3914 (November), National Bureau Keen, Michael, Ian W. H. Parry, and Jon Strand. 2012. of Economic Research, Cambridge, MA. “Market-Based Instruments for International Aviation —— ——. 1995. “Economic Growth and the Environment.” and Shipping as a Source of Climate Finance.” Policy Quarterly Journal of Economics 110 (2): 353–77. Research Working Paper 5950, World Bank, Washing- He, Jie. 2006. “Pollution Haven Hypothesis and Environmen- ton, DC. tal Impacts of Foreign Direct Investment: The Case of Kellenberg, Derek K. 2009. “An Empirical Investigation of Industrial Emission of Sulfur Dioxide (SO2) in Chinese the Pollution Haven Effect with Strategic Environment Provinces.” Ecological Economics 60 (1): 228–45. and Trade Policy.” Journal of International Economics 78 (2): Heine, Dirk, and Susanne Gäde. 2018. “Unilaterally Remov- 242–55. ing Implicit Subsidies for Maritime Fuels.” International Kroeger, Alan, Haseebullah Bakhtary, Franziska Haupt, and Economics and Economic Policy 15 (2): 523–45. Charlotte Streck. 2017. “Eliminating Deforestation from Hertzke, Patrick, Nicolai Müller, Stephanie Schenk, and Ting the Cocoa Supply Chain.” March, World Bank, Washing- Wu. 2018. “The Global Electric-Vehicle Market Is Amped ton, DC. Up and on the Rise.” Our Insights (blog), May. McKinsey Kummritz, Victor, Daria Taglioni, and Deborah Elisabeth and Company, New York. https://www.mckinsey.com Winkler. 2017. “Economic Upgrading through Global Impact on the environment | 133 Value Chain Participation: Which Policies Increase the Agenda: Digital Economy and Society, January 24, World Value Added Gains?” Policy Research Working Paper Economic Forum, Geneva. https://www.weforum.org 8007, World Bank, Washington, DC. /agenda/2019/01/how-a-circular-approach-can-turn-e Lanoie, Paul, Jérémy Laurent-Lucchetti, Nick Johnstone, and -waste-into-a-golden-opportunity/. Stefan Ambec. 2011. “Environment Policy, Innovation, Shapiro, Joseph S., and Reed Walker. 2018. “Why Is Pollution and Performance: New Insights on the Porter Hypoth- from U.S. Manufacturing Declining? The Roles of Envi- esis.” Journal of Economics and Management Strategy 20 (3): ronmental Regulation, Productivity, and Trade.” American 803–42. Economic Review 108 (12): 3814–54. Lepawsky, Josh. 2015. “The Changing Geography of Global Sims, Ralph, Roberto Schaeffer, Felix Creutzig, Xochitl Cruz- Trade in Electronic Discards: Time to Rethink the e-Waste Núñez, Marcio D’Agosto, Delia Dimitriu, Maria Josefina Problem.” Geographical Journal 181 (2): 147–59. Figueroa Meza, et al. 2014. “Transport.” In Climate Change Levinson, Arik. 2010. “Offshoring Pollution: Is the United 2014: Mitigation of Climate Change; Contribution of Working States Increasingly Importing Polluting Goods?” Review Group III to the Fifth Assessment Report of the Intergovernmen- of Environmental Economics and Policy 4 (1): 63–83. tal Panel on Climate Change, edited by Ottmar Edenhofer, Levinson, Arik, and M. Scott Taylor. 2008. “Unmasking the Ramón Pichs-Madruga, Youba Sokona, Jan C. Minx, Pollution Haven Effect.” International Economic Review 49 Ellie Farahani, Susanne Kadner, Kristin Seyboth, et al., (1): 223–54. 599–670. New York: Cambridge University Press. Mekonnen, Mesfin M., and Arjen Y. Hoekstra. 2011. “National Stern, David I. 2017. “The Environmental Kuznets Curve after Water Footprint Accounts: The Green, Blue, and Grey 25 Years.” Journal of Bioeconomics 19 (1): 7–28. Water Footprint of Production and Consumption.” Value UNCTAD (United Nations Conference on Trade and Devel- of Water Research Report 50 (May), IHE Delft Institute opment). 2017. Review of Maritime Transport 2017. Report for Water Education, Delft, the Netherlands. UNCTAD/RMT/2017 (October 25). Geneva: UNCTAD. Muntean, Marilena, Diego Guizzardi, Edwin Schaaf, Monica —— — —. 2018. Review of Maritime Transport 2018. Report Crippa, Efisio Solazzo, Jos G. J. Olivier, and Elisabetta UNCTAD/RMT/2018 (October 3). Geneva: UNCTAD. Vignati. 2018. Fossil CO2 Emissions of All World Countries, UNU (United Nations University). 2018. “E-Waste Rises 2018 Report. JRC Science for Policy Report EUR 29433 EN. 8% by Weight in 2 Years as Incomes Rise, Prices Fall.” Luxembourg: Joint Research Center, European Union. Press release, December 14, UNU, Tokyo. https://unu Roosevelt, Margot. 2018. “Can Southern California’s Electric .edu/media-relations/releases/ewaste-rises-8-percent-by Bike Industry Survive Trump’s Tariffs?” Orange County -weight-in-2-years.html. Register, September 10. https://www.ocregister.com/2018 van der Werf, Hayo M. G., and Jean Petit. 2002. “Evaluation /09/10/can-southern-californias-electric-bike-companies of the Environmental Impact of Agriculture at the Farm -survive-trumps-tariffs/. Level: A Comparison and Analysis of 12 Indicator-Based Ryder, Guy, and Houlin Zhao. 2019. “The World’s e-Waste Is Methods.” Agriculture, Ecosystems, and Environment 93 (1–3): a Huge Problem: It’s Also a Golden Opportunity.” Global 131–45. 134 | World Development Report 2020 6 Technological change Key findings  rade costs are likely to continue to fall. New digital technologies enhance opportunities • T for global value chain (GVC) participation. Developing countries, which exhibit the highest costs and biggest impediments to trade, stand to gain the most.  latform firms and e-commerce generate uneven benefits across firms and households. • P Platform firms facilitate participation but also foster concentration, which affects the distribution of gains from participation in GVCs.  nxiety that automation will hinder export-led industrialization may not be warranted. • A Evidence of reshoring is limited. New production technologies have promoted North– South trade, although the effects are heterogeneous across countries and sectors. • I ncreased automation in manufacturing is likely to have distributional impacts. Adoption of robots is driving down the labor share of income and increasing the demand for skilled workers, thereby exacerbating inequality in the labor market and increasing the need for adjustment policies to support disrupted workers.  estricting trade to promote manufacturing is counterproductive. It lowers efficiency, • R raises prices of both inputs and outputs, and undermines incentives to innovate. 136 | World Development Report 2020 S upply chains are rapidly changing under the to verify seller and buyer quality may foster concen- pressure of digital innovation. Robotics, 3D tration, which makes it harder for entrants to compete. printing, big data, blockchain technologies, Platform firms also pose new challenges for regulators cloud computing, the Internet of Things, and the rise seeking to ensure fair competition and prevent abuse of platform firms are transforming production and of market power. Meanwhile, because of technological distribution processes in many industries. Digital progress more goods and services, as well as new ones, technologies raise productivity but are also disruptive, are likely to become tradable over time. especially when they lead to a reduction in demand Anxiety that automation will hinder export-led for workers. Meanwhile, a substantial share of exports industrialization may not be warranted. Evidence from low-wage developing countries is in sectors being of companies moving operations back to their home rapidly automated by their trade partners. These devel- country (reshoring) is very limited, and new produc- opments have sparked fears that industrialization led tion technologies such as industrial robots and 3D by labor-intensive exports may no longer be a viable printing have promoted North–South trade, although model for developing economies seeking to develop the effects are heterogeneous across countries and by joining and then moving up the value chain—and sectors. Those that mainly compete with robot- that labor costs are becoming a less important deter- adopting countries in output markets are at risk of minant of competitiveness. Moreover, changing skills being outcompeted by foreign robots and may suffer demands associated with technological progress could substantial reductions in employment. Adoption of place developing countries at a disadvantage. robots is driving down the share of income accruing This chapter reviews the evidence on how emerg- to labor and increasing the demand for skilled workers ing digital technologies, including advanced robotics that perform tasks that complement those performed and 3D printing, are affecting global value chains by robots, thereby exacerbating inequality. (GVCs), trade flows, and the prospects for export-led Robot adoption improves productivity, which leads industrialization. In doing so, it reviews the channels to an expansion in output and increased demand for through which technological progress could have material inputs. It also leads to the creation of new impacts on GVCs—reducing trade costs, inducing tasks. In spite of these benefits, robot adoption will quality upgrading and product churning, and chang- likely entail substantial labor market pain. ing productivity and relative costs across countries Increasing tariffs to shield domestic industries and sectors, thereby changing comparative advantage. from intensified competition associated with the It then explores how changes in trade policy might adoption of new production technologies in other alter these effects1 and offers a tentative assessment of countries is likely counterproductive because it lowers the potential for continued expansion of global supply efficiency, raises the prices of both inputs and outputs, chains and export-led development. New technologies and undermines incentives to innovate. will likely change GVCs and the trade and jobs they create. But forecasting exactly how is fraught with uncertainty, not least because technological progress Declining trade costs is difficult to predict. The Internet facilitates GVC participation Trade costs are likely to continue to fall because The information and communication technology of new digital technologies, offering greater oppor- (ICT) revolution that emerged in the mid-1990s has tunities for GVC participation. Developing countries been an important enabler of the expansion of GVCs. may stand to gain the most from emerging digital The share of the global population using the Internet technologies because they face the highest trade grew from less than 1 percent in 1993 to 46 percent costs and biggest distortions. Extending access to in 2016. By 2014, almost all firms (with at least five high-speed Internet and expanding e-commerce will employees) in high-income Organisation for Eco- facilitate greater GVC participation. But the gains from nomic Co-operation and Development (OECD) coun- e-commerce are unevenly distributed across house- tries used a broadband Internet connection. Among holds, and not all firms benefit equally from Internet firms in lower-income countries, broadband usage access. Artificial intelligence applications, such as remains lower, but it is rising rapidly.2 At the same machine translation, can further reduce trade and time, the cost at which information can be transmitted logistics costs, and might also help reduce red tape. via an optical network has fallen dramatically. In fact, Platform firms make it easier to participate in global today the time it takes to download a high-definition markets. But the reputation mechanisms they rely on movie through a modem connected to fiber optics is Technological change | 137 almost imperceptible. This ICT revolution has not only Meanwhile, cloud computing offers a pay-as- reduced trade costs by lowering the cost of processing you-go subscription model for storage and software, and transmitting information over long distances, but facilitating file sharing between cross-country teams it also has enabled firms to improve productivity and and lowering the fixed costs of investments in IT has led to a new range of information technology (IT)– infrastructure. related services. These advances have contributed to Some robotics and artificial intelligence applica- a rise in global trade and production sharing because tions might further reduce logistics costs, the time firms are increasingly spreading their production pro- to transport, and the uncertainty of delivery times cess across borders and sourcing more intermediate (box 6.1). At ports, autonomous vehicles might unload, inputs and services from abroad.3 stack, and reload containers faster and with fewer High-speed Internet enables firms in developing errors. Blockchain shipping solutions may lower countries to link to GVCs. The introduction of fast transit times and speed up payments. The Internet of Internet in Africa and China has spurred employ- Things has the potential to increase the efficiency of ment and export growth, as recent studies of the eco- delivery services by tracking shipments in real time, nomic effects of the rollout have shown.4 In Africa, while improved and expanded navigation systems the gradual arrival of submarine Internet cables led may help route trucks based on current road and to faster job growth (including for low-skilled work- traffic conditions. Although the empirical evidence on ers) in locations that benefited from better access to these impacts is limited, it is estimated that new logis- fast Internet relative to those that did not, with little tics technologies could reduce shipping and customs or no job displacement across space. Increased firm processing times by 16 to 28 percent.9 entry, productivity, and exporting are among the drivers of the higher net job creation in these loca- Investments in digital technologies may tions. Similarly, in China provinces experiencing an be especially beneficial for developing increase in the number of Internet users per capita countries also witnessed faster export growth, with more firms Ongoing technological progress, more widespread competing in international markets and a higher adoption of existing digital technologies, and invest- share of provincial output sold abroad.5 These exam- ments in transport infrastructure are likely to reduce ples attest to the potential of ICTs to help countries trade costs, promote trade, and lead to a continued become part of international supply chains. They expansion of GVCs. These developments may espe- also show that the uneven provision of ICT infra- cially benefit developing countries, which currently structure can aggravate spatial inequalities if already face higher trade and transport costs and have com- productive regions are the prime beneficiaries of paratively limited ICT infrastructure. For example, 4G infrastructure upgrading. network coverage remains low in large parts of Africa compared with that in richer countries (map 6.1). Digital technologies are lowering logistics Tariffs and nontariff measures continue to pose a sig- and coordination costs nificant restriction to trade by low-income countries, Digital technologies can improve customs perfor- despite preferential access programs.10 In addition, mance by automating document processing and mak- developing countries face large intranational trade ing it possible to create a single window for stream- costs, which determine the extent to which producers lining the administrative procedures for international and consumers in remote locations are affected by trade transactions. In Costa Rica, a one-stop online changes in trade policy and international prices. For customs system increased both exports and imports.6 example, the effect of distance on trade costs within Similarly, in Colombia computerizing import proce- Ethiopia or Nigeria is four to five times larger than in dures increased imports, reduced corruption cases, the United States. Intermediaries capture most of the bolstered tariff revenues, and accelerated the growth surplus from falling world prices, especially in more of firms most exposed to the new procedures.7 distant locations. Therefore, consumers in remote Digital technologies also facilitate trade in exist- locations see only a small part of the gains from falling ing services and may promote new services (such as international trade barriers.11 Despite recent advances videoconferencing and telecommuting) supporting in the provision of ICT infrastructure, the scope for GVCs. The services trade is becoming more important, further expanding access to high-speed Internet in and the World Trade Organization projects it will rise developing countries remains huge. from approximately 21 percent of world trade today to In part because of high trade costs, firms in low- 25 percent by 2030.8 income countries tend to operate on a small scale and 138 | World Development Report 2020 Box 6.1  Digital innovation and agricultural trade Distributed ledger technologies (DLTs) are decentralized with detailed information on purchased chicken, such as systems for recording transactions of assets in which the veterinary treatments, freshness, and other metrics.c Simi- transactions and their details are recorded in multiple larly, Barilla, an Italian pasta and pesto sauce manufacturer, places at the same time. DLTs could increase efficiency and uses blockchain technology to improve transparency and transparency in agricultural supply chains by improving traceability in its pesto production cycle along the entire product traceability and integrity, contract certainty, veri- supply chain—from farm to fork. fication of geographic origin, and compliance with sanitary Meanwhile, many start-ups are aiming to shorten agri- and phytosanitary requirements. They could also improve culture value chains and reduce the role of intermediaries. the implementation and monitoring of provisions of World INS, an e-commerce platform, uses DLTs to directly connect Trade Organization agreements relevant to the agricultural producers and consumers through data integration. And trade. DLTs can ensure that gains from trade accrue more AgriDigital, an Australian company, uses blockchain- directly to producers and consumers.a Meanwhile, the food enabled contracts to facilitate interactions among the vari- losses in food systems could be reduced by up to 30 million ous players in the grain supply chain. tons a year if blockchains monitored information in half the To ensure their scalability and accessibility, DLT solutions world’s supply chains.b require the appropriate ecosystems. Although some elements Blockchain technology is still in its infancy, but pilots of such ecosystems are technology-specific, they also largely testing its use are rapidly spreading. One of the most suc- rely on enabling policy, regulatory, and institutional condi- cessful initiatives is the Food Trust consortium run by IBM. tions, as well as basic requirements for infrastructure, literacy It uses blockchain technologies to improve the traceability (including digital), and network coverage.d As one example, of food, and it has brought together large retail and food according to a recent PricewaterhouseCoopers survey,e reg- industry companies from across the world, including Dole, ulatory uncertainty around blockchain-based solutions was Driscoll’s, Golden State Foods, Kroger, and McCormick. As identified as a major scale-up challenge across various sectors. part of this consortium, Carrefour, a supermarket chain in Other major challenges are interoperability and the potential France, uses blockchain technology to provide consumers failure of different blocks within the chain to work together. a. Jouanjean (2019). b. WEF (2018). c. OECD (2019). d. Tripoli and Schmidhuber (2018). e. PwC (2018). are less likely to export or import. A typical modal Digital marketplaces are on the rise, manufacturing firm in the United States has 45 fostering GVC participation—and workers, and larger firms tend to be more productive concentration and pay higher wages and are more likely to export Greater access to (and more extensive use of) broad- and import.12 By contrast, a modal firm in most devel- band Internet and digital-enabled devices would also oping countries has one worker, the owner. Among connect more consumers and firms in low-income firms that do hire additional workers, most hire countries to online markets and business-to-business fewer than 10. In India, Indonesia, and Nigeria, firms platforms. with fewer than 10 workers account for more than Digital marketplaces and online retailers are on the 99 percent of the total. rise. Platforms such as Alibaba, Amazon, eBay, Taobao, Developing countries tend to have a smaller num- and Mercado Libre are becoming an increasingly ber of exporters and a lower concentration of export important interface between global manufacturers revenue in their top exporters, suggesting that these and consumers. At the same time, manufacturers and firms face greater distortions.13 Investments in reduc- traditional retailers are seeking to achieve a stronger ing barriers to competition and minimizing frictions online presence, alongside their standard distribution may thus be especially beneficial for developing channels. Consumers worldwide purchased approx- countries. imately $2.86 trillion in goods and services online in Technological change | 139 Map 6.1  4G network coverage, 2018 % of population with coverage 100 90 70 50 30 6 No data IBRD 44668 | SEPTEMBER 2019 Source: GSMA Intelligence (https://www.gsmaintelligence.com). 2018, up from $2.43 trillion in 2017. The share of online are predominantly from North America and East sales in total retail sales increased from 11.3 percent in Asia; Africa and Latin America are greatly under­ 2016 to 13.3 percent in 2017.14 represented. The role of first-mover advantages in the E-commerce is growing especially rapidly in establishment of platform firms may make it difficult China. The United States and China—the world’s two for Africa, Latin America, and even Europe to bridge largest economies—accounted for more than half of the gap. global e-commerce sales in 2017. China is the largest A limited number of e-commerce platforms domi- e-commerce market, with sales of $877 billion in nate most markets (figure 6.1). Amazon ranks first by 2017, up 28 percent from 2016.15 In China, the share of traffic share in North America, Western Europe, and online sales in total retail sales reached 15 percent in parts of the Middle East and India; Alibaba is the most 2017, up from 12.6 percent in 2016. In the United States, visited site in China and some parts of the Middle consumers spent $449.88 billion on retail sites in 2017, East; and Mercado Libre tops Latin America (map 6.2). up 15.6 percent from 2016; online penetration reached The activities of platform firms are thus highly con- about 13 percent of total retail sales.16 E-commerce centrated among a few large megafirms. sales are likely to continue to rise in developing coun- Platforms enable GVC participation (box 6.2), but tries as Internet access and usage expand. Improve- they may lead to concentration because their busi- ments in enabling infrastructure, such as e-payment ness model relies on building and exploiting network systems, logistics, third-party authenticators, and dis- effects. They reduce transaction costs and help verify pute resolution support services can further augment the quality and reputation of suppliers and match e-commerce. them to potential foreign buyers.17 One study finds Platform firms have emerged as the largest com- that the extent to which distance reduces trade is panies in the world, but geographically they are not 65 percent smaller for eBay than for total trade flows distributed evenly. Seven of the 10 largest global com- (for the same set of goods and countries).18 Although panies by market capitalization in the first quarter platform firms offer opportunities for new actors to of 2019 were platform firms, up from only three in connect and integrate into GVCs, the mechanisms 2015 and two in 2011 (table 6.1). These platform firms that they typically use to overcome information 140 | World Development Report 2020 Table 6.1  Ten largest global companies, by market capitalization, 2011, 2015, and 2019 Market value Year Ranking Company Country (US$, billions) 2019 1 Apple United States 961.3 2 Microsoft United States 946.5 3 Amazon United States 916.1 4 Alphabet United States 863.2 5 Berkshire Hathaway United States 516.4 6 Facebook United States 512.0 7 Alibaba China 480.8 8 Tencent Holdings China 472.1 9 JPMorgan Chase United States 368.5 10 Johnson & Johnson United States 366.2 2015 1 Apple United States 724.8 2 ExxonMobil United States 356.5 3 Berkshire Hathaway United States 356.5 4 Google United States 345.8 5 Microsoft United States 333.5 6 PetroChina China 329.7 7 Wells Fargo United States 279.9 8 Johnson & Johnson United States 279.7 9 Industrial and Commercial Bank of China China 275.4 10 Novartis Switzerland 267.9 2011 1 ExxonMobil United States 417.2 2 PetroChina China 326.2 3 Apple United States 321.1 4 Industrial and Commercial Bank of China China 251.1 5 Petrobras Brazil 247.4 6 BHP Billiton Australia/United Kingdom 247.1 7 China Construction Bank China 232.6 8 Royal Dutch Shell United Kingdom 228.1 9 Chevron United States 215.8 10 Microsoft United States 213.3 Sources: Financial Times Top 500 Companies (https://www.ft.com/ft500); Forbes Global 2000: The 2019 World’s Largest Public Companies (https://www .forbes.com/global2000/). Note: The table lists the top 10 global companies by market capitalization for 2011, 2015, and 2019. Over time, platform firms (shown in bold) have become progressively more important. frictions, such as consumer ratings that help firms revolution in prediction capabilities, with potentially establish a credible reputation, tend to favor concen- broad implications for transaction costs both within tration. Although platforms enable small and medium and across countries. Enabling this transformation enterprises to penetrate export markets, they also are the greater availability of data, significantly make their reputations more widely visible, favoring improved algorithms, and substantially more power- the emergence of superstar exporters.19 They make it ful computer hardware.20 Large firms, multinational easier, then, to connect, but harder to compete. enterprises, and big online retailers such as Alibaba and Amazon are increasingly relying on big data and Artificial intelligence applications are machine learning to understand and forecast con- facilitating e-commerce sumer behavior and manage their supply chain more GVCs and e-commerce may be further supported by efficiently.21 recent advances in machine learning. The current Machine learning also reduces the linguistic bar- generation of artificial intelligence represents a riers to trade and GVC participation. One application Technological change | 141 Figure 6.1 Large platform companies are concentrated in North America and Asia Facebook Apple Naspers Amazon Google Tencent Alibaba SAP Microsoft Africa Asia Europe North America Source: Peter C. Evans, Global Platform Database, Platform Strategy Institute, 2019. Note: The figure shows the concentration of the world’s 75 largest platform firms by region, with bigger circles representing firms with more market capitalization. Map 6.2 Top e-commerce platforms, by traffic share, 2019 Top site family Alibaba allegro.pl Amazon avito.ru blocket.se daraz.pk dba.dk eBay emap.ro finn.no gumtree.co.za heureke.cz vheureka.sk marktplaats.nl njuskalo.hr sahibinden.com tokopedia.com jofogas.hu Mercado Libre OLX shopee.vn trademe.com.nz kupujemprodajem.com mtvuutiset.fi qoo10.sg skroutz.gr No data IBRD 44667 | SEPTEMBER 2019 Source: Alexa, SimilarWeb (https://www.similarweb.com/website/alexa.com#overview). 142 | World Development Report 2020 Box 6.2  GVC linkages and cross-border connections between people move together To operate effectively, GVCs rely on efficient process- educational and career backgrounds, are part of a net- ing of information. This is the point at which platform work and thereby “linked” to other professionals in other firms enter the picture because they enable other firms firms, sectors, and countries. Analysis of the LinkedIn to connect and communicate as well as encourage the data (figure B6.2.1) reveals that exports (panel a) and formation of new linkages. Professional networks enable both backward and forward GVC participation (panels the operation of GVCs. To explore the linkages between b and c, respectively) are strongly correlated with the networks and trade, the World Bank has partnered with number of foreign connections indicated by members of LinkedIn, a professional platform with more than 630 LinkedIn. Although causality is more difficult to estab- million members in over 200 countries and territories. lish, these patterns suggest that professional networks Members of LinkedIn, who provide information on their are complementary to the expansion of GVCs. Figure B6.2.1  Relationship of exports and GVC participation to online foreign connections a. Exports 10.0 9.5 9.0 8.5 8.0 Exports 7.5 7.0 6.5 6.0 5.5 5.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 Foreign connections b. Backward GVC participation c. Forward GVC participation 8.0 Backward GVC participation 7 Forward GVC participation 7.5 7.0 6.5 6 6.0 5 5.5 5.0 4 4.5 4.0 3 3.5 3.0 2 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 10.0 10.5 11.0 11.5 12.0 12.5 13.0 13.5 14.0 14.5 15.0 Foreign connections Foreign connections Sources: World Bank Group–LinkedIn Digital Data for Development, Jobs, Skills, and Migration; OECD’s TiVA database. See appendix A for a description of the databases used in this Report. Note: The graphs show the correlation between the three GVC measures and the foreign connections of members of LinkedIn. The y-axis is based on data from the TiVA data set of the Organisation for Economic Co-operation and Development (OECD) at the sector level (36 sectors) for 64 countries. The variables are the natural log of total exports in millions of U.S. dollars (panel a) and backward and forward participation in GVCs (panels b and c, respectively), also measured in logs of millions of U.S. dollars. The x-axis data are from the Economic Graph at LinkedIn (https://economicgraph.linkedin .com/), showing the natural log of the total number of foreign connections in a given sector in the same 64 countries for 2015–18. Each point in the scatterplot represents the mean of the y-axis variable in each of the 100 chosen bins of the x-axis data. The diagonal line represents the prediction of the dependent variable, calculated using a linear regression with additional country and sector fixed effects. Therefore, its slope represents the elasticity between the y-axis and x-axis measures. Technological change | 143 Figure 6.2  From 2013 to 2015, U.S. exports to Latin America through eBay increased after the introduction of machine translation a. U.S. exports on eBay, Latin America b. U.S. exports to Latin America, and non-Latin America online and offline 1.4 1.4 1.2 1.2 Exports Exports 1.0 1.0 0.8 0.8 0.6 0.6 May November June December July May November June December July 2013 2013 2014 2014 2015 2013 2013 2014 2014 2015 Latin America Non-Latin America eBay Offline Source: Brynjolfsson, Hui, and Liu 2018. Note: Exports in panel a are measured in quantity and normalized to the level in April 2013. Exports in panel b are measured in U.S. dollars and normalized to the level in April 2013. The red vertical line marks the introduction of query translation, and the aqua vertical line marks the introduction of item title translation. of machine learning—machine translation—has also observed in many other developing countries. improved in recent years. For example, the best score Although most of this growth has so far been observed at the Workshop on Machine Translation for English in urban areas, emerging economies such as China, to German rose from 15.7 to 28.3, according to a widely the Arab Republic of Egypt, India, and Vietnam are used comparison metric, the BLEU score.22 The intro- developing policies aimed at expanding e-commerce duction of machine translation from English to Span- to rural areas. But such expansion requires more than ish by eBay has significantly boosted international Internet access alone. It also means overcoming logis- trade between the United States and Latin America on tical and transactional barriers, such as the dearth of this platform, increasing exports by 17.5 percent (figure modern commercial parcel deliveries and rural house- 6.2). These effects reflect a reduction in translation- holds’ lack of familiarity with how to navigate online related search costs and show that artificial intelli- platforms and lack of access to (or trust in) online gence has already begun to boost trade in North and payment services. The sizable welfare gains from South America. The results further suggest that con- e-commerce stem predominantly from reductions in sumers benefit more than sellers because consumers consumer prices and access to new products. In Japan, gain both from reduced language frictions and lower e-commerce has driven down overall prices, raising prices. Although the evidence refers to online trade, aggregate welfare by 1 percent. Meanwhile, new vari- machine translation may also facilitate communica- eties available through online shopping have raised tion offline—for example, within multinational firms welfare by 0.7 percent, and increased intercity price or across trading partners. arbitrage has raised welfare by 0.06 percent.23 The gains from e-commerce are unevenly dis- Platform firms and e-commerce have tributed across households. A recent study looked at uneven benefits the effects of a program that invests in the logistics Besides fueling GVCs and cross-border trade, deeper needed to ship products to and sell products from integration of e-commerce may also help it reach more tens of thousands of Chinese villages that were largely firms and households in rural markets in developing unconnected to e-commerce.24 Between the end of countries. In China, the largest e-commerce market, 2014 and middle of 2016, nearly 16,500 villages in 333 the number of people buying and selling products counties and 27 provinces in China were connected online grew from essentially zero in 2000 to more to e-commerce through the program. The sizable than 400 million in 2015. A clear upward trend was gains from e-commerce trading in both number of 144 | World Development Report 2020 Figure 6.3  Effects of an e-commerce program on the number of buyers and online transactions in Chinese villages a. Number of buyers b. Number of online transactions 100 350 300 Number of transactions 80 Number of buyers 250 60 200 40 150 100 20 50 0 0 –2 0 2 4 6 8 10 12 14 16 18 20 22 24 –2 0 2 4 6 8 10 12 14 16 18 20 22 24 Number of months before and after program entry Number of months before and after program entry Source: Couture et al. 2018. Note: The figure shows point estimates from a regression of depicted outcomes on months since program entry with village and month fixed effects. Outcomes are the number of buyers (panel a) and the number of online transactions (panel b). The data are from a major e-commerce firm’s internal database and contain the universe of village purchase transactions from November 2015 to April 2017 in five provinces: Anhui, Guangxi, Guizhou, Henan, and Yunnan (roughly 11,900 villages in total). The last point estimate of each plot pools months 24–28. The graphs show 95 percent confidence intervals based on standard errors that are clustered at the village level. Overall, the figure indicates that the introduction of e-commerce was associated with an increase in both the number of buyers and the number of online transactions. buyers and number of online transactions (figure 6.3) the scope for raising consumer prices appears to be have, however, tended to accrue to a minority of fairly limited. Online platforms still account for a rural households who are younger, richer, and better fairly small share of the overall retail market. Recent positioned to take advantage of the opportunities evidence points to strong substitution between online e-commerce offers. Importantly, the gains have been and offline sales for personal computers, news, and significantly stronger among villages not previously advertising.26 Meanwhile, services such as Google serviced by commercial parcel delivery, suggesting Shopping facilitate price comparisons across online that the impacts of the program stem mainly from merchants and marketplaces, many of which are still overcoming a logistical barrier rather than from addi- in their infancy. tional investments aimed at adapting e-commerce to The interdependencies between platforms’ third- transactional barriers specific to rural households. party sales for retailers and their own online retail E-commerce benefits consumers by reducing the operations can result in potential conflicts of interest cost of living, especially in remote rural areas. On the and may enable anticompetitive conduct.27 Hybrid income side, e-commerce has displacement effects. platforms such as Amazon, JD.com, and Flipkart sell In the United States, the growth of e-commerce from their own inventory and also act as an online market- 3.8 percent of retail sales in 2010 to 8.3 percent in 2017 place for other retailers to sell their products, taking was associated with a reduction in employment in a commission for each order. Operating as both an brick-and-mortar retail stores. In counties with retail upstream intermediation market for other firms and a fulfilment centers, the labor income of retail workers downstream retail market for its final customers may fell by 2.4 percent after the establishment of such a give rise to conflicts of interest. Online shoppers may center, with both younger and older workers experi- not be able to tell the difference between a platform’s encing sharper decreases in labor income.25 Consump- own retail services and its marketplace activities for tion gains thus come at the expense of labor market other merchants. Moreover, hybrid platforms may use adjustments. the data they collect while operating as a marketplace to identify successful products in the marketplace so Platforms create new regulatory that they can then market their own branded version challenges in the same platform. As platform firms grow, gain access to more private Another, more traditional, form of potential abuse data, and wield market power, so do concerns about is predatory pricing, whereby platforms use their priv- anticompetitive behavior. At least for now, however, ileged access to third-party data to temporarily charge Technological change | 145 prices below cost on their own products to gain a per- digital-intensive sectors than in others, with the growth manent competitive edge over other merchants. The driven by firms at the top end of the distribution. These concern is not that platforms offer their own products superstar firms are thus accounting for a higher share at a lower price than that offered by the original seller, of profits, which increasingly are unevenly divided. thereby benefiting consumers. It is that hybrid plat- forms may be able to offer such prices only because of their use of third-party data. They could then adopt New products temporary pricing strategies to gain more permanent Since the 1990s, many new types of products have advantages over their competitors and subsequently entered global trade, primarily intermediate goods, raise prices. At the same time, it is important to rec- further demonstrating the increasing fragmentation ognize that pricing structures are complex. Subsidies of production and the emergence of entirely new across users can help a platform increase its volume of products (figure 6.4). Indeed, the trade in new prod- transactions and benefits. In other words, a platform ucts has grown dramatically. In 2017, 65 percent of can charge prices below marginal cost to some par- trade was in categories that either did not exist in 1992 ticipants, which does not necessarily mean that it is or were modified to better reflect changes in trade. engaged in predatory pricing. Alternatively, charging Trade in intermediate goods (parts and components prices above marginal cost to other participants does and semifinished goods) expanded, and entirely new not necessarily mean market power is at work. products entered global trade. For example, trade in IT Concerns about anticompetitive behavior are not products tripled over the past two decades, as trade in unique to platform firms. Markups have been rising digitizable goods such as CDs, books, and newspapers in many sectors of the economy, and especially so in steadfastly declined from 2.7 percent of the total goods digital-intensive sectors.28 The average U.S. markups trade in 2000 to 0.8 percent in 2018.29 Technological have risen from 18 percent above marginal cost in developments are likely to continue to produce prod- the 1980s to the present 67 percent. Similar trends in uct churning. markups have been documented in other countries. Because of technological progress, more goods and According to OECD, markups have grown more in services are likely to become tradable over time. For Figure 6.4  Globally, the number and trade share of new products increased from 1996 to 2017 a. New products, by share of HS codes b. New products, by share of total goods trade 30 70 Share in number of HS six-digit codes (%) 60 25 Share in total goods trade (%) 50 20 40 15 30 10 20 5 10 0 0 1996 2002 2007 2012 2017 1996 2002 2007 2012 2017 Capital goods Consumption goods Parts and components Semifinished goods Primary goods Other Source: UN Comtrade (International Trade Statistics, Import/Export Data). Note: Products are classified by a Harmonized System (HS) six-digit code. New products are classified relative to the set of products in the first HS classification in 1988/1992. New codes are either genuinely new products, or old product codes that split into two new codes, or two old codes that merged into one new code. Products are further classified as final (consumption and capital), intermediate (parts and components and semifinished), or primary and other goods using the Broad Economic Categories revision 4 classification from the United Nations Conference on Trade and Development. The figure shows that over time trade in new products has grown dramatically. 146 | World Development Report 2020 example, platforms such as Upwork and Mechanical Figure 6.5  Robot adoption is greater in high-income Turk make it easier for businesses to outsource tasks countries and in sectors in which tasks are easily to workers who can perform them virtually. And new automated goods and services are likely to be developed, includ- a. Robot adoption and income per capita ing ones not even imaginable today, thereby boosting 1.4 the incentives to trade. DEU 1.2 BEL DNK FIN SWE Automation anxiety 1.0 ITA NLD FRA Robotization USA Robotization is on the rise, raising 0.8 SVN concerns about the future of GVCs ESP The spread of new production technologies, such as 0.6 HUN GBR CZE PRT advanced robotics and 3D printing, has raised con- 0.4 IRL cerns about the future of trade and of GVCs. Robotics SVK technology, having advanced greatly in the last two 0.2 POL BGR ROU LVA GRC decades, is predicted to develop further in the coming HRV EST 0 years. The average price of an industrial robot has 0 10,000 20,000 30,000 40,000 50,000 60,000 fallen by half in real terms and even more relative to Initial GDP per capita (constant 2010 US$) labor costs. Global sales of industrial robots reached a record 387,000 units in 2017, up 31 percent from b. Robot adoption and feasibility of automation across sectors 2016. Figure 6.5 shows that robotization is higher in 2.5 countries with higher income per capita, where wages Automotive are higher, and in sectors in which robotization is feasible. Robots are used predominantly in high-wage 2.0 countries in Asia, North America, and Western Europe Rubber and plastics (panel a). In recent years, China saw the largest growth in demand for industrial robots and was projected to Robotization 1.5 have the largest operational stock of robots by the end of 2018, but still relatively low robot density.30 Roboti- Electronics Metal 1.0 zation is most pronounced in the automotive, rubber Machinery and plastics, metals, and electronics sectors, reflecting Education Chemicals Food products differences in the feasibility of automation (panel b). It Agriculture 0.5 Other manufacturing is still limited in traditionally labor-intensive sectors Other nonmanufacturing Wood, paper, and printing such as textiles, suggesting that export-led industri- Mining Textiles alization in these sectors is still a viable development Utilities Construction path. Robot adoption is projected to increase greatly 0 10 20 30 40 50 60 70 over the coming decade, reflecting further reductions Share of replaceable jobs (%) in quality-adjusted robot prices.31 Source: Artuc, Bastos, and Rijkers 2018. Modern industrial robots can be programmed to Note: Robotization is the logarithm of 1 plus the ratio of the average stock of robots to the number of perform a variety of repetitive tasks with consistent working hours (in millions) between 1993 and 2015 (or the subsample of years over this period for which precision, and they are increasingly used in a wide robot data from the International Federation of Robotics [IFR] are available). The stock of robots is estimated using the perpetual inventory method based on the observed stock of robots in the IFR data range of industries and applications. If tasks previ- and using a depreciation rate of 10 percent. The share of jobs that is potentially replaceable by robots ously performed by low-skilled workers in the South is based on the task makeup of the job. See Artuc, Bastos, and Rijkers (2018) for a detailed explanation of how replaceability is measured. For country abbreviations, see International Organization for (low-wage developing countries) are performed by Standardization (ISO), https://www.iso.org/obp/ui/#search. relatively inexpensive robots in the North (industrial countries), there may be a reversal in North–South trade flows and a greater reliance on domestic pro- change as well, with low labor costs becoming a less duction. Moreover, the skill and capital content of important determinant of competitiveness (at least in inputs that countries in the North demand from sectors in which automation is feasible), and comple- the South may increase now that the North can use mentary factors, such as the availability of skills and robots and other technologies more intensively, as sound infrastructure, becoming more important.33 discussed in more depth shortly.32 The criteria for Although the risk of displacement of jobs or exports becoming an attractive production location may currently seems low, middle-income countries such Technological change | 147 Map 6.3  A substantial share of exports from developing countries is in goods that can be produced by robots % of total exports to OECD countries 29.7–53.7 17.0–29.7 11.5–17.0 6.7–11.5 0–6.7 No data IBRD 44671 | SEPTEMBER 2019 Source: WDR 2020 team, based on Artuc, Bastos, and Rijkers (2018). Note: The map shows exports (by quintile) as a percentage of total exports to high-income OECD (Organisation for Economic Co-operation and Development) countries, weighted by the share of jobs in sectors that produce the exported goods that are potentially replaceable by robots based on their task makeup. See Artuc, Bastos, and Rijkers (2018) for a detailed explanation of how replaceability is measured. as Mexico, Tunisia, and Pakistan would seem most produced in 2017. Adidas’s competitor Nike has several exposed to the threat of robotization-induced reshor- automated platforms under development. ing because their exports are heavily concentrated in goods that robots can help produce (map 6.3). Com- Robotization and 3D printing have modity exporters, however, seem somewhat shielded promoted North–South trade with from the threat of robotization-induced reshoring. heterogeneous impacts across countries The advent of 3D printing led to predictions that Despite the concerns about the effects of automati- many goods would be printed locally, shortening zation, the evidence that reshoring will result is very GVCs and limiting trade. The concern is that if 3D limited.34 Moreover, these technologies may enhance printing becomes cheap, then firms capable of creat- GVCs and boost trade. The spread of automation ing a solid 3D object from a digital file will prefer to 3D in richer countries can improve productivity and print products at home rather than import them. 3D income, thereby raising the demand for inputs and printers may therefore perform the tasks previously final goods from countries with large pools of low- performed by workers engaged in production and wage labor as a comparative advantage. Furthermore, assembly activities located abroad. developed countries with similar factor endowments These concerns are in part predicated on a few and technologies trade a great deal among themselves. high-profile examples. For example, the sporting Even if the labor advantage of low-income countries is goods manufacturer Adidas recently established two (partially) canceled out by robotization, there will still “speedfactories” in Germany and the United States be opportunities for trade in differentiated goods and that use robots and 3D printing to more quickly pro- for specialization in some stages of production. duce customizable running shoes for high-income Thus far, the rising adoption of industrial robots domestic consumers. Adidas hopes the two factories and 3D printing seems to have promoted North–South can produce 1 million pairs of shoes a year by 2020, trade. Greater robot intensity in production has led to which is still a tiny share of the 403 million pairs it more imports sourced from lower-income countries 148 | World Development Report 2020 in the same broad industry—and to an even stronger Figure 6.6  Automation in industrial countries has increase in gross exports (which embody imported boosted imports from developing countries inputs) to those countries. The surge in imports from 10 the South has been concentrated in intermediate goods % change in imports of parts from developing countries such as parts and components. The positive impact of 8 automation on imports, particularly on imports of intermediates, attests to the importance of examining 6 the effects of robotization on trade through a GVC 4 framework. More-traditional trade models would pre- dict the increase in exports by the North but fail to fore- 2 see the surge in imports from the South in the same industry.35 Rather than reducing North–South trade, 0 robotization seems to have been boosting it, although str es ric on uc e M n g m pr s Fo fac ng he ts a c ls ec ry cs l to ics e g pr ing d eta d tile Ed ltur iv io in in M ica C duc El ine ni on iliti Ag ucti ti Au last ot at in ur od tur it is uncertain whether this trend is likely to continue. an M an in th an Tex tro u m m h t o ct U p a uf u These average effects mask heterogeneity across an C nm er O r, countries and sectors (figure 6.6). The biggest pe bb er no pa Ru automation-induced increase in trade has been in er d, th oo O W the quick-to-automate automotive sector. Countries already supplying inputs to automating producers Industries in order of increasing automation in the North are well positioned to benefit from the Source: Artuc, Bastos, and Rijkers 2018. higher demand for their exports. But countries directly Note: The figure depicts the automation-induced increase in imports of parts by developed countries (North) from developing countries (South) by broad sector from 1995 to 2015. The change in imports competing with them in output markets could lose of parts is measured in log points; a 0.10 increase in log points is roughly equivalent to a 10 percent export revenue and manufacturing employment if increase in imports. their workers are outcompeted by foreign robots. The negative effects of reduced manufacturing employ- Automation is compressing labor’s ment could outweigh the welfare gains associated with income share but not necessarily reducing the lower import prices resulting from automation in employment the North, at least in the short run. But these countries As automation improves productivity, it also com- might benefit from automation-induced increases in presses labor’s share of income in advanced economies. global productivity and income, which could translate Higher robot density at the industry level is associated into more exports and activity in sectors where they with a lower labor share of income, defined as total retain a comparative advantage. labor compensation over sales (figure 6.8). This pattern A related dynamic of innovation-induced trade has implications for inequality because it suggests that can be observed in goods that can be produced using the primary beneficiaries of automation are the own- 3D printers, such as hearing aids (box 6.3). In 2007 ers of capital. Moreover, technological progress and the hearing aids shifted almost entirely to 3D printing, accompanying cost reductions in the relative price of and trade increased when compared with similar capital goods may be contributing to the global decline goods (figure 6.7). Estimates that take into account in labor’s share of income observed across countries industry growth and the standard determinants of over the past few decades.36 Although the jury is out trade reveal that trade in hearing aids was boosted by on the drivers of this decline, the fall in the labor share 60 percent following the introduction of 3D printing. across sectors is highest in sectors undergoing concen- Other industries producing goods that were partially tration and the emergence of superstar firms. These 3D printed have demonstrated that the technology firms make high profits and typically have a lower has similar positive effects on trade. The results are share of labor in sales and value added, in part because at odds with the view that 3D printing will shorten they are harnessing technological innovations.37 supply chains and reduce trade, at least for this set of Robot adoption among OECD countries has products. The findings do suggest that gains may dis- reduced the employment share of low-skilled workers proportionately accrue to middle- and high-income in robot-intensive industries. Across local labor mar- countries, and thus they serve as a reminder that the kets in Mexico and the United States, workers with gains from the introduction of new production tech- a high exposure to domestic robotization have wit- nologies are likely to be unevenly distributed across nessed a reduction in employment and wages relative countries. to those with more limited exposure.38 Technological change | 149 Box 6.3  Fully automating the production of hearing aids A common refrain is that automating production, such as costs, prices fell by about 25 percent.b Meanwhile, the with 3D printing, will allow companies to produce goods product underwent improvements: 3D printing allowed closer to markets. Companies will drastically shorten their for high levels of customization and cosmetic improvements value chains, which will reduce international trade. Lower- in hearing aids, which reduced discomfort and the stigma income countries will be most affected because their for users. Demand increased and trade expanded. exports are often intermediate products based on abun- There is no evidence that 3D printing shifted the product dant, low-cost labor. One attempt to quantify and predict closer to consumers or displaced trade—the comparative the trade impacts of 3D printing stated it could eliminate advantages of different countries in the hearing aid value as much as 40 percent of trade by 2040.a By contrast, new chain remained the same. Nor does this trend seem to research on the production of, and trade in, hearing aids be exclusive to hearing aids. A preliminary analysis of 35 suggests quite the opposite. other products c that are partially 3D printed found similar Similar to a standard ink printer, 3D printing uses very positive effects on trade, although to a smaller degree. Per- little labor and can generate customized products from the haps 3D printing had not yet been fully adopted for those same machine. In 2007, following a series of inventions in 3D products across the entire industry. Unlike the results of the scanning, software development, and biocompatible mate- hearing aids analysis, the results of this analysis point to a rials, the production of hearing aids shifted almost entirely reshuffling of comparative advantage from labor-abundant to 3D printing. In the decade that followed, trade increased countries to countries that adopted 3D printing technolo- overall by 60 percent, and because of lower p ­ roduction gies for each product. a. Leering (2017). b. Freund, Mulabdic, and Ruta (2018). c. Freund, Mulabdic, and Ruta (2018). Figure 6.7  Trade in hearing aids increased with the Figure 6.8  Higher robot density is adoption of 3D printing in 2007 associated with lower shares of income for labor 500 0.25 450 400 0.23 Exports (2000 = 100) 350 300 0.21 Labor share 250 0.19 200 150 0.17 100 0.15 50 0 0.13 0 1 2 3 4 19 5 19 6 19 7 19 8 20 9 20 0 20 1 20 2 20 3 20 4 20 5 20 6 20 7 20 8 20 9 20 0 20 1 20 2 20 3 20 4 20 5 16 9 9 9 9 9 0 0 0 0 0 0 0 0 0 0 1 1 1 1 1 1 19 Robot density Product category: Fitted Hearing aid 902140 Chapter 90 High-tech High-tech chapter 90 95% confidence interval (upper bound) 95% confidence interval (lower bound) Source: Freund, Mulabdic, and Ruta 2018. Source: Artuc, Bastos, and Rijkers 2018. Note: The Harmonized System (HS) code for hearing aids is 902140. Three additional categories are included for comparison. Chapter 90 covers optical, photographic, cinematographic, measuring, Note: The figure shows the association between labor’s share of income, checking, precision, medical or surgical instruments and apparatus, and parts and accessories thereof. defined as total labor compensation over sales, and robot density, defined as High-tech are other goods similar to hearing aids found both in and outside chapter 90. High-tech the number of robots per million work-hours, for industries in the EU-KLEMS chapter 90 includes high-tech products selected from chapter 90. data set for the period 1993–2015. 150 | World Development Report 2020 Although automation is no doubt causing pain automation ultimately helps or hurts net job creation, in the labor market, it would be incorrect to assume it certainly causes significant, and costly, labor market that because robots replace workers they always adjustments. reduce aggregate employment. Robots are a labor- saving form of technological progress and may directly Automation is changing the demand for displace jobs, but their adoption can in fact spur job skills and comparative advantages creation through three indirect channels that are The intuition behind these findings is that automating challenging to measure. First, the productivity gains tasks that can be performed by robots almost surely in supplier industries can yield steep increases in the raises the economic value of the complementary tasks demand for labor because of input–output linkages, as and thus the demand for laborers to perform them.39 shown earlier. Second, productivity growth can boost Automation may also lead to the creation of new tasks final demand. And, third, adoption of robots may lead and products in which human labor has a compara- to compositional shifts in the structure of the econ- tive advantage both at home and abroad. These forces omy and could create jobs by spurring the growth of give rise to a reinstatement effect, raising the demand sectors with high labor shares. Across member coun- for labor by expanding the set of tasks allocated to tries of the OECD, industry-level productivity growth workers.40 For example, in industrial sectors where has been associated with job losses in the industries in robotization is more prevalent in the United States, which it originates, but these losses have been more low-skilled occupations such as assemblers and pro- than compensated by indirect gains in customers and duction workers experienced sizable job losses over supplier industries and growth in final demand. Since the past decades, while occupations such as sales the early 1970s, aggregate employment in OECD coun- representatives, engineers, and programmers experi- tries has grown, even though relative employment enced strong increases in net employment (figure 6.9). in industries experiencing the fastest growth in pro- Meanwhile, rising incomes due to automation may ductivity has fallen. Although it is not clear whether lead not only to new tasks, but also to new products Figure 6.9  Change in U.S. employment in robot-intensive industries, by occupation, 1990–2010 Other metal and plastic workers Sales representatives, wholesale and manufacturing Other engineering technicians Mechanical engineers Computer control programmers and operators Other engineers Secretaries and administrative assistants Other assemblers and fabricators Electrical, electronics, and electromechanical assemblers Other production workers Other life, physical, and social science technicians Other managers 0 0 0 0 00 00 00 00 00 00 00 ,0 ,0 ,0 ,0 0, 0, 0, 10 20 30 40 –3 –2 –1 Number of workers (per million) Source: WDR 2020 team, based on tabulations of IPUMS-USA data using the 2010 Harmonized Occupation Classification Scheme. Note: Data refer to the automotive, machinery, electronics, rubber and plastics, and metal industries. The figure depicts changes in employment for the five occupations with higher and lower net employment creations. The total number of workers in these sectors is normalized to 1 million per year. Occupations labeled as “Other” refer to those not listed separately. Other metal and plastic workers include electrical discharge machine setup operators, metal rivet machine operators, and tin recovery workers. Other engineering technicians include agricultural, biomedical, metallurgical, and optical engineering technicians. Other engineers include optical, ordinance, photonics, and salvage engineers. Other assemblers and fabricators include air bag builders, crate builders, and doll makers. Other production workers include chemical processing machine setters, operators, and tenders; crushing, grinding, polishing, mixing, and lending workers; and cutting workers. Other life, physical, and social science technicians include meteorological aides and polygraph examiners. Other managers include clerks of court, social science managers, and utilities managers. Technological change | 151 and services. Greater product customization may Changing Nature of Work, which documents how new require tasks that robots cannot perform.41 A glance technologies are changing the demand for skills and back at the U.S. economy reveals that between 1990 the nature of work.42 and 2010 the occupational category “retail salesper- Future automation in developed and emerging son” experienced greater net employment gains in the economies will likely affect worker groups differently, U.S. economy, along with other service occupations and it may exacerbate inequality. Low-skilled workers such as food preparation (which includes restaurant performing repetitive tasks are more likely to be dis- chefs and sandwich makers). These findings align placed by robots. In developing countries, however, with those of the World Development Report 2019: The middle-skill jobs may also be at risk (box 6.4). Women Box 6.4  Mexico and technological change Global value chains link the fates of workers living in dif- Figure B6.4.1  Automation reduces ferent countries because technological progress in one the wage employment of high school country can affect employment in others. graduates Over the last two decades, car manufacturers in Detroit 0.06 % change in employment share have gradually incorporated the use of robots to automate the production of engines, thereby displacing workers. 0.04 Because some of the engine components are produced 0.02 elsewhere within GVCs, workers living thousands of miles 0 from Detroit in cities such as Chihuahua, Mexico, where U.S. companies assemble car parts, are exposed to the threat –0.02 of robotization. In other words, automation in the United –0.04 States could produce unemployment in Mexico by bringing –0.06 jobs back to . . . U.S. robots. –0.08 But the story is not quite so simple: robots have also Change in wage Change in informal increased U.S. productivity, which has led to greater employment employment demand for intermediate and consumer products from Less than high school High school College Mexico and created new jobs for Mexicans (although not Source: WDR 2020 team, based on Artuc, Christiaensen, and Winkler necessarily in Chihuahua). For example, roughly 70 percent (2019). of the electrical wiring components of U.S. cars are currently Note: Figure shows the estimated percentage change in wage employment and informal employment of different skill groups between produced in Mexico, and their production process cannot be 2011 and 2016 that can be attributed to local automation in Mexico. The automated. After automation induces a productivity spike, impact is statistically significant for high school graduates, who constitute a larger share of employment in robotized industries, such as automotive, the demand for electrical wiring produced in Mexico could compared with other industries. be expected to increase. This productivity boost in the U.S. car industry also increases aggregate income and enhances overall demand. Thus the demand for consumer products, linked by large corporations, foreign direct investment, and in addition to car parts, from Mexico expands. In the end, GVCs. The relationship between domestic and foreign firms it is difficult to predict the size and direction of the impact as subsidiaries or as arm’s-length suppliers of parts accel- of high-income country automation on developing country erates transfers of technology and eases access to capital workers operating through international trade channels. in developing countries. And even when different parts Recent evidence indicates that the overall impact of U.S. are produced by different firms, using similar technologies automation on Mexican workers has been negligible. ensures compatibility. Does this mean that Mexican workers have been immune As producers in Mexico have begun to use industrial to the negative distributional effects of robotization? No. robots in the footsteps of their counterparts in the United The use of industrial robots is not limited to high-income States, Mexican workers, like U.S. workers, are beginning countries. In the last 15 years, manufacturers in Mexico to be displaced (figure B6.4.1). However, contrary to spec- have also adopted new automation technologies, but less ulation, the impact has not been through reshoring, but intensively than manufacturers in the United States. Pro- through the diffusion of technological shocks with the duction technologies in Mexico and the United States are global integration of production processes. 152 | World Development Report 2020 also tend to perform more routine tasks than men would be much more expensive to import goods across all sectors and occupations—tasks most prone produced in developed countries using these tech- to automation. Female workers thus face a higher nologies. But this does not mean that protectionism risk of automation than male workers, with signifi- stimulates innovation. Instead, it likely prevents cant heterogeneity across sectors and countries. Less efficiency-enhancing specialization across countries. well-educated older female workers are dispropor- By contrast, by opening up opportunities in new mar- tionately exposed to automation, even though the kets and fostering competition in domestic markets, gender pay gap weakens incentives to automate tasks trade liberalization tends to incentivize competition performed by women, who tend to be paid less than and scale and, by implication, innovation. About men.43 The potentially dis-equalizing effects of auto- 7 percent of the increase in knowledge creation during mation are likely to be compounded by the increase in the 1990s was attributable to trade reforms lowering the relative returns to capital that automation is likely barriers to foreign markets.45 to entail, at least in the short run. Recent firm-level studies point out that interna- This evidence is well aligned with results from tional sourcing strategies could serve as a conduit model-based counterfactual simulations of the impact to innovation. For example, evidence from Denmark of further reductions in robot prices.44 As robot prices suggests that offshoring allows firms to devote a decline, increased automation displaces workers in larger share of their labor force to innovation-related the North in a wider range of tasks, which initially activities, thereby facilitating technological upgrad- depresses wages. Welfare nevertheless increases ing. These findings align with evidence from Norway because the income losses associated with lower labor showing research and development (R&D) and inter- income are more than offset by the higher income national sourcing to be complementary.46 Cheaper from the rental rate of robots and lower consumer access to imported intermediate inputs raises the prices. The adverse impacts of automation on labor returns to R&D. These estimates are also in line with markets may eventually be overturned by further broader cross-country evidence pointing to greater reductions in robot prices. As robot adoption proceeds functional specialization in trade: high-income coun- in the North, production continues to expand and tries tend to specialize in R&D, lower-income coun- may raise the labor demand for the tasks in which tries tend to specialize in fabrication, and specializa- robotization is technologically unfeasible. This situ- tion in management and marketing is unrelated to ation potentially leads to an increase in the demand income.47 for labor and in real wages. Workers in the South may Inflating the costs of international sourcing by benefit from robotization. raising trade protection could thus undermine gains That robot adoption can at times go hand in hand from specialization and stunt productivity growth. with job creation is illustrated by the U.S. automotive Put differently, openness stimulates innovation. The industry, which in recent decades has adopted more positive impacts of trade openness on technological robots than any other sector in the United States, both progress are an often overlooked source of gains from in absolute terms and per worker. From 2010 until trade. 2016, the operational stock of U.S. robots in the auto- motive sector rose by 52,000 units. At the same time, the number of jobs increased by 260,600, according to Export-led industrialization the Bureau of Labor Statistics, partly recovering from Although predicting the future is a treacherous exer- the steady decline in the previous decade. cise, new technologies will likely reduce trade costs and make it easier to participate in global markets. Such outcomes may offer developing countries new Openness and innovation opportunities to link into GVCs. However, the atten- How are these patterns affected by trade policy? dant intensification of competition may make it Inflating trade costs by, for example, imposing tar- more challenging for countries to succeed. Platform iffs will not only diminish trade, but also influence firms, for example, are making it easier to connect, patterns of technology adoption. Model simulations but their reputation mechanisms for verifying sup- suggest that developing countries may themselves be plier quality tend to foster concentration and make more likely to adopt labor-saving technologies when it harder for entrants to grow. They are creating new trade costs are high. They would then be somewhat challenges for regulators both because they wield shielded from foreign competition in sectors where market power and because their interactions with these technologies are used more intensively as it agents in different parts of the value chain may Technological change | 153 create potential conflicts of interest and enhance the 17.  See Chen and Wu (2018) and Garicano and Kaplan (2001). scope for anticompetitive conduct. 18.  Lendle et al. (2016). Automation anxiety is not warranted for all devel- 19.  Chen and Wu (2018). 20.  Agrawal, Lacetera, and Lyons (2016); Brynjolfsson and oping countries. Although some countries are likely to McAfee (2017); Mullainathan and Spiess (2017). lose manufacturing employment because of greater 21. Columbus (2018). competition in output markets, countries that are part 22.  Brynjolfsson, Hui, and Liu (2018), citing http://matrix of GVCs and supplying inputs to other countries that .statmt.org/matrix. are automating may see an increase in the demand 23.  Jo, Matsumura, and Weinstein (2019). for their goods, and consumers everywhere will enjoy 24.  Couture et al. (2018). lower prices. The primary challenge arising from new 25.  Chava et al. (2018). production technologies is to ensure that the bene- 26.  Duch-Brown, Martens, and Mueller-Langer (2017); Gertner and Stillman (2001); Goldfarb and Tucker (2011); fits are shared and that losers are compensated both Goolsbee (2001); Prince (2007); Seamans and Zhu (2014). across and within countries. Among the countries 27.  Höppner and Westerhoff (2018). adopting these technologies, labor market disruptions 28.  Calligaris, Criscuolo, and Marcolin (2018). are likely to be significant, skill premiums are likely to 29. WTO (2018). rise, and labor’s share of income may decline further. 30. IFR (2018). These outcomes point to the importance of sound 31.  A study by the Boston Consulting Group predicts that social safety nets and redistributive and tax policies growth in installed robotic systems will rise from the current 3 percent annually to around 10 percent annually to ensure that gains are widely shared without dis- in the next decade (BCG 2016). torting incentives to innovate. These policies will be 32.  See Rodrik (2018) for a detailed discussion of this discussed in chapter 8. “technological-compatibility” channel. 33.  Hallward-Driemeier and Nayyar (2018). 34.  Oldenski (2015) provides evidence that reshoring is not Notes widespread in the United States. 1. Artuc, Bastos, and Rijkers (2018); Bown et al. (2017); 35.  Artuc, Bastos, and Rijkers (2018). de la Torre et al. (2015); Dutz (2018); Lopez-Acevedo, 36.  Karabarbounis and Neiman (2014). Medvedev, and Palmade (2017). 37.  Autor et al. (2017). 2. In lower-middle-income countries, the share of firms 38.  Acemoglu and Restrepo (2018); Artuc, Christiaensen, and (with at least five employees) using broadband Inter- Winkler (2019); Graetz and Michaels (2018). net rose from 39 percent in 2006–09 to 68 percent in 39.  World Bank (2019). 2010–14. According to the World Development Report 2016: 40.  Acemoglu and Restrepo (2019). Digital Dividends, the share in low-income countries in 41.  For example, in 2016 the car manufacturer Mercedes- 2010–14 was about 38 percent (World Bank 2016). Benz decided to replace some of its assembly-line robots 3. Antràs, Garicano, and Rossi-Hansberg (2008); Costinot, with more capable humans at its Sindelfingen plant Vogel, and Wang (2013); Fort (2017); Freund and Wein- in Germany. The wide variety of options for the cars hold (2002, 2004); Grossman and Rossi-Hansberg demands adaptability and flexibility, two attributes for (2008). which humans currently outperform robots. Skilled 4. See, for example, Hjort and Poulsen (2019). humans can change a production line in a weekend, 5. Fernandes et al. (2019). whereas weeks are required to reprogram and realign 6. Carballo et al. (2016). robots. 7. Laajaj, Eslava, and Kinda (2019). 42. World Bank (2019). 8. The McKinsey Global Institute argues that traditional 43.  Brussevich et al. (2018). trade statistics do not duly account for the rising 44.  Artuc, Bastos, and Rijkers (2018). importance of trade in services, notably by under­ 45.  Coelli, Moxnes, and Ulltveit-Moe (2016). estimating (1) the services embodied in goods; (2) the 46. Bøler, Moxnes, and Ulltveit-Moe (2015). intangibles sent by firms to foreign affiliates; and Timmer, Miroudot, and de Vries (2019). 47.  (3) the proliferation of free digital services made avail- able to global users (MGI 2019). 9. MGI (2019). References 10. Hoekman and Nicita (2008). Acemoglu, Daron, and Pascual Restrepo. 2018. “Low-Skill and 11. See Atkin and Donaldson (2015). High-Skill Automation.” Journal of Human Capital 12 (2): 12. Hsieh and Klenow (2014); Hsieh and Olken (2014); 204–32. McKenzie (2017). ——— —. 2019 “Automation and New Tasks: How Technology 13. Fernandes, Freund, and Pierola (2016). Displaces and Reinstates Labor.” Journal of Economic 14. Young (2019). Perspectives 33 (2): 3–30. 15. Young (2019). Agrawal, Ajay, Nicola Lacetera, and Elizabeth Lyons. 2016. 16. Young (2019). “Does Standardized Information in Online Markets 154 | World Development Report 2020 Disproportionately Benefit Job Applicants from Less Chava, Sudheer, Alexander Oettl, Manpreet Singh, and Developed Countries?” Journal of International Economics Linghang Zeng. 2018. “The Dark Side of Technological 103 (November): 1–12. Progress? Impact of E-Commerce on Employees at Antràs, Pol, Luis Garicano, and Esteban Rossi-Hansberg. Brick-and-Mortar Retailers.” Paper presented at Northern 2008. “Organizing Offshoring: Middle Managers and Finance Association’s 30th Annual Conference, Charle­ Communication Costs.” In The Organization of Firms in a voix, Quebec, September 21–23. Global Economy, edited by Elhanan Helpman, Dalia Marin, Chen, Maggie Xiaoyang, and Min Wu. 2018. “The Value of and Thierry Verdier, 311–40. Cambridge, MA: Harvard Reputation in Trade: Evidence from Alibaba.” Paper pre- University Press. sented at Workshop on Trade and the Chinese Economy, Artuc, Erhan, Paulo S. R. Bastos, and Bob Rijkers. 2018. King Center on Global Development, Stanford University, “Robots, Tasks, and Trade.” Policy Research Working Stanford, CA, April 12–13. Paper 8674, World Bank, Washington, DC. Coelli, Federica, Andreas Moxnes, and Karen Helene Artuc, Erhan, Luc Christiaensen, and Hernán Winkler. 2019. Ulltveit-Moe. 2016. “Better, Faster, Stronger: Global “Does Automation in Rich Countries Hurt Develop- Innovation and Trade Liberalization.” NBER Working ing Ones? Evidence from the U.S. and Mexico.” Policy Paper 22647 (September), National Bureau of Economic Research Working Paper 8741, World Bank, Washington, Research, Cambridge, MA. DC. Columbus, Louis. 2018. “10 Ways Machine Learning Is Revo- Atkin, David, and Dave Donaldson. 2015. “Who’s Getting lutionizing Supply Chain Management.” Forbes, June 11. Globalized? The Size and Implications of Intra-national https://www.forbes.com/sites/louiscolumbus/2018/06/11 Trade Costs.” NBER Working Paper 21439 (July), National /10-ways-machine-learning-is-revolutionizing-supply Bureau of Economic Research, Cambridge, MA. -chain-management/#360c6d223e37. Autor, David H., David Dorn, Lawrence F. Katz, Christina Costinot, Arnaud, Jonathan Vogel, and Su Wang. 2013. “An Patterson, and John Van Reenen. 2017. “The Fall of the Elementary Theory of Global Supply Chains.” Review of Economic Studies 80 (1): 109–44. Labor Share and the Rise of Superstar Firms.” NBER Couture, Victor, Benjamin Faber, Yizhen Gu, and Lizhi Liu. Working Paper 23396, National Bureau of Economic 2018. “E-Commerce Integration and Economic Devel- Research, Cambridge, MA. opment: Evidence from China.” NBER Working Paper BCG (Boston Consulting Group). 2016. “Inside OPS: Are Your 24384 (August), National Bureau of Economic Research, Operations Ready for a Digital Revolution?” July, BCG, Cambridge, MA. Boston. de la Torre, Augusto, Tatiana Didier, Alain Ize, Daniel Bøler, Esther Ann, Andreas Moxnes, and Karen Helene Lederman, and Sergio L. Schmukler. 2015. Latin America Ulltveit-Moe. 2015. “R&D, International Sourcing, and the and the Rising South: Changing World, Changing Priorities. Joint Impact on Firm Performance.” American Economic Latin American and Caribbean Studies Series. Washing- Review 105 (12): 3704–39. ton, DC: World Bank. Bown, Chad P., Daniel Lederman, Samuel Pienknagura, Duch-Brown, Nestor, Bertin Martens, and Frank Mueller- and Raymond Robertson. 2017. Better Neighbors: Toward Langer. 2017. “The Economics of Ownership, Access, and a Renewal of Economic Integration in Latin America. Latin Trade in Digital Data.” JRC Digital Economy Working American and Caribbean Studies Series. Washington, DC: Paper 2017–01, Joint Research Center, European Commis- World Bank. sion, Seville, Spain. Brussevich, Mariya, Era Dabla-Norris, Christine Kamunge, Dutz, Mark A. 2018. Jobs and Growth: Brazil’s Productivity Pooja Karnane, Salma Khalid, and Kalpana Kochhar. Agenda. International Development in Focus Series. 2018. “Gender, Technology, and the Future of Work.” IMF Washington, DC: World Bank. Staff Discussion Note SDN/18/07 (October), International Fernandes, Ana Margarida, Caroline L. Freund, and Martha Monetary Fund, Washington, DC. Denisse Pierola. 2016. “Exporter Behavior, Country Size Brynjolfsson, Erik, Xiang Hui, and Meng Liu. 2018. “Does and Stage of Development: Evidence from the Exporter Machine Translation Affect International Trade? Evi- Dynamics Database.” Journal of Development Economics 119 dence from a Large Digital Platform.” NBER Working (March): 121–37. Paper 24917 (August), National Bureau of Economic Fernandes, Ana Margarida, Aaditya Mattoo, Huy Le Nguyen, Research, Cambridge, MA. and Marc Tobias Schiffbauer. 2019. “The Internet and Brynjolfsson, Erik, and Andrew McAfee. 2017. “The Business Chinese Exports in the Pre-Ali Baba Era.” Journal of of Artificial Intelligence: What It Can—and Cannot—Do Development Economics 138 (May): 57–76. for Your Organization.” Harvard Business Review (July 18). Fort, Teresa C. 2017. “Technology and Production Fragmen- Calligaris, Sara, Chiara Criscuolo, and Luca Marcolin. 2018. tation: Domestic versus Foreign Sourcing.” Review of “Mark-Ups in the Digital Era.” OECD Science, Technology, Economic Studies 84 (2): 650–87. and Industry Working Paper 2018/10, Organisation for Freund, Caroline L., Alen Mulabdic, and Michele Ruta. 2018. Economic Co-operation and Development, Paris. “Is 3D Printing a Threat to Global Trade? The Trade Carballo, Jerónimo, Alejandro Graziano, Georg Schaur, and Effects You Didn’t Hear About.” Working paper, World Christian Volpe Martincus. 2016. “The Border Labyrinth: Bank, Washington, DC. Information Technologies and Trade in the Presence of Freund, Caroline L., and Diana Weinhold. 2002. “The Internet Multiple Agencies.” Working Paper IDB-WP-706 (June), and International Trade in Services.” American Economic Inter-American Development Bank, Washington, DC. Review 92 (2): 236–40. Technological change | 155 ————. 2004. “The Effect of the Internet on International Colombia.” Documento CEDE 8 (February), Center for Trade.” Journal of International Economics 62 (1): 171–89. Economic Development Studies, Department of Econom- Garicano, Luis, and Steven N. Kaplan. 2001. “The Effects ics, Universidad de Los Andes, Bogotá, Colombia. of Business-to-Business E-Commerce on Transaction Leering, Raoul. 2017. “3D Printing: A Threat to Global Trade.” Costs.” Journal of Industrial Economics 49 (4): 463–85. September 28, Economic and Financial Analysis Division, Gertner, Robert H., and Robert S. Stillman. 2001. “Vertical ING Bank NV, Amsterdam. Integration and Internet Strategies in the Apparel Indus- Lendle, Andreas, Marcelo Olarrega, Simon Schropp, and try.” Journal of Industrial Economics 49 (4): 417–40. Pierre-Louis Vézina. 2016. “There Goes Gravity: eBay and Goldfarb, Avi, and Catherine E. Tucker. 2011. “Privacy Regu- the Death of Distance.” Economic Journal 126 (591): 406–41. lation and Online Advertising.” Management Science 57 (1): Lopez-Acevedo, Gladys, Denis Medvedev, and Vincent Pal- 57–71. made, eds. 2017. South Asia’s Turn: Policies to Boost Compet- Goolsbee, Austan. 2001. “Competition in the Computer Indus- itiveness and Create the Next Export Powerhouse. South Asia try: Online versus Retail.” Journal of Industrial Economics 49 Development Matters Series. Washington, DC: World (4): 487–99. Bank. Graetz, Georg, and Guy Michaels. 2018. “Robots at Work.” McKenzie, David. 2017. “Identifying and Spurring High- Review of Economics and Statistics 100 (5): 753–68. Growth Entrepreneurship: Experimental Evidence from Grossman, Gene M., and Esteban Rossi-Hansberg. 2008. a Business Plan Competition.” American Economic Review “Trading Tasks: A Simple Theory of Offshoring.” American 107 (8): 2278–2307. Economic Review 98 (5): 1978–97. MGI (McKinsey Global Institute). 2019. Globalization in Hallward-Driemeier, Mary, and Gaurav Nayyar. 2018. Trouble Transition: The Future of Trade and Value Chains. New York: in the Making? The Future of Manufacturing-Led Development. McKinsey. Washington, DC: World Bank. Mullainathan, Sendhil, and Jann Spiess. 2017. “Machine Hjort, Jonas, and Jonas Poulsen. 2019. “The Arrival of Fast Learning: An Applied Econometric Approach.” Journal of Internet and Employment in Africa.” American Economic Economic Perspectives 31 (2): 87–106. Review 109 (3): 1032–79. OECD (Organisation for Economic Co-operation and Hoekman, Bernard, and Alessandro Nicita. 2008. “Trade Pol- Development). 2019. “Digital Opportunities for Trade icy, Trade Costs, and Developing Country Trade.” Policy in Agriculture and Food Sectors.” Report TAD/TC/CA/ Research Working Paper 4797, World Bank, Washington, WP(2018)4/FINAL (January 16), Joint Working Party on DC. Agriculture and Trade, OECD, Paris. Höppner, Thomas, and Phillipp Westerhoff. 2018. “The EU’s Oldenski, Lindsay. 2015. “Reshoring by U.S. Firms: What Competition Investigation into Amazon Marketplace.” Do the Data Say?” PIIE Policy Brief 15–14 (September), Kluwer Competition Law Blog, November 30, Wolters Peterson Institute for International Economics, Wash- Kluwer, Alphen aan den Rijn, the Netherlands. http:// ington, DC. competitionlawblog.kluwercompetitionlaw.com/2018/11 Prince, Jeffrey T. 2007. “The Beginning of Online/Retail /30/the-eus-competition-investigation-into-amazon Competition and Its Origins: An Application to Personal -marketplace/. Computers.” International Journal of Industrial Organization Hsieh, Chang-Tai, and Peter J. Klenow. 2014. “The Life Cycle of 25 (1): 139–56. Plants in India and Mexico.” Quarterly Journal of Economics PwC (PricewaterhouseCoopers). 2018. “Blockchain Is Here: 129 (3): 1035–84. What’s Your Next Move?” August, PwC, London. https:// Hsieh, Chang-Tai, and Benjamin A. Olken. 2014. “The Miss- fcibglobal.com/pdf/handouts/03_Blockchain-DS.pdf. ing ‘Missing Middle.’  ” Journal of Economic Perspectives 28 Rodrik, Dani. 2018. “New Technologies, Global Value Chains, (3): 89–108. and Developing Economies.” NBER Working Paper 25164, IFR (International Federation of Robotics). 2018. World National Bureau of Economic Research, Cambridge, MA. Robotics 2018. Frankfurt: IFR. Seamans, Robert, and Feng Zhu. 2014. “Responses to Entry in Jo, Yoon Joo, Misaki Matsumura, and David Eli Weinstein. Multi-sided Markets: The Impact of Craigslist on Local 2019. “The Impact of E-Commerce on Urban Prices and Newspapers.” Management Science 60 (2): 476–93. Welfare.” Paper presented at International Trade Sem- Timmer, Marcel Peter, Sébastien Miroudot, and Gaaitzen J. inar, Department of Economics, Stanford University, de Vries. 2019. “Functional Specialization in Trade.” Jour- Stanford, CA, April 3. nal of Economic Geography 19 (1): 1–30. Jouanjean, Marie-Agnès. 2019. “Digital Opportunities for Tripoli, Mischa, and Josef Schmidhuber. 2018. “Emerging Trade in the Agriculture and Food Sectors.” OECD Food, Opportunities for the Application of Blockchain in the Agriculture, and Fisheries Paper 122, Organisation for Agri-food Industry.” Issue Paper (August), Food and Economic Co-operation and Development, Paris. Agriculture Organization of the United Nations, Rome; Karabarbounis, Loukas, and Brent Neiman. 2014. “The Global International Center for Trade and Sustainable Develop- Decline of the Labor Share.” Quarterly Journal of Economics ment, Geneva. 129 (1): 61–103. WEF (World Economic Forum). 2018. “Innovation with a Pur- Laajaj, Rachid, Marcela Eslava, and Tidiane Kinda. 2019. pose: The Role of Technology Innovation in Accelerating “The Costs of Bureaucracy and Corruption at Customs: Food Systems Transformation.” Prepared in collaboration Evidence from the Computerization of Imports in with McKinsey and Company, January 23, WEF, Geneva. 156 | World Development Report 2020 http://www3.weforum.org/docs/WEF_Innovation_with WTO (World Trade Organization). 2018. World Trade Report _a_Purpose_VF-reduced.pdf. 2018: The Future of World Trade; How Digital Technologies World Bank. 2016. World Development Report 2016: Digital Are Transforming Global Commerce. Geneva: WTO. Dividends. Washington, DC: World Bank. Young, Jessica. 2019. “Global E-Commerce Sales Grow 18% ————. 2019. World Development Report 2019: The Changing in 2018.” Internet Retailer, August 7. https://www.digital Nature of Work. Washington, DC: World Bank. commerce360.com/article/global-ecommerce-sales/. Technological change | 157 PART IV What domestic policies facilitate fruitful participation? 7 Policies to enhance participation  olicies for inclusion and 8 P sustainability Policies to enhance participation Key findings  actor endowments matter: Eliminating restrictions in factor markets enables countries • F to exploit their comparative advantage. Avoiding overvalued exchange rates and restrictive regulations ensures labor is competitively priced. A favorable business climate and effective investment promotion facilitate foreign direct investment.  arket size matters: Liberalizing trade expands access to markets and inputs. By • M reducing tariffs and eliminating nontariff measures, a country expands its sources of supply. Liberalization in destination markets through trade agreements expands market access.  eography matters: Remoteness can be overcome by improving connectivity and • G lowering trade costs. Costs related to delay and uncertainty can be reduced by customs reform, introducing competition in transport services, and improving port structure and governance. nstitutional quality matters: It can be improved by strengthening contract enforcement, • I protecting intellectual property rights, and improving standards regimes. Deep trade agreements can help lock in institutional reforms.  roactive policies can enhance and upgrade global value chain (GVC) participation. • P Coordinating, informing, and training domestic small and medium enterprises helps link them to GVC lead firms. Investment in education and improvements in management encourage upgrading. Special economic zones can be a shortcut on the GVC development path when they successfully address specific market and policy failures. 160 | World Development Report 2020 W hat needs to be done to reap the benefits Third, because geography matters, countries can of global value chains (GVCs)? And what overcome remoteness by improving their connec- is the role of government policy in facil- tivity and lowering trade costs. Some countries are itating GVC participation and upgrading? Drawing disadvantaged naturally by being landlocked or in on evidence from chapter 2 on the determinants of remote locations. Others are disadvantaged by policy GVCs, as well as from cases from around the world, restrictions on transport services and by bureaucratic this chapter considers policies to enter and enhance actions such as slow, costly, unpredictable border pro- participation in GVCs. It begins by highlighting four cedures. GVCs rely on the fast and predictable move- areas of policy that would support GVC participation. ment of goods. For many goods traded among GVCs, First, because factor endowments matter, coun- a day’s delay is equal to imposing a tariff in excess tries should exploit their comparative advantage by of 1 percent. Improving customs and border proce- eliminating barriers to investment and ensuring that dures, promoting competition in transport services, labor is competitively priced, by avoiding overvalued improving port structure and governance, opening exchange rates and restrictive regulations. the domestic market to global providers of third-party Lead firms in GVCs are often multinational cor- logistics and express delivery services, and improving porations (MNCs), and so policies aimed at attract- information and communication technology (ICT) ing foreign direct investment (FDI) are especially connectivity—all are strategies that can reduce trade important for GVC participation. As a starting point, costs related to time and uncertainty. countries should facilitate the establishment and Fourth, because institutional quality matters, operation of businesses (the agenda is outlined in the countries need to strengthen enforcement of con- World Bank’s Doing Business reports). An investment tracts, protection of property rights, and regulatory policy should facilitate GVC-oriented FDI and sup- standards. GVCs thrive on the flexible formation of port investors throughout the investment life cycle. networks of firms. Contract enforcement ensures that Relying on well-planned investment promotion strat- legal arrangements within a network are stable and egies, countries such as Costa Rica, Malaysia, and predictable. Protecting intellectual property rights Morocco have successfully attracted transformative creates an environment for more innovative and com- GVC investments by large MNCs. plex value chains, and it can be supported through Second, because market size matters, countries deep PTAs. Governments can also facilitate participa- need to liberalize trade to expand access to markets tion in GVCs by strengthening their national certifi- and inputs. By liberalizing imports of inputs and cation and testing capacity to ensure compliance with eliminating unnecessary nontariff measures (NTMs), international standards, public and private. Pakistan’s a country can expand its sources of supply, as well ability to overcome an export ban on fish and expand as the possible roles it can play in the value chain. horticultural exports attests to the value of building a For example, the large unilateral tariff cuts by Peru strong national standards regime. in the first decade of the 2000s are associated with But being in a value chain today does not guarantee lower import costs, faster productivity growth, and that a country will capture significant benefits from expansion and diversification of GVC exports.1 Liber- participation and that those benefits will grow. Many alization in destination markets can expand market of the traditional approaches to industrial policy, access. For example, preferential trade agreements including tax incentives, subsidies, and local content (PTAs) have acted as a catalyst for GVC entry for a policies, are more likely to distort than help in today’s wide range of countries, including Bangladesh, the GVC context, as Brazil’s poor experience of promoting Dominican Republic, Honduras, Lesotho, Madagas- localization in the automotive sector illustrates. How- car, and Mauritius. ever, a range of proactive policies can enhance GVC Goods and services are increasingly linked, and participation. so liberalizing the trade in services is an important Countries can promote linkages between domes- part of any strategy for promoting GVCs. Policies tic small and medium enterprises (SMEs) and GVC should therefore seek to improve the environment for lead firms by coordinating local suppliers, providing e-commerce, liberalize telecommunications services, access to information about supply opportunities, and and promote free movement of data, as well as sup- supporting training and capacity building of SMEs. port access to other important service inputs such as There are many examples of successful supplier link- transport, finance, accounting, and other business age programs such as those in Chile and Guinea in support services. mining, Kenya and Mozambique in agriculture, and Policies to enhance participation | 161 the Czech Republic in the electronics and automotive specific climatic conditions, and mineral resources. sectors. Governments can also help domestic suppli- A move to basic manufacturing GVCs often relies ers gain access to finance and technology to support on access to low-cost labor, while moving into more raising productivity and meeting global standards. advanced and innovative activities requires higher Countries can strengthen sector-specific human levels of human capital. And entry in almost all GVCs capital through targeted workforce development strat- requires access to capital—especially foreign direct egies, involving close coordination between the public investment in most developing countries. But just hav- and private sectors. The Penang Skills Development ing favorable endowments is no guarantee of success. Centre in Malaysia, an industry-led training center, National policies fundamentally shape the price of has played an important role in supporting Malaysia’s factor endowments and how well they are able to upgrading in the electronics and engineering GVCs. contribute to GVC participation. Countries can also support firms in their efforts to upgrade management capabilities and strengthen the Natural resources capacity for innovation. Turkey’s upgrading into the Despite having favorable conditions for agricultural branded segment of the apparel GVC was supported production, many countries have a regulatory and by both government and private sector initiatives, institutional environment that undermines invest- including workforce training, consulting and design ment prospects in the sector. Surveys of agribusiness services, and incentives for investment in research investors2 have identified land acquisition as a special and development (R&D) and technology. concern. Lack of proper land registries and weak Governments can also strengthen national inno- legal systems in many countries make it impossible vation systems to support upgrading in GVCs. Ger­ to enforce land titles. The situation is aggravated in many’s dense networks of public-private collaboration regions such as Sub-Saharan Africa, where countries involving foreign and local industry, academia, and may have parallel (and often conflicting) customary government research institutions is one example of and statutory land tenure systems. In postconflict an effective model. environments, forced displacement, land occupations, This chapter also considers whether and how spe- and loss of official title deeds may make it impossible cial economic zones (SEZs) may be used as a shortcut to secure tenure. For example, in Liberia, despite the on the GVC development path, recognizing that deliv- government granting large concessions to interna- ering on the policies just outlined is a medium-term tional investors in rubber and palm oil, competing agenda. SEZs can be successful when they address land claims and community conflicts have resulted in specific market failures. Getting conditions right, even investors managing to plant only on a small portion in a restricted geographical area, requires careful plan- of the land concession, and the surrounding small- ning and implementation to ensure that the needed holders have been unable to secure finance to plant resources—such as labor, land, water, electricity, and without land titles. The ensuing lack of production telecommunications—are readily available, that there scale has also made it uneconomic for the lead firms are no unnecessary regulatory barriers, and that con- to invest in processing facilities. nectivity is seamless. The relatively few successful Governments should have a clear legal framework zone programs in places such as China, Panama, and for land policy, along with a legal and administrative the United Arab Emirates, and emerging in Ethiopia, apparatus that can enforce land rights, while recogniz- offer important lessons for how best to take advantage ing various acceptable forms of tenure. Such objectives of the instrument to establish an environment for dif- can be supported by adopting a proactive process of ferent types of GVC participation. engagement, beyond simply consultation, with com- munities likely to be affected by large investments in agriculture. For example, in Ghana the government Facilitating participation has published guidelines on community engagement Take advantage of factor endowments and practices to help facilitate large-scale commercial agri- eliminate restrictions in factor markets culture investments. As described in chapter 2, factor endowments matter For countries with large mineral endowments, the for a country’s entry and positioning in GVCs, and main issues revolve around the terms of concession that is not surprising. Investment in most commodity agreements. Most notably, such terms relate to royalty GVCs (as well as the travel and tourism services GVCs) and tax payments, but they also may include local con- depends on access to natural resources such as land, tent requirements, such as requiring investors to hire 162 | World Development Report 2020 a certain share of local staff, to purchase from local Figure 7.1  Manufacturing labor costs are out of line companies, or to carry out value-added processing with national income levels in Sub-Saharan Africa but in the country. Mining investments typically require not in Bangladesh large amounts of capital up front, with returns over 2,500 the long term. Thus investors face many uncertainties, including production costs and future trends in com- 2,000 modity prices. Governments can reduce uncertainty by having legal frameworks such as mining codes or US$ (2010) 1,500 mining laws that clearly establish the terms under which mining concessions will operate. Botswana, 1,000 Chile, and Namibia have high-quality policy environ- ments,3 whereas Zimbabwe, which nationalized and 500 partly nationalized various mining sectors over the past decade and most recently threatened an export 0 ban on platinum, is rated as having one of the least Bangladesh Ethiopia Kenya Senegal Tanzania favorable policy environments for mining investment. Average annual labor cost per GDP per capita manufacturing worker Human capital Source: WDR 2020 team, using data from Gelb et al. (2017). The empirical findings on the importance of low- cost labor for GVC entry in basic manufacturing is They raise the prices of domestic resources relative to supported by the evidence of foreign investment in imports, thereby deterring international investments GVC-intensive sectors such as apparel. The shift of in labor-intensive activities and making domestic manufacturing to China and Vietnam, and now (as investors more likely to import capital equipment to wages rise in these countries) to Bangladesh, Cambo- substitute for high-priced domestic labor. Historically, dia, and Ethiopia, reflects the importance of low-cost countries with competitive or undervalued exchange labor in this sector. At times, investors exploit large rates have undergone greater structural change6— labor cost wedges in local environments—for example, the experiences of China and the Republic of Korea when South Africa’s apparel production moved just stand out here. Because overvalued exchange rates over the border to Lesotho4 and when apparel factory are common in countries that heavily rely on natural clusters emerged on the Dominican Republic’s border resources, they pose a threat when these countries with Haiti. transition into basic manufacturing value chains. For But many countries with low levels of per capita example, an overvalued exchange rate was a major income and large pools of moderately skilled, under- factor in the failure of Trinidad and Tobago to develop employed labor find themselves priced out of the its manufacturing sector.7 Beyond the exchange rate, market for GVC investments in basic manufacturing workers in many countries have high reservation activities because of uncompetitive labor costs. For wages because of the high cost of living in urban example, a recent study using World Bank survey data and peri-urban areas. For example, in urban areas of on 5,500 companies in 29 countries found that for any Sub-Saharan Africa workers often face high costs for given level of GDP, labor is substantially costlier for food and housing, along with high transport costs, manufacturing firms located in Sub-Saharan Africa.5 which can consume up to 50 percent of wages. As shown in figure 7.1, the average labor costs in Ban- Therefore, getting the price of labor right requires gladesh are in line with the average GDP per capita, policies that go well beyond the realm of labor and whereas in comparator countries in Africa they are policies to support urbanization and public services. often almost twice that average. Only Ethiopia has Governments also have to address other fundamental wage levels on a par with those in Bangladesh. investment climate constraints, such as poor infra- Addressing rigid labor regulatory policies, while structure, as well as skill mismatches, which compound ensuring protection of workers and appropriately labor price gaps by suppressing productivity growth. sharing the gains from GVC trade (see chapter 8), is As countries look to upgrade in GVCs, policy pri- one step governments can take toward more compet- orities shift to the quality rather than the quantity of itively priced labor. But regulation is just one contrib- human capital. Higher value-added positions in GVCs utor to labor price gaps. Overvalued exchange rates require both high-level technical skills and adaptability are a significant threat to competitively priced labor. because changing technologies rapidly reshape the Policies to enhance participation | 163 kinds of skills needed. Research in Costa Rica and the flows and repatriation of profits, what form the tax- Dominican Republic has found that large differences ation regime takes, and what fiscal and nonfiscal in the investments of each country in human capital is incentives (such as work permits) are available—are one of the primary explanations for the different devel- central to FDI decision making. But in the GVC world, opment trajectories of the two countries over recent where investments are fundamentally linked to decades. Costa Rica’s success in diversifying away from import–export relationships, trade policy is equally apparel to high-technology exports was supported by important. Similarly, because of the service intensity public social spending that averaged close to 20 per- of GVCs, domestic regulatory policy, including the cent of GDP in the 1980s and 1990s. By contrast, in the role of state-owned enterprises (SOEs) and competi- Dominican Republic, which struggled to move away tion in infrastructural and business services, plays a from low value-added apparel exports, public social big part in defining the attractiveness of a location for spending during this time averaged just 5 percent of GVC-linked investments. Facilitating FDI for GVCs GDP, the lowest in all of the countries in Latin America.8 requires, then, effective coordination of investment, trade, and domestic regulatory policies. Foreign capital Political stability, investor protection, and a In developing countries, foreign capital is especially business-friendly regulatory environment are espe- important for GVC integration.9 Foreign investors cially important in attracting FDI. However, FDI is bring with them the technology, managerial expertise, not homogenous. Investors with different motives and established market relationships needed for GVC consider different factors in their decision to invest. For integration. Thus policies and strategies to attract and example, MNCs that primarily seek access to natural retain FDI are important for countries seeking to par- resources—such as in extractive industries—care about ticipate in GVCs. access to land and resources, whereas market-seeking Attracting and retaining FDI in a GVC context FDI tends to give priority to the size and purchasing requires a well-formulated investment policy. Cer- power of the domestic market. Efficiency-seeking FDI, tainly, the core elements of investment policy—what which characterizes most noncommodity GVC invest- sectors are open to foreign investment, what assets ments, focuses on factors that affect production and may be foreign-owned, what rules exist for capital trade costs (box 7.1). Box 7.1  Determinants of efficiency-seeking investment For multinational corporations (MNCs), what are the most • Incentives. Sixty-three percent of efficiency-seeking important determinants of efficiency-seeking foreign direct investors rate incentives as important or critically import- investment (FDI)? Compared with investors with other moti- ant, in contrast with 43 percent of investors with other vations, efficiency-seeking firms, which connect countries motivations. These firms rated eight different incentive directly to GVCs, find the following factors more important instruments more highly than other investors, with an (figure B7.1.1): a average difference of about 13 percentage points. • Capacity and skills of local suppliers. This factor was • Characteristics of host countries. Most are important, rated important or critically important by 77 percent of especially low-cost labor and inputs, which 66 percent MNCs engaged in efficiency-seeking FDI, compared with of firms involved in efficiency-seeking investment find 70 percent of investors with other motivations. To pro- important or critically important, compared with only mote linkages, 55 percent of MNCs involved in efficiency- 39 percent of investors with other motivations. seeking FDI have internal “talent scouts” to find local • Investment policy factors. These factors include invest- suppliers, compared with only 45 percent of investors ment protection guarantees, owning all equity, hiring involved in other types of FDI. expatriate staff, importing production inputs, ease of • Investment promotion agencies (IPAs). Fifty-two percent obtaining approvals, bilateral investment treaties, and of efficiency-seeking investors identify IPA services as preferential trade agreements (PTAs). PTAs were found to important or critically important, compared with 37 per- be important or critically important by 65 percent of firms cent of investors involved in other types of FDI. involved in efficiency-seeking investment, compared with only 45 percent of investors with other motivations. (Box continues next page) 164 | World Development Report 2020 Box 7.1  Determinants of efficiency-seeking investment (continued) Figure B7.1.1  MNCs involved in efficiency-seeking FDI are more selective Political stability and security Legal and regulatory environment Large domestic market size Macro stability and favorable exchange rate Available talent and skilled labor Good physical infrastructure Low tax rates Low cost of labor and inputs Access to land or real estate Financing in domestic market 0 10 20 30 40 50 60 70 80 90 100 Percent Efficiency-seeking: Critically important Important Somewhat important Not important Don’t know Nonefficiency-seeking: Critically important Important Somewhat important Not important Don’t know Source: World Bank 2018. Note: FDI = foreign direct investment; MNCs = multinational corporations. This overview of locational determinants of FDI is based on findings from the World Bank’s 2017 Global Investment Competitiveness survey on investor a.  perceptions and preferences (World Bank 2018). Beyond policy, strategies and tactics—and their in FDI inflows, for a cost of just $78 to create one job implementation—matter for attracting and retaining in the promoted sectors.12 IPAs can also improve the GVC investors. Proactive efforts to attract and facil- quality of investments and contribute to economic itate foreign investment, through the use of invest- transformation by exploiting comparative advantage. ment promotion agencies (IPAs), can help overcome For example, Costa Rica, Malaysia, and Morocco suc- problems of information asymmetry and coordination cessfully attracted transformative, efficiency-seeking failures that may restrict FDI.10 IPAs typically carry investments by large MNCs using well-targeted out image-building campaigns, undertake investment investment promotion strategies that built off core generation through targeted efforts to identify and policies of macroeconomic stability and skills develop- attract specific investors, help investors to establish ment. These economies saw a boost in revealed com- their businesses, and lobby government for investor- parative advantage and better integration into GVCs.13 friendly policies. Research has shown that IPAs can contribute to larger FDI flows11 (figure 7.2) and can Liberalize trade to expand markets be highly cost-effective, with one study finding that Market size matters because larger markets enable every $1 spent on investment promotion yields $189 firms to benefit from returns to scale in terms of both Policies to enhance participation | 165 Figure 7.2  Better-quality investment promotion agencies attract more FDI inflows 30 USA LUX CHN GBR 25 FRA HKG DEU Average FDI inflow, 2000–10 (log) ITA NLD ESP BRA CAN SWE RUS MEX SAU SGP AUS IND THA JPN CHE HUN POL UKR EGY ROM CHL IRL DNK TUR AUT KAZ NGA BGR COL ISR HRV CZE PRT FIN VNM MAR IDN LBN ZAF MYS ARG KOR SRB NZL DZA PAK PHL PER SVK CYP SDN LBY DOM OMN MNE TUN JOR EST ISL VEN URY GRC PAN AGO BLR BGD ALB TTO CRI LTU COG SYR GEO BHR JAM LVA 20 MOZ IRQ GHA GTM HND BWABIH ECU BOL YEM BHS AZE SLV ARM SVN CMR KSV GIN ZMB ETH TJK KHM TZA MNG LKA UGA FJI SEN MKD NIC GAB PRY ATG SYC MDA KGZ LSO NAM LBR LAO PNG GUY BRB MUS BRN CPV BEN MLI KNA KEN LCA GRD BFA SLE MDV DJI GMB SWZ VCT BLZ NER IRN MDG ERI TGO KWT DMA BTN HTI SLB RWA WBG VUT MWI WSM GNB NPL TON 15 BDI 10 0 20 40 60 80 IPA quality Source: Harding and Javorcik 2012. Note: The IPI quality rating is based on the World Bank’s Global Investment Promotion Benchmarking (GIPB) series. The figure shows the average results of GIPB scores from the 2006, 2009, and 2012 GIPB series. FDI = foreign direct investment; IPA = investment promotion agency. For country abbreviations, see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. production efficiencies and an ability to make the is still scope for an international effort to lower tar- most of knowledge and technologies. But as chapter 1 iffs—bilaterally, regionally, or in a multilateral round describes, domestic market size is less relevant in a (chapter 8)—the scope for countries to engage in uni- GVC world because scale economies can be reaped lateral liberalization remains substantial. through deeper specialization and global market Tariff schedules that place higher duties on pro- integration. This offers a critical shortcut for small cessed goods than on unprocessed goods—a feature developing countries. Taking advantage of this oppor- known as tariff escalation—have particularly negative tunity requires liberalized trade policies that support effects on developing countries in GVCs. Escalation integration. Indeed, as chapter 2 notes, countries with acts as a barrier preventing developing countries from low tariffs and greater market access are more likely upgrading to higher value-added segments of the to participate in GVCs. value chain, potentially locking them into lower-value, limited-processing activities. Trade agreements have Tariffs significantly reduced the extent of tariff escalation Worldwide, most-favored-nation (MFN) tariffs fell in high-income countries, but the process needs to go by about a third between 2001 and 2013.14 Of this lib- further, especially for agricultural products. eralization, more than half was the result of countries High tariffs and tariff escalation can undermine cutting tariffs on their own initiative. This reduction the development of regional value chains. For exam- included unilateral cuts of between 10 and 20 per- ple, in southern Africa, despite the customs union of cent in ad valorem tariffs by India, Morocco, Nige- Botswana, Eswatini, Lesotho, Namibia, and South ria, Peru, and Tunisia, and between 5 and 10 percent Africa, as well as the expressed strategic interests by Bangladesh, Kenya, and Mexico. Although there in developing regional agriculture value chains, 166 | World Development Report 2020 protection of domestic agricultural interests has (now Eswatini); and Mauritian investors established resulted in multiple trade restrictions, including sea- some of the first apparel manufacturing facilities in sonal import bans and quotas, as well as duties of up to Madagascar. Preferential arrangements such as the 40 percent on grain, feed, dairy, and poultry products. African Growth and Opportunity Act (AGOA) and the Moreover, in many parts of the world tariffs Everything but Arms (EBA) initiative of the European and other forms of trade protectionism have seen a Union, along with regional trade agreements such as resurgence over the last two years, fueled in part by the Caribbean Basin Initiative (later the Dominican tensions between the United States and China. In Republic–Central America Free Trade Agreement, the age of GVCs, where hyperspecialization and dis- DR–CAFTA) played a similar role. Recent fragmenta- tribution of tasks across borders ensure that trade tion in the global trading system may in fact create costs are incurred multiple times, this new wave of opportunities for countries to exploit PTAs as a chan- protectionism is likely to have significant negative nel for GVC entry. implications. They will arise not only directly from higher trade costs but also from the costs of trade Nontariff measures policy uncertainty, which can make firms reluctant to The use of NTMs is increasingly widespread. The invest in supply chains and thus result in long-lasting share of tariff lines covered by NTMs averages about disruptions in global investment and production. 40 percent for the least developed and developing Finally, as discussed in detail in box 2.5 in chapter 2, countries and more than 60 percent for developed governments can exploit the opportunities created countries. The trade covered by such measures is by PTAs, particularly when they offer duty-free mar- even higher (figure 7.3, panel a). Moreover, multiple ket access, to catalyze GVC entry. This was apparent NTMs are often applied to the same product category during the period of the Multifibre Arrangement (figure 7.3, panel b). (MFA) quota system, when footloose GVC investors Although it may appear that countries are simply sought opportunities to exploit unused quotas. For substituting tariff protection for NTM protection, this example, Korean investors kickstarted the apparel is not necessarily the case. NTMs such as quantitative GVC in Bangladesh and Honduras; Taiwanese inves- restrictions and nonautomatic licensing have effects tors initiated the sector in Lesotho and Swaziland similar to those of tariffs, and they serve primarily to Figure 7.3  Nontariff measure use increases by development status a. Frequency index and coverage ratio b. Prevalence score and regulatory intensity 100 6 0.8 Frequency index and coverage ratio (%) 80 0.6 Regulatory intensity 4 Prevalence score 60 0.4 40 2 0.2 20 0 0 0 LDC Developing Developed LDC Developing Developed Frequency index Coverage ratio Prevalence score Regulatory intensity (right axis) Source: UNCTAD and World Bank 2018. Note: Panel a: The frequency index captures a country’s share of traded product lines subject to at least one nontariff measure (NTM). The coverage ratio captures a country’s share of trade subject to NTMs. Unlike the frequency index, it is weighted by import values instead of traded product lines. Panel b: The prevalence score indicates a country’s average number of distinct NTMs applied on regulated products. In doing so, it measures the diversity of NTM types applied and provides some indication of the intensity of regulation. The regulatory intensity adjusts the prevalence score for differences in regulatory intensity and trade importance across products. In doing so, it adjusts for the fact that some products are more traded and regulated than others such as medicines. Computed as an average for a country, the regulatory intensity is normalized by the average number of measures for each product around the world and then weighted by its importance in world trade. LDC = least developed country. Policies to enhance participation | 167 restrict trade—indeed, such NTMs can inhibit GVC allocation of new licenses remains opaque and highly formation. But a large share of modern NTMs are discretionary in many countries. regulatory in nature. Technical barriers to trade (TBTs) Across regions, some of the fastest-growing coun- and sanitary and phytosanitary (SPS) measures are at tries in Asia and the oil-rich Gulf states have highly least ostensibly designed to protect human, animal, restrictive policies in services, while some of the and plant life; health; and the environment. Moreover, poorest countries are remarkably open, as measured their application is regulated by World Trade Organi- by the World Bank’s Services Trade Restrictiveness zation (WTO) agreements. Higher-income countries, Index, which takes values from 0 for completely open which tend to have lower tariffs, are more likely to regimes to 100 for completely closed (map 7.1). Across make extensive use of TBTs and SPS measures. sectors, professional and transport services are among However, regulatory measures, even when they the most protected in both industrial and developing have legitimate goals, can pose challenges for low- and countries, whereas retail, telecommunications, and middle-income countries as their producers strive to even finance tend to be more open. meet more stringent standards, which may be costly. National decisions to open markets to certain types For exporters, failure to meet standards—such as of services trade are critical for GVCs. Among those those for quality and traceability in agriculture—can types are third-party logistics providers and express lock firms out of markets. For importers, inappropri- delivery services. In addition, much of the innovation ate standards may exclude firms from some valuable in value chains takes place at the downstream end, opportunities for GVC participation. For example, through retailers. It may be easier for large retailers importers in many South Asian countries find it to take advantage of new supply chain technologies difficult to import synthetic yarn and fabrics, which to enhance GVC productivity than for the more tra- inhibits their apparel producers from serving the mar- ditional small retailers to do so, and even easier for ket for higher value-added segments such as athletic e-commerce firms.16 Thus policies that restrict the wear. entry of large retailers (either domestic or foreign) On the other hand, the emergence of well-defined can have a negative impact on efforts to exploit the product standards can help firms in developing full efficiencies of GVCs. To the extent that advanced countries overcome technical, informational, and supply chain technologies complement e-commerce, reputational barriers to market access and so play an interventions to improve the enabling environment important role in facilitating GVC participation and for e-commerce and policies to enable the free move- upgrading (this issue is discussed in more detail later ment of data are likely to complement the devel- in this chapter). opment of GVCs. Liberalizing telecommunications services, including access to the Internet, is essential Trade in services to facilitating the flow of information between buyers For many developing countries, the best opportunities and sellers needed to promote GVCs (box 7.2). In addi- for GVC integration will not come through natural tion, countries can remove impediments to importing resources or manufacturing, but instead through services.17 Initiatives such as liberalization of pro- integration in services GVCs, notably through sectors fessional licensing are possible subjects for regional such as tourism and business process outsourcing. cooperation. And as discussed in chapter 1, even manufacturing and natural resources–focused GVCs are highly Enhance connectivity to lower trade costs service-intensive. Thus eliminating impediments to Beyond tariffs, the cost of moving goods remains a trade and investment in services is a high priority to substantial impediment to trade. Supply chains go promote GVC participation. where the logistics are smooth. To compete in GVCs, The limited information on trade and investment firms need to respond quickly to any changes in policy for services suggests that much higher bar- demand, which is costly when intermediate inputs riers remain to liberalizing the services trade than face border delays that necessitate maintaining inven- the goods trade. The World Bank’s Services Trade tories. Supply chain efficiency has therefore emerged Restrictions Database reveals that, although public as an important determinant of trade performance. monopolies are now rare and few service markets are Improving supply chain–related trade costs associ- completely closed, numerous restrictions remain on ated with border administration and transport and entry, ownership, and operations.15 Even where there communications infrastructure halfway to global best is little explicit discrimination against foreign provid- practice would, it is estimated, produce global GDP ers, market access is often unpredictable because the gains up to six times larger than the elimination of all 168 | World Development Report 2020 Map 7.1  Services trade remains restricted in many countries Services Trade Restrictiveness Index 0–10 10–20 20–30 30–40 40–50 50–60 60–70 70–80 80–100 No data or under review IBRD 44673 | SEPTEMBER 2019 Source: Borchert, Gootiiz, and Mattoo 2014. Note: The World Bank’s Services Trade Restrictions Database covers 103 countries (79 developing) and financial, basic telecommunications, transport, distribution, and selected profes- sional services. Data were collected between 2008 and 2010. The Services Trade Restrictiveness Index (STRI) takes on values from 0 to 100, where 0 indicates a country is completely open to trade without restrictions, and 100 indicates a country is completely closed to trade. Box 7.2  Foreign services firms in India’s manufacturing value chains India offers a powerful example of how foreign services in fewer locations, to efficiently manage inventories, and to firms help support greater participation in manufacturing coordinate decisions with suppliers and customers. value chains. Conventional explanations of the modest To analyze the link between service reforms and manu- resurgence of Indian manufacturing since the early 1990s facturing productivity in India, Arnold et al. (2016) collected have focused on policy reforms in manufacturing industries. detailed information on the pace of reform across Indian However, a central factor lies outside manufacturing in the services sectors, focusing on entry and operational restric- services sector. Reforms in the 1990s visibly transformed tions. To make this information amenable to econometric services sectors, with greater openness and improved reg- analysis, the investigators aggregated it into time-varying ulation leading to dramatic growth in domestic and foreign reform indexes. They then related the total factor produc- investment. Indian manufacturing firms were no longer tivity (TFP) of about 4,000 manufacturing firms to the state at the mercy of inefficient public monopolies; they could of liberalization in the services sectors, taking into account now source services from a wide range of domestic and other aspects of openness such as tariffs on output and foreign providers operating in an increasingly competitive intermediate inputs, as well as foreign direct investment environment. As a result, they had access to better, newer, (FDI) in the final and intermediate goods sectors. more reliable, and more diverse business services. These The results suggested that pro-competitive reforms in improvements enhanced firms’ abilities to invest in new banking, transport, insurance, and telecommunications business opportunities and better production technology, boosted the productivity of both foreign and locally owned to exploit economies of scale by concentrating production manufacturing firms. A one standard deviation increase in (Box continues next page) Policies to enhance participation | 169 Box 7.2  Foreign services firms in India’s manufacturing value chains (continued) the aggregated index of services liberalization resulted in a manufacturing.b Similarly, another study demonstrates that productivity increase of 11.7 percent for domestic firms and substantial FDI inflows in producer services sectors in Chile 13.2 percent for foreign enterprises. The largest additional had a positive effect on the TFP of Chilean manufacturing effect was for transport reforms, followed by telecommuni- firms.c The same study suggests that foreign investment cations and banking reforms. in services fosters innovation in manufacturing and offers Several other studies have confirmed that access to low- opportunities for laggard firms to catch up with industry cost, high-quality (domestic or foreign) producer services leaders.d These benefits arise not just from foreign invest- can promote productivity and economic growth.a Firm-level ment but also from cross-border trade in services. For data for the Czech Republic for 1998–2003 reveal that ser- example, services offshoring by high-income countries vices sector reforms leading to greater FDI had a positive tends to raise the productivity of their manufacturing effect on the productivity of domestic firms in downstream sectors.e a. Hoekman and Mattoo (2008). b. Arnold, Javorcik, and Mattoo (2011). c. Fernandes and Paunov (2012). d. Similar results have been found for Sub-Saharan Africa (Arnold, Mattoo, and Narciso 2008) and Indonesia (Duggan, Rahardja, and Varela 2015). e. Amiti and Wei (2009a). Although offshoring of services has both positive and negative effects on domestic employment, Amiti and Wei (2009b) show that, at least for the United States, it tends on average to enhance domestic employment. tariffs.18 One aspect of these costs is trade facilitation countries can take measures unilaterally to promote and logistics. Delays due to shipping and border pro- increased connectivity and cost-effectiveness: cedures have a negative effect on trade comparable to that of tariffs. A day’s delay reduces trade by more • Rebalance and repurpose trade infrastructure. For many than 1 percent in Africa,19 and a day’s reduction in developing countries, particularly in Sub-Saharan inland transit times can boost exports by as much as Africa, Central Asia, and parts of Latin America, 7 percent.20 trade infrastructure has been established primar- Figure 7.4 shows the estimated tariff equivalent of ily around extractive sectors. Such infrastructure, a day’s delay in shipping for a wide variety of product built around bulk and direct connections between categories. The time costs in trade are significant for often rural areas (such as mining locations) and products with complex value chains such as motor ports, may not be supportive of the environment vehicles; perishable products such as fruits and vege- needed for value chain–oriented sectors, which tables; and textiles and apparel, both of which involve may require denser, multimodal infrastructure. complex GVCs and changes in fashion that reduce A study of port costs in South Africa found that, their shelf life. By contrast, traders are willing to wait although export charges for mining commodities longer for goods such as live animals, leather goods, were well below the global average in 2014, charges and wood and forestry products. for containerized exports were almost twice the GVCs are impeded not only by the slow movement global average.22 of goods but also by their unpredictable movement, • Improve port infrastructure and governance.23 There which disrupts the ability of a value chain to per- are vast differences between the world’s most and form its steps in the appropriate sequence. In Sub- least efficient ports in terms of the time it takes to Saharan Africa, the slowness and unpredictability unload ships, cargo dwell time (the time it takes for of land transport impeded the formation of GVCs in a container to be available for pickup after being almost all countries until very recently.21 unloaded from a ship), and the adequacy of ware- Many poor, remote, landlocked countries are under- houses and port customs procedures. Technologi- served by international shipping and air cargo ser- cal solutions do exist, such as use of electronics at vices. In part, this is a vicious circle—because of weak customs or improvement in gantry cranes, but the economic activity few shippers schedule service to reforms needed may be obstructed because some such countries, which increases trade costs. However, stakeholders benefit from delays. 170 | World Development Report 2020 Figure 7.4  Shipping delays matter more for products with complex value chains 1.6 Tariff equivalent of a day’s delay (%) 1.4 1.2 1.0 0.8 0.6 0.4 0.2 0 o ts ts s s ts ts s rts s s s y k t l ct ile tic en cc al al ct uc uc re sil uc nu tr pa du du es im et ba as xt pa pm od od d nd nd m Te r ro o ro an pl d Fo to Ap pr pr pr ui la n us ,a rp p nd sa eq nd e al d al rro d oo Liv its he oo ,a oo co sa et cle c W fru Fe at ni M W sf er d ge hi Le tro an s, bb ou ve ra e ec ru ne bl um ve or El ta la s, Be ot le ge al el M tro isc ic Ve Pe m M he C Source: WDR 2020 team, using data from Minor (2013), based on USAID (2007). Note: The tariff equivalent on the y-axis is measured as the percentage of an ad valorem tariff economically equivalent to a day’s delay in shipping. For example, a day’s delay in moving chemicals, rubber, and plastics is equivalent in economic terms to imposing a 1.2 percent tariff on imports of the same goods. Improve connectivity of landlocked countries and of •  long way in this area. In Albania, a risk management remote regions within countries. Although landlocked reform that sharply reduced the number of physical and remote regions tend to be poorer (20 of 54 inspections of shipments shortened clearance times, low-income countries were landlocked in 2011, reduced uncertainty of clearance, and expanded compared with 3 of 35 high-income countries), imports (figure 7.5).26 human action adds to naturally high trade costs. For GVC integration can also be supported by liber- example, road transport cartels emerge in environ- alization of trade and transport services, including ments where roads are of low quality.24 Cooperation opening domestic markets to global providers of between landlocked and transit countries may reduce costs, as well as cooperation between remote Figure 7.5  Customs reform can reduce delay and neighboring countries in the recognition of transit expand imports: Evidence from Albania rights for trucking, harmonization of rules on trans- port (such as axle weight loads and insurance), and 0.30 30 Change in customs clearance time treatment of goods in transit. “Hard” multimodal 0.25 25 Import value (%) (right bar) infrastructure (rail, road, air, and pipeline) should 0.20 20 (left and center bars) complement “soft” initiatives such as pursuing bet- ter border procedures through trade facilitation. 0.15 15 0.10 10 Indeed, unilateral regulatory reforms to improve 0.05 5 trade facilitation could have a significant impact on GVC competitiveness. Such reforms include mod- 0 0 ernization of customs systems and reforms and har- –0.05 –5 monization of customs rules and procedures such as –0.10 –10 implementing effective risk management systems, Median customs Dispersion in Import value (%) replacing paper-based documentation with electronic- clearance time customs clearance time based documentation, and improving transparency (days) (interquartile range, days) through trade information portals and single win- Source: Fernandes, Hillberry, and Mendoza-Alcántara 2019. dows.25 A concerted effort to implement the provisions Note: It is assumed that the probability that a shipment is inspected falls from 50 percent or more to of the WTO’s Trade Facilitation Agreement could go a under 50 percent. Policies to enhance participation | 171 third-party logistics and express delivery services. skills development is pervasive and deep, including Advances in logistics include not only those related through technical and vocational education systems to companies (some of which are engaging directly and through support of firms seeking to invest in ICT in shipping and road and air transport), but also those systems/applications and training. related to freight forwarders, customs brokers, loaders and unloaders, “pick and pack” warehouses, and many Strengthen institutions for contracts, other types of services. At the high end, the coordina- intellectual property protection, and tion of many of these services by a third-party logis- standards tics company can be critical in the design of a local or global supply chain (such as that for the organization Contract enforcement of disc drive manufacturing in Thailand).27 The supply Coordination of a GVC involves managing large net- of such services can be expanded both by liberalizing works of firms, which must share dispersed knowl- FDI in the relevant sectors and by removing imped- edge and often commit assets to relationships with iments to doing business domestically in the same specific partners. It is therefore essential that the part- sectors. ners in a GVC enter and enforce complex contracts. Finally, as discussed in chapter 6, ICT is critical as In an environment in which contract enforcement is a facilitator of information and coordination in value relatively weak, the formation and ongoing conduct of chains, especially for countries that are peripherally GVCs are inhibited. located. The Philippines is an example of a peripheral Litigation between pairs of U.S. firms reveals that country that has utilized ICT to participate in rela- contract enforcement issues are most prevalent in rela- tively high-value segments of services GVCs. How- tionships between firms and their suppliers of profes- ever, many developing countries have an insufficient sional services, including insurance, business services, ICT infrastructure and its pricing is uncompetitive. and financial services (figure 7.6). This finding implies Moreover, the ICT capabilities of many smaller com- that the supply of such services may be lower where panies are limited. Governments can support efforts the legal institutions to enforce contracts are weak. to improve ICT capabilities by investing in infrastruc- Since such institutions are generally weaker in lower- ture (including “last mile” broadband), promoting income countries, this accounts in part for the scarcity competition in ICT markets, and ensuring that ICT of business services in those countries (figure 7.7).28 Figure 7.6  Contract enforcement intensity is higher in services sectors: Evidence from the United States 1.6 1.4 Contract enforcement intensity 1.2 1.0 0.8 0.6 0.4 0.2 0 EC l ts e ts s rts s ts re s ng ty ce s s er s rt EC t ga tic ad en uc ct ge ile uc uc al po ci pa N pa at an hi du as et N pm od Tr xt od od ri ra W d ns fis s, Ap m ct r Te pl ro d an ve su s, pr pr pr ui tra ce e an rp us nd ce nd El In be eq il al vi d al rro vi O ir ,a s he oo ,a er et co e A d c er cl Fe M ss ni at an try W er ls hi d tro Le bb es es an ve ia ts ec r uc sin nc ru fo um or El od na Bu s, ot e, le al pr Fi M r tro tu ic od ul m Pe ic he Fo gr C A Source: WDR 2020 team, using data from Boehm (2018). Note: See Boehm (2018) for method of calculating contract enforcement intensity. NEC = not elsewhere classified. 172 | World Development Report 2020 Figure 7.7  Share of “other business services” in intermediate inputs is low in poor countries 28 SWE 26 NZL 24 FRA IRL 22 NLD FIN ITA AUT 20 Share of “other business services” (%) KAZ AUS MLT DNK CHE 18 LUX BEL NOR 16 GRC PRT SGP CHL ISR CAN BRN 14 CYP CRI HRV USA CZE 12 COL SVK POL 10 GTM ROU JAM MUS LTU 8 BRA TTO HKG ETH PER KOR ARG BWA PAN ARE QAT 6 MYS BHR KWT 4 AZE OMN ECU IRN 2 PRY ARM 0 10 20 30 40 50 60 70 80 90 100 110 120 130 GDP per capita (US$, thousands, PPP-adjusted) Source: WDR 2020 team, based on Boehm (2018). Note: PPP = purchasing power parity. For country abbreviations, see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. A survey of GVC sectors across 14 countries in Sub- contracting relationships with independent suppliers, Saharan Africa found that just 43 percent of lead firms raising the risk of replication of designs, technologies, outsourced critical business and technical services, and processes. How different national systems deal with the majority choosing to bring the required with contractual frictions and incomplete contracts expertise in-house.29 Results from the survey suggest is therefore important in driving firm choices of loca- that this choice is driven in part by lack of access to tion and sourcing, as well as firm boundaries (what a sufficient breadth of quality suppliers (reflecting they outsource) in GVCs.30 According to the evidence, barriers to trade and investment in services, among countries with stronger IP protections tend to attract other things). Weak legal and regulatory enforce- more FDI and receive more technology flows through ment mechanisms also contribute significantly to licensing and royalties.31 Weak intellectual property the underdevelopment of local markets for services. rights (IPR) protection and weak contract enforce- ment more broadly not only limit access to GVCs, but Intellectual property rights protection also are a significant barrier to countries seeking to Complex and innovative GVCs can be influenced by a secure higher value-added activities in GVCs. country’s contract enforcement in the realm of intel- Rules on protection of IPRs have become a com- lectual property (IP). The very nature of outsourcing mon feature of PTAs over the last two decades either involves the application of know-how (such as design, through specific provisions in trade agreements or engineering, production, and business processes) as part of a bilateral investment treaty. However, the and may include formal licensing or some other specificity and strength of IPR provisions vary across form of technology transfer. Outsourcing is based on agreements, with PTAs led by high-income countries Policies to enhance participation | 173 (especially the United States and the European Union) supports entry and upgrading in GVCs by firms most commonly paying significant attention to IPRs. in developing countries.33 The research points out By contrast, many other PTAs lack adequate provi- that although traditional factor endowment and sions for IPR protection.32 demand-based explanations imply low-quality pro- duction from developing countries, in fact quality Standards varies markedly across sectors within countries.34 Standards for health, safety, the environment, labor, Certification of standards offers a way to overcome and quality are imposed by governments primarily information asymmetries and signal the quality and to protect consumers, workers, and the environment. capability of suppliers down the value chain. Without But in a GVC world, lead firms are increasingly apply- compliance, firms have limited opportunities to enter ing standards across global supply chains. Driven by such GVCs. In the absence of a credible authority to national regulatory pressures, but even more by con- enforce warranty contract and certification, repeated sumer and social demands, private standards are grow- interactions—such as through long-term contracts in ing in importance. On the one hand, these standards GVCs—can alleviate a quality signaling problem.35 establish barriers to entry into global supply chains, Case studies and impact evaluations indicate and more so for firms in developing countries that may that small institutional or technological changes can have lower levels of skills, knowledge, and technology. improve the quality of products dramatically in a On the other hand, through standards, knowledge and very short time, and the effects can be long-lasting.36 technology from FDI can be transferred in a codified For example, in only three years the quality of Malian way to firms and workers in developing countries, cotton doubled because of implementation of a cred- offering them a shortcut to GVC entry, even where the ible quality certification program, and the effects of broader policy environment may be weak. the system remained 10 years after the intervention Recent research highlights how, by overcoming (figure 7.8).37 problems of asymmetric information and negative Because adoption of private standards takes advan- reputational effects, the adoption of global standards tage of the relational nature of GVCs (that is, they are organized and governed by lead firms), they are especially attractive as a channel for GVC entry and Figure 7.8  Certification had long-lasting effects on upgrading. But governments can play a critical facili- quality in Mali’s cotton sector tating role through support for standards institutions. 100 97.84 Difference = 14.96 percentage points They can adopt flexible regulatory regimes based on (7.9 percentage points attributable to principles of equivalence, which would help ensure 90 Fairtrade quality certification) 82.88 compatibility between national and global stan- Average % of “premium quality” cotton 80 dards. Governments can also promote the adoption of standards through both regulatory enforcement 70 and advocating the adoption of voluntary standards. Difference = 23.71 60 55.91 percentage points Most important, governments can build the capacity (8.75 percentage for domestic inspection, testing, and certification and 50 points attributable to Fairtrade quality open the domestic market to international agencies. 40 certification) Effective and efficient quality infrastructure, appro- 32.20 priately recognized internationally, is a precondition 30 for delivering such demonstrable compliance. For 20 example, Pakistan’s development of a robust national quality standards regime helped to lift the European 10 Union’s ban on the country’s fish exports and facili- 0 tated rapid growth in mango and mandarin exports 2007 (3 years after quality 2014 (10 years after quality by ensuring full traceability in the supply chain. certification program begins) certification program begins, Many countries reform their national infra- 6 years after it ends) structure institutions in line with their trade, com- Certified cooperatives Noncertified cooperatives petitiveness, and regional integration frameworks. Sources: Auriol, Balineau, and Bonneton, forthcoming; Balineau 2013. Efficient and effective standards institutions and Note: The intervention was implementation of a Fairtrade quality certification in 2004. The figure shows mutual recognition by trading partners are essential the percentage of “premium quality” cotton from cooperatives that participated in the certification program versus those that did not—three years after implementation (2007) and 10 years after enablers of trade facilitation. Some countries find it implementation (2014). more feasible to share quality infrastructure services 174 | World Development Report 2020 within a regional construct. For example, a laboratory weaker bargaining power than that of the global lead for mass and volume in Trinidad and Tobago serves as firms, there is a significant risk that subsidies will a reference laboratory for 12 standards bodies in the amount largely to a transfer of rents to private inves- Caribbean region. tors at the expense of social returns. Second, subsidies may distort market outcomes (even when they seek to address a market failure). And, third, subsidies often Policies to enhance benefits create a political economy problem: once in place they As governments seek to facilitate entry into GVCs are difficult to remove because the beneficiaries lobby and upgrade to higher technology and value-added to maintain them. activities within GVCs, most will seek to go beyond Subsidy-like support for GVC firms, whether for- “getting the basics right” and undertake proactive eign investors or network lead firms, is also likely to policies, including industrial policy. Some of the most have a “beggar thy neighbor” aspect and create trade successful efforts to leverage manufacturing exports tensions. If all countries offered subsidies, the result for development, including those by China, Korea, would be global welfare losses and a race to the bot- and, more recently, Vietnam, have been closely asso- tom.38 In fact, in recent years more than half the poten- ciated with the adoption of government-led industrial tially distortionary trade policy instruments employed policies. On the other hand, outside of these East Asian worldwide have involved subsidies, export-related experiences, industrial policy has been implemented measures (including subsidies), trade-related invest- extensively with limited success. Although many of ment measures, or FDI measures (figure 7.9). Under the traditional approaches to industrial policy are WTO rules, countries that find themselves importing likely to be ineffective in today’s GVC context, that is cheap subsidized goods are allowed to impose coun- not to say that government can do nothing. In fact, a tervailing duties; they may also impose antidumping range of proactive policies show significant promise measures that target specific firms or sectors. Thus for supporting enhanced GVC participation, including: any gains in exports that stem from subsidies (which (1) promoting linkages between domestic suppliers— for the most part are prohibited by the WTO) may be typically SMEs—and GVC lead firms; (2) building reversed by action by the other country. By the end sector-specific skills and management capabilities; of 2018, 218 instances of countervailing duties had and (3) strengthening national and regional innova- been notified to the WTO and were currently in force. tion systems. Minimize the use of “traditional” Figure 7.9  Subsidies account for more than half of distortionary instruments distortionary trade policy instruments worldwide Standard industrial policy approaches of the past relied on tax incentives, subsidies, and other protec- FDI measures, 2% Nonautomatic licensing, tionist measures designed to build domestic supply quotas, etc., 3% Other, 4% chains in targeted sectors. Such instruments may Trade-related investment have a role to play if they help overcome a market measures, 4% failure (such as information asymmetries), address a Government procurement coordination failure (such as requirements for com- restrictions, 4% plementary investments in supply chains), or help Subsidies (excluding export subsidies), capture an externality (such as technology spillovers). 29% Indeed, countries such as Indonesia, South Africa, Contingent and Vietnam have commonly used such subsidies to trade-protective measures, 13% attract FDI. Too often, however, these instruments have proven ineffective or have created efficiency- sapping distortions by contributing to rent seeking Tariff measures, Export-related 17% measures (including and misallocation of capital. They are also increas- export subsidies), ingly problematic in a GVC environment, where full 24% supply chain development is not necessary and trade integration is paramount. These traditional approaches have a number of other drawbacks as well. First, in the GVC context, Source: WDR 2020 team using data from Global Trade Alert (https://www.globaltradealert.org/). which often finds national governments having Note: Data are from November 2018. FDI = foreign direct investment. Policies to enhance participation | 175 Of these, 162 were applied either to metals and metal linkages many countries employ local content require- products or to chemicals, rubber, and plastics and ments either as conditions for foreign investments or products thereof, suggesting that trade in those sec- as requirements that foreign investors must meet to tors is particularly distorted by subsidies. In addition, access public procurement (box 7.3). In the absence of charges of export subsidies can entangle countries in quality local suppliers, however, such requirements WTO disputes about both the subsidies and the coun- can backfire, restricting investment. tervailing duties in response. Similarly, in many natural resource sectors policy Local content policies have been similarly prob- makers may focus on developing forward linkages— lematic. For example, in an effort to develop backward and raising domestic value added—by requiring local Box 7.3  Local content requirements are a mismatch in the global auto industry The global auto industry is characterized by extended value into effect in 2019. Rota 2030 seeks to simplify complex chains, with parts and components produced on a large local content rules and increase R&D spending require- scale and exported worldwide to maximize efficiency. Both ments in part through additional government grants. Brazil and South Africa have invested heavily in and sig- Energy efficiency targets, vehicle identification, structural nificantly protected development of their domestic auto- performance, and incentives for electric cars are also motive sectors over the past two decades. And yet, despite included. Like Inovar-Auto, however, the policy continues to the huge costs, the countries are struggling to maintain focus on the domestic market over exports, and importers competitiveness, and the long-term sustainability of the will be excluded from the program, suggesting that it may sectors remains in question. not be enough to bring Brazil’s auto industry into modern value chains, which thrive on global content. Brazil Notwithstanding already high levels of protection (roughly South Africa 60 percent local content requirement), automotive imports The mixed performance of South Africa’s extensive incen- in Brazil rose in the late 2000s, prompting the domestic tives and policy interventions in the automotive sector industry to lobby the government for further protection. demonstrates how difficult it is to use industrial policies The Inovar-Auto policy (2011–17) imposed additional local in an environment in which the comparative advantage content requirements, this time including incentives for R&D is uncertain. The automotive sector has benefitted from spending, structured primarily around tax benefits. Although state support since its inception, starting with the Motor the policy diminished the effects of Brazil’s 2014 economic Industry Development Programme (MIDP) from 1995 to crisis on the auto sector, it did not boost productivity, nor 2012, which was replaced by the Automotive Production did it improve export competitiveness.a Indeed, a study of and Development Programme from 2013. The program the 12 largest automakers between 2007 and 2015 revealed started with extensive protection from import compe- that average production per automaker declined from tition and local content requirements under the MIDP, 233,186 units to 195,747 units per year. Scale efficiency likely shifting more recently to some liberalization and investor worsened because of the overinvestment that was incentiv- subsidies. Several major automakers operate in South ized by the policy, and employment levels did not change. Africa, and they have created some 150,000 jobs in the Meanwhile, rising costs, declining productivity, and declining industry, but it has never managed to thrive on its own. profit margins continued across the industry. And although Although the auto sector has become more competitive competition among domestic producers increased (the pol- over time, it has not performed nearly as well as those in icy attracted new market entrants and increased investments Mexico and Thailand, which benefit from better connec- from existing producers), prices went up because domestic tivity with both the Asian production hub and global con- automakers were protected from import competition. sumer demand. South African producers export largely to Inovar-Auto is in the process of being replaced by Rota receive duty drawbacks on imports, while linkages to local 2030, a new policy for the automotive industry, which came suppliers remain limited.b a. Sturgeon, Chagas, and Barnes (2017). b. Black, Barnes, and Monaco (2018). 176 | World Development Report 2020 processing or by taxing exports of unprocessed or Figure 7.10  The share of locally supplied inputs in semiprocessed commodities. Such strategies have GVCs varies by sector and country the potential to overcome coordination failures and unlock profitable, value-adding investments, but they Eswatini 1 Apparel are highly context-dependent and are determined by Lesotho 1 a combination of market power and the basic eco- Kenya 4 nomics of production and transport. For example, Mozambique 13 Botswana’s dominant position as a source of high- Mining quality diamonds enabled the government to nego- Ghana 25 tiate a relocation of De Beers’s sorting, aggregation, Chile 30 and sales operations from London to Gaborone, Ghana 19 Agribusiness which has contributed to substantially strengthening Kenya 20 Botswana’s value-added position in the diamond value chain. Elsewhere, export taxes have helped tip the Mozambique 46 balance to expand domestic processing of agricultural Vietnam 53 products such as cashews in India and Vietnam. On 0 10 20 30 40 50 60 the other hand, the literature is filled with examples of poorly designed export bans or taxes that have Percent contributed to collapsing prices for farmers or pro- Source: Farole and Winkler (2014), based on 400 surveys of lead firms and suppliers conducted between March 2012 and October 2012. duction (such as cashews in Mozambique and maize in Malawi and Tanzania) or otherwise created seri- ous distortions across the value chain (for example, suppressed indefinitely. This scenario is apparent Argentina’s 2006 beef export ban). in many countries that have participated in apparel GVCs. For example, although Lesotho and Swaziland Promote domestic supply chain linkages (now Eswatini) experienced the rapid transformation and FDI spillovers of their economies when they attracted foreign invest- Establishing linkages between lead firms and domes- ment in apparel GVCs, after more than 20 years that tic SME suppliers is the starting point for leveraging sector remains almost wholly disengaged from the spillovers and upgrading in GVCs. The extent of domestic economy and no upgrading of the sector supply linkages varies dramatically across countries has taken place. As a result, uncertainties over trade and GVC sectors (figure 7.10). Although some of the preferences and wage pressures constantly threaten variation is structural in nature, there is scope for the viability of investments. significant densification of GVCs in many develop- Governments can play a role in providing the infor- ing countries. Support for building these domestic mation needed to bring local SMEs together with FDI supply linkages would be an important proactive through supplier linkage programs (box 7.4). Where government policy that would help reap the benefits the local supplier base is fragmented and character- of GVCs. ized by very small, often informal, enterprises, coordi- Realizing the potential of GVCs for productivity nation of suppliers through cooperative structures or gains through spillovers of knowledge and technol- associations can be important for helping producers ogy is by no means guaranteed. Indeed, the barriers achieve greater scale, allowing for investments in to spillovers may be even higher than they are in non- common goods, and pooling knowledge and expertise. GVC environments. GVCs, with their global gover- It can also enable suppliers to engage more effectively nance of supply chains and often footloose investing, with lead firms. create an environment in which foreign investors may Governments can help deepen domestic supply have little incentive to invest in research and labor chain relationships through broad reforms of their market integration in host countries and in which country’s investment climate. This is particularly crit- technologies and processes for production may be ical for domestic investors, who may not be in a posi- significantly disconnected from local realities. The tion to benefit from targeted investment incentives implication is that the process of upgrading within or SEZ programs that are available to large foreign GVCs may be curtailed, risking the sustainability of investors. Moreover, and at minimum, governments investment in the first place because the attractive- must be sure to avoid displaying a bias against domes- ness of a location remains reliant on access to inputs tic investors. For example, many SEZs, either by rule (labor or natural resources) whose price cannot remain or de facto, exclude domestic investors, especially Policies to enhance participation | 177 Box 7.4  Supplier development programs help deliver inclusive, sustainable GVCs Guinea Linkages Programa Malaysia’s Industrial Linkages Programc As part of the development of a major iron ore mine in Established in 1996, Malaysia’s Industrial Linkages Program Guinea, the International Finance Corporation (IFC), (ILP) is a cluster-based program centered on fiscal incentives together with lead investors Rio Tinto and Guinea Alu- for both multinational corporations (MNCs) and SMEs. It mina, initiated a pilot supplier linkage programmed at includes components of business matching, support for skills integrating local small and medium enterprises (SMEs) development, access to industrial sites, and financing for into the mining supply chain. The program combined SMEs. SMEs become eligible to participate in the program if informational support of mining procurement teams and they meet certain criteria. Most important, they must supply at comprehensive supply-side support for potential local least one MNC and manufacture a product on the “List of Pro- SME suppliers, including training, managerial capacity moted Activities and Products.” Once accepted, they receive building, support for achieving procurement standards, fiscal benefits, allowing them a tax exemption of 100 percent and assistance in gaining better access to finance. After of statutory income and an investment tax allowance of 60 just a couple years of operation, the program achieved percent on qualifying capital expenditures incurred within significant results: five years. They are also offered “matching services” from SME Corporation Malaysia (the country’s SME agency), which More than 100 local SMEs upgraded their capacity •  facilitates relationships with the MNCs to support upgrading. through the program. In its first decade of operation, more than 900 SMEs were Over $9.1 million in new contracts were signed •  registered with ILP, of which 128 were linked to MNCs. between local businesses and international mining companies. Czech Pilot Supplier Development Programd Over 700 new jobs were created in local businesses •  Through CzechInvest, the Czech investment promotion as a part of the mining sector’s supply chain. agency, the Czech government implemented a pilot Chile’s World-Class Supplier Development Programb National Supplier Development Program from 2000 to Chile’s World-Class Supplier Development Program was 2002 in the electronics and automotive sectors. The moti- launched in 2008 by BHP Billiton, and it has since vation for the program was to raise local content in these expanded to include other mining companies such as sectors to widen foreign direct investment benefits to the Codelco. The program is coordinated by Fundación local economy and strengthen these sectors.e The pro- Chile, a nonprofit corporation that is seeking to support gram, which was demand-driven, sought to improve the technology transfer and innovation and increase the competitiveness of Czech SMEs, thereby enabling them to competitiveness of Chilean firms across the economy. enter GVCs by becoming suppliers to MNCs. A dozen MNCs The project’s goal is to create 250 world-class suppliers in were involved in the project, and 45 SMEs received tar- Chile by 2020. The model encourages mining companies geted training based on needs uncovered during business to identify areas in which innovative solutions could con- reviews. An evaluation revealed that within 18 months of tribute to operational efficiency across their operations completion of the program, one-third of participants had and identify local suppliers who have the capacity to work gained new business, which they attributed to the pro- on the problem. The selection procedure is rigorous—only gram, benefiting from contracts worth $46 million for the 16 percent of identified projects at Codelco reached the period 2000–2003. The share of components sourced from implementation stage. Selection criteria include eco- Czech companies by the MNCs participating in the program nomic benefits, replicability, urgency of the problem, correspondingly increased, from a rate of 0–5 percent at technological risk, and impact on health, safety, and the the start to 2.5–30 percent by 2004. Driven by supply-side environment. Through 2014, more than 70 projects were improvements in export performance, the Czech Republic implemented, and a number of suppliers have expanded experienced significant gains in global market shares and exports as a result. continual improvement in product quality. a. World Bank (2015). b. Farole and Winkler (2014). c. Malaysia Ministry of International Trade and Industry (2019). d. Malinska and Martin (2000–2002). e. The country had been one of the most successful at attracting FDI since the fall of communism in the 1990s, but relatively few of the investments were felt by the local economy. 178 | World Development Report 2020 local SMEs, by imposing minimum size require- For local SMEs to absorb spillovers from GVC ments, mandating establishment of a new business participation, ongoing investments are required in entity, and placing restrictions on mixing domestic technology, process improvements, and training. In and export businesses, among other things. More- fact, lack of financing is one of the main obstacles over, physical (customs gates) and financial controls, to GVC participation among suppliers in developing along with financial incentives (for example, firms economies (figure 7.11). Policies that facilitate access inside SEZs can import inputs duty-free but must pay to credit via financial sector reforms, the provision of the value added tax or deal with complex drawback information, as well as incentives such as matching arrangements when buying from a local supplier) grants and loan guarantees can play an important role. may prove to be barriers for local SMEs taking advan- Beyond pure financing, incentives can be made avail- tage of GVC opportunities. By contrast, in Bangladesh able to support technology transfer and licensing, a the government intervened directly to address two major source of spillovers for local suppliers in GVCs. specific investment climate constraints faced by local New financial technologies are helping GVC sup- manufacturers in the apparel GVC by introducing pliers improve their access to supply chain financing, a bonded warehouse scheme that enabled duty-free effectively leveraging the higher credit rating of their imports for export production and a “back-to-back” let- global buyers to access financing on better terms. Tools ter of credit that would allow manufacturers to obtain such as electronic invoices and e-receivables speed and credit for input purchases secured by export orders. improve communication among customs brokers, Governments also play a central role in building a freight forwarders, transportation carriers, govern- local absorptive capacity. Research shows that direct ment agencies, and banks. For example, seven global technical assistance from lead firms—either through banks recently announced formation of the Trade formal linkage programs or as part of the normal Information Network to digitize trade finance. Other firm-client relationship—is one of the biggest sources examples of financial technology (fintech) innovations of spillovers to local suppliers.39 However, strengthen- include the use of “smart” factory technology, which ing the absorptive capacity of local firms and workers collects frequent data on production and assembly also depends on government policies to support, lines and can be used for credit scoring, and Bluetooth among other things, access to finance and technology, scales, which are used in agribusiness chains to accu- as well as skills development. rately weigh farmers’ harvests and provide real-time Figure 7.11  Lack of financing impedes low-income country suppliers the most from entering or moving up in GVCs Business environment Access to finance Business environment Labor skills Supply chain governance Investment climate Inadequate ICT networks Power supply Transport costs and capacity Trade issues Customs procedures Import duties Licensing requirements Meeting standards 0 5 10 15 20 25 30 35 40 45 50 55 60 65 % citing issue as major obstacle Low-income country suppliers Lead firms Source: Cusolito, Safadi, and Taglioni 2016. Note: ICT = information and communication technology. Policies to enhance participation | 179 lines of credit at the point of sale. In addition, new mod- education system and coordination through works els reward firms that have better sustainability ratings councils, is being adapted to other countries. Other with cheaper financing to support the significant costs examples, such as the Penang Skills Development imposed on SMEs to meet international standards. Centre in Malaysia (box 7.5), illustrate how govern- For example, Puma, BNP Paribas, and the fintech firm ments, in coordination with the private sector, can GT Nexus offer better receivable financing (discount) build strongly territorialized capabilities through an terms to suppliers who score high on Puma’s sustain- industry or cluster-led skills development initiative. ability index. Levi’s has a comparable arrangement Turkey is an example of a country that has man- with its suppliers through the Global Trade Supplier aged to successfully move up the value chain in the Finance program of the International Finance Corpo- apparel GVC. Its firms are assuming design roles and ration (IFC). Investors are also designing “green” bonds even building global brands. This achievement has that pool smaller loans for GVC suppliers to invest in been supported by both the private and public sectors environmentally friendly technology. and their active workforce development efforts. For example, the Istanbul Textile and Apparel Exporter Invest in sector-specific skills, Associations (ITKIB) partnered with the private sector management, and innovation capabilities and government agencies to promote vocational train- ing in fashion design. The Istanbul Fashion Academy Developing sector-specific skills is a partnership of the European Union and ITKIB. The Although human capital development is a long-term Small and Medium Industry Development Organiza- process going back to foundational education and tion (KOSGEB), a quasi-governmental organization, early childhood development, much can be done to has also been involved in workforce development; it build industry-specific skills. In many developing provides marketing support, training, and consulting countries, there are large gaps between the outputs services. The movement into own branding has also of traditional education and skills development been supported by government incentives, including institutions and the needs of employers.40 Targeted reimbursement of up to 60 percent of the cost of workforce development strategies can bridge these personnel expenses for a maximum of three years gaps, ideally linking lead firms and local institutions, (including training and recruiting highly qualified including universities and vocational and technical personnel), machinery, equipment and software, con- centers. The German model, which includes a dual sultancy, and R&D-related materials. Box 7.5  Building a workforce with industry-specific skills: Penang Skills Development Centre The Penang Skills Development Centre (PSDC), the first to provide up-to-date training and educational programs industry-led training center established in Malaysia, was in support of operational requirements and to stay abreast conceived in 1989 in response to an urgent sense that of technology. The center operates on a full-cost basis— if Penang was going to continue to attract foreign direct companies (FDI and local) pay to send employees for train- investment (FDI), its human capital would have to be ing. To ensure that the training meets the needs of industry, trained to keep pace with changes in technology. Although the programs are continually upgraded and adapted to the state and federal governments launched the initiative evolving skill needs. and provided the land and some financial support, Malay- The center has trained more than 200,000 workers by sian and foreign private companies played the leading role means of more than 10,000 courses, pioneered local indus- in establishing the center. Not only did these companies try development initiatives, provided input and helped furnish the initial trainers and equipment, but they also formulate national policies for human capital development, designed the training programs to meet their needs. and contributed directly to the Malaysian workforce trans- PSDC has more than 200 members and operates as a formation initiatives. Meanwhile, the PSDC model has been nonprofit society. Its mission is to pool resources among adopted throughout the country—skills development cen- the free industrial zones and industrial estates in Penang ters operate in almost all states in Malaysia. Source: Adapted from Farole (2011). 180 | World Development Report 2020 Governments can also facilitate access to skilled Figure 7.12  Managerial know-how is associated with labor by ensuring open labor markets and helping greater GVC participation in Mexico match investors’ needs with the available local skills. 8 In many developing counties, lack of skills in techni- cal and managerial positions is a binding constraint 7 to upgrading in value chains. Pervasive skills gaps % of firms that participate in GVCs often result in a large wage premium for these posi- 6 tions, as well as in professions such as accounting and engineering. Nevertheless, explicit policies to promote 5 “localization” of skilled jobs often result in investors 4 facing high barriers to obtaining work permits to bring in skilled workers. By contrast, some countries 3 actively help GVC investors identify skilled labor. For example, the Chengdu Hi-tech Industrial Develop- 2 ment Zone gives priority to talent recruitment, assist- ing companies in the zone with their recruitment 1 efforts both within China and abroad. 0 1 2 3 4 5 6 7 8 9 10 Developing management and firm capabilities Deciles of management score Although most skills development policies target Source: WDR 2020 team, using data from ENAPROCE 2015. See appendix A for a description of the workers, an equally important constituency typically databases used in this Report. undersupported is firms and their managers. Accord- ing to a growing body of research, firms differ greatly boost productivity, but also a useful tool to support in management capabilities and practices, especially GVC integration. in developing countries, where productivity and Another type of market failure takes the form of profitability vary significantly.41 Governments can uncertainty and limited information about demand. support firm upgrading and boost firm productivity Firms are then unwilling to invest in searching for by correcting market failures, including encouraging potential buyers when competitors may also ben- firms to improve their managerial practices and build efit from their investments. This failure especially relationships with buyers. affects young firms, which are often more productive Recent studies point to several market failures than incumbents but less likely to survive adverse that result in firms underinvesting in management. shocks because of underdeveloped relationships with Information asymmetries are manifested in man- buyers.46 In this context, helping firms discover agers who “don’t know what they don’t know,” and markets and building relationships with clients can therefore they systematically misdiagnose the quality improve product quality and raise overall productiv- of the organization and management of their com­ ity. For example, in a randomized controlled trial in pany.42 These asymmetries are further compounded which Egyptian carpet producers were given access by uncertainties about the returns on investments to demand from high-income foreign markets (such in improving management and organization, as well as the United States and the European Union), the as lack of information on the quality of providers treated firms experienced a 16–26 percent increase of management consulting services.43 When firms in profits, driven by higher quality and learning-by- do invest in improving management, they not only doing as their product quality improved over time.47 experience much higher profits, productivity, and The ability of firms to effectively match the needs job growth,44 but also improve product quality and of foreign buyers—a core requirement of participating increase the likelihood of exporting.45 In Mexico, firms in GVCs—requires a combination of good manage- in the top decile of the managerial practices index ment and actively accumulating demand. Successful are more than seven times more likely to participate programs to support supplier development, such as in GVCs than firms in the bottom decile (figure 7.12). those in Chile and the Czech Republic, typically com- This and other evidence from developed and devel- bine interventions that address both supply-side and oping countries indicate that financial incentives demand-side market failures. In Chile, the Chilean or direct support to firms to facilitate improvements Innovation Agency (CORFO) set up a large matching in management is not only a cost-effective way to grant program in which lead firms would apply for Policies to enhance participation | 181 support for their SME suppliers and CORFO would private funding. The Fraunhofer Society is responsi- cofinance a six-month consulting diagnostic and up to ble for applied contract-based research that bridges three years of diagnostic implementation. An evalua- basic research and industrial demand. Emerging new tion of the impact of the program revealed significant forms of cooperation within the innovation system, increases in supplier survival, sales, employment, and mainly privately funded, involve the creation of salaries, as well as positive effects on the sales and institutions to bridge the existing centers of knowl- exporting likelihood of the lead firms two years after edge and skills. Of these, new forms of collaboration joining the program.48 between universities and industry have proliferated. Strengthening innovation systems Consider special economic zones as a The capacity of national and regional innovation sys- possible shortcut to GVC participation tems also needs to expand.49 The range of technical, Delivering on the policy priorities outlined in this engineering, and managerial skills to sustain complex chapter is no easy task, least of all for developing manufacturing, much less innovation-intensive GVCs, countries, which almost by definition face significant is substantial. Although innovation systems—univer- weaknesses across many of these policy areas. What sities, government, firms, and specialized research then can these countries and the firms operating in institutions—vary in their configurations and role, them do to improve their chances for GVC participa- the desired outputs of an innovation system’s capabil- tion in the short term, while taking the steps needed ities are similar. Whatever forms such systems take, to improve the policy environment over the medium knowledge must flow among firms, government, and term? This section discusses the possibility of using universities. Agglomerations of innovation—such as SEZs as a means of shortcutting GVC participation. Silicon Valley in California, Cambridge (U.K.), Banga- SEZs are demarcated geographical areas within a lore, London, Berlin, and Dublin—are a feature of this country’s national boundaries where the rules of busi- stage. Governments can even establish innovation ness are generally more liberal than those that prevail parks to induce agglomerations of innovators. in the national territory. Specifically, most economic The German innovation system primarily focuses zones create a “special” regime (box 7.6) that usually on developing complex innovations along known confers four main advantages to investors relative technology trajectories. The existing knowledge to what they could normally receive in the domestic in auto manufacturing, mechanical and electrical environment: engineering, and chemicals is mature enough that incremental improvements tend to have clear market • Infrastructure (including serviced land, factory shells, applications. In turn, the development of both services and utilities) that is easier to access and more reli- and advanced manufacturing is a central determinant able than is normally available domestically of the long-run rate of economic growth.50 • A customs regime that includes efficient customs The development of a rich national innovation administration and (usually) access to imported system involves a great deal of networking and a inputs free of tariffs and duties wide variety of institutions—in effect, value chains • A regulatory and administrative regime that includes of knowledge. In Germany, knowledge-intensive streamlined procedures for company setup, licens- service sectors include both traditional professional ing, and operations services such as marketing and advertising and • A fiscal regime that includes reduction or elimination technology-based services such as software and of corporate taxes, the value added tax, and other computer systems design and R&D. A wide array of taxes; labor contributions; and sometimes training institutions mediate the relationship among private or other subsidies.51 sector R&D, the university system, and the govern- ment, fulfilling the functions of coordination and SEZs are designed to facilitate trade and attract cooperation. These institutions vary both in their FDI, but governments may also seek to take advan- focus on nonappropriable basic research versus tage of other potential benefits of SEZs. Examples are marketable applied research and in their mix of pri- capturing agglomeration economies,52 which happens vate and public funding. The Max Planck Institutes, through exploiting backward and forward linkages;53 “‘Blue List” institutes, national research centers, and labor pooling, which facilitates matching between subnationally focused “Länder” institutes are largely firms and workers;54 and technology spillovers.55 publicly funded institutions that focus on basic In some countries, SEZs have been used to pilot research. Universities receive a mix of public and experimental policy reforms. In China, for example, 182 | World Development Report 2020 Box 7.6  Clarifying the terminology: SEZs versus industrial parks The term special economic zone (SEZ) may be used to regulatory regime, typically covering customs (such as refer to any one of the similar spatial industrial instruments duty-free imports and exports), fiscal issues (such as tax- known as free zones, free economic zones, export pro- ation), and potentially a broad range of special regulatory cessing zones, industrial zones, economic and technology regimes (such as on company registration and labor). SEZs development zones, high-tech zones, science and innova- may be geared to manufacturing, but often they accommo- tion parks, free ports, and enterprise zones.a Even though date mixed-use development, including services, and also the terms SEZ and industrial park are often used inter- may include commercial and residential activities. changeably, there are important policy and operational These differences matter because they have import- differences between the two. ant implications when choosing between an SEZ and an Industrial parks are property developments that are industrial park. When governments are dealing with land zoned for industry or manufacturing activity. A government constraints, when they need to concentrate infrastructure or a private property developer may prepare services sites investment, or when they are primarily seeking to promote or even build infrastructure, but industrial parks are not agglomeration but do not need to create a policy and reg- necessarily governed by any special fiscal, customs, or regu- ulatory environment that differs from the existing domestic latory regime. Thus industrial parks are not necessarily SEZs. environment, an industrial park is likely to be sufficient. It is An SEZ may take the form of an industrial park, or only when a special regulatory regime is needed and there an industrial park may be located in it. However, what are good reasons why this cannot be done nationally that makes an SEZ “special” is that it operates within a special an SEZ is the appropriate instrument. a. Zeng (2015). financial, legal, labor, and even pricing reforms were manufacturing sectors in economies previously reli- introduced first within its SEZs before being extended ant on agricultural commodities. In the Middle East to the rest of the economy. and North Africa, SEZs have played an important role Whatever the objective, one the main attractions in catalyzing export-oriented diversification in coun- of SEZs as an instrument has always been the idea tries such as the Arab Republic of Egypt, Morocco, and that they can act as a shortcut to infrastructure invest- the United Arab Emirates. And in Sub-Saharan Africa, ments or policy reforms that would take many years SEZs in Mauritius have been a central policy tool to deliver, if at all, across a country. Instead of build- supporting a highly successful process of economic ing infrastructure or enacting a policy everywhere, diversification and industrialization. which could be financially, technically, and politically And yet despite these success stories, SEZs have a infeasible, a country could concentrate its efforts on mixed record (box 7.7). In some countries, the zones one or two specific locations where the environment have failed to attract investors, leaving “white ele- could be designed specifically to meet the needs of phants” that inflicted both fiscal and political damage. GVC investors or where difficult policy reforms could In other countries, SEZs have been exploited by inves- be contained. tors to take advantage of tax breaks without delivering substantial employment or export earnings. And in SEZs: A mixed record many countries, traditional export processing zone In some countries, the SEZ model has delivered programs have been successful in attracting invest- spectacularly, playing a catalytic role in growth and ment and creating employment in the short term but structural transformation. Examples include China have failed to sustain competitiveness in the face of and Korea, which used SEZs as a platform to support rising wages or eroding trade preferences.56 the development of export-oriented manufacturing. Overall, any kind of empirical assessment of SEZs In Latin America, the Dominican Republic, El Sal- (beyond individual zones and country programs) and vador, and Honduras, among other countries, have their determinants has proven difficult. Even the most used free zones to take advantage of preferential serious studies have tended to be plagued by small access to U.S. markets and have generated large-scale sample sizes and difficulty in obtaining comparable Policies to enhance participation | 183 Box 7.7  Comparing SEZ experiences: China, India, and Sub-Saharan Africa China Suppliers and ancillary units co-locating within the SEZ to China’s special economic zones (SEZs) have been a well- supply anchor investors are unable to claim income tax documented global success story. They account for about 22 exemptions. Such tensions, a direct result of competing percent of its GDP, 46 percent of foreign direct investment policy objectives, have limited the development of linkages (FDI), and 60 percent of exports, generating more than 30 between zones and the DTA, further eroding the “special” million jobs,a or about 60 percent of global employment in environment of zones in India. SEZs.b An analysis of panel data for 270 cities at the prefec- The Indian experience suggests that zone performance ture level over 23 years shows that opening a major zone in a depends on operational factors working in tandem rather city led to an increase in GDP of 12 percent on average in the than a single dominant factor. Zone performance, often postreform years, with the effect depending on the type of measured by export growth, ability to attract investors, and zone. The long-term (cumulative) effect of an SEZ could be a other indicators, is a complex function of internal and exter- roughly 20 percent increase in GDP.c Another analysis of 321 nal factors. In India, almost all zones with higher exports prefecture-level cities between 1978 and 2008 finds that on are in states with a supportive regulatory environment, are average an SEZ program increases per capita FDI by 21.7 per- close to seaports, and have access to skilled labor through cent and the growth rate of FDI by 6.9 percentage points.d proximity to urban centers. Several were set up by the Moreover, the average wage of workers in municipalities central government under the previous export promotion with an SEZ increased by 8 percent more than that of the regime, giving investors a sense of confidence. New zones control group, against a 5 percent rise in the cost of living.e set up as public-private partnerships offer superior infra- The performance of Chinese zones has not, however, structure and quick approvals, attracting more investors. been uniformly outstanding. As zones have proliferated, But some zones could not sustain operations because of especially at the provincial level, their marginal impact has inexperienced private developers and underinvestment in diminished.f In addition, many zones have suffered from infrastructure improvements, despite having state support. environmental degradation, as well as from challenges in social services delivery, including inadequate health, Sub-Saharan Africa Several Sub-Saharan countries launched zone programs as education, and transport services. They have also lacked far back as the early 1970s, but most came into being in cultural and recreational activities for workers. In the the 1990s or 2000s. Modern SEZs did not appear until after 2000s, China responded by shutting down a large number 2006. The early SEZ record in Africa is less than spectac- of poorly planned industrial zones, improving the coordina- ular. Except for Mauritius and some modest achievements tion between zones and urban and regional planning, and in Kenya, Lesotho, and Madagascar, most Sub-Saharan SEZ seeking to increase the role of market forces. programs have not had a transformative impact. A 2011 analysis comparing African SEZs with those in other parts Indiag of the world developed several stylized facts: Over time in India, policy decisions have contributed to erosion of the “specialness” of SEZs. For example, the • The takeoff of export growth in African SEZs was less overall incentive and support package available to firms in significant than that outside of Africa. the domestic tariff area (DTA) is often more beneficial and • SEZs accounted for a smaller share of industrial easier to use than the zone-specific incentives. In addition, employment (except in Lesotho) and much smaller firms in the DTA can access the domestic market. With the absolute levels of industrial employment than that proliferation of new free trade agreements with Japan, the enjoyed outside of Africa. Republic of Korea, and member countries of the Association • Although structural transformation of exports, as of Southeast Asian Nations (ASEAN), exporters in the DTA measured by diversification into manufacturing, took can import with reduced or no duties from these countries place fairly rapidly in SEZ-intensive countries outside instead of importing tariffed goods from zones. By con- Africa, it has been more limited in Africa. trast, India’s SEZ policy framework restricts market access • African SEZs have provided weaker enabling condi- to the DTA, thereby constraining value chain development. tions than those in the rest of the world.h (Box continues next page) 184 | World Development Report 2020 Box 7.7  Comparing SEZ experiences: China, India, and Sub-Saharan Africa (continued) Despite relatively weak performance to date, SEZ pro- a public-private partnership involving Olam International, grams remain highly popular across the continent, and pol- the Africa Finance Corporation, and the Republic of Gabon, icy makers seem determined to learn from the lessons of the are showing significant promise. And early evidence from past, both within Africa and globally. As a result, some coun- the rapidly developing SEZ program in Ethiopia, including tries, such as Ghana, Kenya, South Africa, and Tanzania, have at Bole Lemi, Eastern, and especially Hawassa, suggests that revisited and reformed their SEZ programs in recent years. Ethiopia may have the conditions and approach to make Other recently developed zones, such as Gabon’s Nkok SEZ, SEZs a successful instrument of GVC integration. a. Zeng (2010). b. Farole (2011, 43). c. Alder, Shao, and Zilibotti (2013). d. Wang (2013). e. Zeng (2015). f. Wang (2013). g. This section relies on Saurav (n.d.). h. Farole (2011). measures of SEZ performance. More recent work social infrastructure, and connectivity to national examines 346 zones in 22 countries across the devel- and global markets. Yet governments continue to try oping world and Korea using night lights data from (and fail) to use zones as regional development tools. satellite observations as a novel way of measuring The majority of countries with zones decide to locate zone activity.57 One critical finding of the study, which at least one in a “lagging” or remote region, and few reinforces conclusions from previous work,58 is that have done enough to address the infrastructure con- SEZs find it difficult to significantly outperform the nectivity, labor skills, and supply access that these underlying economy. Few of the zones included in regions tend to lack. Not surprisingly, foreign inves- the study experienced growth much higher than the tors typically shun these locations in favor of more national average, and many grew at a rate lower than central ones—a preference that has been on display in the national average. SEZs tend to perform better in Bangladesh, Indonesia, Thailand, and Turkey, among national economies that are open, growing, and com- other countries.59 petitive than in those that are not. Although SEZs are often implemented specifi- cally to catalyze the development of new sectors, a Lessons for successful implementation location’s comparative advantage remains essential. of SEZs An extensive market assessment will reveal what SEZs are not easy to get right. And even successful factors drive investment decision making, and a SEZs usually take a decade or more to start showing realistic assessment of the location will reveal what results. Policy makers should approach SEZs with a it has to offer. Gaps between comparative advantage clear objective, a long-term commitment, and a strong and SEZ targeting may explain why countries that technical team. Among the many lessons that they have specialized in natural resources but do not should take to heart in planning SEZs are concentrat- have competitively priced labor and efficient infra- ing on only the best location; understanding the mar- structure (such as Ghana, Kuwait, Nigeria, and to ket and leveraging comparative advantage; and, most some degree Indonesia and Peru) have struggled to important, ensuring that zones are “special.” develop manufacturing-oriented zones. Mauritius is A consistent finding from empirical research is a good example of a country that has leveraged the that location choice is critical to success. International zone instrument over several phases to exploit evolv- experience supports that finding, with SEZs flourish- ing sources of comparative advantage. The export ing in core areas and around gateway infrastructure processing zone model, so successful in transform- (seaports, airports). Cities offer features that tend to be ing Mauritius from its reliance on sugar and vanilla essential to the success of large-scale, labor-intensive plantations to becoming a major apparel exporter, SEZs, including access to deep and specialized labor eventually became obsolete. However, as its source of pools, specialized suppliers and business services, comparative advantage moved away from low wages, Policies to enhance participation | 185 the government returned to the zone instrument to Attracting FDI would be the quickest way to amass promote emerging industries such as ICT and finan- such capabilities. A country would, then, have to cial services.60 address its business climate constraints and establish With a clear understanding of investors’ needs, simple procedures for registering foreign investors. countries can design and deliver zones that fully Foreign investors will also want to be assured of basic overcome the existing constraints to investment. If political stability and rule of law, but deep institutional investors need reliable electricity, the SEZs should reforms may not be critical at this stage. Competitive guarantee no downtime. If they need smooth cus- labor costs are important at this point, but less so for toms clearance as a priority, SEZs should ensure that GVCs that involve processing of natural resources customs authorities resolve all possible reasons for (such as agriprocessing) and more so for those that delays. Too often SEZs are not, in fact, special. For mainly make use of imported inputs (such as apparel example, a survey of global SEZ investors found that and electronics). infrastructure (especially electricity quality) was Because imported inputs play a large role in basic among their top considerations in choosing an invest- manufacturing GVCs, countries should give priority ment location, and that customs and trade issues were to measures that would support trade, including those also a high priority. However, that survey also found that would improve physical connectivity, in particular that although successful global SEZs in the survey vir- through critical trade-related physical infrastructure tually eliminated downtime from electrical power out- such as ports and first-generation trade facilitation ages, issues with electricity remained fairly frequent reforms. Tariff reforms—at least for selected goods— in the African SEZs, even though there were some may ensure access to competitively priced inputs or improvements compared with the situation in the rest involve the use of a mechanism such as bonded ware- of the country. As for customs clearance, the times at houses, duty drawbacks, or SEZs. Finally, countries seaports were actually worse in the SEZ than outside should seek to secure market access through PTAs. the SEZ in half of the African countries surveyed. Finally, it is important to recognize that SEZs Transitioning to advanced manufacturing cannot overcome all the constraints that may restrict and services access to GVCs. Once outside the gates of an SEZ, prob- Transitioning to advanced manufacturing and ser- lems of poor infrastructure, predatory institutions, and vices GVCs presents a much bigger challenge than lack of safety and security may become binding. Such that to basic manufacturing. Examples of such sectors problems can affect SEZ inputs and outputs traveling are motor vehicles, medical devices, aerospace, and between the zone and the port. They can also affect precision instruments. Countries that have recently the managers and workers who must go in and out of succeeded in one or more of these sectors, though the zone on a daily basis. More broadly, macro factors, not necessarily on the aggregate, include Costa Rica such as a volatile exchange rate, may present problems (box 7.8), Poland, Turkey, and Vietnam. Moving into that are difficult to shield from SEZ investors. these activities requires a step change in the policy environment. Although labor costs still matter for some parts of Policies for upgrading the value chain—for example, in the final assembly of This chapter has highlighted a broad range of policies electronics and in some auto components such as igni- that can help countries to accelerate GVC participa- tion wiring sets—advanced manufacturing GVCs typ- tion, to deepen the levels of participation, and to cap- ically require a more highly educated workforce. The ture more of the gains from GVCs. But some policies range of technical, engineering, and managerial skills are more salient than others, depending on the stage needed to sustain complex manufacturing is substan- of GVC participation. Figure 7.13 is a summary of the tial. Improvements are needed in national education policies that countries should consider as they plan and employability policies and programs, but because their transition to the next stage of GVC participation. many of these skills may be firm-specific, a policy environment that is open to bringing in foreign skilled Transitioning from commodities labor and that incentivizes foreign investors to invest specialization to limited manufacturing in training and transfer tacit knowledge is needed as GVCs well. Policies that prescribe the use of domestic part- To move into downstream manufacturing from a ners or force technology transfers can be inhibiting. commodities base, a country would likely have to Finally, because domestic supply capabilities will be acquire new technological and managerial capabilities. increasingly important for advanced manufacturing 186 | World Development Report 2020 Figure 7.13  Different policy priorities underpin the transitions between types of GVC participation Commodities to limited Limited manufacturing to advanced Advanced manufacturing and manufacturing manufacturing and services services to innovative activities Fundamentals Policy priorities Foreign direct investment: adopt supportive investment policy and improve the business climate Endowments Finance: improve access to banks Finance: improve access to equity finance Labor costs: avoid rigid regulation and Technical and managerial skills: Advanced skills: educate for exchange rate misalignment educate, train, and open to foreign skills innovation and open to foreign talent Access to inputs: reduce tariffs and NTMs; Standardization: harmonize or mutually accept standards reform services Market size Market access: pursue trade agreements Market access: deepen trade agreements to cover investment and services Trade infrastructure: reform customs; liberalize transport services; invest in Advanced logistics services: invest in multimodal transport infrastructure ports and roads Geography Basic ICT connectivity: liberalize ICT services; invest in ICT infrastructure Advanced ICT services: expand high-speed broadband Governance: promote political stability Governance: improve policy predictability; pursue deep trade agreements Institutions Standards certification: establish Contracts: enhance enforcement Intellectual property rights: conformity assessment regime ensure protection Source: WDR 2020 team. Note: ICT = information and communication technology; NTMs = nontariff measures. and services GVCs, policies that promote linkages, state-owned or other domestic firms. One particular build managerial capabilities, and facilitate upgrading area in which services inputs matter is transport and of domestic SMEs come into play as countries look logistics. At this stage of GVC development, trade facil- toward making this transition. itation becomes more complex and critical, requiring The demand for lower trade costs is even greater the development of a competitive logistics services for complex manufactures than for simpler ones. sector. Linked to this is the need for high-quality, Lower tariffs are important, including zero-tariff competitively priced ICT infrastructure and services treatment of regional partners through trade agree- to help coordinate increasingly complex activities and ments (see chapter 9). Trusted trader programs, value chains. which expedite customs procedures for shipments of At the institutional level, the shift to advanced established value chain firms, are also useful at this manufacturing GVCs demands that greater attention stage. But access to low-cost inputs must go beyond a be paid to contract enforcement and protection of limited range of goods inputs. Countries at this stage intellectual property. The capacity of national inno- must liberalize access to competitive services inputs, vation systems also must expand. Although univer- including through trade and investment reforms. sities, government, firms, and specialized research They must ensure that the domestic regulatory envi- institutions play various roles in national innovation ronment does not restrict competition by either limit- systems, the desired outputs of an innovation system’s ing access by foreign services providers or protecting capabilities are similar. Policies to enhance participation | 187 Box 7.8  Costa Rica moves into the medical devices GVC As part of a concerted strategy to upgrade beyond basic school graduates, whereas the university system provides light manufacturing exports (notably apparel), Costa Rica specialized workers such as material handlers, engineers, sought integration into higher value-added GVCs. The coun- and microbiologists. try has been highly successful, achieving a 10-fold increase in foreign direct investment (FDI) and GVC participation in Technology and management practices less than 30 years. Costa Rica’s shift to higher value-added The technology required to produce medical devices is GVCs has included semiconductors (the country famously proprietary. Similarly, the management practices required attracted large-scale investment from Intel), global shared to secure regulatory approval for such devices in foreign services, and medical devices, a value chain in which Costa markets are mostly found in firms with prior experience. Rica has been particularly successful in upgrading its posi- Because foreign firms bring with them “follow-on” suppliers tion over the last two decades (figure B7.8.1). Its success in the medical devices GVC (who are also foreign investors), can be attributed to effective public policy on issues such this GVC activity has grown rapidly in Costa Rica. Linkages as workforce development, technology acquisition, and to Costa Rican domestic firms have been concentrated in regulatory alignment, supported by high-quality trade and areas such as packaging but are gradually deepening to investment institutions.a include manufacture of parts and components. Workforce development Regulatory alignment The number of workers required to produce medical The regulatory systems of the European Union, Japan, and devices to standard specifications is unusually high the United States categorize medical devices according to compared with that in other manufacturing sectors the risk facing the consumer: more stringent regulations because of the fatal consequences of human error and apply to higher-risk devices. Items such as elastic bandages, the potential for liability suits. Although Costa Rica is blood pressure cuffs, and X-ray film may be regulated lightly not the lowest-cost source of labor, the training of its as Class I, whereas more stringent Class III regulations are workforce more than offsets this factor. Direct labor for applied to devices implanted in the human body such as medical devices tends to be drawn from technical high pacemakers, artificial heart valves, and silicone breast Figure B7.8.1  Costa Rica’s medical device exports have increased in volume and sophistication since 2000 3.5 3.0 Exports (US$, billions) 2.5 2.0 1.5 1.0 0.5 0 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 18 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 Disposables Medical and surgical instruments Therapeutic devices Diagnostic equipment Source: WDR 2020 team, based on data from Bamber and Gereffi (2013). (Box continues next page) 188 | World Development Report 2020 Box 7.8  Costa Rica moves into the medical devices GVC (continued) implants. An increasing number of Class III products are the government actors are PROCOMER, the export pro- being fabricated in Costa Rica, reflecting the growing con- motion organization, and COMEX, the Ministry of Foreign fidence in the capabilities of producers to follow strict reg- Trade, responsible for trade policy and trade negotiations ulatory protocols. Regulatory cooperation across borders is and for fulfilling an investment promotion role. CINDE has also important. The Latin American headquarters of the U.S. enjoyed a high level of government support and strong Food and Drug Administration opened in San José in 2011 to partnerships since the late 1990s when it landed Intel, enable access to regulatory information and to work with Abbott Laboratories (now Hospira), P&G, and other anchor local regulatory authorities, industries, and academics. investors in the country’s most dynamic sectors. It has continually sharpened its focus, from an all-purpose devel- Trade and investment institutions opment agency when it was founded in the mid-1980s to Costa Rica has a unique nongovernmental organization, a full-blown investment promotion agency attracting and CINDE, that is devoted to promoting economic growth expanding FDI projects by the turn of the century. CINDE through FDI. CINDE provides a forum in which firms can has also improved its service offering beyond attracting share information and address challenges in coordinating investors; it now accompanies strategic investors through with other government and nongovernment actors. Among their investment cycle. a. Bamber and Gereffi (2013). Transitioning to innovative activities that supports start-ups and SMEs is also essential— As countries move toward high-income status, inno- many service firms are small start-ups that are “born vation becomes the main determinant of GVC par- global.”62 ticipation. Such status is normally delivered either Innovation and advanced services GVCs also entirely through services or in GVCs that are highly require a high-quality institutional environment that services-intensive. Because of the growing intertwin- includes intellectual property rights protection and ing of services and manufacturing, the development strong contract enforcement capabilities. They require, of services in GVCs is not entirely autonomous.61 as well, policies that support a high-quality and flexi- Nevertheless, high-income countries have been able ble innovation ecosystem, including advanced ICT to establish leading positions in services value chains infrastructure and services; strong academic, private ranging from research and consulting to motion sector, and government partnerships; and a support- pictures to software design. Some middle-income ive R&D policy that incentivizes collaborative research countries have established positions in services value and development. chains as well—Nollywood in Nigeria, call centers in the Philippines, and software, call centers, and Bolly- Notes wood in India. 1. Pierola, Fernandes, and Farole (2018). Overall, the policy priorities needed to support 2. UNCTAD and World Bank (2018). innovation and advanced services GVCs are similar to 3. Stedman and Green (2018). those needed for advanced manufacturing, although 4. Morris and Staritz (2017). some policies are of even greater importance such 5. Gelb et al. (2017). as those for technical and managerial skills and for 6. Rodrik (2008). access to advanced services. To maintain international 7. Conrad and Jagessar (2018). collaboration in services, national markets must be 8. Sánchez-Ancochea (2006). 9. This section relies on material from Echandi (2019). open to foreign participation—not only for cross- 10. World Bank (2020). border supply and commercial presences but also for 11. Charlton and Davis (2007); Morisset and Andrews- the temporary movement of national persons. Work- Johnson (2004); Wells and Wint (2000). ers holding higher education degrees are needed to 12. Harding and Javorcik (2011). produce innovative services. A business environment 13. Freund and Moran (2017). Policies to enhance participation | 189 14. Bureau, Guimbard, and Jean (2016). References 15. See appendix A for a description of the databases used in this Report. Alder, Simon Francis, Lisheng Shao, and Fabrizio Zilibotti. 16. Ferrantino and Koten (2019). 2013. “Economic Reforms and Industrial Policy in a 17. Huria et al. (2019). Panel of Chinese Cities.” CEPR Discussion Paper DP9748, 18. WEF (2013). Center for Economic Policy Research, London. 19. Djankov, Freund, and Pham (2010). Amiti, Mary, and Shang-Jin Wei. 2009a. “Service Offshoring 20. Freund and Rocha (2011). and Productivity: Evidence from the U.S.” World Economy 21. Christ and Ferrantino (2011). 32 (2): 203–20. 22. Pieterse et al. (2016). —— — —. 2009b. “Does Service Offshoring Lead to Job Losses? Evidence from the United States.” In International Trade 23. Londoňo-Kent and Kent (2003); Raballand et al. (2012). in Services and Intangibles in the Era of Globalization, edited Some issues with transport may require regional or mul- by Marshall Reinsdorf and Matthew J. Slaughter, tilateral coordination, such as anticompetitive conduct in 227–43. National Bureau of Economic Research Studies shipping and air markets (Clark, Dollar, and Micco 2004; in Income and Wealth Series 69. Chicago: University of Fink, Mattoo, and Neagu 2002) and low connectivity in Chicago Press. remoter developing countries (Arvis and Shepherd 2011). Antràs, Pol, and Stephen Ross Yeaple. 2013. ““Multinational 24. Arvis et al. (2011). Firms and the Structure of International Trade.” NBER 25. World Bank (2012). Working Paper 18775 (February), National Bureau of 26. Fernandes, Hillberry, and Mendoza-Alcántara (2019). Economic Research, Cambridge, MA. 27. Hiratsuka (2006). Arnold, Jens Matthias, Beata Smarzynska Javorcik, Molly 28. Boehm (2018). Lipscomb, and Aaditya Mattoo. 2016. “Services Reform 29. Farole (2016). and Manufacturing Performance: Evidence from India.” 30. Antràs and Yeaple (2013). Economic Journal 126 (590): 1–39. 31. Smith (2001). Arnold, Jens Matthias, Beata Smarzynska Javorcik, and 32. Fink (2011). Aaditya Mattoo. 2011. “Does Services Liberalization 33. Auriol, Balineau, and Bonneton (forthcoming). Benefit Manufacturing Firms? Evidence from the Czech 34. Henn et al. (2017). Republic.” Journal of International Economics 85 (1): 136–46. 35. Auriol, Balineau, and Bonneton (forthcoming). Arnold, Jens Matthias, Aaditya Mattoo, and Gaia Narciso. 36. Balineau (2013); Bold et al. (2017). 2008. “Services Inputs and Firm Productivity in Sub- 37. Auriol, Balineau, and Bonneton (forthcoming). Saharan Africa: Evidence from Firm-Level Data.” Journal 38. See the discussion in chapter 3 on the consequences of of African Economies 17 (4): 578–99. tax competition and the discussion in chapter 9 on tax Arráiz, Irani, Francisca Henríquez, and Rodolfo Stucchi. cooperation. 2011. “Impact of the Chilean Supplier Development Pro- 39. Farole and Winkler (2014). gram on the Performance of SMEs and Their Large Firm 40. Fernández-Stark, Bamber, and Gereffi (2012). Customers.” OVE Working Paper OVE/WP–04/11 (May), 41. See for example, Bloom and Van Reenen (2006, 2010); Office of Evaluation and Oversight, Inter-American Bloom et al. (2019). Development Bank, Washington, DC. 42. Bloom et al. (2013, 2019). Arvis, Jean-François, Robin Carruthers, Graham Smith, and 43. Cirera and Maloney (2017). Christopher Willoughby. 2011. Connecting Landlocked 44. Bloom et al. (2013); Bloom, Dressel, and Yam (2018); Developing Countries to Markets: Trade Corridors in the 21st Bruhn, Karlan, and Schoar (2018). Century. Directions in Development: Trade Series. Wash- 45. Bloom et al. (2018). ington, DC: World Bank. 46. Foster, Haltiwanger, and Syverson (2016). Arvis, Jean-François, and Ben Shepherd. 2011. “The Air Con- 47. Atkin, Khandelwal, and Osman (2017). nectivity Index: Measuring Integration in the Global 48. Arráiz, Henríquez, and Stucchi (2011). Air Transport Network.” Policy Research Working Paper 49. Fagerberg and Srholec (2008). 5722, World Bank, Washington, DC. 50. Koschatzky and Stahlecker (2010). Atkin, David, Amit Kumar Khandelwal, and Adam Osman. 51. FIAS (2008). 2017. “Exporting and Firm Performance: Evidence from 52. UNIDO (2009). a Randomized Experiment.” Quarterly Journal of Economics 53. Ottaviano and Puga (1998). 132 (2): 551–615. 54. Combes and Duranton (2006). Auriol, Emmanuelle, Gaëlle Balineau, and Nicolas Bonneton. 55. Rodríguez-Pose and Crescenzi (2008). Forthcoming. “The Economics of Quality in Developing 56. Farole (2011). Countries.” 57. Frick, Rodríguez-Pose, and Wong (2019). Balineau, Gaëlle. 2013. “Disentangling the Effects of Fair 58. See Farole (2011). Trade on the Quality of Malian Cotton.” World Develop- 59. See Farole, Norman, and Kilroy (2014); Rothenberg et al. ment 44 (April): 241–55. (2017). Bamber, Penny, and Gary Gereffi. 2013. “Costa Rica in the 60. Farole (2011). Medical Devices Global Value Chain: Opportunities for 61. Kommerskollegium (2012). Upgrading.” Research Report (August 20), Global Value 62. Knight and Cavusgil (2004). Chains Center, Duke University, Durham, NC. 190 | World Development Report 2020 Black, Anthony, Justin Barnes, and Lorenza Monaco. 2018. Uncertainty in Sub-Saharan Africa.” World Development 39 “Structural Transformation in the Auto Sector: Industrial (10): 1749–59. Policy, State-Business Bargaining, and Supply Chain Cirera, Xavier, and William F. Maloney. 2017. The Innovation Development.” Draft Project Report (March 15), Industrial Paradox: Developing-Country Capabilities and the Unrealized Development Think Tank, Center for Competition, Reg- Promise of Technological Catch-Up. Washington, DC: World ulation, and Economic Development, and SARChI Chair Bank. in Industrial Development, University of Johannesburg, Clark, Ximena, David Dollar, and Alejandro Micco. 2004. Johannesburg, South Africa. “Port Efficiency, Maritime Transport Costs, and Bilateral Bloom, Nicholas, Erik Brynjolfsson, Lucia Foster, Ron Trade.” Journal of Development Economics 75 (2): 417–50. Jarmin, Megha Patnaik, Itay Saporta-Eksten, and John Combes, Pierre-Philippe, and Gilles Duranton. 2006. “Labour Van Reenen. 2019. “What Drives Differences in Man- Pooling, Labour Poaching, and Spatial Clustering.” agement Practices?” American Economic Review 109 (5): Regional Science and Urban Economics 36 (1): 1–28. 1648–83. Conrad, Daren, and Jaymieon Jagessar. 2018. “Real Exchange Bloom, Nicholas, Leonie Dressel, and Emilie Yam. 2018. Rate Misalignment and Economic Growth: The Case of “Management Delivers: Why Firms Should Invest in Trinidad and Tobago.” Economies 6 (52): 1–23. Better Business Practices.” IGC Growth Brief 14 (March), Cusolito, Ana Paula, Raed Safadi, and Daria Taglioni. 2016. Inclu- International Growth Center, London. sive Global Value Chains: Policy Options for Small and Medium Bloom, Nicholas, Benn Eifert, Aprajit Mahajan, David Enterprises and Low-Income Countries. Directions in Devel- McKenzie, and John Roberts. 2013. “Does Management opment: Trade Series. Washington, DC: World Bank and Matter? Evidence from India.” Quarterly Journal of Econom- Organisation for Economic Co-operation and Development. ics 128 (1): 1–51. Djankov, Simeon, Caroline L. Freund, and Cong S. Pham. Bloom, Nicholas, Kalina Manova, John Van Reenen, Stephen 2010. “Trading on Time.” Review of Economics and Statistics Teng Sun, and Zhihong Yu. 2018. “Managing Trade: 92 (1): 166–73. Evidence from China and the U.S.” NBER Working Paper Duggan, Victor, Sjamsu Rahardja, and Gonzalo J. Varela. 24718 (June), National Bureau of Economic Research, 2015. “Revealing the Impact of Relaxing Service Sector Cambridge, MA. FDI Restrictions on Productivity in Indonesian Manu- Bloom, Nicholas, and John Van Reenen. 2006. “Measuring facturing.” Policy Note 5, World Bank, Jakarta, Indonesia. and Explaining Management Practices across Firms and Echandi, Roberto. 2019. “Connecting the Dots between Inter- Countries.” NBER Working Paper 12216 (May), National national Trade and Investment Regulation, Investment Bureau of Economic Research, Cambridge, MA. Climate Reform and Development.” Revised working —— ——. 2010. “Why Do Management Practices Differ across paper, World Bank, Washington, DC. Firms and Countries?” Journal of Economic Perspectives 24 Fagerberg, Jan, and Martin Srholec. 2008. “National Innova- (1): 203–24. tion Systems, Capabilities, and Economic Development.” Boehm, Johannes. 2018. “The Impact of Contract Enforce- Research Policy 37 (9): 1417–35. ment Costs on Value Chains and Aggregate Productiv- Farole, Thomas. 2011. Special Economic Zones in Africa: Compar- ity.” Sciences Po Economics Discussion Paper 2018–12, ing Performance and Learning from Global Experience. Direc- Department of Economics, Sciences Po, Paris. tions in Development: Trade Series. Washington, DC: Bold, Tessa, Kayuki C. Kaizzi, Jakob Svensson, and David World Bank. Yanagizawa-Drott. 2017. “Lemon Technologies and Adop- —— — —. 2016. “Factory Southern Africa? SACU in Global tion: Measurement, Theory, and Evidence from Agricul- Value Chains.” World Bank, Washington, DC. tural Markets in Uganda.” Quarterly Journal of Economics Farole, Thomas, Martin Norman, and Austin Kilroy. 2014. 132 (3): 1055–1100. “Special Economic Zones (SEZs) and Cities: Opportuni- Borchert, Ingo, Batshur Gootiiz, and Aaditya Mattoo. 2014. ties for Mutual Benefit in South Africa.” Reimbursable “Policy Barriers to International Trade in Services: Evi- Advisory Services, World Bank, Washington, DC. dence from a New Database.” World Bank Economic Review Farole, Thomas, and Deborah Elisabeth Winkler, eds. 2014. 28 (1): 162–88. Making Foreign Direct Investment Work for Sub-Saharan Bruhn, Miriam, Dean Karlan, and Antoinette Schoar. 2018. Africa: Local Spillovers and Competitiveness in Global Value “The Impact of Consulting Services on Small and Chains. Directions in Development: Trade Series. Wash- Medium Enterprises: Evidence from a Randomized Trial ington, DC: World Bank. in Mexico.” Journal of Political Economy 126 (2): 635–87. Fernandes, Ana Margarida, Russell Henry Hillberry, and Bureau, Jean-Christophe, Houssein Guimbard, and Sébastian Alejandra Mendoza-Alcántara. 2019. “Trade Effects of Jean. 2016. “Competing Liberalizations: Tariffs and Trade Customs Reform: Evidence from Albania.” Paper pre- in the 21st Century.” CEPII Working Paper 2016–12 (May), sented at Department of Economics Spring Seminar, Centre d’Etudes Prospectives et d’Informations Interna- University of Nebraska, Lincoln, January 18. tionales, Paris. Fernandes, Ana Margarida, and Caroline Paunov. 2012. “For- Charlton, Andrew, and Nicholas Davis. 2007. “Does Invest- eign Direct Investment in Services and Manufacturing ment Promotion Work?” B. E. Journal of Economic Analysis Productivity: Evidence for Chile.” Journal of Development and Policy 7 (1): 1–21. Economics 97 (2): 305–21. Christ, Nannette, and Michael Joseph Ferrantino. 2011. “Land Fernández-Stark, Karina, Penny Bamber, and Gary Gereffi. Transport for Export: The Effects of Cost, Time, and 2012. “Upgrading in Global Value Chains: Addressing Policies to enhance participation | 191 the Skills Challenge in Developing Countries.” Research Hoekman, Bernard, and Aaditya Mattoo. 2008. “Services Report (September 26), Global Value Chains Center, Trade and Growth.” Policy Research Working Paper 4461, Duke University, Durham, NC. World Bank, Washington, DC. Ferrantino, Michael Joseph, and Emine Elcin Koten. 2019. Huria, Sughanda, Ruchita Manghnani, Sebastian Saez, and “Understanding Supply Chain 4.0 and Its Potential Eric van der Marel. 2019. “Servicification of Indian Man- Impact on Global Value Chains.” In Global Value Chain ufacturing.” World Bank, Washington, DC. Development Report 2019: Technological Innovation, Supply Knight, Gary A., and S. Tamar Cavusgil. 2004. “Innovation, Chain Trade, and Workers in a Globalized World, 103–19. Organizational Capabilities, and the Born-Global Firm.” Geneva: World Trade Organization. Journal of International and Business Studies 35 (2): 124–41. FIAS (Facility for Investment Climate Advisory Services). Kommerskollegium (National Board of Trade). 2012. “Every- 2008. “Special Economic Zones: Performance, Lessons body Is in Services: The Impact of Servicification in Learned, and Implications for Zone Development.” Manufacturing on Trade and Trade Policy.” Kommerskol- Report 45869 (April), World Bank, Washington, DC. legium, Stockholm. Fink, Carsten. 2011. “Intellectual Property Rights.” In Prefer- Koschatzky, Knut, and Thomas Stahlecker. 2010. “The Emer- ential Trade Agreement Policies for Development: A Handbook, gence of New Modes of R&D Services in Germany.” edited by Jean-Pierre Chauffour and Jean-Christophe Services Industry Journal 30 (5): 685–700. Maur, 387–405. Washington, DC: World Bank. Londoňo-Kent, María del Pilar, and Paul E. Kent. 2003. “A Tale Fink, Carsten, Aaditya Mattoo, and Ileana Cristina Neagu. of Two Ports: The Cost of Inefficiency.” Research Report 2002. “Trade in International Maritime Services: How (December), World Bank, Washington, DC. Much Does Policy Matter?” World Bank Economic Review Malaysia Ministry of International Trade and Industry. 2019. 16 (1): 81–108. “SME Development,” accessed at miti.gov.my/index.php Foster, Lucia, John Haltiwanger, and Chad Syverson. 2016. /pages/view/1801 October 3, 2019. “The Slow Growth of New Plants: Learning about Malinska, Jana, and Stephen Martin. 2000–2002. “Czech Sup- Demand?” Economica 83 (329): 91–129. plier Development Programme in Electronics and Auto- Freund, Caroline L., and Theodore H. Moran. 2017. “Multi- motive. http://www3.weforum.org/docs/Manufacturing national Investors as Export Superstars: How Emerging- _Our_Future_2016/Case_Study_13.pdf. Market Governments Can Reshape Comparative Advan- Minor, Peter J. 2013. “Time as a Barrier to Trade: A GTAP tage.” PIIE Working Paper 17–1 (January), Peterson Insti- Database of Ad Valorem Trade Time Costs.” ImpactEcon, tute for International Economics, Washington, DC. Boulder, CO. Freund, Caroline L., and Nadia Rocha. 2011. “What Constrains Morisset, Jacques, and Kelly Andrews-Johnson. 2004. “The Africa’s Exports?” World Bank Economic Review 25 (3): Effectiveness of Promotion Agencies at Attracting 361–86. Foreign Direct Investment.” FIAS Occasional Paper 16, Frick, Susanne A., Andrés Rodríguez-Pose, and Michael D. Facility for Investment Climate Advisory Services, World Wong. 2019. “Toward Economically Dynamic Special Eco- Bank, Washington, DC. nomic Zones in Emerging Countries.” Economic Geography Morris, Mike, and Cornelia Staritz. 2017. “Industrial Upgrad- 95 (1): 30–64. ing and Development in Lesotho’s Apparel Industry: Gelb, Alan, Christian J. Meyer, Vijaya Ramachandran, and Global Value Chains, Foreign Direct Investment, and Divyanshi Wadhwa. 2017. “Can Africa Be a Manufac- Market Diversification.” Oxford Development Studies 45 turing Destination? Labor Costs in Comparative Per- (3): 303–20. spective.” CGD Working Paper 466 (October), Center for Ottaviano, Gianmarco, and Diego Puga. 1998. “Agglomeration Global Development, Washington, DC. in the Global Economy: A Survey of the ‘New Economic Harding, Torfinn, and Beata Smarzynska Javorcik. 2011. “Roll Geography.’ ” World Economy 21 (6): 707–31. Out the Red Carpet and They Will Come: Investment Pierola, Martha Denisse, Ana Margarida Fernandes, and Promotion and FDI Inflows.” Economic Journal 121 (557): Thomas Farole. 2018. “The Role of Imports for Exporter 1445–76. Performance in Peru.” World Economy 41 (2): 550–72. ————. 2012. “Investment Promotion and FDI Inflows: Qual- Pieterse, Duncan, Thomas Farole, Martin Odendaal, and ity Matters.” Department of Economics Discussion Paper Andre Steenkamp. 2016. “Supporting Export Competi- 612 (June), University of Oxford, Oxford, U.K. tiveness through Port and Rail Network Reforms: A Case Henn, Christian, Chris Papageorgiou, Jose Manuel Romero, Study of South Africa.” Policy Research Working Paper and Nikola Spatafora. 2017. “Export Quality in Advanced 7532, World Bank, Washington, DC. and Developing Economies: Evidence from a New Data Raballand, Gaël, Salim Refas, Monica Beuran, and Gözde Set.” Policy Research Working Paper 8196, World Bank, Isik. 2012. Why Does Cargo Spend Weeks in Sub-Saharan Washington, DC. African Ports? Lessons from Six Countries. Directions in Hiratsuka, Daisuke. 2006. “Vertical Intra-Regional Produc- Development: Trade Series. Washington, DC: World tion Networks in East Asia: A Case Study of the Hard Disc Bank. Drive Industry.” In East Asia’s De Facto Economic Integration, Rodríguez-Pose, Andrés, and Riccardo Crescenzi. 2008. edited by Daisuke Hiratsuka, 181–99. Institute of Devel- “Research and Development, Spillovers, Innovation Sys- oping Economies–Japan External Trade Organization tems, and the Genesis of Regional Growth in Europe.” Series. London: Palgrave Macmillan. Regional Studies 42 (1): 51–67. 192 | World Development Report 2020 Rodrik, Dani. 2008. “The Real Exchange Rate and Economic In and Moving Up: New Industrial Challenges for the Bottom Growth.” Brookings Papers on Economic Activity 39 (2): Billion and the Middle-Income Countries. Vienna: UNIDO. 365–439. USAID (U.S. Agency for International Development). 2007. Rothenberg, Alexander D., Samuel Bazzi, Shanthi Nataraj, “Calculating Tariff Equivalents for Time in Trade.” USAID, and Amalavoyal V. Chari. 2017. When Regional Policies Fail: Washington, DC. An Evaluation of Indonesia’s Integrated Economic Development Wang, Jin. 2013. “The Economic Impact of Special Economic Zones. Santa Monica, CA: RAND Corporation. https:// Zones: Evidence from Chinese Municipalities.” Journal of www.rand.org/pubs/working_papers/WR1183.html. Development Economics 101 (March): 133–47. Sánchez-Ancochea, Diego. 2006. “Development Trajectories WEF (World Economic Forum). 2013. Enabling Trade: Valuing and New Comparative Advantages: Costa Rica and the Growth Opportunities. Geneva: WEF. Dominican Republic under Globalization.” World Develop- Wells, Louis T., Jr., and Alvin G. Wint. 2000. “Marketing a ment 34 (6): 996–1015. Country: Promotion as a Tool for Attracting Foreign Saurav, Abhishek. No date. “Performance of Special Economic Investment.” FIAS Occasional Paper 13 (March). Facility Zones in India: Experience and Lessons.” Background for Investment Climate Advisory Services, World Bank, note prepared for this Report. Washington, DC. Smith, Pamela J. 2001. “How Do Foreign Patent Rights Affect World Bank. 2012. “Developing a Trade Information Portal.” U.S. Exports, Affiliate Sales, and Licenses?” Journal of Inter- Report 83273 (July), World Bank, Washington, DC. national Economics 55 (2): 411–39. ——— —. 2015. “A Practical Guide to Increasing Mining Local Stedman, Ashley, and Kenneth P. Green. 2018. “Fraser Insti- Procurement in West Africa.” Report AUS6324 (February), tute Annual Survey of Mining Companies, 2018.” Fraser World Bank, Washington, DC. Institute, Vancouver, Canada. ——— —. 2018. Global Investment Competitiveness Report 2017/2018: Sturgeon, Timothy J., Leonardo Lima Chagas, and Justin Foreign Investor Perspectives and Policy Implications. Wash- Barnes. 2017. Inovar Auto: Evaluating Brazil’s Automotive ington, DC: World Bank. Industrial Policy to Meet the Challenges of Global Value Chains. ——— —. 2020. Foreign Direct Investment as a Key Driver for Washington, DC: World Bank. Global Value Chains. Washington, DC: World Bank. UNCTAD (United Nations Conference on Trade and Zeng, Douglas Zhihua. 2010. Building Engines for Growth and Development) and World Bank. 2018. “The Unseen Competitiveness in China: Experience with Special Economic Impact of Non-tariff Measures: Insights from a New Zones and Industrial Clusters. Directions in Development: Database.” Report UNCTAD DC:/DITC/TAB/2018/2, UNC- Countries and Regions. Washington, DC: World Bank. TAD, Geneva. ——— —. 2015. “Global Experiences with Special Economic UNIDO (United Nations Industrial Development Organi- Zones: Focus on China and Africa.” Policy Research Work- zation). 2009. Industrial Development Report 2009; Breaking ing Paper 7240, Washington, DC, World Bank. Policies to enhance participation | 193 8 Policies for inclusion and sustainability Key findings  eveloping countries would benefit from policies that spread the jobs and earnings • D gains from global value chain (GVC) participation across society. Access to child care and training programs support jobs for women and youth, respectively. Smallholders need assistance, such as extension services and access to finance, to integrate into agricultural value chains. GVC lead firms, labor, and governments can work together to protect workers’ safety and rights. • I ndustrial countries would benefit from adjustment policies for workers displaced by technology, trade, and the expansion of GVCs. Placement services, training, and mobility support can help workers transition to more productive jobs. • P  olicy can mitigate negative environmental consequences and promote the adoption of environmentally friendly technologies. Pricing the environmental costs of production and distribution appropriately will encourage conservation and cleaner technologies. In addition, regulation is needed for specific pollutants and industries.  hese national measures can be complemented by global cooperation on the • T environment and working conditions. Standardized international data will help expose poor production practices and induce firms to improve. 194 | World Development Report 2020 T he policies and institutions that maximize the chains. Yet a role remains for national governments to aggregate gains from participation in global safeguard the protection of workers in GVCs through, value chains (GVCs) will not necessarily ensure for example, collective bargaining, freedom of asso­ that these gains are shared—across locations, across ciation, and social dialogue. skill levels, and across different groups in society such Second, in advanced countries the welfare of work- as women and youth (chapter 3). They may even exac- ers left behind in communities where factories have erbate the negative environmental consequences of closed is the primary threat to the sustainability of GVCs (chapter 5). In this context, ensuring the inclu- trade and GVCs. In response, labor adjustment poli- siveness and sustainability of the GVC model calls cies can be used to ensure that workers have the skills for considering policies in three areas that support to move to new industries and places. By contrast, broader gains among workers and mitigate their neg- using trade restrictions or rigid labor policies to pro- ative social and environmental consequences. tect existing jobs is unsustainable and will slow eco- First, developing countries would benefit from nomic transformation and long-run income growth. policies that spread the jobs and earnings gains from Place-based interventions should take into account GVC participation across society, thereby helping to local endowments and favor targeted initiatives to lift the bottom 40 percent. For countries participating address coordination failures over broad-based invest- in agriculture value chains, policies that support the ment subsidies. integration of smallholders are particularly import- Third, policies can mitigate negative environmen- ant. As countries move into basic manufacturing, tal consequences and promote the adoption of envi- lower-skilled poor workers will benefit most from pol- ronmentally friendly technologies. An important first icies that support comparative advantage and incen- step is to set a price on environmental degradation. tivize investment in labor-intensive activities. Prices of goods should reflect both their economic and Policies should also support the inclusion of socioenvironmental costs, and trade should be carried women and youth in GVCs, including by providing out based on comparative advantage that accounts for access to child care and training programs and by these costs. In addition to pricing, there is also a role addressing legal and social barriers to employment for regulation, especially for specific pollutants and and earnings. For example, in Côte d’Ivoire a project industries. to develop the cashew value chain—supported by These national measures can be complemented by investments in feeder roads, reform of agricultural global cooperation to deal directly with the environ- extension services, and improved access to finance— ment, as well as to ensure that trade agreements are is expected to raise earnings for 225,000 smallholder consistent with environmental goals. International farmers and create at least 12,000 processing jobs, treaties such as the 2015 Paris Agreement on climate with half going to women. change include requirements on environmental pro- Significant shortfalls in worker safety standards tection for signatories at all income levels. Recent are still common in the supply chains of many global major trade agreements, such as the Comprehensive brands, particularly among their second- and third- and Progressive Agreement for Trans-Pacific Partner- tier suppliers, as incidents such as the 2013 Rana ship (CPTPP) and the new European Union (EU) free Plaza garment factory disaster in Bangladesh attest. trade agreements, include environmental provisions. Although most global brands have developed social Finally, standardized international data on the and environmental compliance standards for their environmental consequences of firms in GVCs can global supply chains, broader initiatives bringing lead expose poor production practices and incentivize firms together with suppliers, trade unions, and civil firms to improve. society are becoming more common. For example, the International Labour Organization-International Sharing the gains Finance Corporation (ILO-IFC) Better Work program, which covers nearly 2.5 million workers in 1,700 Exploit comparative advantage to ensure GVC-linked garment factories in eight countries, jobs-rich investment in GVCs has demonstrated that improved compliance with Jobs are the most direct and important channel labor standards can lead to higher productivity and through which GVCs contribute to poverty reduction profits. Meanwhile, governments are moving toward and shared prosperity. Chapter 3 describes the strong “binding due diligence,” whereby lead firms are legally links between GVCs and job creation at both the responsible for compliance across their entire supply national and the firm level. Importantly, it shows that Policies for inclusion and sustainability | 195 job growth in GVCs is associated with both greater Therefore, governments have an important role use of imported inputs and greater use of technology. to play both in facilitating lead firm investment and Thus, although the higher imported inputs and capital in supporting smallholder integration in agricultural intensity of GVC production may mean less labor is GVCs. Policies for the former include many of those needed per unit of output, the output boost induced by discussed in chapter 7—notably, trade and investment GVC participation means more jobs are created overall. policy and infrastructure. Policies for the latter include Chapter 3 also points out that GVCs are, on the whole, the provision of agricultural extension services, access inclusive; they are both pro-poor and a significant to risk management instruments such as insurance, source of jobs for women. These positive outcomes and assistance with convening and coordinating can be facilitated by supportive domestic policies. smallholders to exploit scale through cooperatives and other producer organizations. Create jobs in sectors that absorb poor and Finally, as countries seek to move downstream low-skilled workers from natural resources and integrate into manu- For many developing countries, particularly those facturing and services value chains, the objective selling only commodities, ensuring GVCs benefit of delivering jobs for the current stock of poor the poorest will come primarily through integrating workers calls for policies that reinforce comparative smallholders into agriculture value chains and home- advantage.3 This means, for example, a relatively based workers into manufacturing and services GVCs. small, agriculturally rich country would focus on Integration of smallholders is particularly important agriprocessing, or a large, low-skilled labor surplus for Africa, where 55 percent of jobs and more than 70 country would implement policies conducive to percent of the earnings of the poor are reliant on the attracting light manufacturing GVCs. For example, agriculture sector.1 Côte d’Ivoire and Rwanda adopted strategies to As discussed in chapter 2, foreign direct invest- expand agriculture value added and increase pro- ment (FDI) may play a critical role in supporting the cessing to raise returns to smallholder coffee farm- development of agriculture value chains. Lead firms ers (box 8.1). Ethiopia, by leveraging FDI in industrial help solve many of the challenges of raising small- parks, developed labor-intensive light manufactur- holder productivity by providing access to inputs, ing to absorb labor transitioning away from agricul- technical support, finance, and markets. The integra- ture. And Morocco upgraded to high-value manufac- tion of smallholders with offtakers or directly with turing to create jobs for an underemployed skilled processors supports greater value addition at the population. These strategies offer a contrast with farmgate through a range of services such as tech- strategies that attempt to promote the development nology transfer, quality or certification premiums, of high-technology, innovation-driven value chain and continual access to the market. For example, the nodes—strategies that in the same countries, even if rapid development of floriculture value chains across successful, would unlikely have significant impacts East Africa, which opened up many jobs and earnings on lower-skilled workers and could contribute to opportunities for smallholder farmers and women in wage polarization. packing and distribution, was made possible through At the heart of policies that reinforce compara- subcontracting models organized by lead firms. tive advantage are those that minimize distortions Smallholder integration in GVCs is, however, not of market prices—of land, labor, and capital—so that a panacea. A case study in Côte d’Ivoire and Ghana on factors flow smoothly to the sectors and places where participation in the pineapple and cocoa value chains comparative advantage can be best exploited. These found that, although participation leads to better include economywide policies to support land mar- growing processes, larger yields, and higher incomes ket reforms, competition, open labor markets, and for successful commercial farmers, it is also associated access to finance, along with investments in critical with an increase in casual labor hiring, as well as dis- infrastructure. placement of farmers from land because of their weak In low-income, labor surplus countries with large bargaining positions and scant knowledge of their pools of unskilled labor transitioning from the agri- rights to land ownership.2 Moreover, the near-collapse culture sector, externalities arising from the diver- of Ghana’s export pineapple sector in the mid-2000s gence between the market price and the opportunity was due in part to smallholders’ lack of organization, cost of labor may call for additional targeted incen- which contributed to overproduction and inflexibility tives for the private sector to invest in labor-intensive in response to changing market demand. activities.4 196 | World Development Report 2020 Box 8.1  Taking advantage of comparative advantage: Agribusiness GVCs deliver more and better jobs in Côte d’Ivoire and Rwanda Côte d’Ivoire’s cashew value chain sector was historically the country’s main export crop and a Cashews are Côte d’Ivoire’s third-ranking export after major source of earnings for up to half a million rural Rwan- cacao and refined petroleum products, and they are an dans. But at the end of the 1990s, fallout from the civil war important source of cash for smallholders and processors in helped put the sector on the verge of collapse because of the poorer north of the country. Although Côte d’Ivoire pro- the low quantity and quality of its product. To address this duces 23 percent of the world’s cashew supply, fewer than challenge, the country put in place a strategy, completed 7 percent of raw cashew nuts are processed domestically. in 2002, to raise production and move to a higher value- Low yields and low quality are a result of poorly maintained added position in the coffee GVC. Working closely with plantations, lack of quality stock and inputs, weak exten- the private sector and nongovernmental organizations, sion services, losses in postharvest handling and storage, the government introduced a two-pronged approach: (1) and lack of finance for improvements. With coordinated upgrade technology and increase production and (2) boost support from the World Bank and the International Finance skills and improve quality. Corporation, and working closely with the private sector, These interventions proved to be a critical turning point a comprehensive program to upgrade the cashew sector for the sector and spurred upgrading along the coffee and increase domestic value addition was put in place. In value chain. The upgrading was manifested in more skilled 2017 the program established four cashew “platforms” and farming techniques, better use of technologies, and higher eight satellite hubs that provide training, access to inputs, productivity. During the first five years of implementing and market information, along with processing demon- the National Coffee Strategy, private investment in coffee stration units. The program was supported with access to washing stations grew by an average of 120 percent a year new sources of finance for smallholders, notably through in locations with the highest cherry availability (the fruit the introduction of a warehouse receipts system that that contains the coffee bean), water supplies, and road enables processors to use unprocessed nuts as collateral linkages. The total number of coffee washing stations for working capital loans. About 225,000 cashew farmers rose from just two in the entire country in 1998 to 299 as are expected to benefit from the upgrading and improved of early 2015. Meanwhile, the higher-quality coffee began value chain integration. to merit higher prices, with Rwandan coffee now fetching a premium in international markets. According to the U.S. Source: World Bank (2018). Agency for International Development (USAID), as a result Rwanda’s coffee value chain of the reforms in the coffee GVC, approximately 50,000 In the late 1990s, as Rwanda looked to rebuild its economy rural households have seen their incomes from coffee pro- and create jobs and earnings opportunities after a civil war, duction more than double, and some 2,000 jobs have been it faced structural challenges to developing internationally created in coffee washing stations. competitive tradables. For one thing, it is small and land- locked and in a poor and fragile neighborhood. The coffee Source: Adapted from Karuretwa (2016). Create jobs for women and youth (not in employment, education, or training). In Ban- The propensity of GVCs to employ women and youth gladesh, more than 3 million women, mainly young is partly related to the sectors and activities that lend rural–urban migrants, gained employment in the themselves to outsourcing and global relocation, garment sector as it integrated into GVCs in the early which in turn are associated with some of the negative 2000s, contributing to an almost 10 percentage point consequences of GVCs, especially those around low rise in the rate of female labor force participation in wages and poor working conditions. Nevertheless, the just a decade. potential of GVCs to employ large numbers of young Education and skills development policy is the female workers means they may play a major role in starting point for helping youth to take advantage supporting many countries’ efforts to increase female of the opportunities for employment in GVCs. The labor force participation and reduce youth NEETs World Development Report 2019: The Changing Nature of Policies for inclusion and sustainability | 197 Work highlighted the rapidly changing demand for 28 apparel factories. Of the 144 women who attended skills, along with the growing importance of advanced the training program, 92 were offered a promotion cognitive skills, sociobehavioral skills, and, most and higher salary within weeks of completing the important, adaptability to changing circumstances program. The evaluation also found that average effi- and to “unlearn and relearn quickly.”5 Research shows ciency increased by 5 percent, and absenteeism fell in increasingly large payoffs from such adaptability—for line where trained female supervisors worked. example, in Armenia and Georgia the ability to solve problems and learn new skills yields a wage premium Balance adequate wages with sustaining of nearly 20 percent.6 competitiveness GVCs are at the forefront of these changing The inclusivity and social sustainability of GVCs demands for skills, but in many if not most coun- depend not only on the scale and distribution of jobs tries there remains a large gap between the needs of in GVCs, but also on the quality of those jobs. Here employers and the approaches of education and skills the concept of job quality incorporates both wages development institutions. Countries need to work (or earnings more broadly) and working conditions, toward a system that emphasizes the employability including working hours, benefits, the health and of youth and facilitates the transition from study to safety environment, treatment of workers, and the work. Promising policy directions include adoption degree to which workers have voice and agency to help of dual education systems that provide flexibility for shape employers’ decisions on issues that affect work- combining general and vocational education, devel- ers. This concept is in line with Sustainable Develop- opment of vocational training curricula with private ment Goal 8 of United Nations Agenda 2030, which sector participation to ensure relevance to employer highlights the importance of ensuring improvements needs, and expanded use of innovative apprentice in working conditions—combining aspects such as models that give youth an opportunity to learn from productive employment, social protection, social dia- working. Public-private models are often used to logue, and rights at work—together with economic develop pathways to GVC-specific employment. For growth. The issue of job quality is particularly relevant example, in Kenya the national coffee board and indus- to labor-intensive GVCs, where outsourcing to devel- try bodies have teamed up with Kimathi University of oping country locations is fundamentally motivated Technology to develop a coffee diploma program that by the desire to access low-cost labor.7 combines classroom training on technology and qual- Because many of the most prominent GVCs involve ity management with an industry placement of three outsourcing of low-skill, labor-intensive activities, to six months. the very low nominal wages in some countries often Governments can help facilitate women’s access grab the world’s attention. For example, recent news to jobs in GVCs by enacting policies that support articles have noted that t-shirts are being produced for women’s participation in the economy. Such policies charities or high-profile brands in factories paying less could establish the legal and regulatory environment than 50 cents an hour. Certainly, to readers in high- for access to quality child care, facilitate access to safe income countries where even the lowest-skilled fac- transport, as well as ensure that women are protected tory jobs pay 20–30 times that level, this is a shock- from unfair treatment. For example, recent research ingly low wage. However, this does not necessarily by the World Bank’s Women, Business, and the Law mean that low wages are a problem in GVCs. As dis- project shows that in the Middle East and North Africa cussed in chapter 3, firms operating in GVCs tend to region women have on average less than half the legal pay higher wages than firms operating in direct trade rights of men in measured areas. Box 8.2 highlights only.8 What matters more is whether the wages on two contrasting approaches to integration of female offer are in line with productivity and whether they workers in GVC factories—one that brought women offer a reasonable “living wage” for workers. workers to the factories and another that brought fac- GVCs can be problematic if they contribute to the tories to women workers. emergence of “low wage traps”—that is, where wage Private enterprises have a role to play through suppression is used to maintain international compet- training and development programs and ensuring fair itiveness. Although low wage traps are not inherent promotion practices. In Bangladesh, for example, an to GVCs, the globalized and footloose nature of GVC initiative in partnership with the ILO-IFC Better Work production in some sectors may make them more program developed, implemented, and evaluated an likely, particularly where lead firms in GVCs use inter- innovative training program for women operatives in national production cost comparisons to maintain 198 | World Development Report 2020 Box 8.2  A tale of two economic zones: Initiatives to promote women’s employment in garment GVCs in Bangladesh and Jordan Bangladesh opening up new opportunities for other women in these In Bangladesh, women’s integration into the workforce northern villages.a of garment factories in export processing zones (EPZs)— mostly in Dhaka and Chittagong—was almost immediate. Jordan Because the EPZs were located in Bangladesh’s largest Jordan’s qualified industrial zones (QIZs), established in the cities, urban women faced somewhat lower transport and late 1990s, were expected to not only generate exports by social barriers to working in GVC factories. The challenge integrating Jordan into the garment GVC, but also create in Bangladesh was how to make these same opportunities large-scale employment for women in a country in which the available to rural women for whom these constraints were female labor force participation is among the lowest in the binding. An innovative pilot project, the Northern Areas world. The QIZs were able to attract investment and create Reduction of Poverty Initiative, supported by the World jobs, but manufacturers faced large barriers in integrating Bank in cooperation with the Bangladesh Export Process- local women into the factories in the QIZs because of lack ing Zones Authority, brought women from the poorest of transport and perceptions of the safety and social accept- regions of northern Bangladesh into Dhaka for training and ability of working in these factories. As a result, virtually all employment in the EPZ-based garment factories. The pro- the jobs created for women in the initial stages of the QIZs gram gave women and the local community the informa- were taken up by migrant workers, mostly from South and tion and awareness they needed to overcome social stig- Southeast Asia. In response, the government created satellite mas. Women also received transport, living stipends, and production units in rural areas around the villages in which comprehensive technical and life skills training, followed by women resided, supported by substantial financial incentives employment. The results of the pilot were positive: more for manufacturers to hire through these satellite units. The than 6,000 women (two-thirds of those who completed initiative, which was launched in 2010, has shown positive training) took up employment in the garment factories at results, even if on a small scale: as of August 2017, approx- earnings above the industry average. And positive spill- imately 3,300 jobs had been created in 12 satellite factories, overs from the pilot are evident. Many of the constraints with a 90 percent female workforce.b of information and social norms have been overcome, a. World Bank (2017). b. Davis (2017). pricing pressure on suppliers in developing countries. levels. Minimum wages also play an important role. Moreover, because GVCs can emerge as enclaves or Virtually all countries have some minimum wage for dominant sectors in developing country economies, regular workers, although it varies dramatically (even there is a risk that employers take advantage of mon- considering differences in purchasing power) from opsony and political power in labor bargaining. For just $2 a month in Burundi and Uganda to more than example, in Bangladesh garment factory owners have $2,900 a month in Norway.9 Minimum wages should managed, despite repeated large-scale protests, to be set at a level that, at the very least, protects workers avoid any real term increase in garment factory wages. from poverty and vulnerability,10 while also keeping an Depressed wages can be particularly problematic for eye on firm competitiveness. Perhaps most important, low-income workers in developing countries where they should be raised at regular intervals and through GVC integration is associated with rapid urbanization a systematic and transparent process,11 which includes and where housing and transport costs are rising far tripartite social dialogue. Minimum wage indexation more quickly than overall inflation rates. should be linked to both productivity growth and the In this context, policies should protect workers’ cost of living, avoiding excessively sharp increases earnings while maintaining competitiveness to attract during significant economic downturns. Meanwhile, GVC investment. Collective bargaining can be an effec- the impact of minimum wage on workers is unequal tive mechanism for negotiating the appropriate wage and depends on compliance and enforcement, as well Policies for inclusion and sustainability | 199 as the degree of segmentation between formal and ranging from violations of core labor standards such informal workers. Thus a minimum wage should be as child labor, forced labor, lack of freedom of associa- seen as just one mechanism for supporting inclusivity tion, and exploitative and abusive practices to unsafe in GVCs. working conditions, low wages, excessive working In some countries, distortions in the domestic hours, and precarious contracts. Although serious market may drive a significant wedge between a liv- breaches of standards are becoming less common in ing wage for workers and the wage at which firms the direct supplier networks of multinationals, they can remain competitive in international markets. remain a problem and are rife in second- and third-tier Government policy can help bridge this gap—over suppliers. Widely publicized examples, such as the the medium term by addressing the market failure Rana Plaza disaster in Bangladesh in 2013, in which and over the short term by undertaking interven- more than 1,100 garment workers lost their lives, and tions that change relative prices. In South Africa, the the Baldia textile factory fire, which killed close to 300 government introduced a wage subsidy for youth workers in Pakistan in 2012, are well known. How- workers and later extended it to all workers based in ever, below the radar millions of workers in globally special economic zones (SEZs). Governments may also linked industries work daily in vulnerable situations. seek to raise net returns to workers by, for example, Although poor working conditions in Dickensian fac- lowering the cost of transport to access jobs through tories tend to garner the most attention, similar prob- transport subsidies or investments in public transport lems exist in global commodity chains such as agri- services or by lowering the costs of housing through culture13 and even in high-technology value chains, social housing schemes or unlocking the constraints as confirmed by recent news reports documenting to private housing construction. Such instruments are the casualization, discrimination, harassment, and often incorporated directly into SEZs, with SEZ-based retaliation encountered in some of the world’s largest employers routinely providing transport for work- technology multinationals. ers and in some cases providing housing in on-site Many of the specific features of poor working con- dormitories. ditions (particularly around workplace safety) are less As countries shift from commodity and basic a feature of GVCs themselves than of the labor markets manufacturing GVCs to advanced manufacturing and in countries to which GVC activities are outsourced. innovation-based GVCs, wages are less fundamental In fact, like wages, working conditions in GVC-linked to competitiveness. However, in many developing enterprises tend to be better on average than in those countries the problem is lack of a sufficient base enterprises in the same country operating outside of of skilled workers (particularly technical workers), GVCs. And yet aside from the fact that the GVC model which in turn creates large wage premiums that not enables global enterprises (and consumers) to profit only contribute to polarization but also undermine from offshoring to avoid the costs of protecting work- competitiveness. Because skilled workers in these ers, GVCs may also exacerbate the problems of poor GVCs typically complement unskilled workers, the working conditions by creating incentives for GVC- lack of skilled workers also has a negative impact on linked suppliers in developing countries to similarly inclusion. Aside from the obvious role of education seek to cut these costs. For example, an underlying and skills-development policy, countries should be factor in Bangladesh’s Rana Plaza disaster was the open to the immigration of skilled workers as a strat- common practice of first-tier suppliers subcontracting egy for both competitiveness and inclusion. work to smaller, often informal, producers to reduce costs and avoid scrutiny from lead firms. Protecting the well-being of workers is Even where developing countries have robust about more than wages national policy regimes in place to support interna- But workers care about more than just wages. Evi- tional labor standards, those regimes are ultimately dence from Vietnam shows that workers’ reported only as good as their enforcement capacity. It is well-being is affected also by incentive structures, here that many developing countries come undone. benefits packages, training, absence of sexual harass- Lack of technical capacity and financial resources, ment, strikes, and health and safety. Beyond wages, corruption, and distorted incentives are all powerful occupational health and safety affect well-being at forces that undermine both national and private stan- four times the rate of any other measure of working dards designed to promote quality jobs in GVCs. As conditions, such as number of hours worked.12 Yet countries engage more deeply in GVCs, investing in working conditions are commonly found to fall short upgrading the capacity and governance of their labor of international standards in GVC supplier countries, regulatory regimes will be increasingly critical to both 200 | World Development Report 2020 protect workers and maintain the “national brand” for standards. In recent years, the limitations of unilateral supply chain compliance. brand initiatives have been addressed somewhat by In response to concerns about working conditions multistakeholder initiatives that bring together lead in GVC supply chains, over the last decade or more firms and suppliers, trade unions, civil society, and, most global brands have developed social and envi- in some cases, national governments. One example ronmental compliance standards for supply chains. is the ILO-IFC Better Work program noted earlier and Such initiatives, which increasingly involve global described in box 8.3.14 framework agreements between trade unions and Despite the success of private and public-private multinational enterprises, have succeeded in improv- initiatives, they are limited in scope and coverage. ing working conditions, particularly those related to At the very least, national governments must play a occupational health and safety and other measurable supportive role in facilitating compliance through Box 8.3  Transparency promotes compliance with labor standards and improves working conditions According to global evidence from the International Labour Greater transparency on working conditions in GVCs Organization-International Finance Corporation (ILO-IFC) plays a role in promoting compliance with labor standards. Better Work program, garment factories in GVCs are more In Vietnam, for example, the noncompliance rates of firms productive and more profitable when they comply with labor in the apparel sector declined with each additional year standardsa—especially those aimed at ensuring freedom of of program participation (figure B8.3.1, panel a), and the association and collective bargaining,b improving workers’ introduction of a policy to publicly disclose the firms that sense of physical security and assurance of wage payments, fail to meet key labor standards has also improved firm and eliminating sexual harassmentc and verbal abuse.d compliance (figure B8.3.1, panel b).e Figure B8.3.1  Working conditions improved in apparel sector firms participating in the ILO-IFC Better Work Vietnam program a. Labor standards noncompliance b. Labor standards noncompliance and years of program participation and public disclosure 25 25 Noncompliance rate (%) Noncompliance rate (%) 20 20 15 15 10 10 5 5 0 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 20 20 20 20 20 20 20 20 20 Years of participation Source: Hollweg (2019). Note: Panel a plots the average noncompliance rate of firms for each year of program participation in Better Work Vietnam (2010–18). Panel b plots the average noncompliance rate of firms over time since Better Work Vietnam was launched in 2010. The policy of public disclosure of firms that failed to meet key labor standards was announced in 2015 and implemented in 2017. IFC = International Finance Corporation; ILO = International Labour Organization. a. ILO and IFC (2016). b. Brown et al. (2015). c. Brown and Lin (2014). d. Rourke (2014). e. Hollweg (2019). Policies for inclusion and sustainability | 201 regulatory enforcement, even if their capacity is often development stages, as well as the places where those weak. In addition, they can, in partnership with the workers are concentrated. private sector, support their GVC-linked enterprises, and especially small and medium enterprises (SMEs), “Flexicurity” approaches can help in meeting the international standards on wages and manage adjustment while maintaining working conditions.15 A starting point is for countries competitiveness to adopt international labor standards—notably ILO’s Labor market policies can provide a cushion when core conventions.16 By doing so, countries send an workers lose jobs and offer assistance in finding new important signal to GVC investors that they will not employment. Income protection policies, such as engage in a race to the bottom on wages and working unemployment and disability insurance, along with conditions. To make the standards effective, however, other forms of social protection, aim to mitigate the countries need to strengthen the monitoring and income losses of workers without taking steps to enforcement capacity of labor inspection regimes and return them to work. And active labor market policies build robust labor market institutions, including sup- (ALMPs), including employment services (such as port for collective bargaining, freedom of association, counseling, job search assistance, and intermedia- and social dialogue. tion), training, wage subsidies, and entrepreneurship In addition, a shortcoming of most of the initiatives programs are designed to match displaced workers targeting working conditions in GVC supply chains is with income-earning opportunities.20 that they are nonbinding on the lead firm and tend to Because GVC employers need to compete in abso- rely on a “name and shame” approach. Increasingly, lute terms in global markets, restrictive labor market however, governments in countries where lead firms policies can be a barrier to investment. To balance are based are responding to demands from civil soci- inclusion and competitiveness, countries may com- ety to introduce “binding due diligence”—that is, lead bine greater labor market flexibility (that is, limiting firms are legally responsible for standards across their labor regulations that significantly restrict employers, supply chains (including subcontractors), particularly while maintaining adequate protection of workers) around issues of labor and human rights, but poten- with highly supportive social protection and comple- tially also with respect to the environment, industrial mentary social insurance. relations, consumer protection, and corporate gov- This approach, proposed in World Development ernance. Most notably, in 2017 France enacted the Report 2019,21 calls for a stronger and expanded socially Duty of Vigilance Law, which requires large French supported minimum level of income, complemented companies to publish and implement a vigilance plan by mandated social insurance and more flexibility in in order to identify and prevent human rights risks labor markets. For example, Denmark’s “flexicurity” linked to their activities. Other European countries are model gives businesses the freedom to hire and fire considering similar measures. workers with relatively limited restrictions, while Some international trade agreements include providing a generous, broad-based unemployment specific provisions on labor rights.17 For example, fol- benefit system that cushions the negative income lowing its participation in the U.S.–Cambodia Bilateral effects on displaced workers. A key feature of Den- Textile Trade Agreement (USCBTTA), Cambodia rati- mark’s system is the significant investment in active fied the ILO conventions on forced labor, freedom of labor market programs to enhance employability and association, collective bargaining, discrimination, and connect workers to jobs. child labor, with positive spillovers for labor conditions Despite having relatively low levels of unemploy- in apparel factories.18 ment, Denmark invests more heavily in labor market policies than other countries (figure 8.1). A large share of this spending is devoted to ALMPs to help workers Managing adjustment in sectors or regions undergoing adjustments. By con- As highlighted in chapter 3 and recent research,19 trast, in the United States minimal, broad-based pro- because GVC trade tends to strongly reinforce com- tection is the norm, and ALMPs are deployed narrowly parative advantage, lower-skilled workers in high- for specific “trade adjustment assistance.” income countries, and the places in which they are Designing and delivering labor market programs concentrated, typically lose out as countries upgrade remain a challenge, particularly in low- and middle- in GVCs. In this context, there is a role for policy to income countries. In these countries, lower formal manage the adjustment process for workers who education in the workforce limits the benefits of voca- may be displaced during the transitions across GVC tional training, higher labor market informality limits 202 | World Development Report 2020 Figure 8.1  Denmark invests more than other OECD countries to support workers 3.5 Spending on total and active labor market policies 3.0 2.5 (% of GDP, 2016) 2.0 1.5 1.0 0.5 0 itz ain G land Fr d Sw ce De den k Ze ic ze Es nd Re nia re blic Po . Po nd Ire al nd N Rep a N ly ay rla y Be ds m m a Hu urg Fi ry o es n l e ng ia Sl om Au nia ov Ca lia ae p ar et an l pa ite xic xe tri an d Ita bl hi iu g a d Latv Re at n an a w a la la ak na Isr d ch to Sw Sp e nm rtu ng Lu Aus bo C lg ew u Ko pu N m e str nl al Ja St Un Me or ov er a, er he d Ki ite Un C Sl Total spending Active spending Source: Adapted from Bown and Freund 2019. Note: Data for France, Italy, and Spain are for 2015, and data for the United Kingdom are for 2011. the reach of adjustment beyond the formal economy, infrastructure gaps and regulatory barriers prevent and a weak institutional capacity limits the ability to the integration of domestic markets.22 ascertain eligibility and control fraud. Because such As a result, the dislocations that come with GVC programs are expensive, they are also difficult to adjustments do not happen just to people; they also implement in countries facing significant fiscal con- happen to places. Examples are easy to find. Detroit straints. However, evidence suggests that they pay off and other Rust Belt cities in the United States have to both protect workers and maintain political support seen automotive and other machinery and equipment for open trade. Denmark, for example, has not experi- manufacturing offshored to lower-cost locations like enced the kind of backlash against trade experienced Mexico. Similarly, industrial production has shifted by many industrial countries. out of places like northeastern France and Ger- many’s Ruhr Valley and toward locations in central Support for people in left-behind places and eastern Europe. In many developing countries, can target services, skills training, and both former industrial hubs and agricultural hinter- mobility lands may no longer be able to rely on captive domes- The challenge of labor adjustment is fundamentally tic markets—examples are Bulawayo in Zimbabwe linked to the spatial distribution of economic activity and East London in South Africa. The effects can be in GVCs both across and within countries. These spa- long-lasting. Research in Brazil has found that regions tial patterns of development are relevant at the initial affected by trade liberalization faced sharply lower stages of GVC integration and throughout the stages formal sector employment and earnings even after 20 of upgrading. As countries integrate globally, GVC years.23 Wherever places are facing adjustment, there investment tends to concentrate in the places within has emerged a strong political imperative for targeted, countries that are well connected to regional and place-based policies designed to create new sources of global markets (figure 8.2). This concentration may demand to absorb displaced workers in the short term aggravate existing disparities by reinforcing the com- and to shift the local economy onto a more sustainable petitiveness of leading regions, especially where large path for the future. Policies for inclusion and sustainability | 203 Figure 8.2  Industrial development was uneven is a major risk that, with fiscal incentives, regions across regions of Mexico during the period of strong within the same country will engage in territorial international market integration competition and a race to the bottom. The fiscal wars fought by Brazilian states competing for global 50 automotive investment in the 1990s are instructive. 45 Investment subsidies amounted to massive transfers Regional share of manufacturing 40 to private—mainly foreign—investors at a huge fiscal 35 cost but with a limited impact on job creation, no employment (%) 30 guarantees of sustainability through development of 25 local supply chains, and the potential of undermining the competitiveness of the overall sector.25 Moreover, 20 governments face risks in offering large incentives to 15 attract investment in footloose activities such as gar- 10 ment manufacturing, which may relocate quickly in 5 response to political factors, global market conditions, 0 or the removal of subsidies such as recently for Fox- Border North Center Mexico City South conn and Amazon in the United States. 1970 1980 1985 1993 Recent proposals have called for shifting from tar- geting investment broadly to deploying a geographi- Source: Hanson 1998. cally targeted wage subsidy or a hiring tax credit that directly targets job creation in regions that are affected Policies addressing these increasing regional by trade adjustment or are otherwise lagging.26 Such inequalities must target the people in them, a princi- approaches may indeed be effective for supporting ple that is set out clearly in World Development Report increased investment and targeted employment, as 2009: Reshaping Economic Geography.24 Policy measures discussed previously in this Report. However, because might include launching retraining programs and they do nothing to change the structure of the local removing barriers to labor mobility, including lack of economy, they are likely to be effective only in the short access to information on job opportunities and subsi- run in the absence of complementary interventions. dies for relocation. China’s relaxation of strict controls on internal mobility and the integration of Germany Environmental sustainability after 1989 are natural experiments that prove the posi- tive effects of promoting labor mobility, not just on the Chapter 5 highlights how GVCs can have significant individuals who migrate, but also on the places both negative environmental impacts. These impacts receiving and sending migrants. include both the aggregate global effects of GVC- Spatially targeted interventions work best when oriented production and distribution systems and designed with a clear understanding of the structural the place-specific effects of the concentration of GVC conditions of the region. Regions that are periph- activities in a country or region. And yet, as outlined eral and sparse in population may well have niche in chapter 5, GVCs also offer opportunities to support opportunities that can be exploited, but interventions environmentally sustainable production models, espe- should give priority to raising welfare, strengthening cially if countries adopt appropriate policies. Building human capital, and promoting mobility. By contrast, environmental sustainability directly into both the regions that have sufficient density and relatively production and governance models guiding GVCs good market access are candidates for place-based will be increasingly critical to their ongoing viability. interventions that aim to overcome coordination fail- That effort will require a combination of appropriate ures blocking investment and preventing the forma- pricing, regulations, and cooperative arrangements. tion of productive agglomerations. Both pricing environmental degradation and man- The traditional subsidies used to attract (usually dating sustainable production, in particular for local foreign) investors to regions lagging or suffering pollutants, are needed to counter the negative impacts from trade adjustment have largely been ineffective, on the environment of the scale and composition for several reasons. For one thing, the level of subsidy effects discussed in chapter 5. Effective policy support required to overcome constraints such as poor infra- in the form of carbon taxes or tradable emissions per- structure, distance to markets, and lack of agglom- mits, regulations such as low-carbon fuel standards, eration is usually infeasibly large. For another, there and restrictions on the use of fossil fuels are needed. 204 | World Development Report 2020 For example, the International Maritime Organiza- harming external competitiveness and in compliance tion (IMO) has committed to an energy transition in with trade law.31 shipping to zero-emissions fuels.27 The combination of The same policy design could be used for taxing taxes on shipping fuels and regulations could support the environmental costs of traded commodities. For fully exploiting the existing energy efficiency poten- example, a country could tax carbon-intensive alumi- tial and developing alternative fuels and other inno- num by setting an excise tax according to the social vative solutions. The challenge, however, is that ships cost of the emissions typically released per ton of the are highly mobile: they travel mostly in international metal. The tax would be levied at the same rate for all waters and can easily be registered anywhere. Thus aluminum used in the country, but not for exports. regional or—ideally—global cooperative solutions are The efficiency of the tax scheme could be further preferable. improved by granting output-based tax rebates to domestic and overseas firms that adopt low-emissions Pricing environmental degradation production techniques. Pigouvian taxes equivalent to the environmental harm Despite the clear efficiency of a carbon tax, its inflicted by an activity (for example, a tax on carbon) feasibility from a political perspective is far from guar- would reduce the use of energy and lead to more inno- anteed. Unlike regulations, which hide the additional vation in energy-efficient products. Not pricing the costs to consumers, carbon taxes are very visible. In environmental costs implies a subsidy for fuels. Using late 2018, when France announced an increase in fuel the difference between existing and efficient (inclu- taxes in part to help the country transition to renewable sive of environmental costs) prices, the International energies, it stoked a reaction from the gilets jaunes (or Monetary Fund (IMF) estimates that the unpriced yellow vests) movement (according to French law, all externality caused by fossil fuels is more than 10 times vehicles must carry yellow vests in case of emergency). the direct financial subsidy.28 Large, at times violent, protests broke out across the What can countries do? Many still have energy country, led by commuters—many from rural areas— subsidies in place that lead to excess pollution. The and those in the transport industry. Protestors argued subsidies are often regressive because the poor tend that the rise in fuel taxes imposed a disproportionate to have smaller houses and fewer appliances and burden on the poorest in society, particularly those thus use less energy. Removing explicit fuel subsi- living in nonurban areas who were already suffering dies shifts business incentives away from energy- from stagnating incomes and poor public transport intensive production toward labor. Shifting from a services. Meanwhile, large multinationals—deemed to system that is subsidizing carbon emissions or is neu- bear greater responsibility for rising emissions world- tral to taxing those emissions is optimal. wide—could find ways to minimize their tax burden To minimize distortions of trade, the most efficient and still reap large profits. Ultimately, the government implementation of a carbon price would be through was forced to withdraw the proposed tax increase. an international agreement on a “carbon price floor.”29 Universal adoption may, however, suffer from free- Mandating more sustainable production rider problems for public goods. Some observers have Countries can also use regulatory (command-and- suggested that incentives to join the agreement could control) policy to deal with externalities from traded be strengthened if participating countries agree to commodities, especially to maintain clean water, grant preferential access to each other’s markets.30 prevent overfishing and overfarming, curb emissions Countries can also act unilaterally. To maintain and specific pollutants, and reduce the production of competitiveness, they can tax the consumption of disposable, single-use goods. For example, agricul- pollution-intensive goods rather than their produc- tural runoff contaminated with high levels of pesti- tion (box 8.4). Different forms of consumption-based cide and fertilizer residue, as well as organic matter carbon pricing are available, but a simple design and sediment, is the primary source of water pollu- consistent with trade law is to tax carbon in much tion in many countries, particularly high-income the same way that countries use corrective taxes for and middle-income ones. A 2018 report by the Food tobacco or alcohol—they apply a consumption-based and Agriculture Organization (FAO) on the topic excise tax. In many countries, both imported and identifies the regulatory measures needed to reduce domestically produced alcohol are taxed alike when agriculture-related water pollution, such as water they are consumed within the country, but alcohol quality standards and pollution discharge permits.32 headed for export is exempt from the tax. This way, China, the world’s largest pork producer, adopted new corrective taxes can be applied unilaterally without laws in 2015 to manage runoff from pig farms and to Policies for inclusion and sustainability | 205 Box 8.4  Cost-effectiveness and equitability of environmental regulation The year 2020 marks the 100th anniversary of British That general argument—that pollution taxes are worse economist Arthur Pigou’s description of environmental for poor households—fails to consider the case in their pollution as “externalities” and his suggestion that they favor for two further reasons. First, it ignores what happens be addressed with taxes.a Pollution taxes are more cost- to the tax revenues. If revenues are distributed to rich effective than other types of environmental regulations— households or used to fund programs that mostly benefit that is, such taxes reduce pollution the most per dollar of rich households, they would, of course, be regressive. cost or, equivalently, taxes cost the least per ton of pollu- But if the revenues are paid to or fund programs for poor tion reduced. They are also more robust to tax evasion than households, that can offset the higher tax burden on poorer direct taxes and are a fiscally efficient means of domestic people. If the revenues are divided evenly, there would be resource mobilization because their coverage extends to a net benefit to poorer households. Poor consumers would the informal sector. pay more in taxes as a share of their income, but would Despite that cost-effectiveness and despite the advocacy receive an even larger share of the dividends. of such taxes by economists, policy makers worldwide have A second shortfall of the regressivity argument is that largely chosen other types of regulations over pollution policies enacted in lieu of pollution taxes can be worse for taxes. Many claim that the burden of paying pollution taxes poor households, no matter what is done with the revenue. would be unfair, or regressive, falling disproportionately on Consider energy efficiency mandates—the type of technical poor households and poor countries. But the evidence that rules that require appliances, buildings, and vehicles to use pollution taxes harm poor people is not straightforward. less energy in their operation. For vehicles, these take the Richer people are indirectly responsible for more pollu- form of fuel economy standards that amount to a tax on gas tion because they spend more money and consume more guzzlers and a subsidy for efficient cars. Whereas a gas tax goods whose production generates pollution. Thus if one targets fuel directly, a fuel economy standard effectively follows Pigou’s 100-year-old advice and taxes pollution, taxes vehicles based on their fuel-consuming attributes. rich households would pay more than poor households in If U.S. households were taxed based on the vehicles absolute terms. But that spending on polluting goods may they own in a way designed to mimic a fuel economy stan- constitute a larger share of poor households’ incomes. In dard and raise the same revenue as a $0.29 a gallon gas relative terms, then, those tax payments could fall dispro- tax, poor households would pay an extra $92 a year and portionally on the poor. Which effect prevails varies across rich households an extra $260. Even if all the revenue were economies. In lower-income countries, these taxes tend refunded evenly, or if the tax subsidy combinations were to be regressive, with overall positive effects on equity.b designed to be revenue neutral, poor households’ net tax Furthermore, a proportion of the environmental tax burden rebates would be lower with the fuel economy standard may fall onto capital factor incomes. Since capital is highly than the gas tax. Fuel economy standards are therefore unequally distributed, studies considering these General both less cost-effective and less progressive than Pigou’s Equilibrium effects find a progressive impact.c 100-year-old, mostly disregarded suggestion. Source: WDR 2020 team, based on Levinson (2019). Note: Data are from the 2009 National Household Travel Survey, conducted by the U.S. Federal Highway Administration (https://nhts.ornl.gov/). Data include all 101,000 households surveyed, including those without vehicles. a. Pigou (1920). b. Dorband et al. (2019). c. Beck et al. (2015); Dissou and Siddiqui (2014); Metcalf and Hassett (2012); Rausch, Metcalf, and Reilly (2011). institute more efficient, sustainable farming meth- growing concern. The European Union announced ods. In fisheries, quota systems are used to prevent in March 2019 that single-use plastics will be banned overfishing, although there is concern that quotas, as of 2021 and has implemented a target to recycle which are based on commercial considerations, may 90 percent of plastic beverage bottles by 2029. Canada be too high to prevent the depletion of certain fish announced a similar measure in May 2019, banning stocks. The European Union has committed to basing single-use plastic items such as bags, straws, cutlery, all fishing quotas on scientific advice (the “maximum and stirring sticks as of 2021. Many more countries sustainable yield”) by 2020.33 have long imposed bans on single-use plastic bags, How to reduce and eliminate the consumption including Bangladesh (2002), Kenya (2017), and of disposable goods such as single-use plastics is a Rwanda (2008). In Kenya, the acts of manufacturing 206 | World Development Report 2020 and importing plastic bags incur penalties ranging industries, such as fuel, tend to be located upstream between $19,000 and $38,000 and jail terms of up to in the value chain and face lower tariffs (as discussed four years. Although some large firms have announced in terms of tariff escalation in chapter 7). The greater the elimination of certain single-use plastics in their environmental protection of downstream industries supply chain, government measures are needed to is explained by the fact that downstream industries achieve broad-based change across society. lobby for relatively low tariffs on their inputs and rel- Developing countries sometimes worry that envi- atively high tariffs on competing goods. If countries ronmental policies would be to their economic disad- applied similar trade policies to clean and dirty goods, vantage. However, the economic literature over the global CO2 emissions may decline substantially with- last 30 years on “double dividends” and the “economic out a fall in global real income.42 co-benefits of environmental policies” finds that inter- In addition to greater market access, countries nalizing external costs through fiscal policy raises upstream in the chain may be reluctant to raise economic development more often than it deflects environmental standards because of fear of losing it.34 It is precisely in those economic circumstances investors. This reluctance leads to a “regulatory chill” characteristic of developing countries—informality,35 in which regulatory progress stalls across policy difficulty in raising domestic tax revenue,36 a highly areas that affect foreign investors. The extent of a distorted preexisting tax system,37 and high air pollu- regulatory chill is a function not of whether firms tion levels,38 among others—that the probability of a will actually relocate, but of whether governments double dividend is higher. believe their threats to do so.43 Evidence from the Other policies can be used to further stimulate economic literature (and the World Bank’s opera- sustainable production and consumption. For exam- tional experience) indicates that governments tend ple, a number of governments are exploring feebates, to believe these threats, even when they are not which combine a surcharge for energy-inefficient credible.44 International investment agreements, production with rebates for energy-efficient produc- especially those with investor-state dispute set- tion.39 Policy makers also need to take into account tlement (ISDS) provisions, seem to be particularly behavioral biases to changing habits. People may stick vulnerable to regulatory chill.45 A solution is not to to old habits for lack of sufficient incentives to switch forgo ISDS provisions, which can help compensate to more sustainable production or consumption. They for weak institutions, but to exclude environmental may also not be aware of better alternatives. In this lat- and health provisions. ter case, labels and certification schemes for sustain- Both international climate agreements and trade ability standards can help. Standards, and in particular and investment agreements can be used to help address private standards, play an increasingly important role the risk of regulatory chill and to implement environ- in GVCs. mental regulations. International treaties such as the A private sector solution to externalities from Paris Agreement include ambitious commitments for traded products is sustainability certification. Such environmental protection and emissions reduction. certificates were first issued in the timber market, Recent trade agreements have taken into account the specifically as a solution for trade, and now they need for environmental policy. Countries that do not are spreading to other commodities. Sustainability adhere to environmental commitments risk losing the certificates can greatly improve the sustainability of preferential market access in the agreements. Deep trade, but they also have their limits. Governments are trade agreements also increasingly include environ- increasingly recognizing the importance of working mental provisions. For example, the Comprehensive with large international corporations and other pri- and Progressive Agreement for Trans-Pacific Partner- vate sector actors to help them establish and phase ship (CPTPP) has an environmental chapter that pro- in higher standards in their production networks. In motes sustainable fisheries (see chapter 9). And new some cases, governments and stock exchanges are agreements include trade-exposed industries in envi- making mandatory sustainability reporting by large ronmental policies without the need for exemptions multinationals and their suppliers.40 for competitiveness problems and without violations of trade law.46 Using trade policy and agreements? In most countries, import tariffs and nontariff bar- Encouraging green goods and riers are substantially lower on dirty industries than environmentally friendly production? on clean industries, measured as carbon dioxide Production subsidies are typically considered distor- (CO2) emissions per dollar of output.41 These dirtier tionary because they encourage production above the Policies for inclusion and sustainability | 207 efficient level. They are especially worrisome in indus- the targeted goods and the agreement would have tries such as agriculture and mining, where overuse is required more liberalization from them. In addition, particularly harmful to the environment. However, if the large countries could not agree on the final list of the good or production process has a positive external- products. The agreement therefore remains stalled. ity, the standard argument changes. And yet the goals of the agreement remain desirable: In particular, there is a possible argument for sub- increase trade, reduce the price of environmental sidizing green goods, especially in industries where goods, and reduce CO2 emissions. If the agreement costs fall with higher production. The electric vehicle one day becomes a multilateral one, the magnitude of sector, for example, was fostered by government inter- the identified impacts would increase substantially.47 ventions tailored to stimulate supply and demand in Chapter 7 discusses the role that SEZs and indus- both China and the United States. Indeed, such sub- trial parks could play in stimulating GVC production. sidies hastened the economic viability of the sector. For their part, governments could induce GVC firms Similarly, deployment incentives have been important to opt for industrial parks that encourage the use of for solar photovoltaic firms in China, Europe, and environmentally friendly production techniques. Latin America, especially for new projects. But such Worldwide, more than 300 industrial parks now con- incentives must be weighed carefully because they sider themselves to be eco-industrial parks (EIPs)—a can distort trade and become fiscally burdensome to number that is expected to rise. In many countries, governments. governments have become more conscious of green Trade agreements on environmental goods can approaches to manufacturing, and lead firms, con- also promote their use, effectively lowering their cost cerned about their reputation, are eager to improve relative to other goods. In the summer of 2014, a group the sustainability of production (box 8.5). of nearly 50 members of the World Trade Organiza- Finally, better metrics are needed for monitor- tion launched negotiations to reduce tariffs on green ing environmental practices to understand prob- goods. Relatively few developing countries signed on lem firms, industries, and regions and encourage because their tariffs tend to be relatively higher on upgrading. Transparent, consistent, and standardized Box 8.5  Green industrial parks support sustainable production and attract better investors In 2018, 40 countries, including Argentina, Bangladesh, Figure B8.5.1  The number of China, Colombia, the Arab Republic of Egypt, India, Indo- eco-industrial parks grew rapidly nesia, the Republic of Korea, Lebanon, Mauritania, Morocco, from 1985 to 2015 Thailand, Turkey, and Vietnam, were home to more than 160 300 eco-industrial parks (EIPs) and special economic zones Cumulative number of EIPs 140 (SEZs).a (Figure B8.5.1 shows the worldwide growth in the 120 number of EIPs from 1985 to 2015.) Because in many coun- tries a high share of export-oriented industrial production 100 is located in industrial parks located in SEZs, a correspond- 80 ingly high share of industrial emissions originates from 60 them—not only air and water pollution but also greenhouse 40 gases. For three reasons, then, SEZs and industrial parks 20 are relevant to pollution control and GVCs.b 0 First, GVCs are now in a position to create strong incen- 1985 1990 1995 2000 2005 2010 2015 tives for more sustainable production in SEZs. A major issue Non-OECD OECD for many developing countries is attracting foreign invest- Source: Kechichian and Jeong 2016. ment, diversifying export baskets, and creating better jobs. Note: EIPs = eco-industrial parks; OECD = Organisation for Economic But in many old-style SEZs, the environmental standards Co-operation and Development. (Box continues next page) 208 | World Development Report 2020 Box 8.5  Green industrial parks support sustainable production and attract better investors (continued) were low, with an industrialization model based on the developed in 2017 an international framework to guide attractiveness of low production costs and taxes. Under policy makers in establishing environmentally sustainable emerging laws on sustainability reporting, companies with EIPs. Meanwhile, EIPs are becoming increasingly important headquarters in many industrialized countries are liable for for overcoming sustainability challenges within the scope risks along their value chains. To reduce those risks and of the United Nations’ Sustainable Development Goals. ensure the traceability and quality of final products, compa- Countries such as Denmark, France, Japan, and Korea, nies are now seeking more transparency along value chains. among many others, have leveraged EIPs to promote more Second, as in other policy areas, SEZs offer an avenue inclusive and sustainable action to improve industrial com- for policy experimentation in making industrial parks sus- petitiveness in line with climate change goals.c tainable. New environmental policies and disciplines can be Third, from the perspective of an industrial park oper- implemented in a more manageable environment, such as ator or developer, offering environmentally sustainable promoting recycling, provisioning renewable energy and facilities is an opportunity to attract higher-quality and other green infrastructure, constructing environmentally higher-paying tenants, which GVC firms tend to be. With friendly buildings, and reusing and commercializing waste thousands of industrial parks globally, operators are seek- products. Because waste reuse and energy cogeneration ing a more sustainable and competitive operating environ- can be designed to link firms within the same SEZ, some of ment in order to differentiate themselves from the more these policies can take advantage of the ecosystem aspects basic industrial parks. For example, Hawassa industrial park of SEZs (industrial symbiosis). in Ethiopia adopted zero-liquid discharge technologies for In a bid to address the negative environmental impacts wastewater treatment to attract high-end apparel manu- of the concentration of industrial production, the World facturers. Vietnam recently issued guidelines for improved Bank, in partnership with the United Nations Indus- environmental performance in its industrial parks. These trial Development Organization (UNIDO) and Deutsche approaches improve socioenvironmental performance Gesellschaft für Internationale Zusammenarbeit (GIZ), without the need for more regulations. a. Kechichian and Jeong (2016); UNIDO, World Bank, and GIZ (2017). b. Special economic zones are spaces in a country intended to attract industrial production by offering companies locating there special concessions on taxes, tariffs, and regulations. Chapter 7 describes SEZs in more detail. c. UNIDO, World Bank, and GIZ (2017). information on how firms produce is not available. 9. World Bank’s Doing Business database (2018). See Multiple sets of sustainability standards exist, from appendix A for a description of the databases used in this public to private and from mandatory to voluntary, Report. 10. Increasingly, many jurisdictions are including calcula- but they are not standardized, and ratings of the same tions of a living wage in efforts to establish the level of a firms often differ widely across them. An international minimum wage. agency tasked with these ratings could shed light on 11. Kuddo, Robalino, and Weber (2015). firms’ activities and offer incentives for changes in 12. Domat et al. (2013). behavior. 13. Barrientos et al. (2016). 14. Hollweg (2019). 15. ILO (2016, 2017). Notes 16. Covering freedom of association, right to collective 1. Beegle, Coudouel, and Monsalve (2018). bargaining, elimination of forced and compulsory labor, 2. Elbehri and Benali (2013). abolition of child labor, and elimination of discrimina- 3. Lin (2012). tion with respect to employment and occupation. 4. Robalino and Walker (2017). 17. Evans (2019). 5. World Bank (2019b, 72). 18. Wetterberg (2011). 6. World Bank (2019b). 19. Farole, Hollweg, and Winkler (2018). 7. ILO and IFC (2016). 20. Bown and Freund (2019). 8. McMillan and Verduzco-Gallo (2011); Shingal (2015). 21. World Bank (2019b). Policies for inclusion and sustainability | 209 2. Farole (2013). 2 Beck, Marisa, Nicholas Rivers, Randall Wigle, and Hidemichi 23. Dix-Carneiro and Kovak (2017). Yonezawa. 2015. “Carbon Tax and Revenue Recycling: 24. World Bank (2009). Impacts on Households in British Columbia.” Resource and 25. Rodriguez-Pose and Arbix (2001). Energy Economics 41: 40–69. 26. Austin, Glaeser, and Summers (2018); Hendrickson, Beegle, Kathleen, Aline Coudouel, and Emma Monsalve, eds. Muro, and Galston (2018). 2018. Realizing the Full Potential of Social Safety Nets in Africa. 27. IMO (2018a, 2018b). Africa Development Forum Series. Washington, DC: 28. The estimates are based on a price gap analysis that Agence Française de Développement and World Bank. takes as the reference price a calculation of externalities Bento, Antonio M., Mark R. Jacobsen, and Antung A. Liu. based on detailed country data. This calculation con- 2018. “Environmental Policy in the Presence of an trols for the true comparative advantages for hosting Informal Sector.” Journal of Environmental Economics and polluting industries, such as having a low population Management 90 (July): 61–77. density or (more controversially) a low value of sta- Böhringer, Christoph, Knut Einar Rosendahl, and Halvor tistical life. For estimates, see Coady et al. (2017, 2019), Briseid Storrøsten. 2017. “Robust Policies to Mitigate and for the method for calculating them, see Parry et al. Carbon Leakage.” Journal of Public Economics 149 (May): (2014). 35–46 29. Cramton et al. (2017); Farid et al. (2016); Weitzman (2017). Bown, Chad P., and Caroline L. Freund. 2019. “Active Labor 30. Gollier and Tirole (2015); Nordhaus (2015). Market Policies: Lessons from Other Countries for the 31. Trachtman (2017). United States.” PIIE Working Paper 19–2 (January), Peter- 32. Mateo-Sagasta, Zadeh, and Turral (2018). son Institute for International Economics, Washington, 33. “Managing Fisheries,” Common Fisheries Policy, DC. Directorate-General for Maritime Affairs and Fisheries, Brown, Drusilla, Rajeev Dehejia, Raymond Robertson, European Commission, Brussels, https://ec.europa.eu George Domat, and Selven Veeraragoo. 2015. “Are Sweat- /fisheries/cfp/fishing_rules_en. shops Profit-Maximizing? Answer: No; Evidence from 34. Pigato (2019). Better Work Vietnam.” Better Work Discussion Paper 17 35. Bento, Jacobsen, and Liu (2018). (March), International Labour Office, Geneva. https:// 36. Liu (2013). users.nber.org/~rdehejia/papers/DP%2017%20web.pdf. 37. Parry and Bento (2000). Brown, Drusilla, and Xirong Lin. 2014. “Sexual Harassment 38. Parry, Veung, and Heine (2015). in the Workplace: How Does It Affect Firm Performance 39. Fay et al. (2015). and Profits?” Better Work Discussion Paper 16 (Novem- 40. See World Business Council for Sustainable Reporting ber), International Labour Office, Geneva. https://better “Reporting Exchange” for a database on mandatory regu- work.org/blog/portfolio/sexual-harassement-in-the lation in sustainability reporting. https://www.reporting -workplace-how-does-it-affect-firm-performance-and exchange.com/ and https://www.cdsb.net/sites/default -profits/. filescdsb_report_1_esg.pdf. Brown, Julia G. 2013. “International Investment Agreements: 41. Shapiro (2019). Regulatory Chill in the Face of Litigious Heat?” Western 42. Shapiro (2019). Journal of Legal Studies 3 (1): 47–74. 43. Neumayer (2001). Coady, David P., Ian W. H. Parry, Nghia-Piotr Trong Le, and 44. Zarsky (2006). Baoping Shang. 2019. “Global Fossil Fuel Subsidies 45. Examples include countries allowing mining in forest Remain Large: An Update Based on Country-Level Esti- protection areas (Brown 2013; Gross 2003) and repealing mates.” IMF Working Paper WP/19/89 (May 2), Interna- science-based environmental regulations on oil mining tional Monetary Fund, Washington, DC. because of threats of investor-state disputes (Tienhaara Coady, David P., Ian W. H. Parry, Louis Sears, and Baoping 2011). Shang. 2017. “How Large Are Global Fossil Fuel Subsi- 46. Böhringer, Rosendahl, and Storrøsten (2017); Trachtman dies?” World Development 91 (March): 11–27. (2017). 47. European Commission (2016). Cramton, Peter, David J. C. MacKay, Axel Ockenfels, and Steve Stoft, eds. 2017. Global Carbon Pricing: The Path to Climate Cooperation. Cambridge, MA: MIT Press. References Davis, Michelle. 2017. “Satellite Factories Create More Austin, Benjamin, Edward L. Glaeser, and Lawrence H. Jobs for Women in Rural Jordan.” Jobs and Development Summers. 2018. “Saving the Heartland: Place-Based Pol- (blog), December 7. https://blogs.worldbank.org/jobs icies in 21st Century America.” BPEA Conference Draft /satellite-factories-create-more-jobs-women-rural-jordan. (March 8–9), Brookings Papers on Economic Activity, Dissou, Yazid, and Muhammad Shahid Siddiqui. 2014. “Can Brookings Institution, Washington, DC. Carbon Taxes Be Progressive?” Energy Economics 42: Barrientos, Stephanie Ware, Peter Knorringa, Barbara Evers, 88–100. Margareet Visser, and Maggie Opondo. 2016. “Shifting Dix-Carneiro, Rafael, and Brian K. Kovak. 2017. “Trade Lib- Regional Dynamics of Global Value Chains: Implications eralization and Regional Dynamics.” American Economic for Economic and Social Upgrading in African Horticul- Review 107 (10): 2908–46. ture.” Environment and Planning A: Economy and Space 48 (7): Domat, George, Paris Adler, Rajeev Dehejia, Drusilla Brown, 1266–83. and Raymond Robertson. 2013. “Do Factory Managers 210 | World Development Report 2020 Know What Workers Want? Manager-Worker Infor- ILO (International Labour Organization). 2016. Assessment mation Asymmetries and Pareto Optimal Working of Labour Provisions in Trade and Investment Arrangements. Conditions.” Better Work Discussion Paper 10 (June), Studies on Growth with Equity Series. Geneva: Inter- International Labour Office, Geneva. https://betterwork national Labour Office. https://www.ilo.org/wcmsp5 .org/global/wp-content/uploads/DP-10-web.pdf. /groups/public/---dgreports/---inst/documents/publi Dorband, Ira Irina, Michael Jakob, Matthias Kalkuhl, and cation/wcms_498944.pdf. Jan Christoph Steckel. 2019. “Poverty and Distributional — —— —. 2017. Handbook on Assessment of Labour Provisions in Effects of Carbon Pricing in Low- and Middle-Income Trade and Investment Arrangements. Growth with Equity Countries: A Global Comparative Analysis.” World Devel- Series. Geneva: International Labour Office. https:// opment 115: 246–57. www.ilo.org/global/publications/books/WCMS_564702 Elbehri, Aziz, and Marwan Benali. 2013. “A Historical Com- /lang—en/index.htm. parative Analysis of Commodity Development Models ILO (International Labour Organization) and IFC (Inter- in West Africa and Implications for Staple Food Value national Finance Corporation). 2016. “Progress and Chains.” In Rebuilding West Africa’s Food Potential: Policies Potential: How Better Work Is Improving Garment and Market Incentives for Smallholder-Inclusive Food Value Workers’ Lives and Boosting Factory Competitiveness.” Chains, edited by Aziz Elbehri, 43–81. Rome: Food and International Labour Office, Geneva. https://betterwork Agriculture Organization of the United Nations. .org/dev/wp-content/uploads/2016/09/BW-Progress-and European Commission. 2016. Trade Sustainability Impact -Potential_Web-final.pdf. Assessment on the Environmental Goods Agreement. Final IMO (International Maritime Organization). 2018a. “Marine Report. Brussels: Directorate-General for Trade, Euro- Litter.” In Focus, IMO, London. http://www.imo.org/en pean Commission. /MediaCentre/HotTopics/marinelitter/Pages/default.aspx. Evans, Alice. 2019. “Incentivising Pro-Labour Reforms.” CID — —— —. 2018b. “UN Body Adopts Climate Change Strategy for Faculty Working Paper 349 (March), Center for Inter- Shipping.” Press Briefing 06, April 13, IMO, London. http:// national Development, Harvard University, Cambridge, www.imo.org/en/MediaCentre/PressBriefings/Pages MA. /06GHGinitialstrategy.aspx. Farid, Mai, Michael Keen, Michael G. Papaioannou, Ian W. Karuretwa, Kaliza. 2016. “Case 14: Rwanda Coffee Sector.” H. Parry, Catherine A. Pattillo, and Anna Ter-Martirosyan. Case Studies on the Future of Manufacturing, World 2016. “After Paris: Fiscal, Macroeconomic, and Financial Economic Forum Global Agenda Council on the Future Implications of Global Climate Change.” IMF Staff Dis- of Manufacturing, World Economic Forum, Geneva. cussion Note 16/01 (January), International Monetary Kechichian, Etienne, and Mi Hoon Jeong. 2016. “Mainstream- Fund, Washington, DC. ing Eco-Industrial Parks.” World Bank, Washington, DC. Farole, Thomas, ed. 2013. The Internal Geography of Trade: Kuddo, Arvo, David Alejandro Robalino, and Michael Weber. Lagging Regions and Global Markets. Directions in Develop- 2015. “Balancing Regulations to Promote Jobs: From ment: Trade Series. Washington, DC: World Bank. Employment Contracts to Unemployment Benefits.” Farole, Thomas, Claire H. Hollweg, and Deborah Elisabeth World Bank, Washington, DC. Winkler. 2018. “Trade in Global Value Chains: An Assess- Levinson, Arik. 2019. “Energy Efficiency Standards Are More ment of Labor Market Implications.” Jobs Working Paper Regressive than Energy Taxes: Theory and Evidence.” 18 (July 16), World Bank, Washington, DC. Journal of the Association of Environmental and Resource Fay, Marianne, Stephane Hallegatte, Adrien Vogt-Schilb, Economists 6 (1): 7–36. Julie Rozenberg, Ulf Narloch, and Tom Kerr. 2015. Decar- Lin, Justin Yifu. 2012. The Quest for Prosperity: How Developing bonizing Development: Three Steps to a Zero-Carbon Future. Economies Can Take Off. Princeton, NJ: Princeton Univer- Climate Change and Development. Washington, DC: sity Press. World Bank. Liu, Antung Anthony. 2013. “Tax Evasion and Optimal Envi- Gollier, Christian, and Jean Tirole. 2015. “Negotiating Effec- ronmental Taxes.” Journal of Environmental Economics and tive Institutions against Climate Change.” Economics of Management 66 (3): 656–70. Energy and Environmental Policy 4 (2): 5–27. Mateo-Sagasta, Javier, Sara Marjani Zadeh, and Hugh Turral. Gross, Stuart G. 2003. “Inordinate Chill: BITS, Non-NAFTA 2018. More People, More Food, Worse Water? A Global Review of MITS, and Host-State Regulatory Freedom; An Indone- Water Pollution from Agriculture. Colombo, Sri Lanka: Inter- sian Case Study.” Michigan Journal of International Law 24 national Water Management Institute; Rome: Food and (3): 893–960. Agriculture Organization of the United Nations. http:// Hanson, Gordon H. 1998. “North American Economic Inte- www.fao.org/3/ca0146en/CA0146EN.pdf. gration and Industry Location.” Oxford Review of Economic McMillan, Margaret S., and Íñigo Verduzco-Gallo. 2011. Policy 14 (2): 30–44. “New Evidence on Trade and Employment: An Over- Hendrickson, Clara, Mark Muro, and William A. Galston. view.” In Trade and Employment: From Myths to Facts, 2018. “Countering the Geography of Discontent: Strat- edited by Marion Jansen, Ralf Peters, and José Manuel egies for Left-Behind Places.” Brookings Institution, Salazar-Xirinachs, 23–60. Geneva: International Labour Washington, DC. Office. Hollweg, Claire H. 2019. “Firm Compliance and Public Disclo- Metcalf, Gilbert, and Kevin Hassett. 2012. “Distributional sure in Vietnam.” Policy Research Working Paper 9026, Impacts in a Comprehensive Climate Policy Package.” World Bank, Washington, DC. In The Design and Implementation of U.S. Climate Policy, Policies for inclusion and sustainability | 211 edited by Don Fullerton and Catherine Wolfram. Chi- Shingal, Anirudh. 2015. “Labour Market Effects of Integration cago: University of Chicago Press. into GVCs: Review of Literature.” R4D Working Paper Neumayer, Eric. 2001. “Do Countries Fail to Raise Environ- 2015/10, Swiss Programme for Research on Global Issues mental Standards? An Evaluation of Policy Options for Development, Swiss National Science Foundation, Addressing ‘Regulatory Chill.’  ” International Journal of Bern, Switzerland. Sustainable Development 4 (3): 231–44. Tienhaara, Kyla. 2011. “Regulatory Chill and the Threat of Nordhaus, William. 2015. “Climate Clubs: Overcoming Arbitration: A View from Political Science.” In Evolution Free-Riding in International Climate Policy.” American in Investment Treaty Law and Arbitration, edited by Chester Economic Review 105 (4): 1339–70. Brown and Kate Miles, 606–28. Cambridge, U.K.: Cam- Parry, Ian W. H., and Antonio M. Bento. 2000. “Tax Deduc- bridge University Press. tions, Environmental Policy, and the ‘Double Dividend’ Trachtman, Joel P. 2017. “WTO Law Constraints on Border Tax Hypothesis.” Journal of Environmental Economics and Man- Adjustment and Tax Credit Mechanisms to Reduce the agement 39 (1): 67–96. Competitive Effects of Carbon Taxes.” National Tax Journal Parry, Ian W. H., Dirk Heine, Eliza Lis, and Shanjun Li. 2014. 70 (2): 469–94. Getting Energy Prices Right: From Principle to Practice. Wash- UNIDO (United Nations Industrial Development Organiza- ington, DC: International Monetary Fund. tion), World Bank, and GIZ (Deutsche Gesellschaft für Parry, Ian W. H., Chandara Veung, and Dirk Heine. 2015. Internationale Zusammenarbeit). 2017. “An International “How Much Carbon Pricing Is in Countries’ Own Inter- Framework for Eco-Industrial Parks.” December, World ests? The Critical Role of Co-benefits.” Climate Change Bank, Washington, DC. Economics 6 (4): 1–26. Weitzman, Martin L. 2017. “How a Minimum Carbon-Price Pigato, Miria A., ed. 2019. Fiscal Policies for Development and Commitment Might Help to Internalize the Global Climate Action. International Development in Focus Warming Externality.” In Global Carbon Pricing: The Path Series. Washington, DC: World Bank. to Climate Collaboration, edited by Peter Cramton, David Pigou, Arthur Cecil. 1920. The Economics of Welfare. London: J. C. MacKay, Axel Ockenfels, and Steven Stoft, 125–48. Macmillan. Cambridge, MA: MIT Press. Rausch, Sebastian, Gilbert Metcalf, and John Reilly. 2011. Wetterberg, Ana. 2011. “Public-Private Partnership in Labor “Distributional Impacts of Carbon Pricing: A General Standards Governance: Better Factories Cambodia.” Equilibrium Approach with Micro-Data for Households.” Public Administration and Development 31 (1): 64–73. Energy Economics 33 (S1): S20–S33. World Bank. 2009. World Development Report 2009: Reshaping Robalino, David Alejandro, and David Ian Walker. 2017. Economic Geography. Washington, DC: World Bank. “Guidance Note: Economic Analysis of Jobs Investment —— ——. 2017. “In Bangladesh, Empowering and Employ- Projects.” Jobs Working Paper 7 (August), World Bank, ing Women in the Garments Sector.” World Bank, Washington, DC. Washington, DC. https://www.worldbank.org/en/news Rodríguez-Pose, Andrés, and Glauco Arbix. 2001. “Strategies /feature/2017/02/07/in-bangladesh-empowering-and of Waste: Bidding Wars in the Brazilian Automobile Sec- -employing-women-in-the-garments-sector. tor.” International Journal of Urban and Regional Research 25 —— ——. 2018. “Maximizing Finance for Development: Côte (1): 134–54. d’Ivoire, Improving Opportunities through Cashew Rourke, Emily L. 2014. “Is There a Business Case against Value Chains.” MFD Brief 06/2018 (June), World Bank, Verbal Abuse? Incentive Structure, Verbal Abuse, Pro- Washington, DC. ductivity, and Profits in Garment Factories.” Better Work —— ——. 2019a. Women, Business, and the Law 2019: A Decade of Discussion Paper 15 (September), International Labour Reform. Washington, DC: World Bank. Office, Geneva. https://betterwork.org/global/wp-content —— ——. 2019b. World Development Report 2019: The Changing /uploads/2014/09/DP-15-web.pdf. Nature of Work. Washington, DC: World Bank. Shapiro, Joseph S. 2019. “The Environmental Bias of Trade Zarsky, Lyuba. 2006. “From Regulatory Chill to Deepfreeze?” Policy.” Paper presented at the Association of Environ- International Environmental Agreements: Politics, Law and mental and Resource Economists’ Summer Conference, Economics 6 (4): 395–99. Lake Tahoe, NV, May 30–31. 212 | World Development Report 2020 PART V How can international cooperation help? 9 Cooperation on trade 10 Cooperation beyond trade 9 Cooperation on trade Key findings  eveloping countries have benefited from the rules-based trade system, with its • D guarantees against trade discrimination, incentives to reform, assured market access, and dispute settlement.  he international trade system is especially valuable in a global value chain (GVC) world. • T Policy action or inaction in one country can affect producers and consumers in other countries. ncreasing pressure on the global trading system, manifested in protectionism and policy • I uncertainty, puts these benefits at risk. These pressures arise, first, from the growing symmetry in the economic size of countries and the persistent asymmetry in their levels of protection; second, from the failure to use domestic policies to address labor market dislocation and growing inequality in some advanced countries.  o sustain beneficial trade openness, countries need to deepen traditional trade • T cooperation to address remaining barriers to trade in goods and services, as well as other measures that distort trade, such as subsidies and the activities of state-owned enterprises.  eaningful outcomes may be possible if the major developing country traders engage • M as equal partners and even leaders instead of seeking special and differential treatment; if the large advanced countries continue to place their faith in rules-based negotiations instead of resorting to unilateral protection; and if countries together define a negotiating agenda that reflects both development and business priorities. 216 | World Development Report 2020 D eveloping countries have benefited enor- and new industries in advanced countries. Although mously from the rules-based multilateral the rapid trade growth of the 1990s and early 2000s trade system. In fact, it is hard to imagine any supported overall income growth, it also created win- current global value chain (GVC) operating outside ners and losers. Those forces were magnified with the of the membership of the World Trade Organization expansion of GVCs because of the hyperspecialization (WTO). The trade system has provided countries with that GVCs produced. Some manufacturing communi- incentives to reform, market access around the globe, ties in advanced countries experienced large job losses and recourse in case of disputes, even against the trade as imports took market shares from domestic firms. heavyweights. Estimates suggest that acceding to the And as developing country production grew rapidly, WTO boosts a developing country’s growth rate by 2 exporters from advanced countries—the traditional percent a year for five years after joining, if the coun- supporters of open trade policies—also experienced try made reforms upon accession.1 The tariffs they face more intense competition at home and in other fall significantly. For example, 90 percent of U.S. tariff markets. Because some of the new developing coun- lines applied to WTO members are below 10 percent, try markets were still relatively protected and their whereas for nonmembers 50 percent of products are exporters were supported by the state, trust in the subject to tariffs of more than 30 percent. Developing trade system to ensure equal treatment eroded. countries also have had success in WTO dispute set- In addition to the challenges presented by the tlement, even against the WTO's largest members. For growing competition, the new global economy pro- example, Indonesia recently won a case against the duced other significant risks that led to disenchant- European Union (EU) about antidumping measures ment (discussed in more detail in chapter 10). A greater for biodiesel products. share of the burden for resource mobilization shifted Supporting the rules-based trade system is there- to workers as capital became much harder to tax in a fore important for development, but a series of events GVC world. Because firms operate around the world have weakened it. The failure of the WTO’s Doha and a high share of value added has become virtual, Round, which began in 2001, was the first strike, they can easily shift profits to low-tax jurisdictions. and recent disputes among members have further The new global economy also sparked concerns about damaged the system. Regional initiatives such as the market failure in international markets where regula- European Union and the North American Free Trade tion remains mostly national. Concerns ranged from Agreement (NAFTA) have also been hurt by disagree- abuse of privacy in data-based services to anticompet- ments among member countries. In view of this trade itive practices in platform-based services. Some devel- climate, this chapter and the next argue that (1) the oping countries also became disenchanted with the multilateral trade system matters profoundly in a international trade system, especially in light of the GVC world; (2) the system is under stress because of failed Doha Development Agenda because the areas tensions between the existing rules and the forces of that matter the most to them, such as agriculture and economic convergence; and (3) revival of the system apparel, have failed to be liberalized. will depend on deepening trade cooperation and The path forward will require more cooperation extending cooperation to new areas. between the new players in global trade, the large The multilateral trade system is especially import- developing countries, and the incumbents, the large ant for GVCs because the costs of protection are mag- advanced countries. The large developing countries nified when goods and services cross borders multiple were mostly inactive during earlier episodes of recip- times. Similarly, the gains from a coordinated reduc- rocal liberalization, but they have now grown to a size tion of barriers to trade are even larger for GVCs than where their exports and their markets matter. Tradi- for conventional trade. Trade and investment policies tional trade negotiations may deliver more meaning- must be known and predictable to encourage firms ful outcomes if the major developing country traders to invest in long-term international relationships. To engage as equal partners, and even leaders, instead of address this need, international trade agreements seeking special and differential treatment (box 9.1); if include rules to enhance the transparency of national the large industrial countries continue to place their policies and help reduce policy uncertainty through faith in rules-based negotiations instead of resorting legally binding commitments. Trade agreements and to unilateral protection; and if all countries together WTO commitments can also help to discipline the pro- define a negotiating agenda that reflects both devel- tectionist impact of differences in regulatory regimes. opment and business priorities. But rapidly growing trade, especially with low- To sustain trade openness, the first priority is to income countries, has put pressure on both existing deepen traditional trade cooperation to address the Cooperation on trade | 217 Box 9.1  Special and differential treatment for developing countries An important feature of the World Trade Organization not been able to agree on criteria to differentiate between (WTO) is the approach it takes to the disparities in the countries and determine when graduation should occur. economic size and capacity of its members. This approach Notwithstanding the rhetoric by opponents and pro- is encapsulated in the principle of special and differential ponents of SDT, building blocks for a more differentiated treatment (SDT) for developing countries—a feature of the approach toward addressing economic development dis- trade system almost since its origins. SDT arose because parities have gradually emerged. In practice, differentiation the export earnings of developing countries were insuffi- has been negotiated on an issue-specific basis. An important cient for development needs and unpredictable because of example is the classification of developing countries based the fluctuations in commodity prices. The solution was to on per capita GDP and export competitiveness in the WTO’s give developing countries more flexibility in tariff setting Agreement on Subsidies and Countervailing Measures. Other and more access to markets in developed countries. examples include the flexible approach taken in the WTO’s SDT also served a purpose for developed countries; it Trade Facilitation Agreement (TFA) to scheduling commit- made negotiations easier because those countries could ments by developing countries and the ability of developing exchange market access among a small group without countries to link implementation of specific TFA provisions having to reach consensus with the full membership of to technical assistance. The TFA embodies a new approach the predecessor of the WTO, the General Agreement toward SDT that is not centered on exemptions for developing on Tariffs and Trade (GATT). At a time when developing countries. Instead, it lets countries decide on the sequencing countries accounted for less than a third of global exports, of implementation, depending on which elements of the this approach made sense to developed countries. How- agreement are priorities from a national perspective and ever, times have changed, and with developing countries commitments by high-income countries to assist those coun- accounting for nearly 45 percent of global exports, it is no tries that request it to implement specific provisions. longer palatable to developed nations. Studies reveal that traditional SDT has not served devel- A peculiar feature of SDT is that countries can declare oping countries well.a Their trade interests, such as agricul- themselves developing countries on a particular issue ture and apparel, have been liberalized slowly or not at all. to avoid full commitments. For example, Japan and the It has also lessened the ability of the trade system to act as Republic of Korea used SDT to postpone commitments an external force for domestic reform. As a result, tariffs in to changing from a quota system to a tariff system on rice developing countries are on average bound at the WTO at in the Uruguay Round. The WTO does not define what 30 percentage points above actual levels. Meanwhile, tariff constitutes a developing country, leaving it to members liberalization among developing countries has been largely to self-determine their status. Outside of the group of unilateral; it has not occurred from external negotiations. 47 (UN-defined) least developed countries (LDCs)— Studies also find that developing countries have had lim- the only distinct group of developing countries formally ited gains from trade preferences, another dimension of identified in the WTO—there are no criteria that allow differ- SDT, because of their unilateral and uncertain nature and entiation between developing countries. WTO members have associated conditions, such as restrictive rules of origin. a. See, for example, Ornelas (2016). remaining barriers to trade in goods and services, in other countries. International cooperation can help as well as other measures that distort trade. Along- address the policy spillovers and achieve better devel- side such an effort, cooperation should be widened opment outcomes in several ways. beyond trade policy to include taxes, regulation, First, because the costs of protection are magnified and infrastructure, as discussed in more detail in when goods and services cross borders multiple times, chapter 10. the gains from a coordinated reduction of barriers to trade are even larger for GVCs than for standard trade. Because foreign investment and GVCs are inextrica- The case for cooperation bly linked, creating an open and secure climate for GVCs span boundaries, and policy action or inaction investment is vital for GVC participation, especially in one country can affect producers and consumers by capital-scarce countries. International cooperation 218 | World Development Report 2020 has so far delivered greater openness in goods and policy uncertainty through legally binding commit- services, but significant barriers remain. ments. But the failure of countries to honor WTO Second, access to information about trade and requirements that they provide regular notifications investment policies and their predictability is import- of subsidies and other measures that affect trade has ant for firms, especially when investing in interna- led to policy opaqueness and has caused trade ten- tional relationships. To address this need, interna- sions. Similarly, large wedges between legal bindings tional trade agreements include rules to enhance the and applied policies in both goods and services have transparency of national policies and to help reduce perpetuated policy uncertainty (box 9.2). Box 9.2  A story of the demise of most-favored-nation status foretold? This is not the first time the world economy has confronted was largely maintained until the early 20th century. That a situation in which the most powerful country moves away commitment first faltered when the United States and Ger- from a policy of nondiscriminatory openness. A surpris- many threatened British dominance toward the end of the ing aspect of British trade policy in the 19th century was 19th century, causing its share of world trade to dip below its nonexclusivity. With a share of world exports of more 15 percent, and collapsed around the time of the Great than 20 percent, Britain sought and obtained not preferred Depression when Britain’s share fell below 10 percent, lead- access to resources and markets but a commitment to ing to a policy of imperial preferences as well as increased nondiscriminatory trade (figure B9.2.1). Combined with protection. its unilateral adoption of a free trade policy applied on a Figure B9.2.1 shows that the events during Pax Britan- most-favored-nation (MFN) or nondiscriminatory basis, nica bear an uncanny resemblance to the U.S. role as a pillar this approach defined the “free trade imperialism” that of the multilateral trading system during Pax Americana in prevailed during Pax Britannica, beginning in the early 19th the 20th century. The U.S. share of world trade had reached century and peaking in the mid-19th century. This stance 20 percent before World War II. In 1947 the United States Figure B9.2.1  Shifts in trade shares and changes in policy stances of the United Kingdom and the United States since 1800 60 UK: UK: first UK: US: US: first US: confident, concerns departs confident, concerns departs chooses about from chooses about from 50 openness German, US MFN openness Japan MFN? and catch-up and MFN catch-up Share of world exports (%) MFN Pax Pax Britannica Americana 40 30 20 10 0 1800 1850 1900 1950 2000 2016 United Kingdom (UK) United States (US) Source: Hoekman and Mattoo 2019. Note: MFN = most-favored-nation. (Box continues next page) Cooperation on trade | 219 Box 9.2  A story of the demise of most-favored-nation status foretold? (continued) was, unquestioned, the dominant power in the world econ- the successful expansion of policy coverage of the trade omy and played a central role in the creation of the General system to areas in which the United States had a compara- Agreement on Tariffs and Trade (GATT). It accommodated tive advantage—services and innovation. the formation of the European Economic Community (EEC) As the U.S. share of global exports declined, the United without departing from its MFN policy itself, and when it States retreated from nondiscrimination. Unexpectedly began to feel the discriminatory effects of European inte- rapid growth by China and other emerging economies in gration, it pushed for reductions in MFN tariffs through the late 1990s and the 2000s drove the U.S. share of global multilateral GATT negotiations (in the Kennedy and Tokyo exports below 10 percent, which seems to be a critical Rounds) rather than through unilateral action. The U.S. threshold inducing the incumbent power to depart from commitment first wavered at the end of the 1970s when an MFN policy. The result was first an attempt to negotiate Japan emerged as a major trader and the U.S. share of world modern-day “imperial preferences” under the Comprehen- trade fell below 15 percent. But the United States relaxed sive and Progressive Agreement for Trans-Pacific Partner- when Japan did not threaten to cause a further decline in ship (CPTPP) beginning in 2008, before the recent recourse share and because the Uruguay Round negotiations led to to discriminatory tariffs and bilateralism. Third, many of the policies affecting GVCs are international order, potentially exacerbating tensions regulatory, including technical regulations, sanitary between countries (box 9.2). and phytosanitary measures, and a range of regula- A consequence is growing political sensitivity to tions for services. Trade agreements and WTO com- the plight of industrial workers in advanced coun- mitments have made some progress in disciplining the tries, whose incomes have stagnated during periods protectionist impacts of these measures, but they tend of rapid globalization. There is evidence that trade to view the measures primarily through a producer- contributed to job loss in some countries, but tech- centric market-access lens. Accordingly, countries nological change reduced the number of jobs in man- have attempted to harmonize or mutually recognize ufacturing for unskilled workers to a much greater product standards and other regulations in the con- extent. At the same time, the emergence of winner- text of regional agreements, seeking to emulate the takes-all industries concentrated income growth in progress in the European Union, especially in goods. the top 1 percent. Progress has been limited, however, because of the Even though trade may not have been the only significant divergence in social preferences across source of the problem, globalization makes reme- countries on regulatory issues. dial action difficult. The winners from globaliza- tion—internationally mobile capital and skills—are Why the system is under stress increasingly hard to tax. Therefore, workers bear not The current retreat from globalization is most obvi- only the burden of adjustment, but also, increasingly, ous in industrial countries: many workers feel they the burden of taxation (figure 9.2). And governments have not benefited from it (figure 9.1); firms feel are tempted to use trade policy as an instrument of they face unfair competition; and consumers worry social protection. about environmental and social standards asso­ To sustain beneficial trade openness, it is essen- ciated with imports. Low-skilled workers in some tial to “walk on two legs.” The first priority is to advanced-country manufacturing communities have deepen traditional trade cooperation to address the seen job opportunities disappear as imports of com- remaining barriers to trade in goods and services, peting goods from developing countries grow. Mean- as well as other measures that distort trade such as while, market- and private enterprise–based policy subsidies and the activities of state-owned enter- regimes tend to be not well suited to softening the prises (SOEs). In parallel, cooperation should be wid- pain associated with adjustment. Sharp adjustments ened beyond trade policy to include taxes, regulation, in trade patterns have also threatened the existing and infrastructure (the subject of the next chapter). 220 | World Development Report 2020 Figure 9.1  Attitudes toward trade differ in the sluggish North and the dynamic South 80 UGA LBN BGD TUN VNM 70 KEN 60 % who say trade increases wages IDN CHN TZA NIC 50 UKR SEN NGA PAK EGY GHA ZAF MYS IND SLV KOR BRA JOR PER 40 VEN ISR THA PHL POL GBR MEX 30 TUR ESP RUS ARG CHL DEU GRC COL 20 USA FRA 10 JPN ITA 0 –6 –4 –2 0 2 4 6 8 10 Average GDP growth, 2008–13 (%) Advanced economies Developing economies Emerging markets Sources: WDR 2020 team, based on Pew Research Center, Spring 2014 Global Attitudes Survey, Q28 (https://www.pewresearch.org/global/2014/06/05/spring -2014-survey-data/); GDP annual growth: International Monetary Fund’s World Economic Outlook database, April 2014. See appendix A for a description of the databases used in this Report. Note: For country abbreviations, see International Organization for Standardization (ISO), https://www.iso.org/obp/ui/#search. Table 9.1 lists the policy areas in which national Figure 9.2  Corporate tax rates and personal income incentives can produce an outcome that is bad for tax rates for the top 1 percent have fallen, but the rate all or most countries and a cooperative solution that for the median worker increased in 65 economies is better for all. between 1980 and 2007 50 18 Deepening trade cooperation Corporate and top 1% income tax rate (%) Because the costs of protection are magnified when 45 Median income tax rate (%) goods and services cross borders multiple times, the 17 gains from coordinated reduction of barriers to trade are even larger from GVCs than from standard trade. 40 And because foreign investment and GVCs are linked, 16 creating an open and secure climate for investment 35 is vital for GVC participation, especially for capital- scarce countries. International cooperation has so far 15 delivered greater openness, if unevenly: 30 For goods, multilateral and preferential initiatives •  have worked in tandem to reduce the tariffs on goods 25 14 and to greatly enhance market access for the poorest 80 83 86 89 92 95 98 01 04 07 19 19 19 19 19 19 19 20 20 20 countries. But problems remain from a GVC per- spective: high tariffs in many of the poorest develop- Top 1% income tax rate (left axis) ing countries hurt GVC participation by increasing Corporate tax rate (left axis) the transaction costs of acquiring inputs even when Median income tax rate (right axis) they are notionally tariff-exempt. Tariff escalation in Source: Egger, Nigai, and Strecker 2019. Cooperation on trade | 221 Table 9.1  Policy rationale, externalities, and cooperative solutions Policy area National motive International externality Cooperative solution Tariffs and other Improve terms of trade; Negative impact on trading Mutually agreed reduction in restrictions on trade protect special interests; partners and possible prisoner’s protection plus legal binding to and investment gain revenue dilemma reduce policy uncertainty Subsidies Support infant, senescent, Negative impact on trading Disciplines on use of specific types or strategic industries partners’ industries but positive of subsidies and other forms of or stages of production; impact on foreign consumers—at assistance such as tax incentives address market failures least in the short run (e.g., positive environmental externalities) Regulatory Protect consumers, the Industries in trading partners face Regulatory cooperation in the form of requirements environment, and intellectual higher costs for compliance, but harmonization, mutual recognition, property rights benefit from enhanced supply of or exporter regulatory commitments public goods Corporate taxes, Attract investment Negative impacts on other Tax cooperation (e.g., the existing investment incentives, investment locations and tax BEPS initiative at the OECD); FDI policies jurisdictions, potential tax destination-based taxes competition, and a race to the bottom Competition law, Promote contestable Abuse of market power; foreclosure Cooperation and common disciplines public ownership and markets; provide public of ability of firms to compete on a to control firm behavior control goods level playing field Investment in Reduce trade costs Positive externality for trading Investment coordination to exploit trade-facilitating partners; potential coordination synergies across countries and forms infrastructure failure and underinvestment of infrastructure Source: WDR 2020 team. Note: BEPS = base erosion and profit shifting; FDI = foreign direct investment; OECD = Organisation for Economic Co-operation and Development. important destination markets inhibits processing Tariffs and tariff preferences activities in agroindustry and other labor-intensive A new International Trade Centre (ITC)–World Bank areas such as apparel and leather goods. And restric- Database on Deep Integration Agreements reveals tive rules of origin curtail sourcing options. that unilateral, multilateral, and preferential liberal- •  For services, international negotiations have not ization has reduced trade-weighted average tariff rates delivered much liberalization beyond that under- to less than 5 percent for most industrial countries.2 taken unilaterally. Important GVC-relevant services, Preferential liberalization has reduced the applied such as air and maritime transport, for which lib- tariffs confronting many countries to a fraction of eralization needs to be coordinated, have typically the most-favored-nation (MFN) rate. Although pref- been excluded from negotiations. erential liberalization has targeted highly protected For investment in goods, there are no multilateral •  sectors, pockets of protection remain for agricultural rules, and the relevant policies are covered by a products, textiles, and footwear—areas of export inter- patchwork of preferential trade agreements (PTAs) est for developing countries (figure 9.3). and bilateral investment treaties (BITs). There is greater room for further liberalization A s for subsidies, trade rules have sought to allow •  in lower-income countries. Low-income and lower- space for legitimate use while preventing protec- middle-income countries still have average trade- tionist abuse, but recent frictions suggest that they weighted preferential tariff levels of over 5 percent have not succeeded. (figure 9.4 , panel a). When preferential tariffs are split by level of development of the importing and export- Two policy areas in which international coopera- ing countries, trade-weighted preferential tariffs tion can help developing countries engage in GVCs imposed by countries in the South on other countries are in reducing tariffs and restrictions in services both (in both the South and North) are more than double at home and abroad. those imposed by the North (figure 9.4, panel b). 222 | World Development Report 2020 Figure 9.3  Tariffs have been liberalized across sectors, but pockets of protection remain 20 Average trade-weighted tariff (%) 15 10 5 0 stu 24 64 odu ble ge ar s tio 9 fu , bb nd s str nd ou 7 gl and 85 uct d 25 trica ry uc al uc al d es ile al od an in 41 orta 6-8 68 ane 0-9 od er od m ec ne ad e ru s a du a ffs s ar n rs s s s w W s s l ts ts et 39 an id od 16- pr ta xt w pr Ani ct er ie as pr in el hi M in ls e pr d p 8 el 9 h ge Te he ot tic M on d oo d ica d ac he aw d Fo 3 Ve as im -05 3 St 7 an M -8 lie m -6 at R -2 l an 7 Pl Fo 72 5 al he 1 an 01 50 -6 r, s le 3 9 al isc -1 -7 0 an s, -4 -4 C oo -4 06 4- M 44 Tr 8 8 -3 28 d an sk ur al so ur s ce Agriculture Manufacturing re at N Average notional MFN rate Average applied tariff Source: Espitia et al. 2018. Note: MFN = most-favored-nation. The numbers on the x-axis are Harmonized System two-digit industrial codes. Figure 9.4  There is room for further liberalization a. Especially in lower-income countries… b. …in their trade with both developing and developed nations 12 12 10.7 Average trade-weighted preferential tariff (%) Average trade-weighted preferential tariff (%) 10 10 9.5 8.0 8 7.6 7.6 8 6 5.3 6 5.0 5.1 5.0 4.8 4.0 4 3.4 3.6 4 2.4 1.9 2 1.4 2 0 0 Low-income Lower-middle- Upper-middle- High-income South–South South–North North–South North–North income income MFN Preferential tariff Source: Espitia et al. 2018. Note: MFN = most-favored-nation. Cooperation on trade | 223 North–South tariffs are on average higher than against competing imported goods, but they also are North–North tariffs because many of the goods relatively cheap to produce because tariffs on inter­ developing countries export, such as agriculture and mediates are below the average tariffs on other goods.4 apparel, face tariff peaks. However, within product All countries and groups have some degree of tariff categories, low income countries do receive higher escalation. It is particularly pernicious in middle- preference margins, averaging 3 percentage points income countries, where processed goods face average above other countries.3 Some countries, such as tariffs of over 10 percent (figure 9.5). From a GVC per- Lesotho and Afghanistan, receive preference margins spective, tariff escalation tends to push countries into as much as 10 percentage points. In contrast, several backward participation. countries outside the global trade system, such as Examining industrial and agricultural goods sepa- Cuba and the Democratic People’s Republic of Korea, rately reveals distinct patterns (figure 9.6). High tariffs face tariffs on their goods about 5 percentage points on raw materials in low-income countries can prevent higher than other countries. The variation highlights them from joining the later stages of supply chains. how the trade system supports developing countries By contrast, middle- and high-income countries tend with market access through preferences, but also to have high tariffs on processed nonagricultural how it penalizes developing countries because their and agricultural goods, preventing other countries export products tend to face higher tariffs. It also from accessing their markets. These patterns hit low- shows the additional hurdles countries outside the income countries twice. First, they suffer a self- trade system face. inflicted wound from the relatively high domestic tariffs on raw materials and the semifinished goods Tariff escalation needed for production of most final goods. Second, if A goal of many developing countries is to move into they are able to produce final goods, their exports face higher value-added production. For example, coffee higher levels of protection abroad. bean producers would like to sell roasted coffee, and cocoa bean producers would like to export chocolate. Trade restrictions on services One difficulty, though, is that tariffs on processed As for services, trade agreements have not done much goods tend to be higher than tariffs on raw materials to deliver liberalization. The General Agreement on or semiprocessed goods in many of the largest mar- Trade in Services (GATS) emerged from the Uruguay kets. This tariff escalation is designed to protect the Round as a framework for negotiating liberalization, high value-added industries, while allowing produc- but there was limited liberalization of access to mar- ers access to imported inputs. Tariff escalation implies kets. In telecommunications services, however, the especially high rates of effective protection on final GATS did have a mutually reinforcing relationship goods because not only are these goods protected with a broader liberalization trend. For example, Figure 9.5  Most countries impose higher tariffs on semifinished and finished goods 12 10 MFN tariff, 2017 (%) 8 6 4 2 0 High-income Middle-income Low-income China European Japan United States OECD Union Raw materials Semifinished goods Finished goods Source: WDR 2020 team, based on World Bank’s WITS database. Note: MFN = most-favored-nation; OECD = Organisation for Economic Co-operation and Development. 224 | World Development Report 2020 Figure 9.6  Low-income countries are penalized by tariff escalation both at home and in their destination markets a. MFN tariffs on nonagricultural goods b. MFN tariffs on agricultural goods 14 30 12 25 MFN tariff, 2017 (%) MFN tariff, 2017 (%) 10 20 8 15 6 10 4 2 5 0 0 e e e na n n es EC e e e na n n s te O om m m O om m m io pa io pa at hi hi a D D co co co co Un Un Ja Ja St St c c C C C in in -in in in -in E d d an an h- e- h- e- w w ite ite dl dl ig ig pe pe Lo Lo Un Un id id H H ro ro M M Eu Eu Raw materials Semifinished goods Finished goods Source: WDR 2020 team, based on World Bank’s WITS database. Note: MFN = most-favored-nation; OECD = Organisation for Economic Co-operation and Development. several countries that were not ready to open markets counterfeit imports; and rules to establish the origin immediately nevertheless chose to commit them- of products needed in applying trade preference selves legally to opening up at specific points in the programs and PTAs. Both WTO and PTA disciplines future—an exercise that lent credibility to reform ensure that traders know what the rules are and programs. Unfortunately, the Doha negotiations in that enforcement procedures are predictable. Govern- services fell victim to the broader negotiating inertia, ments are increasingly cooperating to facilitate trade and the initial offers did not promise any meaningful by agreeing on good practices to reduce trade costs liberalization. without undermining regulatory goals such as prod- Typically excluded from services agreements are uct safety and tax collection. air and maritime transport services—two services Complying with standards is critical to participat- vital for connectivity and participation in GVCs. In ing in GVCs. Two WTO agreements—one on sanitary international transport, it takes two to liberalize. Zam- and phytosanitary measures and one on technical bia cannot unilaterally introduce greater competition barriers to trade—encourage the adoption of inter- on the Lusaka–London or Lusaka–Johannesburg air national standards where they exist and require that routes. Both the United Kingdom and South Africa national product standards have a scientific basis, do also need to agree to allow entry by third-country not restrict trade unnecessarily, and are applied on a airlines on each route. Both industrial and developing nondiscriminatory basis. countries use restrictive bilateral air service agree- International standards are being developed not by ments to fragment the international market into a the WTO but by specialist organizations. For example, series of route-specific duopolies. The WTO would international standards for phytosanitary measures, have been a natural platform to negotiate liberaliza- which are particularly significant for agriculture tion, but powerful members have ensured that air GVCs, are developed and adopted by contracting par- traffic rights are excluded from its scope. ties to the International Plant Protection Convention. These standards provide countries with harmonized Trade-related regulatory costs guidance on the implementation of regulations in the An important area of traditional trade cooperation trade of plants, plant products, and conveyances that relevant to GVC participation is the concerted action may carry pests and diseases of plants. to reduce the trade costs associated with trade- related regulation. Examples are customs clear- Gaps in rules ance procedures; enforcement of product health, The gaps in multilateral rules are in at least two import- safety, and environmental standards; control of ant GVC-relevant areas: investment and subsidies. Cooperation on trade | 225 Investment preestablishment or entry phase of investment, The WTO has uneven rules for policies affecting including national treatment, which requires the investment. Policies for foreign investment in goods host state to remove all discriminatory market access are not covered. The existing national treatment rule5 barriers and allow foreign investors to invest on the on the goods trade does not allow governments to give same terms as domestic investors. Investor protec- incentives or require firms, including those benefit- tions in PTAs generally grant national treatment to ing from foreign investments, to source inputs locally other members of PTAs and MFN treatment once the instead of importing them. But governments are free investment has been made (in the postestablishment to restrict or provide investment incentives for foreign phase) and cover direct and indirect forms of expro- direct investment (FDI). Policies affecting the estab- priation (figure 9.7). Finally, dispute settlement plays lishment of a commercial presence by foreign firms in a prominent role in the investment chapters of PTAs, services are covered in the GATS. WTO members may particularly investor–state dispute settlement provi- make commitments on access to markets through sions, which allow investors to bring disputes relating FDI, but this is not a general obligation—it is up to to the treaty’s substantive provisions. Almost all PTAs each WTO member to decide whether to do so, sector that cover this area provide for a mechanism for con- by sector. sultations and state-to-state dispute settlement, and International cooperation in the treatment of for- 77 percent provide for investor–state dispute settle- eign investment has mainly taken the form of bilateral ment provisions. investment treaties. These are not always instruments of liberalization in terms of market access; instead, Subsidies they provide foreign investors with protection against Subsidies, like taxes, are an important policy tool that governments taking action against them once they governments can use to pursue a number of legiti- have entered the country. The main goals are to ensure mate goals. Often, they are the best way to address that foreign investors are treated the same as domestic market failures that lead to the underprovision of investors and to put in place international arbitration certain goods. They can also be used to promote social mechanisms to determine the appropriate compen- objectives such as supporting access to basic services sation for a foreign investor if the host government in marginalized areas. But subsidies also can have takes actions to expropriate the investment. The distortive effects, including on trade. They may under- arbitration dimension of BITs has been contested in mine the benefits of trade and investment by distort- recent years, resulting in revisions of the regime by ing international prices or limiting market access, some jurisdictions. such as when they are granted with the condition Increasingly, PTAs are providing for both that local content be used. Such a condition can have investment liberalization and investment protec- negative welfare effects on other trading partners and tion.6 Liberalization may include access during the the global economy. Ensuring that subsidies pursue Figure 9.7  A majority of PTAs protect investors from discrimination and expropriation National treatment (postestablishment) MFN treatment (postestablishment) Covers direct expropriation Covers indirect expropriation Includes provision on fair and equitable treatment (FET) Covers currency transfers 0 10 20 30 40 50 60 70 80 90 100 Percent Source: Mattoo, Rocha, and Ruta, forthcoming. Note: National treatment requires imported products to be treated no less favorably than “like domestic products.” MFN = most-favored-nation; PTAs = preferential trade agreements. 226 | World Development Report 2020 desirable goals and are not captured by special groups What types of subsidies generate the greatest adverse to further their own interests is a challenge. Trade effects for other countries and for the trade system? rules have sought to allow space for legitimate goals Are subsidies achieving government objectives, or while preventing protectionist ones, but it is not clear are they likely to be captured by special interests? All they have succeeded. these questions require better information and fur- The impact of a subsidy is less clear in a GVC world ther analysis. in terms of the resulting distortion, as well as who As discussed in chapter 8, about half of all trade- benefits from a “subsidy” and who might be hurt.7 related policy measures imposed by governments The most obvious feature of a subsidy is that it can since 2009 take the form of subsidies or some type of be targeted to specific stages of production or types support for exports. These subsidies are only partially of economic activity—presumably associated with covered by WTO disciplines. immediate or future spillover benefits—rather than entire industries. That feature may imply that location WTO subsidy rules decisions are more responsive than others to financial WTO subsidy rules have significant gaps—they do incentives. In relational GVCs, subsidies can help not cover investment incentives or support received overcome a market failure in which investment in spe- by services activities, and only partially do they disci- cific goods is too low because of incomplete contracts. pline the behavior of SOEs. Most PTAs do little more than the WTO on subsidies, but the European Union The two sides of a subsidy is a major exception. For SOEs, however, several recent The first order of business in considering a subsidy deep PTAs, such as the Comprehensive and Progres- is to identify and define its spillovers. Subsidies used sive Agreement for Trans-Pacific Partnership (CPTPP) by a country to support local firms may have adverse and the United States–Mexico–Canada Agreement effects on the firms producing similar goods or ser- (USMCA), do go beyond the WTO. vices. Therefore, the potential for welfare-reducing Much of the focus of WTO members has been on subsidy competition between jurisdictions is signifi- agricultural subsidies, but their views have changed cant. U.S. states “spend” some $80 billion a year on tax in recent years. Many high-income countries have incentives and subsidies of investments, reflecting vig- long supported their agriculture sectors through a orous competition to attract investment.8 This compe- variety of policy instruments, including border bar- tition increases state-level welfare by attracting firms, riers and production subsidies. The WTO Agreement increasing employment, and raising wages, but it on Agriculture negotiated during the Uruguay Round generates beggar-thy-neighbor effects. Although large significantly reduced the ability of members to use potential gains can accrue at the state level from sub- agricultural subsidies and encouraged governments sidizing investment, such subsidies distort resource to decouple support from production. In 2015 WTO allocation by making inputs too cheap and generat- members agreed to ban agricultural export subsidies. ing excessive entry. The cost to the United States as Although other agricultural support continues to be a whole is significant—if states were to refrain from trade-distorting, it is much less so than in the 1980s subsidy competition, manufacturing real income in and 1990s because of the shift to decoupling support the United States would be 3.9 percent higher.9 from production and linking it to achievement of Although investment subsidies may have negative equity, environmental, and sustainability goals as welfare spillovers, they can also achieve outcomes opposed to increasing output. Since the early 2000s, sought by governments, such as generating local the Organisation for Economic Co-operation and employment. A U.K. program that offers investment Development (OECD) has seen a remarkable reduc- subsidies to firms in depressed areas on the condition tion in production support (figure 9.8), but there has they create or safeguard manufacturing jobs in these been an increase in support in large emerging econo- areas has positive effects on employment, investment, mies such as China. These trends illustrate the value and net entry. A 10 percent investment subsidy gen- and feasibility of cooperation to reduce the negative erates about a 7 percent increase in manufacturing spillovers created by subsidies. But further coopera- employment. The “cost per job” has been estimated at tion is needed to address the increase in farm support $6,300, suggesting that investment subsidies can be not decoupled from production in countries such as cost-effective.10 China and the United States.11 These examples illustrate the trade-offs associated Nondistorting forms of support are positively with subsidies and raise several questions from a trad- associated with agri-food GVC participation and the ing system perspective. How large are any spillovers? generation of domestic value added.12 Conversely, Cooperation on trade | 227 Figure 9.8  Agricultural producer support converged the Chinese economy has resulted in a substantial across some high-income and lower-income countries increase in the relative weight of SOEs in the global from 2000 to 2017 economy. In 2006, 4 percent of the world’s top 1,000 firms were Chinese, and by 2014, 14 percent were Chi- 35 nese, of which 70 percent were state-owned.18 SOEs % Producer Support Estimate 30 are also active in other emerging and developed coun- tries, often in cross-border mergers and acquisitions, 25 engaging in outward FDI. Concerns are frequently 20 expressed about the potential of SOEs to distort com- petition, reflecting views that SOEs are effectively 15 subsidized through soft loans, guarantees, and direct subsidies, among other things. They also may ben- 10 efit from indirect subsidies for factor inputs such as 5 energy and land, as well from protection from foreign competition (reflected, for example, in FDI restric- 0 tions, joint venture requirements, and preferential access to public procurement).19 Many SOEs operate in 00 01 02 03 04 05 06 07 08 09 10 11 12 13 14 15 16 17 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 20 GVC-intensive sectors, both upstream in energy and Brazil China EU-28 OECD downstream in transport. Source: WDR 2020 team, using data from the Organisation for Economic Co-operation and Develop- Disciplines on SOEs are included in recent PTAs ment’s (OECD’s) Producer and Consumer Support Estimates database. such as the CPTPP and USMCA, and the relevant pro- Note: EU-28 refers to the 28 member countries of the European Union. visions are enforceable through dispute settlement procedures. These disciplines require SOEs to make subsidies linked to output and market price support purchases and sales on the basis of commercial consid- measures lower the benefits of GVC participation. erations, and specify that subsidies granted to SOEs, Distortionary payments increase forward GVC partic- both direct fiscal transfers and indirect subsidies, are ipation in OECD member countries but decrease the actionable and that signatories may not discriminate domestic returns to participation in agri-food GVCs in favor of SOEs (that is, they must apply the national because the subsidy acts as a tax on other contributing treatment principle). The agreements also include sectors. Cooperation to limit subsidies and distortions provisions requiring signatories to list their SOEs and in agri-food sectors may thus enhance the domestic publish data on measures used to assist them. As just value added captured through participation in GVCs.13 noted, incentives to attract investment are not covered A separate WTO Agreement on Subsidies and by WTO rules. Countervailing Measures (ASCM) pertaining to sub- Current WTO rules on countervailing action are sidies on nonagricultural goods seeks to limit their directed at the domestic industry: if a sufficiently use while granting flexibility to developing countries. large share of the industry agrees it is being injured The ASCM has a twofold objective: (1) to prevent the by a foreign subsidy, action can be initiated. In a GVC use of subsidies to circumvent market access (tariff) setting, the high import content of total value added concessions and (2) to regulate countervailing duties embodied in a final good means subsidies will benefit (CVDs) used to offset the harmful effects on domes- foreign interests as well as local ones. The current con- tic producers of the foreign subsidization of goods.14 cept of injury may need to be reconsidered. Because Export subsidies are prohibited. All other subsidies any GVC spans firms in different countries, it may be can be used, but they could lead to the imposition more appropriate to focus on the effects of subsidies of CVDs in destination markets.15 De minimis provi- on GVCs as a whole. sions allow developing countries to use subsidies subject to certain thresholds.16 However, the WTO Strengthening subsidy rules rules are not concerned with why a government has Concerns and conflicts about the effects of subsidies implemented a subsidy, such as whether it can be jus- and the potential competition-distorting role of SOEs tified by a market failure.17 in the international economy call for revisiting the WTO disciplines on SOEs are limited, with only a WTO rules. Such efforts can take different forms, provision for state trading enterprises to require firms ranging from “soft law”—agreement on guidelines— granted exclusive or special privileges in trading to to enforceable treaty commitments. In 2018 the Euro- abide by the nondiscrimination rules. The growth of pean Union, Japan, and the United States launched a 228 | World Development Report 2020 trilateral process to identify ways to strengthen disci- Substantive disciplines plines on subsidies, suggesting expansion of the list A precondition for considering how and where to of prohibited subsidies in the WTO to include SOEs, revisit WTO rules is agreement on what types of sup- open-ended financial guarantees, subsidies to insol- port are a problem and where there should be a pre- vent or failing companies with no credible restruc- sumption that a measure is not trade-distorting or not turing plan, and preferential pricing for inputs. A large enough to matter. It is desirable to move toward necessary condition for meaningful outcomes is that an approach that devotes more attention to the aims developing countries, especially the larger emerging and effects of subsidies and prioritizes rule making economies, participate in such deliberations. for subsidies that are more likely to have adverse spill- overs on low-income countries, while enabling the use Transparency, transparency, transparency of subsidy instruments to address market failures. A first step—and a core part of any revision of sub- There may also be lessons from the European sidy rules—is transparency. Cooperation to ensure Union because it is the only international integration transparency and allow assessments of the effects of effort that ensures a level playing field for firms in subsidies can benefit both the subsidizing country the integrated market. Subsidies are covered by EU and the trade system. The WTO requires members to competition policy disciplines, and four criteria deter- regularly notify subsidy programs, but often compli- mine whether state aid is illegal: (1) state resources ance is neither timely nor comprehensive. In part this (a subsidy or tax expenditure) lead to (2) a selective may reflect capacity constraints; in part it may reflect advantage for a firm or activity that (3) distorts com- a decision to not notify subsidies. petition and (4) affects trade between member states. New rules could build on the EU experience. EU This also applies to undertakings to which member member states must comply with transparency obli- states have granted special or exclusive rights (such gations for state aid allocations of more than €500,000, as to SOEs). Subsidies falling under a General Block including the name of the beneficiary and the amount Exemption Regulation are deemed to raise few or no of aid granted. These data are accompanied by eval- concerns about distorting competition in the EU mar- uation of selected large state aid schemes to assess ket. These include regional aid (including for ports their impact and guide possible improvements in the and airports); aid for small and medium enterprises design of programs as well as the subsidy rules. Les- (SMEs); and aid for research and development and sons learned from the processes used by EU member innovation, broadband infrastructures, energy and states and the European Commission to report data on the environment, employment and training, natural subsidies could inform changes by the WTO. disasters, sports, and culture. Transparency could be bolstered through a collec- In 2017 EU member states spent €116.2 billion, or tive effort to compile information on subsidies (going 0.76 percent of the European Union’s GDP, on state aid. beyond reliance on notifications by countries) and to More than 90 percent of total state aid was allocated launch a process of dialogue and deliberation in the to horizontal objectives of common interest, such as WTO to define a negotiating agenda. This effort may environmental protection; regional development; and be more effective if undertaken plurilaterally, centered research, development, and innovation. Agreeing to on the major trading powers, but any initiative in this a set of subsidies deemed not to cause spillover con- area should be open to all countries and be informed cerns along the lines of the European Union could by economic analysis of the (spillover) effects of dif- help differentiate between subsidies that are not ferent types of subsidies. An important challenge in considered to have harmful trade spillover effects and defining possible rules and related cooperation is those that may have such consequences and should be to agree on what in principle constitutes desirable actionable. (globally welfare-enhancing) policies and what types The elements of progress are already embodied in of subsidies are more likely to generate undesirable WTO agreements, including the green box approach spillover effects, based on empirical analysis and evi- used in the Agreement on Agriculture, which exempts dence. In the WTO working group on investment set subsidies that cause minimal distortion to trade and up after the WTO’s Singapore ministerial meeting in includes social and environmental programs. The 1996, it became clear early on that many governments agreement also gives developing countries additional were not willing to discuss and consider disciplines to flexibility in providing domestic support. The green address the spillover effects of investment incentives box approach was also incorporated on a provisional and subsidies, removing much of the potential ratio- basis in the WTO agreement on subsidies and coun- nale for a multilateral agreement. tervailing measures that expired in 1999. Revisiting it Cooperation on trade | 229 is one possible factor in balancing stronger disciplines such as OECD and the Asia-Pacific Economic Coop- on subsidies with recognition that many types of eration forum. subsidies fulfill an important function in addressing Deep integration agreements can fill some of the market failures. Moreover, the various de minimis gaps in the WTO on investment-related policies, SOEs, provisions included in these WTO agreements for and services. They do so, however, on a preferential developing countries are a way of recognizing that basis: the benefits of market access are limited to the spillover effects created by subsidies used by low- partners. They may also offer a way of bundling dis- income countries are likely to be small from a sys- ciplines on a range of GVC-relevant issues. Evidence temic perspective. suggests that these bundles affect the joint evolution All this suggests that any new subsidy rules should of GVCs and FDI. consider, in a way that current WTO rules do not, the motivation for a policy that may give rise to negative Deep trade agreements boost GVC spillovers. Such rules should cover all subsidy-like participation policies to encompass services and investment incen- There is a strong correlation between GVC-related tives, as well as the agricultural domestic support trade and the depth of PTAs (figure 9.9). Adding a policies that have long been an interest of the WTO provision to a PTA boosts the domestic value added membership—and that matter most for many devel- of intermediate goods and services exports (forward oping countries. GVC linkages) by 0.48 percent, while an additional provision in a PTA increases the foreign value added Deep integration agreements and GVCs of intermediate goods and services exports (backward Trade cooperation can be characterized as either “shal- GVC linkages) by 0.38 percent.22 Although deep PTAs low” or “deep.”20 Shallow cooperation is limited to com- boost trade between their members,23 this effect is mitments to enhance the transparency and visibility stronger for GVC trade, which is consistent with the of extant trade policies and reduce or eliminate trade view that policy spillovers are more relevant to GVCs barriers such as tariffs and quotas. It gives countries than to standard trade. Indeed, deep trade agreements discretion in setting nontariff measures that could improve forward linkages, especially for more com- affect trade. Its basic requirement is “national treat- plex GVCs, which export intermediates across borders ment,” which requires imported products to be treated two or more times.24 Conversely, the unraveling of no less favorably than “like domestic products.” deep trade agreements can have an adverse effect on Deep agreements go beyond national treatment by GVCs (box 9.3). including commitments on the substance of nontariff measures. Examples include agreements to protect Figure 9.9  Deep trade agreements are certain types of intellectual property, to adopt com- associated with GVC integration mon approaches to regulating the services sectors, 13 or to implement a competition law that embodies criteria that mirror those of trading partners. A feature 12 Average GVC-related trade (log) of deep trade agreements is that many provisions 11 are enforceable: they specify precise legally binding obligations, and trading partners can raise objections 10 and take action if a signatory does not live up to its 9 commitments.21 In some situations, cooperation may not require 8 binding disciplines. If the problem is a coordination 7 failure, all that may be required is information and agreement to apply a given norm at the national 6 level. An example is an agreement on technical 5 standards to allow equipment, vessels, or network No PTA Low High infrastructure to connect. In many circumstances, depth of depth of agreement agreement soft law cooperation will center on international Source: Laget et al. 2018. monitoring and mechanisms that elicit dialogue and Note: The estimator is the Poisson Pseudo Maximum Likelihood (PPML). analysis to allow learning and identification of good GVC-related trade is defined as trade in parts and components. PTA = practices. This is an important role of institutions preferential trade agreement. 230 | World Development Report 2020 Box 9.3  The impact of Brexit on GVC trade How will Brexit affect trade between the United Kingdom added in U.K. exports (backward linkages) increased by and the European Union (EU)? One difficulty in addressing 37 percent. this question lies in the lack of systematic information on In a second step, the study examines the impact that the content of trade agreements, which makes it difficult Brexit can have on U.K.–EU trade relations going forward. to assess precisely the impact that a set of common rules Three distinct scenarios are considered, varying by the has on trade flows.a A recent study uses the information decreasing depth of the post-Brexit agreement between available on the content of trade agreements to assess the the United Kingdom and the rest of the European Union. impact of Brexit on goods, services, and value-added trade.b The first scenario is a “soft” Brexit, assuming that the Specifically, it augments a standard gravity model of trade post-Brexit arrangement between the United Kingdom and to quantify the effect that the “depth” of the EU agreements the European Union will be as deep as the agreement the has on U.K. trade and then use the estimates from this European Union has with Norway. In the second scenario, analysis to evaluate the future of U.K.–EU trade relations the United Kingdom and the European Union will sign an under different post-Brexit scenarios. In a first step, the agreement as deep as the average agreement the Euro- study examines the extent to which EU membership con- pean Union currently has with third countries. The third tributed to boosting U.K. trade, notably GVC trade.c It finds “hard” scenario has no agreement. that EU membership increased goods, services, and value- Under all scenarios, bilateral U.K.–EU trade declines, added trade for member countries and that this impact and the drop is sharper the lower the depth of the has been even stronger for the United Kingdom (figure post-Brexit arrangement relative to the depth of the EU B9.3.1). Following its membership in the European Union, agreement. In terms of value-added trade, the decline the United Kingdom’s services trade more than doubled; ranges from 6 percent for the “softer” scenario to 28 per- its intermediates value added in gross exports (forward cent for the “harder” Brexit scenario. The largest declines linkages) increased by 31 percent; and the foreign value are for U.K. services trade. Figure B9.3.1  Trade impacts of membership in the European Union on the United Kingdom and other EU members 1.2 Marginal trade impact 0.8 0.4 0 Goods Services Domestic value GVC forward GVC backward added linkages linkages Other EU members United Kingdom Source: Mulabdic, Osnago, and Ruta 2017. Note: The figure reports the coefficients and confidence intervals of an augmented gravity equation, capturing the marginal impact on trade of a deep trade agreement and its statistical significance. In each category, the blue bar represents the coefficient for all countries in the sample, except the United Kingdom (for which the red bar is the coefficient). For example, a coefficient of 0.5 for goods trade indicates that country-pairs that signed the deepest trade agreement increased their total bilateral trade in goods by 69 percent (exp 0.5–1.0). The United Kingdom was not affected more than the average in terms of goods trade. The estimator is the Poisson Pseudo Maximum Likelihood (PPML). All specifications include bilateral fixed effects and country-time fixed effects. Ninety percent confidence intervals are constructed using robust standard errors, clustered by country-pair. (Box continues next page) Cooperation on trade | 231 Box 9.3  The impact of Brexit on GVC trade (continued) Table B9.3.1  Changes in the United Kingdom’s bilateral trade with the European Union under three Brexit scenarios Percent “Norway” “No agreement” Type of trade (or “soft”) scenario “Average PTA” scenario (or “hard”) scenario Goods –12 –38 –50 Services –16 –48 –62 Domestic value added –6 –20 –28 GVC forward linkages –5 –18 –26 GVC backward linkages –7 –25 –34 Source: Mulabdic, Osnago, and Ruta 2017. Note: The depth of the post-Brexit arrangement falls from a score of 44 to 36 (the number of legally enforceable policy areas covered by the agreement) in the “Norway” scenario, to 14 in the “average PTA” scenario, and to 0 in the “no agreement” scenario. PTA = preferential trade agreement. Table B9.3.1 summarizes results of the study. These pre- entry and exit from an agreement are symmetric processes dictions are average effects because it takes time for trade because the predictions are based on the impact that EU flows to respond to changes in trade costs, and so the impact membership had on U.K.–EU trade. Because firms have in the short run is expected to be smaller than in the longer already paid sunk costs to enter the market, the trade effect term. Moreover, they are made under the assumption that of a breakup can be less than what a gravity model predicts. a. A way around this problem is to make assumptions about how different scenarios will lower trade costs (Dhingra et al. 2016) or to identify trade agreements that have diverse content (Baier et al. 2008). b. Mulabdic, Osnago, and Ruta (2017). c. The analysis uses data from the World Bank’s WIOD database on goods, services, and value-added trade and World Bank data on the content of deep agreements (Hofmann, Osnago, and Ruta 2019) to estimate a gravity equation augmented with a measure of depth for the period 1995–2011. The effect of common trade rules on U.K. imports and exports of goods, services, and value added is quantified by interacting the depth of trade agreements with dummies identifying the United Kingdom. Deep PTAs have an indirect effect on third Consistent with a model of contractual frictions and countries’ trade along the value chain global sourcing,26 the depth of PTAs is correlated with In a world in which production is fragmented across vertical FDI.27 This relationship is driven by areas in countries, the depth of PTAs affects their members trade agreements (such as regulatory cooperation) as well as GVC trade with nonmembers. Intuitively, that improve the process of contracting with suppli- deeper trade agreements in third countries lower trade ers for inputs provided by suppliers. costs along the entire value chain, thereby encourag- ing trade in intermediates among countries that are Addressing traditional trade barriers still not part of the agreement. The estimated impact from matters for South–South GVC trade augmented gravity regressions are larger than those of The impact of deep agreements on GVC trade may be a standard gravity model, suggesting that signing deep heterogeneous across countries at different levels of PTAs has indirect effects through third-country trade.25 development. South–South GVCs tend to be impeded by traditional barriers, such as high tariffs and long Deep PTAs affect the structure of delays at the border, more than GVC trade between international production North and South economies. Evidence suggests that Deep trade agreements also affect FDI and, more gen- PTAs boost South–South GVC integration by going erally, the way in which goods are traded internation- further in policy areas under the current WTO man- ally (within firms or at arm’s length). The underlying date (such as tariffs, customs, and services), whereas idea is that deep PTAs affect firms’ make-or-buy deci- North–South GVCs are primarily affected by commit- sions—that is, whether producers outsource to trad- ments in areas such as investment, competition, and ing partners’ suppliers or vertically integrate produc- intellectual property rights protection not covered by tion processes with affiliates in foreign economies. the WTO.28 232 | World Development Report 2020 Table 9.2  Existing trade agreements in Africa are relatively shallow Common Economic Southern West African African Market for Community Southern African Economic and Continental Eastern and East African of West African Development Monetary Free Trade Southern Africa Community African States Customs Community Union Policy area Area (AfCFTA)a (COMESA) (EAC) (ECOWAS) Union (SACU) (SADC) (WAEMU) Tariffs on        manufacturing goods Tariffs on agricultural        goods Export taxes      Customs       Competition policy     State aid ?   Antidumping      Countervailing    measures Agreement on Trade-   Related Aspects of Intellectual Property Rights (TRIPS) State trading  enterprise (STE) Technical barrier to     trade (TBT) General Agreement    on Trade in Services (GATS) Sanitary and     phytosanitary (SPS) Movement of capital      Public procurement ? Intellectual property  rights (IPRs) Investment   Environmental laws ? Labor market ? regulation Agreement on Trade- ? Related Investment Measures (TRIMs) Source: WDR 2020 team, based on Hofmann, Osnago, and Ruta 2019. a. The depth of AfCFTA is based on the text of the Agreement Establishing the African Continental Free Trade Area (“Kigali Draft Text,” March 2018). Several AfCFTA details are still being negotiated. It is unknown if any commitments will be included in the areas of state aid, public procurement, environment, labor market regulation, and TRIMs. These findings provide useful guidance for South– countries affects backward and forward participation South integration initiatives, such as the African in agriculture and food GVCs,29 the immediate chal- Continental Free Trade Area (AfCFTA). It is far more lenge for AfCFTA negotiations for GVC integration is ambitious than the existing PTAs in Africa (table 9.2). to address the distortions created by traditional barri- Because bilateral trade protection among African ers to trade within Africa (box 9.4). Cooperation on trade | 233 Box 9.4  How the African Continental Free Trade Area (AfCFTA) can support integration into GVCs AfCFTA will likely boost trade and deepen regional integra- (NTBs) to trade in goods and services and implementing tion in Africa, with positive effects on growth and poverty trade facilitation measures. World Bank staff estimates reduction. The agreement, now ratified by 27 countries, has indicate that reduction of tariffs alone is expected to become legally binding and entered into force in May 2019. increase the welfare of AfCFTA members by 0.2 percent. The first phase of negotiations will consider trade in goods Reducing NTBs in goods and services by half would and services, as well as procedures for dispute settlement. increase welfare gains by 1.6 percent. Full implementa- This phase will include negotiations on rules of origin, tion of the World Trade Organization's Trade Facilitation which are likely to have an important role in enabling the Agreement (TFA) would bring the overall welfare gains development of regional value chains. The second phase to 5 percent by 2035 (compared with the baseline). How- will cover the rules defining investment, competition, and ever, the aggregate results mask heterogeneity of impacts intellectual property rights. across countries. Even though AfCFTA is expected to ben- There is widespread optimism throughout the continent efit all members, welfare gains by 2035 will range from 0.4 that increased trade integration will strengthen the emerg- percent to 19 percent (figure B9.4.1). Thus the impact of ing regional value chains and enable firms throughout the agreement will depend on its depth and the extent to Africa to participate in GVCs. Creating an integrated and which it covers NTBs and services, especially in backbone much larger market can attract market-seeking foreign sectors such as transport and logistics, and on the respec- direct investment, especially if some of the deeper inte- tive export basket and economic structure of each country. gration ambitions are also realized. Similarly, a well-staffed AfCFTA will also provide an opportunity to build on AfCFTA secretariat with clear monitoring and enforcement efforts by the many regional economic communities to capacities can help ensure that commitments are fully develop integrated regional value chains (RVCs) to support implemented, leading to greater policy predictability. Some growth and industrial development. In the recent past, institutions such as the African Export-Import Bank are these efforts have suffered from the fragmented and piece- seeking to develop facilities to help governments address meal engagement of the private sector and the capacity, adjustment costs. However, it is unclear whether such political economy, and coordination challenges that lead efforts will be sufficient. As for most free trade agreements to the “implementation gap” in regional commitments.a (FTAs), governments will have to look for ways to support Ongoing initiatives by regional communities, national gov- those workers who may lose from the adjustment-related ernments, and donors are seeking to identify and address aspects of greater trade openness. policy and regulatory constraints to cross-border trade, There is, however, reason for caution at this stage. such as in the soya RVC in southern Africa, the dairy RVC Despite a long history of hope for greater integration in East Africa, and the leather RVC in the Common Market in Africa, the efforts to date have fallen short. Here, the for Eastern and Southern Africa (COMESA) region. The development of integrated trade and production networks benefit of structuring interventions and support around in Asia provides some lessons. Implementing trade facilita- individual RVCs is that it allows participants to focus on tion commitments and improving border management can required policy reforms and needed investments to address reduce trade costs within Africa and also reduce distances market failures and to create mutually beneficial outcomes. to global hubs. The impact of AfCFTA depends, then, on This in turn can create demonstration effects and reduce much more than tariff reduction; some of the largest gains cross-cutting barriers across sectors that can be scaled up would come from effectively tackling nontariff barriers across RVCs spanning subgroups of countries. (Box continues next page) 234 | World Development Report 2020 Box 9.4  How the African Continental Free Trade Area (AfCFTA) can support integration into GVCs (continued) Figure B9.4.1  AfCFTA members benefit from reductions in tariffs, nontariff measures, and implementation of the World Trade Organization’s Trade Facilitation Agreement 20 (%, 2035 compared with baseline) 15 Welfare gains and losses 10 5 0 –5 en A o Zi Af e am e ia , D en in . R al st e . C SA a nz n Eg ast or nia t, A co rk un . a a Et aso au ia Bo ha s tsw na So ig na h ia an a M Rw da ag da M Ma car Af mb i TA e st ô A a b a e U ab ca Za rica N d'I ca w ep Bu T Rep G tiu l th ir N abw u Ta roo C ern Tog of ny in isi Ug mbi Re C tral fric m ric ta am C go S Ben em eg ib M iop ut er oz la CF iq yp ern oc Ar fri of te fri or vo a ad an E M a as ri to F Af K h N a of st Re st e Re of W on of C st st Re Re Tariffs, NTMs, and TFA Tariffs and NTMs Tariffs Source: Maliszewska, Osorio-Rodarte, and van der Mensbrugge, forthcoming. Note: The figure shows the percentage change in welfare in 2035 compared with the baseline. Tariffs refers to full elimination of tariffs in trade within the African Continental Free Trade Area (AfCFTA); NTMs refers to halving the nontariff measures (NTMs) in goods and services; and TFA refers to full implementa- tion of the World Trade Organization’s Trade Facilitation Agreement. Rest of Central Africa includes Angola, Central African Republic, Chad, Republic of Congo, Equatorial Guinea, Gabon, and São Tomé and Príncipe; rest of Eastern Africa includes Burundi, Comoros, Djibouti, Eritrea, Mayotte, Seychelles, Somalia, South Sudan, and Sudan; rest of North Africa includes Algeria, Libya, and Western Sahara; rest of SACU (South African Customs Union) includes Eswatini and Lesotho; rest of Western Africa includes Cabo Verde, The Gambia, Guinea-Bissau, Liberia, Mali, Mauritania, Niger, Saint Helena, and Sierra Leone. a. Vanheukelom and Bertelsmann-Scott (2016); World Bank (2017, 2019). Notes 1. Tang and Wei (2009). 8. Ossa (2015). 2. The website for this database is still under construction. 9. Ossa (2015). 3. Competition adjusted preference margins are calculated 10. Criscuolo et al. (2012). as the difference between the weighted average tariff 11. OECD (2018). rate applied to the rest of the world and that applied to 12. Greenville, Kawasaki, and Beaujeu (2017). the beneficiary country, holding weights constant based 13. Greenville, Kawasaki, and Beaujeu (2017). on preference-granting country imports. 14. Adverse effects include injury to a domestic industry, 4. Corden (1971). nullification or impairment of tariff concessions, or 5. National treatment is specified by the WTO (in Article serious prejudice to the country’s interests. Serious prej- III of the General Agreement on Tariffs and Trade), as udice arises if subsidies are used to cover the operating well as in most preferential trade agreements. It requires losses of a firm or industry or if debt relief is granted for imported products to be treated no less favorably than government-held liabilities. Serious prejudice may arise “like domestic products.” if the subsidy reduces exports of other WTO members, 6. Crawford and Kotschwar (2018). results in significant price undercutting, or increases 7. Hoekman (2016). the world market share of the subsidizing country in a Cooperation on trade | 235 primary product. WTO disciplines focus on the amount References of assistance given and not on the extent to which a sub- sidy harms trading partners. Subsidies below 5 percent Antràs, Pol, and Elhanan Helpman. 2008. “Contractual ad valorem are not actionable. See Hoekman and Frictions and Global Sourcing.” In The Organization of Kostecki (2009). Firms in a Global Economy, edited by Elhanan Helpman, 15. In response to subsidies, other countries can impose Dalia Marin, and Thierry Verdier, 9–54. Cambridge, MA: CVDs on subsidized imports that injure a domestic Harvard University Press. industry (with duties up to the amount of the subsidy Baier, Scott L., Jeffrey H. Bergstrand, Peter Egger, and Pat- paid) or request a government to withdraw a prohibited rick A. McLaughlin. 2008. “Do Economic Integration subsidy or withdraw or modify an actionable subsidy. In Agreements Actually Work? Issues in Understanding the case of noncompliance, the injured WTO member can Causes and Consequences of the Growth of Regional- take countermeasures against the subsidizing state up to ism.” World Economy 31 (4): 461–97. the amount of the subsidy paid (in the case of a prohib- Balié, Jean, Davide Del Prete, Emiliano Magrini, Pierluigi ited subsidy) or up to the amount of the injury suffered Montalbano, and Silvia Nenci. 2017. “Agriculture and by the domestic industry (in the case of an actionable Food Global Value Chains in Sub-Saharan Africa: Does subsidy). Bilateral Trade Policy Impact on Backward and Forward Participation?” DISSE Working Paper 4/17, Dipartimento 16. If the subsidy is less than 2 percent of the per unit value di scienze sociali ed economiche, University of Rome– of products exported, developing countries are exempt Sapienza, Rome. from CVDs (for less developed countries the threshold is Corden, W. M. 1971. The Theory of Protection. Oxford, U.K.: 3 percent). De minimis also applies if the import market Oxford University Press. share of a developing country is below 4 percent, and Crawford, Jo-Ann, and Barbara Kotschwar. 2018. “Invest- the aggregate share of all developing countries is below ment Provisions in Preferential Trade Agreements: Evo- 9 percent of total imports. The ASCM also exempts lution and Current Trends.” WTO Staff Working Paper nations with per capita incomes below $1,000 from the ERSD-2018–14 (December), Economic Research and WTO prohibition on the use of export subsidies and Statistics Division, World Trade Organization (WTO), precludes CVDs on associated exports if global market Geneva. shares are less than 3.5 percent for a product. De minimis Criscuolo, Chiara, Ralf Martin, Henry Overman, and John provisions are also included in the WTO Agreement on Van Reenen. 2012. “The Causal Effects of an Industrial Agriculture, permitting support up to 10 percent of out- Policy.” NBER Working Paper 17842, National Bureau of put in developing countries. Economic Research, Cambridge, MA. 17. In the Uruguay Round, a third category, nonactionable Dhingra, Swati, Rebecca Ann Freeman, and Eleonora Mav- subsidies, was included in the ASCM spanning envi- roeidi. 2018. “Beyond Tariff Reductions: What Extra ronmental, R&D, and regional subsidies. This provision Boost from Trade Agreement Provisions?” CEPR Discus- was time-bound and lapsed at the end of 1999 because a sion Paper 12795 (March), Centre for Economic Policy consensus could not be reached to extend it. Research, London. 18. Freund and Sidhu (2017). Dhingra, Swati, Gianmarco Ottaviano, Thomas Sampson, 19. See, for example, USTR (2018) for arguments to this and John Van Reenen. 2016. “The Consequences of Brexit effect. Empirical evidence suggests that SOEs are less for UK Trade and Living Standards.” CEP Brexit Analysis profitable and less productive than private firms in their 2, Centre for Economic Performance, London School of respective sectors. See European Commission (2016) for Economics, London. the European Union; Harrison et al. (2019) for China; and Ederington, Josh, and Michele Ruta. 2016. “Non-Tariff Mea- Kowalski et al. (2013) and OECD (2016) for a broad sample sures and the World Trading System.” Policy Research of countries. Working Paper 7661, World Bank, Washington, DC. 20. Lawrence, Bressand, and Ito (1996). Egger, Peter H., Sergey Nigai, and Nora Margot Strecker. 2019. 21. Ederington and Ruta (2016). “The Taxing Deed of Globalization.” American Economic 22. Laget et al. (2018). For other research on the relationship Review 109 (2): 353–90. between deep PTAs and GVCs see Dhingra, Freeman, Espitia, Alvaro, Aaditya Mattoo, Mondher Mimouni, Xavier and Mavroeidi (2018); Johnson and Noguera (2017); Pichot, and Nadia Rocha. 2018. “How Preferential Is Pref- Orefice and Rocha (2014). erential Trade?” Policy Research Working Paper 8446, 23. Mattoo, Mulabdic, and Ruta (2017). World Bank, Washington, DC. 24. Laget et al. (2018). European Commission. 2016. “State-Owned Enterprises in 25. Laget et al. (2018) build on the approach by Noguera the EU: Lessons Learnt and Ways Forward in a Post-Crisis (2012) to investigate this mechanism through a gravity Context.” European Economy Institutional Paper 31 (July), model augmented to account for third-country effects. European Union, Luxembourg. 26. Antràs and Helpman (2008). Freund, Caroline L., and Dario Sidhu. 2017. “Global Compe- 27. Osnago, Rocha, and Ruta (2017, 2019). tition and the Rise of China.” PIIE Working Paper 17–3 28. Laget et al. (2018). (February), Peterson Institute for International Econom- 29. Balié et al. (2017). ics, Washington, DC. 236 | World Development Report 2020 Greenville, Jared, Kentaro Kawasaki, and Raphaël Beaujeu. Mulabdic, Alen, Alberto Osnago, and Michele Ruta. 2017. 2017. “How Policies Shape Global Food and Agriculture “Trading Off a ‘Soft’ and ‘Hard’ Brexit.” VOX CEPR Value Chains.” OECD Food, Agriculture, and Fisheries Policy Portal, January 23, Centre for Economic Policy Paper 100, Organisation for Economic Co-operation and Research, London. https://voxeu.org/article/trading-soft Development, Paris. -and-hard-brexit. Harrison, Ann E., Marshall Meyer, Peichun Wang, Linda Noguera, Guillermo. 2012. “Trade Costs and Gravity for Zhao, and Minyuan Zhao. 2019. “Can a Tiger Change Its Gross and Value Added Trade.” Unpublished job market Stripes? Reform of Chinese State-Owned Enterprises in paper, Columbia University, New York. the Penumbra of the State.” NBER Working Paper 25475 OECD (Organisation for Economic Co-operation and Devel- (January), National Bureau of Economic Research, Cam- opment). 2016. State-Owned Enterprises as Global Competitors: bridge, MA. A Challenge or an Opportunity? Paris: OECD. Hoekman, Bernard M. 2016. “Subsidies, Spillovers, and WTO ——— —. 2018. Agricultural Policy Monitoring and Evaluation 2018. Rules in a Value-Chain World.” Global Policy 7 (3): 351–59. Paris: OECD. Hoekman, Bernard M., and Michel M. Kostecki. 2009. Orefice, Gianluca, and Nadia Rocha. 2014. “Deep Integration The Political Economy of the World Trading System: The and Production Networks: An Empirical Analysis.” World WTO and Beyond, 3rd ed. Oxford, U.K.: Oxford University Economy 37 (1): 106–36. Press. Ornelas, Emanuel. 2016. “Special and Differential Treatment Hoekman, Bernard M., and Aaditya Mattoo. 2019. “Altered for Developing Countries.” In Handbook of Commercial States: Rethinking International Trade Cooperation.” Policy, vol. 1A, edited by Kyle Bagwell and Robert W. Unpublished working paper, World Bank, Washington, Staiger, 369–432. Handbooks in Economics. Amsterdam: DC. Elsevier. Hofmann, Claudia, Alberto Osnago, and Michele Ruta. 2019. Osnago, Alberto, Nadia Rocha, and Michele Ruta. 2017. “Do “The Content of Preferential Trade Agreements.” World Deep Trade Agreements Boost Vertical FDI?” World Bank Trade Review 18 (3): 365–98. Economic Review 30 (1): 119–25. Johnson, Robert Christopher, and Guillermo Noguera. 2017. ——— —. 2019. “Deep Trade Agreements and Vertical FDI: The “A Portrait of Trade in Value-Added over Four Decades.” Devil Is in the Details.” Canadian Journal of Economics 52 Review of Economics and Statistics 99 (5): 896–911. (November). Kowalski, Przemyslaw, Max Büge, Monika Sztajerowska, Ossa, Ralph. 2015. “A Quantitative Analysis of Subsidy and Matias Egeland. 2013. “State-Owned Enterprises: Competition in the U.S.” NBER Working Paper No. 20975, Trade Effects and Policy Implications.” OECD Trade Pol- National Bureau of Economic Research, Cambridge, MA. icy Paper 147, Organisation for Economic Co-operation Tang, Man-Keung, and Shang-Jin Wei. 2009. “The Value and Development, Paris. of Making Commitments Externally: Evidence from Laget, Edith, Alberto Osnago, Nadia Rocha, and Michele Ruta. WTO Accessions.” Journal of International Economics 78 (2): 2018. “Deep Trade Agreements and Global Value Chains.” 216–29. Policy Research Working Paper 8491, World Bank, Wash- USTR (U.S. Office of the Trade Representative). 2018. Findings ington, DC. of the Investigation into China’s Acts, Policies, and Practices Lawrence, Robert Z., Albert Bressand, and Takatoshi Ito. Related to Technology Transfer, Intellectual Property, and Inno- 1996. A Vision for the World Economy: Openness, Diversity, vation under Section 301 of the Trade Act of 1974. Washington, and Cohesion. Integrating National Economies Series. DC: USTR. Washington, DC: Brookings Institution Press. Vanheukelom, Jan, and Talitha Bertelsmann-Scott. 2016. “The Maliszewska, Maryla, Israel Osorio-Rodarte, and Dominique Political Economy of Regional Integration in Africa: The van der Mensbrugge. Forthcoming. “Making the Most Southern African Development Community (SADC).” of the Africa Continental Free Trade Area: Assessment of European Centre for Development Policy Management, the Impacts on Trade, Growth and Poverty Reduction.” Maastricht, the Netherlands. World Bank, Washington, DC. World Bank. 2017. “Supporting Africa’s Transformation: Mattoo, Aaditya, Alen Mulabdic, and Michele Ruta. 2017. Regional Integration and Cooperation Assistance “Trade Creation and Trade Diversion in Deep Agree- Strategy for the Period FY18–FY23.” Report 121912-AFR ments.” Policy Research Working Paper 8206, World (December 15), World Bank, Washington, DC. Bank, Washington, DC. ——— —. 2019. Africa’s Pulse: Analysis of Issues Shaping Africa’s Mattoo, Aaditya, Nadia Rocha, and Michele Ruta. Forthcom- Economic Future. Vol. 19. Washington, DC: World Bank. ing. Handbook of Deep Integration Agreements. Washington, DC: World Bank. Cooperation on trade | 237 Cooperation beyond trade Key findings  ustaining openness to trade and global value chains (GVCs) requires cooperation • S beyond trade policy on taxes, regulation, competition policy, and infrastructure.  VCs exacerbate the problems of tax avoidance and tax competition between potential • G host countries. International cooperation is necessary to enable countries to raise tax revenues and to ensure that conditions of competition are not distorted. Ultimately, a joint approach to greater use of destination-based corporate taxation could eliminate the incentive to shift profits and compete over taxes. Meanwhile, other measures against tax base erosion and income shifting could enhance domestic resource mobilization. • D  omestic regulation is insufficient to address international market failures, such as privacy concerns related to cross-border data transfers. Cooperation by data-destination countries to protect foreign consumer data could reassure data-source countries that their commitments to openness will not put their citizens’ data at risk.  nticompetitive behavior by GVC firms can affect the distribution of gains from GVC • A participation. Enhanced international cooperation around competition law enforcement would enable countries to overcome jurisdictional and capacity constraints to combat anticompetitive practices.  oordination between countries on investment in transport and communication • C infrastructure can improve international connectivity. Gains are larger when governments collaborate to expedite trade simultaneously. 238 | World Development Report 2020 T o sustain trade openness, it is essential to “walk from abuse of privacy in data-based services to anti- on two legs.” The previous chapter looked at competitive practices in platform-based services. the first leg—deepening traditional liberal- International market failures could be addressed ization and removing distortions. This chapter looks cooperatively in several areas that matter for GVCs. at the second leg—widening cooperation beyond For example, for cross-border, data-based services, trade policy to include taxes, regulation, competition addressing market failures efficiently is not possi- policy, and infrastructure. Enhanced cooperation ble without the cooperation of the regulator in the among countries on taxes is needed to reduce both the data-destination country. Governments may fear incentives for governments to engage in inefficient opening markets if the gains from liberalization are tax competition and the opportunities for firms to likely to be eroded by anticompetitive practices in shift profits to low-tax jurisdictions. Such steps will both goods and services—practices for which there is help governments mobilize the resources necessary growing evidence. Cooperative solutions that support to pay for labor adjustment programs and build the innovation and efficiency while protecting consum- infrastructure needed for economic growth. Cooper- ers will be needed to maintain an open trade system ation among countries on regulation and competition in these goods and services. But developing countries policy can reassure consumers that greater openness must not be left out of such cooperation; multilateral need not imply vulnerability to fraud or anticompet- trade rules require that they be given an opportunity itive practices. Finally, cooperation and assistance on to join any such agreements. infrastructure can help poorer countries remedy the Finally, more cooperation is needed on infrastruc- energy and connectivity gaps that have limited their ture gaps. Coordination failures in infrastructural participation in trade and global value chains (GVCs). investment affect GVC investment, expansion, and Tax competition and profit shifting may be affect- upgrading. Multinational agreements can help ing both the ability of countries to join GVCs and address this problem. Consider the Trade Facilitation their benefits. Multinationals encourage competition Agreement (TFA) of the World Trade Organization between potential hosts, which results in countries (WTO), which encourages countries to coordinate using fiscal incentives to win them. GVCs have thus improvements in trade facilitation. Each country does made it hard for countries to tax profits, especially not fully internalize the benefits to foreign traders of those of firms reliant on patents for their profits, reductions in domestic trade costs, and gains are larger which can easily be shifted to low-tax jurisdictions. when governments on both sides of the border invest As a result, a greater share of the burden for resource in expediting trade simultaneously. The WTO agree- mobilization has fallen on workers. International ment addresses this coordination problem and pro- cooperation may be needed to enable states to raise vides low-income countries with financial assistance tax revenues in a GVC world and to ensure that con- for the necessary investments. A similar approach ditions of competition are not distorted. The Organi- could exploit synergies in other infrastructure invest- sation for Economic Co-operation and Development ments in transport, energy, and communications. (OECD) has already taken steps to address tax base erosion and profit shifting (BEPS) by multinationals, including changes in the transfer pricing of interme- Taxes diate inputs, especially for intangibles such as services Although GVCs are not the cause of the tax compe- and intellectual property. These problems, as well as tition between governments or the tax avoidance by tax competition, may ultimately best be addressed by firms, they do magnify the challenges facing the inter- a destination-based corporate tax, similar to a value national tax system (see chapter 3). Firms are more added tax (VAT), in all countries, which would elim- sensitive to tax differences when factors of production inate the incentive to shift profits and compete for are mobile and production processes are fragmented taxes. The consequences of such a tax for revenue in across countries. Cross-border trade between corpo- small developing countries would, however, have to be rate affiliates creates opportunities for tax avoidance considered. A destination-based tax may not be imme- because multinational enterprises can reduce their diately feasible, but transitional arrangements could tax burden by manipulating transfer prices and other begin to alleviate the resource mobilization problem. artificial mechanisms. Profit shifting has become eas- The new economy has also raised concerns about ier for firms and harder for governments to identify market failure in international markets where reg- as the importance of intangible assets and the digital ulation is still mostly at the national level, ranging delivery of services has grown. Cooperation beyond trade | 239 A global consensus is emerging around the need to the West African Economic and Monetary Union reform international corporate taxation. As long as the (WAEMU) has instituted legal arrangements for tax current system relies on the physical location where coordination that are among the most advanced in the value is created (and booked) for tax purposes, it is world, but because of gaps in implementation they are open to abuse and compromises the revenue collection ineffective in many areas.13 efforts of governments.1 As elaborated in chapter 3, an Tax competition is a legitimate fiscal policy estimated 30 percent of global cross-border corporate tool that countries can use in aligning their tax sys- investment stocks are routed through offshore hubs, tems with development priorities to, for example, and the associated tax losses for developing countries attract foreign direct investment (FDI) that supports amount to about $100 billion.2 Overall, non-OECD high-quality and sustainable jobs, as well as technol- countries lose out on approximately 1.3 percent of GDP ogy transfers that spur productivity spillovers. Yet as a result of profit shifting.3 often tax competition results in inefficient outcomes International efforts are already well under way with costs exceeding benefits.14 This situation leads to address tax avoidance by large multinational firms. to negotiated tax breaks to attract foreign investment New measures are contained in the OECD/G20 Inclu- that benefit favored businesses and economic sectors, sive Framework on BEPS, including updated guidance while undermining competition and producing little on transfer pricing.4 Transparency in international in terms of jobs added or productivity enhanced. tax matters is being enhanced by an OECD initia- Finally, the BEPS package does not extend taxing tive that supports the exchange of data between tax rights over corporate income to countries where a administrations.5 And overall coordination between firm has no presence but makes sales (market coun- governments in implementing the BEPS measures tries). Traditionally, the income of affiliates of a mul- is supported through the Multilateral Convention to tinational corporation (MNC) is taxed in the country Implement Tax Treaty Related Measures to Prevent where production takes place, with the “residence” BEPS, which enables quick updates to international country in which the MNC’s headquarters is physi- tax treaties between signatory governments.6 cally located taxing the residual profits. Safeguards Although countries have made significant progress (antiabuse measures) are in place in many jurisdic- within the BEPS framework in reducing opportunities tions to prevent profit shifting between them purely for corporate profit shifting and base erosion, imple- for lowering an MNC’s aggregate tax bill. However, mentation of the relevant measures by developing the digitalization of the economy has spurred many countries is still lagging.7 Guidance on when and how market countries to contest this distribution of taxing to apply transfer pricing methods leaves firms and rights. The Internet makes it possible for companies tax administrations with significant discretion.8 The to generate vast profits in countries in which they complexity of many BEPS rules and lack of data, par- have no physical presence and are not liable for cor- ticularly on segments of GVCs located in other juris- porate income taxes. Profit may be generated out of dictions, pose further obstacles. As a result, developing intangible assets that are difficult to tax, such as cus- countries find it difficult to implement key parts of tomer data. In the absence of a coordinated solution at the BEPS package. the global level, countries are threatening to impose More important, however, the current BEPS pack- income taxes on companies that generate income age fails to address inefficient tax competition. The from economic activities in their country even if revenue losses from tax competition are estimated to they do not have a physical presence in that country outweigh those of tax avoidance.9 Indeed, reducing (so-called destination-based income taxes).15 the opportunities for tax avoidance by firms increases Against this backdrop, the OECD/G20 Inclusive the incentives for tax competition between govern- Framework is negotiating larger reforms of the ments.10 For example, analysis suggests that the 2017 international corporate tax architecture.16 To further U.S. federal corporate income tax reform, which com- advance the agenda, other proposals and analyses bined a cut in the headline rate with tighter rules to have been developed by the International Monetary prevent profit shifting, provoked other countries to Fund (IMF),17 the World Bank,18 and academia.19 The reduce their headline rates by about four points to various reform options come with different costs and compete.11 Meanwhile, regional coordination could be benefits for developing countries from both an admin- helpful for aligning policy makers’ incentives on taxes, istrative and a revenue generation perspective.20 but in practice such efforts fall short in eliminating Two of the OECD/G20 Inclusive Framework undesirable forms of tax competition.12 For example, reform proposals embody stronger antiabuse rules 240 | World Development Report 2020 within the current international tax framework: physical presence there. Again, however, the benefits the income inclusion rule and the base-eroding payment reaped by developing countries depend on the specific rule, together known as the global antibase erosion design. For example, by focusing on where consump- (GLoBE) proposal. The income inclusion rule allows tion takes place, these two proposals may disadvan- countries in which MNCs are headquartered (the resi- tage countries with production- or resource-based dence country) to tax income held by MNC subsidiar- economies. These options are also highly complex, ies in low-tax jurisdictions abroad. This rule does not creating implementation challenges for low-capacity directly benefit developing countries, which typically tax administrations.23 are not residence countries for major MNCs. How- The reform options currently under consider- ever, it does offer those countries an indirect benefit ation by the OECD/G20 Inclusive Framework would by reducing the incentive for tax competition between go a long way toward correcting the distortions countries. The base-eroding payment rule would present in the current system. However, alternatives, not allow MNCs to take deductions for payments to such as a destination-based cash flow tax (DBCFT), related parties abroad if those payments are suspected could eliminate tax competition and avoidance more of being motivated by tax avoidance and are not sub- completely.24 With a DBCFT, taxes are collected in the ject to a minimum effective tax rate in the foreign destination country, thereby extending the OECD/ country. Although such a rule is relatively straight- G20 Inclusive Framework proposals that focus on forward to enforce by means of MNC self-assessment the reallocation of residual profits. This tax would and disclosure obligations, it is difficult to identify eliminate the incentive for firms to shift profits base eroding payments if they go first through inter- between affiliates and for governments to lower tax mediate countries that meet the minimum effective rates to compete for investment. Moreover, unlike rate. Where successful, however, the rule can directly the OECD/G20 Inclusive Framework’s GLoBE pro- help developing countries to raise revenue.21 posal, governments would not need to agree on a A third option, the diverted profits rule, would pro- minimum tax rate. vide developing countries with a more direct benefit The DBCFT would replace the existing corporate and could be adopted as part of any reform package income tax with a new tax on the receipts of corpo- that includes antiabuse measures. This option is not rations less their expenditures, similar to a VAT. It currently under consideration by the OECD/G20 would tax all cash inflows (from sales of products, Inclusive Framework. A diverted profits rule would services, and real assets, borrowing, and the receipt reallocate profits posted in (very) low-tax jurisdictions of interest, but excluding injections of equity) with a over and above that allocable to any productive activ- deduction for all cash outflows (purchase of materials, ities in those entities. These residual profits would products, labor, and other services, real assets, lend- then be allocated more fairly across jurisdictions in ing, repayment of borrowing, and interest payments, which MNCs operate based on a formula using a set of but excluding equity repurchases and dividends).25 factors that indicate profit generation such as assets, However, to eliminate the incentive for tax competi- labor, and sales. A main advantage of this rule is that tion it would include a “border adjustment”: receipts it would allocate low-taxed profits to all countries in from exports would not be included, but imports the same GVC instead of to the parent entity. A main consumed locally would be taxed at the domestic rate. obstacle will be reaching agreement between coun- Like the VAT, it is a domestic tax based on the location tries on a formula for distributing low-taxed profits.22 of sales to consumers (the “destination” of the prod- Two proposals considered by the OECD/G20 uct) rather than on the location of profits, production, Inclusive Framework—user value and marketing intan- or corporate residence.26 As such, the DBCFT removes gibles—grant greater taxing rights to the destination incentives for tax competition and tax avoidance by countries in which goods and services are consumed. MNCs.27 But by exempting the labor element of value These proposals would allocate to source countries added from taxation, it provides an incentive for job only a portion of residual profits, with the allocation creation. formula based on the value of the market. Destination Based on the prevailing tax rates, global adoption countries would then have the right to tax businesses of a DBCFT system could have significant redistribu- that interact with their economies—either through tive effects on revenues across countries.28 Countries the location of users or through links to certain with trade deficits, limited revenues from natural marketing intangibles such as market research or resources, and low per capita income would be more brands/trademarks—even if those businesses have no likely to benefit under such a tax, at least initially.29 Cooperation beyond trade | 241 Importantly, countries that lose from a switch to a Regulation destination-based system can raise tax rates to com- pensate because pressures from profit shifting and tax In the conventional producer-centric view, regulatory competition are removed.30 cooperation is a complement to liberalization. In the Skepticism about the feasibility of a destination- alternative consumer-centric view, regulatory coop- based system is valid. MNCs that currently engage eration is a precondition for liberalization. Both are in aggressive tax planning would lose and are important in facilitating the operation of GVCs.36 likely to resist such a system, as occurred recently in the United States.31 Unilateral adoption would in Producer-centric cooperation to address the immediate term increase the prices of imported regulatory heterogeneity items and lower export prices, which should result in Regulatory heterogeneity can impede the compat- an exchange rate adjustment that would fully offset ibility of parts that is vital for GVCs. It arises when such price effects. But the need for such a large and requirements differ across countries because of differ- immediate appreciation presents an important risk ences either in institutions (leading typically to “hori- of major economic distortions. Border adjustment zontal” differentiation, such as in electrical plugs and for direct taxes may also raise questions about WTO legal services) or in social preferences (leading to “ver- consistency, compared with that for indirect taxes tical” differentiation, such as in the stringency of food, such as the VAT where it is explicitly allowed. How- paint, or financial regulations). The traditional case for ever, because the DBCFT is economically equivalent regulatory cooperation arises from the fact that regu- to a VAT plus a wage subsidy, both of which are latory heterogeneity segments international markets WTO-compatible, technical adjustments in the form in a way that prevents the exploitation of economies of the tax may be made to achieve compliance.32 Fur- of scale in production. For example, because each East thermore, a globally coordinated switch to a DBCFT African country has its own regulatory requirements may be more generally acceptable. Another concern is for service professionals, compliance costs cannot be that administering and enforcing such a tax could be spread out over the provision of professional services complex, but perhaps not much more so than current in other East African countries but must be incurred rules or those experienced under a VAT.33 separately in each market. According to one estimate, Notwithstanding the OECD/G20 Inclusive Frame- the European Union (EU) stock of FDI could increase work’s ongoing negotiations, governments in devel- by 20–35 percent if regulatory heterogeneity were oping countries can take immediate steps to address reduced in response to a common services regulation issues related to profit shifting and tax competition, directive.37 primarily by adopting stronger antiabuse rules— Such regulatory heterogeneity cannot be addressed mechanical, simple, and transparent. Countries can by imposing traditional trade disciplines because greatly benefit from the application of mechanical the problem is not due to protectionist or explicit rules for transfer pricing in some GVCs where appli- anticompetitive intent. But there is an economic cation of the arm’s-length principle is straightfor- cost of such heterogeneity because each country is ward.34  Countries also need to revise their tax treaty independently choosing its regulations without con- networks to renegotiate or cancel cost-ineffective sidering their negative impacts on foreign producers tax treaties.35 Depending on how the ongoing efforts and thus on competition. There are, then, potential unfold to reach a consensus on rule design by 2020, gains from international cooperation in which each developing countries should also consider adopting country forgoes the benefits of maintaining differ- the antiabuse GLoBE proposals, supplemented with a ent nationally optimal regulations for the benefits of diverted profit rule. integrating markets through some form of regulatory Observers are optimistic that the final solution convergence. proposed by the OECD/G20 Inclusive Framework In some cases, regulatory cooperation could be will move toward granting greater taxing rights to far-reaching and lead to harmonization or mutual jurisdictions where users and markets are located and recognition, which would eliminate the costs of regu- incorporating stronger antiabuse rules. Such propos- latory heterogeneity for firms and liberate them from als are a step in the right direction—but only when the uncertainty of discretionary licensing.38 In other low-capacity countries can implement them easily cases, regulatory cooperation could be valuable even and allocation rules do not compromise the taxing if it only involves greater mutual understanding of ability of producer and resource countries. how regulatory discretion in each jurisdiction will be 242 | World Development Report 2020 Figure 10.1 exercised because that, too, would lend predictability Figure 10.1 Regulatory commitments by exporters can to commitments. be exchanged for import liberalization commitments Consumer-centric regulatory cooperation Country Country to address international externalities The alternative case for regulatory cooperation arises because regulators in the jurisdiction of the exporter do not consider the consequences of market failure for consumers in the jurisdiction of the importer. Reciprocal For example, weak data protection in a country that liberalization exports data processing services can compromise the privacy of citizens of other countries. An increase in Import policy Import policy the concentration and anticompetitive practices of producers in one market can lead to exploitation of r downstream consumers in another market. And poor fo rn Exporter regulatory regulation of medicines, hospitals, and universities in re tu import commitments one country can hurt the health and human capital of in liberalization (government or private) foreign citizens who receive or visit for treatment or education. Regulatory Conventional trade negotiations and rulemaking cooperation are primarily concerned with reciprocal liberalization of import policy (figure 10.1). Accordingly, rules and Regulation Regulation commitments focus on tying the hands of importers: Harmonization, tariffs are bound; quotas are prohibited or restrained; mutual recognition discrimination against imports and trading partners agreements, equivalence is prohibited or restrained; and further disciplines Source: WDR 2020 team. may be imposed on importing country product stan- dards—such as the requirement that rules be “nec- essary” to achieving a legitimate objective. For the international trade. The McKinsey Global Institute most part, trade rules do not concern themselves with has estimated that cross-border data flows were 45 exporter disciplines or commitments. The rare exam- times larger in 2015 than in 2014, and about 12 percent ples for goods include prohibitions or restraints on of the international trade in goods was through global export subsidies, quotas, and agricultural assistance. e-commerce platforms such as Alibaba and Amazon.40 This asymmetric structure of trade rules in The U.S. International Trade Commission estimates which rules and commitments are directed entirely that in 2014 global digital trade, including data pro- toward importing countries and none (or very cessing and other data-based services, led to a more few) toward exporting countries is not conducive than 3.4 percent increase in U.S. GDP by increasing to consumer-centric regulatory cooperation. The productivity and lowering the costs of trade.41 Recent result is importing countries’ unwillingness to give empirical research finds that restrictions on data up protection or regulatory discretion, or both. The flows have significant negative consequences on the solution may be mutually binding commitments by productivity of local companies using digital tech- exporting and importing countries.39 The exporting nologies and, in particular, on trade in services. These countries would make regulatory commitments to estimates underscore the importance of cross-border looking after the interests of consumers in import- data flows for diffusing knowledge and technology ing countries, and in return the importing countries and for enabling the fragmentation of production of would make commitments to allowing access to goods and services across countries. their markets (represented by the diagonal arrow in But international data flows also raise concerns. figure 10.1). The provision online of search, communication, health, education, retail, and financial services relies Data flows on, or could lead to, the collection of personal data. The ability to move data freely across borders under- Because of the global nature of the Internet, such data pins a growing range of economic activity and can be quickly and easily transferred to third parties Cooperation beyond trade | 243 in other jurisdictions. This transfer can undermine In May 2018, the European Union implemented the domestic privacy goals when the personal data of world’s most comprehensive data protection regime, citizens flow to jurisdictions that do not offer compa- the General Data Protection Regulation (GDPR), which rable levels of privacy protection, prompting domestic replaced its 1995 Data Protection Directive. Under the regulators to limit the free flow of data across borders. GDPR, personal data are allowed out of the European These concerns are prompting governments to Union only under strict conditions. One option is for apply new regulatory policies to digital trade and data the non-EU country to adopt a privacy regime whose flows, severely dampening the positive impact that level of protection is “essentially equivalent” to that digital trade has on the economy.42 Meanwhile, policy guaranteed within the European Union.44 In other makers are paying special attention to cross-border options, firms can accept Binding Corporate Rules data flows, and so restrictions on data flows have (BCRs) or use Standard Contractual Clauses (SCCs), been trending upward in recent years (figure 10.2). which are mechanisms to authorize companywide or Burdensome data policies can be split into two types: transaction-specific data transfers, respectively. those affecting the cross-border mobility of data, such These new regulations are likely to especially as data localization or local storage requirements, and affect services GVCs that depend on data flows. Such those affecting how data are treated domestically. In data flows drive the most dynamic exports of develop- both cases, the pattern emerging from a wide swath of ing countries—digitally delivered data processing and countries is rising policy restrictiveness. data-related business services. These services, ranging Restrictions on data flows have large negative from financial accounting and tax returns to health consequences on the productivity of local companies transcriptions and diagnostics, contributed to the using digital technologies and especially on trade more than $50 billion in developing country exports to in services. Studies show that countries would gain the European Union in 2015—one-fifth of them from on average about 4.5 percent in productivity if they Africa. removed their restrictive data policies, whereas the Here, developing countries face a dilemma: either benefits of reducing data restrictions on trade in ser- they must adopt EU-like national privacy regulations vices would on average be about 5 percent.43 or their firms must incur the firm-specific costs of using BCRs or the transaction-specific costs of using SCCs. The GDPR offers a balance between privacy and Figure 10.2  Countries’ restrictions on the economic and trade opportunities from data flows data flows increased from 2006 to 2016 that may not be optimal for developing countries.45 A GDPR-based national privacy law would impose 0.5 the same high standard on all firms, even when they Digital Trade Restrictiveness Index (0–1) sell at home, leading to higher economywide costs 0.4 of doing business. The adoption of tough standards is likely to reduce the scope to use personal data to 0.3 improve access to domestic services, such as by open- ing new credit bureaus, and to reduce the competi- tiveness of digital exports in third markets such as the 0.2 United States that do not require GDPR-like privacy standards. Overall, then, BCRs and SCCs have proved 0.1 costly and time-consuming. A survey in India of the impact of the earlier, less-stringent EU Data Protection 0 Directive revealed that the process to ensure that firms 2006 2008 2010 2012 2014 2016 complied took over six months, and 90 percent of the Data policies respondents used transaction-specific contracts that Cross-border data flows involved on average a complex process lasting more Domestic data flows than three months.46 As many as two-thirds of the sur- Source: Ferracane, Lee-Makiyama, and van der Marel 2018. veyed services exporters claimed a significant loss of Note: This figure is based on the ECIPE Digital Trade Restrictiveness Index business opportunities because of the requirements. (DTRI), which ranges from 0 (completely open) to 1 (virtually restricted), Because privacy regulations affect the inter- with higher levels indicating increasing data restrictiveness. The index covers 64 countries representing more than 95 percent of the value-added national data transfers on which the digital trade content of gross exports. depends, developing countries could in principle 244 | World Development Report 2020 challenge at the WTO the consistency of the GDPR In 2016 the United States and the European Union with EU trade commitments. But WTO litigation is concluded the Privacy Shield—an arrangement that unlikely to address the underlying challenge raised the EU Commission has deemed “adequate” under the by the GDPR: how to preserve digital trade opportu- EU Data Protection Directive—thereby enabling the nities while maintaining nationally desired privacy transfer of personal information from the European standards. Even so, WTO litigation could induce the Union to U.S. participating businesses.48 Under the European Union to be more flexible in its application Privacy Shield, U.S. companies self-certify individu- of the GDPR and offer other countries opportuni- ally or through an industry body to the U.S. Depart- ties to negotiate arrangements like the one with the ment of Commerce that they will protect personal United States. data consistent with the Privacy Shield, which largely The EU–U.S. Privacy Shield offers a way of resolv- reflects the main elements of the EU Data Protection ing the conflict between regulatory heterogeneity and Directive.49 U.S. businesses are required to publish international data flows (a subject discussed in more their privacy policies, and the Privacy Shield gives detail shortly). Whereas traditional trade agreements the U.S. Federal Trade Commission jurisdiction over are geared toward an exchange of market access such businesses if they breach their own policies. In commitments, the Privacy Shield is an innovative bar- addition, the United States provides various means of gain: the destination country for the data promises to redress for people whose personal data has been com- protect the privacy of foreign citizens consistent with promised, including a direct complaint to the business their own national standards. In return, the source or a complaint to the U.S. Department of Commerce. country commits to not restricting the flow of data. Such an agreement with the European Union The rules on digital trade in the Comprehensive and gives participating U.S. firms two big advantages Progressive Agreement for Trans-Pacific Partnership over the existing options. First, unlike in the case (CPTPP) reflect a similar bargain in a multicountry of BCRs and SCCs, the firms are not required to context.47 In conjunction with progress toward devel- establish a costly presence in the European Union oping common privacy standards in OECD countries because domestic regulators assess conformity with and the Asia–Pacific Economic Cooperation (APEC) EU standards at home. Second, unlike in the case of forum, such cross-border commitments can help cre- a national adequacy determination by the European ate a framework for global privacy protection that also Union, firms are not obliged to adopt more stringent supports digital trade. and costlier standards for data involving transactions The approaches described here, however, risk at home or with countries less demanding than the excluding some developing countries that may not European Union. be able to make credible regulatory commitments The CPTPP provision on data flows requires that in the near term, leading to a pattern of trade based “each Party shall allow the cross-border transfer of on existing mutual trust rather than comparative information by electronic means, including personal advantage. Fortunately, the existing multilateral rules, information, when this activity is for the conduct of notably provisions on mutual recognition agreements the business of a covered person.” It also prohibits in the WTO’s General Agreement on Trade in Services data localization, stating that “no Party shall require (GATS), can help protect the interests of excluded a covered person to use or locate computing facilities countries. Some developing countries are participat- in that Party’s territory as a condition for conducting ing in the CPTPP, in which the provisions on data business in that territory.” At the same time, the CPTPP flows are matched by provisions on protecting pri- breaks new ground by obligating data-destination vacy and preventing fraud. Developing countries also countries to prevent fraud and deception and protect should take advantage of the U.S. Clarifying Lawful personal information. In particular, “each Party shall Overseas Use of Data (CLOUD) Act, which has created adopt or maintain a legal framework that provides the basis for new agreements to supplement older and for the protection of the personal information of the slower mutual legal assistance treaties. users of electronic commerce.” Moreover, “each Party Table 10.1 is an overview of the different approaches shall endeavor to adopt non-discriminatory practices to cross-border data flows of some of the major pri- in protecting users of electronic commerce from vacy arrangements in place. Each privacy mechanism personal information protection violations occurring relies on some convergence toward common privacy within its jurisdiction.” principles (whether in the European Union or among Such reciprocal obligations on data source and des- a set of countries). tination countries are a perfect example of the type of Cooperation beyond trade | 245 Table 10.1  Regulation of international transfers of personal information, by privacy regime EU Data Protection Regime Directive and EU–U.S. OECD privacy attribute EU GDPR Privacy Shield CPTPP and USMCA APEC CBPR principles Privacy Determined by EU Determined by EU, Determined by each Common APEC privacy Common OECD principles but recognizes that party, taking into principles based on privacy principles U.S. promise of account “principles and OECD privacy floor, privacy protection guidelines of relevant which domestic privacy for EU citizens is international bodies” regimes can go beyond equivalent to that of EU Scope Applies to all firms Applies to U.S. firms Requires each party Applies to APEC CBPR– Applies to data collecting data participating in the to “endeavor to adopt complianta organizations controllers— on EU citizens no Privacy Shield and non-discriminatory collecting personal entities that matter where the collecting data on practices in protecting information from APEC decide about the firms are located EU citizens users of electronic economies content and use commerce from personal of personal data, information protection without regard violations occurring for location of within its jurisdiction” data Enforcement In case of a United States (data- Unspecified—depends Data source country Data source national adequacy destination country) on national privacy law enforces through APEC country enforces finding, the data enforces—that is, Accountability Agentb against data destination country EU recognizes and Privacy Enforcement controller enforces U.S. enforcement Authority (PEA),c with procedures cross-border enforcement In case of BCRs cooperation facilitated by and SCCs, the APEC Cross-Border data source EU Privacy Rulesd country enforces against local entity Source: Mattoo and Meltzer 2018. Note: APEC = Asia–Pacific Economic Cooperation; BCRs = Binding Corporate Rules; CBPR = Cross-Border Privacy Rules; CPTPP = Comprehensive and Progressive Agreement for Trans- Pacific Partnership; EU = European Union; GDPR = General Data Protection Regulation; OECD = Organisation for Economic Co-operation and Development; SCCs = Standard Contractual Clauses; USMCA = United States–Mexico–Canada Agreement. An entity is CBPR-compliant when its self-assessment of compliance with its own data privacy policies compared with those of the APEC Privacy Framework has been reviewed by an a.  APEC-recognized Accountability Agent. An APEC Accountability Agent has met the APEC recognition criteria to the satisfaction of the APEC economies. b.  A Privacy Enforcement Authority is any public body responsible for enforcing privacy law that can conduct investigations or pursue enforcement proceedings. c.  Endorsed by APEC ministers in 2009, a Cross-border Privacy Enforcement Arrangement (CPEA) is a voluntary framework that aims to facilitate cooperation among PEAs in d.  enforcing the CBPR, such as parallel or joint investigations or enforcement actions. Information sharing and cooperation are also encouraged with privacy enforcement authorities outside of APEC. regulatory cooperation needed to reassure data-source agreements can help ensure that the emerging arrange- countries that their commitments to openness will ments between sets of countries are fully transparent. not place their consumers at the mercy of indifferent More important, GATS Article VII can help ensure that foreign regulators. any such arrangements do not discriminate against, Countries can be expected to self-select into these and are open to participation by, third countries. arrangements, as members of APEC, OECD, the Euro- pean Union, and the United States are already doing, and gradually widen and deepen them. The African Competition policy Union’s Convention on Cyber Security and Personal Anticompetitive practices in international markets Data Protection, adopted in 2014, has been ratified into can affect the distribution of gains from participating domestic law by five members to date and signed by in GVCs. Because GVCs span many markets, action nine others.50 In the transitional phase, multilateral against anticompetitive practices must take into rules can fulfil two important roles. GATS Article account the behavior that reduces the availability or III on transparency and Article VII on recognition raises the prices of the end product (to the detriment 246 | World Development Report 2020 Figure 10.3  Cartel episodes and significant overcharges have been observed across all regions 160 50 48.4 140 Number of cartel episodes 40 36.8 120 Overcharge (%) 33.0 32.7 32.2 100 30 25.0 25.1 80 20.0 20 17.7 17.2 60 40 10 20 0 0 North America Asia and Oceania Western Europe Africa, European Union Latin America, and Eastern Europe Median average overcharge Mean average overcharge Number of cartel episodes (right axis) Sources: World Bank and OECD (2017), with elaboration on data (485 decisions) from Connor (2014). of buyers) as well as of intermediates (to the detriment benefits accruing to participants at another stage of of rivals). An examination of 1,530 cartel cases involv- production. The cross-border nature of GVCs means ing overcharges across countries reveals that the that restrictive practices often also have a cross-border mean average overcharge is at least 49 percent, and dimension. For example, in 2009 the South African 80 percent where cartels are strongest.51 The largest Competition Commission detected a cartel among overcharges have been observed in North America four large cement producers involving market allo- (figure 10.3). The price-raising effect of such behavior cation and price-fixing in South African provinces may foreclose the ability of producers to participate as well as in Botswana and Namibia. Since the cartel in a value chain or limit their profits and thus their was broken up, prices and margins for downstream opportunities for expansion.52 firms in the region have declined by 7.5–9.7 percent.55 The effects of these practices can fall outside In 2015 Colombia’s Superintendence of Industry and the jurisdiction of national competition authorities Commerce fined 12 sugar mills, 14 individuals, two where the firms are based. And the firms can be companies, and three business associations a total outside the jurisdictions of the authorities where of $91 million for agreeing to prevent sugar imports the effects are felt. Overall, then, one set of author- from Bolivia, Costa Rica, El Salvador, and Guatemala ities is not mandated to address the effects, and the and for allocating clients.56 Food processing associa- other set—even if it could in principle enforce the tions in Colombia had reported sugar overcharges of law under the effects doctrine53—is, in practice, not 45 percent57 affecting food value chains and confec- able to do so without collaboration among jurisdic- tionary exports that account for 13 percent of overall tions. Meaningful international cooperation on the food exports.58 enforcement of competition policy would reassure In 2016 the European Commission prosecuted a countries facing jurisdictional constraints or limited cartel case against major European truck producers enforcement capacity that the gains from GVC par- that had colluded on pricing and the timing to intro- ticipation will not be appropriated by firms behaving duce new emissions technologies—and had agreed to anticompetitively.54 pass on the cost of such systems to buyers of trucks.59 Intermediate input suppliers may also collude to raise The negative spillovers of anticompetitive prices for parts needed by lead firms. Automotive parts practices makers in Europe were first investigated in 2010–12, Anticompetitive behavior by companies at one stage and eventually more than a dozen specific cartels for of production, whether abuse of a dominant posi- a range of car parts were identified by the authori- tion or restrictive business practices, can reduce the ties. The European Union alone imposed more than Cooperation beyond trade | 247 Figure O.24 Figure 10.4 The European Commission has imposed large fines on car parts cartels since 2013 Airbags (2017 and 2019) Flexible foam (2014) Seat belts Wire harnesses (2017 and 2019) (2013) Parking heaters (2015) Air conditioning and engine cooling (2017) Brakes (2018) Steering wheel (2017 and 2019) Spark plugs Automotive bearings (2018) (2014) Lights (2017) Alternators and starters (2016) Source: European Commission 2019. €2 billion in fines in 15 separate rulings pertaining to and in 2012 the European Union uncovered four car- various car parts producers (figure 10.4). tels in the international air freight forwarding market Anticompetitive practices have also been identi- (between six and nine of the same companies were in fied in services sectors central to global production each cartel).62 The companies were charged with coor- networks, such as finance and transport, as well as dinating conduct such as currency adjustment and in new digital services in search, advertising, com- peak season surcharges. munication, and distribution. For example, fines of Anticompetitive behavior is also commonly found $1 billion or more were levied by the United King- among services related to port transport, such as han- dom’s Financial Conduct Authority, the United States’ dling and towage services. In 2019 the Guyana Com- Commodity Futures Trading Commission, and Swiss petition and Consumer Affairs Commission (CCAC) regulators on the world’s biggest banks—Barclays, fined five terminal operators almost $4 million each JPMorgan Chase, Royal Bank of Scotland, Citigroup, for colluding to fix prices for the haulage of contain- and Creditbank—for manipulating foreign exchange ers—an arrangement facilitated by the national ship- markets. The rigging apparently took place through ping association.63 And in 2017 the German and Dutch information sharing and coordinated trading. competition agencies collaborated in investigating Various cartels involving as many as 16 freight a cartel in harbor and towage services dating back forwarders have been discovered in key destination to 2000/2001. Four companies were fined a total of markets for GVCs. Between 2002 and 2007, freight for- €13 million for allocating orders between them in warding companies were investigated in the United accordance with turnover targets.64 States for price-fixing, with 16 companies pleading In 2010 the European Commission fined 11 air cargo guilty by 2011. The total fines levied by the U.S. Depart- carriers nearly €1 billion for operating a worldwide ment of Justice amounted to $100 million.60 In 2009 cartel that affected cargo services within the Euro- Japan issued a cease and desist order to 12 of the same pean Economic Area (EEA).65 The carriers coordinated freight forwarding companies for similar conduct their action on surcharges for fuel and security with- during the same period, with a joint fine of ¥9 billion,61 out discounts over a six-year period. The European 248 | World Development Report 2020 Commission’s fines on the air cargo carriers were chain Whole Foods in 2017 raised concerns about the reduced by 50 percent in relation to sales between use of consumer data.70 In 2019 Mexico’s Federal Eco- the EEA and third countries to take into account the nomic Competition Commission (COFECE) blocked a fact that the harm of the cartel fell outside of the EEA’s merger between Walmart and Cornershop, a Mexican jurisdiction. International maritime transport has platform that delivers groceries from online retailers also regularly been a target of enforcement: in 2018 such as Costco, Chedraui, and Walmart.71 According to the European Union levied $458 million in fines on COFECE’s decision, the merger could unduly displace four maritime car carriers for customer allocation and competitors in the provision of logistical services, and price-fixing for deep-sea transport of vehicles.66 the market power of the merged economic entity could Digital companies are also attracting attention inhibit the development of new platforms. from national competition authorities. Large multi­ It is not known how much cartels affect GVCs and sided markets created through the inherent net- cost consumers in developing countries, but the spill- work effects of individual platforms are vulnerable over effects of foreign cartels clearly can be significant. to monopolistic behavior, and platform firms can exploit user data to stifle competition. In early 2019, The empowered: Not very concerned the United States launched a Technology Task Force In 2018 over 130 jurisdictions had a competition law to monitor competition in U.S. technology markets, in place, up from fewer than 50 in the early 1990s.72 particularly those in which platforms compete.67 Aus- The growth in the number of competition agencies tralia, the European Union, Germany, and the United has been associated with an increase in the number of Kingdom have also initiated multistakeholder inqui- cartels prosecuted each year. Between 1989 and 2016, ries into competition in digital markets. Meanwhile, 953 cartel investigations led to fines totaling $112 bil- competition authorities continue to police specific lion. While large, the number is much less than total instances of anticompetitive behavior. For example, overcharges to buyers, which are estimated to exceed the European Commission is pursuing a review of $1 trillion.73 smartphone chargers that could have implications for A central feature of competition law, however, is Apple because the iPhone’s charger departs from the that it is directed at the effects of anticompetitive prac- micro-USB connectors used by the rest of the industry tices on national consumers and markets. Addressing through voluntary agreement.68 Abuse of dominance the effects of behavior by national firms on a foreign was a recurring theme in three EU investigations market is not part of the mandate of national com- into Google between 2017 and 2019, resulting in fines petition agencies. For example, Section 3 of South totaling $9.3 billion (EU regulations permit fines of up Africa’s Competition Act states that it “applies to all to 10 percent of a company’s annual global turnover). economic activity, within, or having an effect within, Brazil (2013), the Russian Federation (2015), and India the Republic.” It does allow foreign agencies to inves- (2018), among other countries, have also launched tigate anticompetitive behavior that has an impact investigations into Google for abuse of dominance in both on South Africa and the region and share infor- web search advertising and bundling of search results mation, but only if the companies concerned agree to as the default on Android mobile devices. this. For the most part, however, countries must rely The ability of platform companies to use data col- on self-defense to combat anticompetitive behavior lected through their platforms to stifle competition is with effects on their markets, whether it involves also a concern. Brazil’s 2013 investigation of Google locally established firms or companies headquartered also examined whether the company was scraping in foreign countries or MNCs. (extracting) relevant competitive content (such as product reviews) held by rival search websites in The concerned: Not fully empowered order to strengthen Google’s own search services. In Competition laws generally permit action against 2018 the European Union opened a preliminary inves- anticompetitive practices that have effects on the tigation into how Amazon uses data on third-party domestic market, but developing countries may not vendors operating on its platform because of concerns have adequate capacity or jurisdiction to act. The that the data allow Amazon to identify product trends effectiveness of this “effects doctrine” depends on the early and promote its own brands.69 capacity of authorities to identify, investigate, and Mergers and acquisitions create similar concerns if necessary fine foreign firms for anticompetitive about their effects on market competition. In the behavior. Small or low-income countries may not be United States, Amazon’s acquisition of the supermarket able to do so. Competition law enforcement capacities Cooperation beyond trade | 249 vary widely across developing countries. Whereas in including chapters on competition policy. In free trade Latin America, agencies in Brazil, Chile, Colombia, agreements, these generally establish a basic frame- Mexico, and Peru apply sophisticated investigative work of principles such as transparency, due process, tools to detect several major cartel agreements each assistance (for example, exchange of nonconfidential year, many of their regional peers have fined only a information), and nondiscrimination. Moreover, tech- few firms for such conduct in over a decade. Cartel nical competition commitments are directed at anti- enforcement in Africa and Asia is quite limited with trust enforcement and merger control, even though in very few exceptions. In 2017–18, only the Arab Repub- some PTAs they are not binding and cannot be chal- lic of Egypt and South Africa imposed significant lenged through the dispute settlement provisions of cartel fines on the African continent.74 In one case, PTAs. Competition commitments in PTAs have helped foreign exchange firms without a local presence were to promote regional market integration. In common excluded from prosecution in South African courts markets, regional secretariats may investigate cases. for colluding on fixing exchange rates.75 Similarly, An example is the COMESA Competition Commission, differences in capacity to act imply that many devel- which has the mandate to investigate cases that affect oping countries are less able than more advanced two or more COMESA members, and it has vetted countries to defend the interests of their consumers merger cases. But besides antitrust, PTAs have included from anticompetitive behavior. a number of sector-specific commitments to eliminate Few, if any, complementary investigations have domestic rules that facilitate anticompetitive prac- been pursued of South African firms that have tices. These provisions typically target domestic rules engaged in anticompetitive behavior in other coun- that reinforce dominance or discrimination in favor of tries in southern Africa, nor have any claims been domestic firms, such as in the case of agribusiness and made for damages, even though in many of these investment chapters.78 cases the firms operate in neighboring countries.76 Cooperation can increase the effectiveness of Zambia is a notable exception. Its competition author- enforcement through sharing information and ity has jurisdiction over Zambian markets and can enhancing the joint capacity to investigate and act. investigate and sanction foreign companies that have The car parts cartel cases described earlier involved Zambian operations. In 2013 it prosecuted a fertilizer cooperation by 13 jurisdictions, including Brazil, cartel that was uncovered in South Africa and fined the Canada, China, the European Union, India, Japan, the participants $20 million.77 For companies domiciled in Republic of Korea, South Africa, and the United States, foreign countries, it collaborates with other national with some 70 companies investigated for price-fixing and regional authorities, such as the Common Market and bid rigging for more than 100 products.79 The for Eastern and Southern Africa (COMESA) Competi- same is true of large or complex merger cases. The tion Commission, to sanction those companies in the acquisition of Lafarge (France) by Holcim (Swiss), event their anticompetitive practices have an effect on two large cement and concrete producers with global the Zambian market. operations, involved seven competition agencies in countries outside the European Economic Area: Bra- The limited scope of existing multilateral zil, Canada, India, Mauritius, the Philippines, South and plurilateral cooperation Africa, and the United States.80 However, Holcim- In 2003 efforts to launch negotiations on a multilateral Lafarge operates in some 80 countries, and most agreement on competition policy in the WTO failed did not investigate the merger or require remedies to attain the needed consensus. The WTO services even though many may be negatively affected.81 agreement does contain a provision on anticompeti- tive practices (GATS Article IX), but it provides only for International agreements on cross-border an exchange of information and consultation. Since regulatory cooperation then, the International Competition Network, in con- Despite the efforts to cooperate in investigations of junction with deliberations in OECD and the United anticompetitive practices, what has not changed is Nations Conference on Trade and Development the explicit nationalist focus of competition laws. (UNCTAD), has established a basis for international Competition law enforcement is premised on self- cooperation between agencies. This effort has been help. There are no examples of international coopera- complemented with bilateral agreements between tion among countries to enforce competition rules to agencies to cooperate in different areas. Moreover, protect the interests of foreign consumers, although preferential trade agreements (PTAs) are increasingly foreign consumers may be incidental beneficiaries of 250 | World Development Report 2020 case-specific collaboration between agencies in two the Eastern Caribbean Telecommunications Author- affected jurisdictions. ity (ECTEL), the world’s first regional telecommuni- One corrective step would be to provide foreign cations authority. Although the member countries jurisdictions with information on the foreign effects retained their sovereign power over licensing and reg- of anticompetitive practices under investigation ulation, ECTEL provides technical expertise, advice, when such effects are identified. Agreeing to explic- and support for national regulations. Apart from the itly assess such effects could also be an element of a economies of scale in establishing a common regula- plurilateral agreement to assist developing countries tor, there are at least three other advantages of such an in addressing restrictive business practices that arrangement. It promotes the development of harmo- harm their consumers or firms. In one further step, nized and transparent regulation in the region, allows countries would end existing exemptions for export for greater independence (and thus credibility) in cartels from the scope of their national competition regulatory advice, and enhances bargaining power in laws.82 For example, the United States and the Euro- negotiations with incumbents and potential entrants. pean Union, which are home to many services multi­ In fact, there is evidence that the creation of ECTEL, nationals, could begin by ending exemptions from along with other reforms, prompted a decline in the the scope of their competition law collusive practices price of a daytime call to the United States of between whose effects are felt outside their jurisdiction. This 24 and 42 percent in these countries. could be pursued through a plurilateral agreement— However, creating a supranational competition in the WTO, OECD, or UNCTAD—among the largest law regime should be a mechanism for strengthening jurisdictions. In a more ambitious step, countries competition rather than weakening national com- could change national legislation to require nationals petition law regimes. The competition legal regime not to harm foreigners abroad by conduct that is ille- of the West African Economic and Monetary Union gal at home.83 Such a change could be accompanied (WAEMU) prohibits parallel national competition by recognition of the right of foreign consumers to rules. Thus Côte d’Ivoire and Senegal, among others, challenge anticompetitive practices by services firms are barred from implementing their national compe- in the national courts of countries whose citizens tition laws. Meanwhile, WAEMU itself has limited own or control these firms. resources to implement competition law enforce- Such a deal could be part of a broader trade agree- ment, which is therefore mostly ineffective in the ment obliging importing countries to liberalize and entire region.84 exporting countries to regulate. For example, Zambia could assert that opening its market to South African firms would be conditional on a commitment by South Infrastructure African authorities to investigate anticompetitive The failure of countries to coordinate the provision of behavior in Zambia by firms based in South Africa, or infrastructure impedes GVC investment, expansion, to assist the local authorities in doing so. In principle, and upgrading. Each country does not fully internalize it would be in South Africa’s interest to provide such the benefits to foreign traders of reductions in domes- reassurance. tic trade costs, and so gains are larger when govern- ments on both sides of the border invest in expediting Regional cooperation between developing trade simultaneously. The WTO Trade Facilitation countries Agreement addresses the coordination problem and In parallel, deepening regional cooperation enforce- provides low-income countries with financial assis- ment of competition policy offers a mechanism for tance for the necessary investments. many developing countries to protect their consum- Coordinated efforts to develop infrastructure can ers and firms from foreign anticompetitive behavior. enhance international connectivity (box 10.1). For An option is to form a regional competition agency any country, building a railway or a road has some to which national competition agencies could pass value, but it also has value to the countries around it jurisdiction in specific circumstances, just as EU because improvements in one part of the transport member states pass jurisdiction to the European Com- network reduce shipping times for all countries in the mission when circumstances warrant. For example, in network. If each country alone decided how to invest a cost-saving move Dominica, Grenada, St. Kitts and in infrastructure, spillovers to other countries would Nevis, St. Lucia, and St. Vincent and the Grenadines not be taken into account. This is even truer for trans- established in May 2000, with World Bank support, port infrastructure that crosses one or more borders. Cooperation beyond trade | 251 Box 10.1  International cooperation on transport infrastructure Of the many examples of international cooperation on Figure B10.1.1  Impact of China’s Belt transport infrastructure, the two most well known are and Road Initiative transport projects the European Union’s Trans-European Transport Network with and without input–output linkages (TEN-T) and China’s Belt and Road Initiative (BRI). TEN-T is an effort to develop a Europe-wide network of roads, railway 0.45 lines, inland waterways, maritime shipping routes, ports, 0.40 airports, and railroad terminals. The TEN-T will require build- 0.35 ing new physical infrastructure; adopting innovative digital 0.30 technologies, alternative fuels, and universal standards; Density 0.25 and modernizing and upgrading the existing infrastructure 0.20 and platforms. Although the scope of the BRI is still taking 0.15 shape, it is structured around two main components, under- pinned by significant infrastructure investments: the Silk 0.10 Road Economic Belt (the “Belt”) and the New Maritime Silk 0.05 Road (the “Road”). The overland Belt will link China to Cen- 0 tral and South Asia and onward to Europe, and the maritime 0 2 4 6 8 10 12 14 Road will link China to Southeast Asia, the Gulf countries, % of GDP East Africa and North Africa, and on to Europe. BRI project with I–O linkages Transport infrastructure that improves international BRI project without I–O linkages connectivity can have a significant impact on international Non-BRI project with I–O linkages trade and GVC integration. Time delays are a barrier to Non-BRI project without I–O linkages international trade. This is even truer for goods and services produced in GVCs because their production relies on the Source: de Soyres, Mulabdic, and Ruta 2019. timely delivery of time-sensitive inputs.a The importance of Note: In this figure, de Soyres, Mulabdic, and Ruta (2019) build on time as a trade barrier is well established in the literature.b Caliendo and Parro (2015)—a Ricardian model with sectoral linkages, trade in intermediate goods, and sectoral heterogeneity—to allow for By one estimate for a sample of 126 countries, a one-day changes in trade costs stemming from improvements in transportation delay in shipping time reduces trade by at least 1 percent.c infrastructure connecting multiple countries—improvements financed through domestic taxation. The model highlights the impact on trade and The World Trade Organization (WTO) finds that delays and GDP of infrastructure investments linked to the Belt and Road Initiative border costs can be equivalent to a 134 percent ad valorem (BRI) through cross-border input–output linkages. tariff on a product in high-income countries and a 219 per- cent tariff equivalent in developing countries.d projects have major impacts on public finances and gen- An analysis of the impacts of transport projects linked erally have asymmetric effects on the trade and GDP of to the Belt and Road Initiative reveals the relevance of individual countries. Countries that build and pay for large international cooperation in infrastructure for GVCs (fig- sections of a project may not gain the most from it. Indeed, ure B10.1.1). For economies along the Belt and the Road, analysis suggests that the BRI transport project increases as well as for non–Belt and Road countries, the effects overall welfare for the economies along the Belt and Road of infrastructure investment on GDP are larger when the by up to 2.8 percent, but three countries (Azerbaijan, Mon- model accounts for cross-border input–output linkages. golia, and Tajikistan) will experience welfare losses because When a sector experiences a decrease in the price of its the infrastructure costs will outweigh gains through trade.e imported inputs as shipping times and trade costs fall, it This raises the difficult question of equitable financing of passes on the associated reduction in production costs to common infrastructure projects. Furthermore, the welfare downstream industries, propagating the benefits across effects of BRI transport projects would increase by a factor the world. These input–output linkages lead to a potentially of four if participating countries would reduce by half the complex reallocation of comparative advantage, produc- delays at borders and tariffs, which highlights the impor- tion, and trade, thereby increasing welfare. tance of complementary policy reforms. Put differently, International cooperation on infrastructure also comes lack of such reforms severely limits the gains from inter­ with its challenges. Large cross-border infrastructure national cooperation on infrastructure. a. Baniya, Rocha, and Ruta (2019). b. Djankov, Freund, and Pham (2010); Hummels and Schaur (2012, 2013). c. Djankov, Freund, and Pham (2010). d. Baniya, Rocha, and Ruta (2019). e. de Soyres, Mulabdic, and Ruta (2019). 252 | World Development Report 2020 For any country, the timing of investments by neigh- are two sets of checks: one for exit and one for entry. boring countries in their infrastructure is relevant The same is true for goods, but the delays tend to be because the value of one’s investment depends on the even longer than those for people because of complex investment decisions of others. The ultimate impact regulations and taxes that differ across products and of a country’s investments also depends on the policy countries. However, there is little a government can choices of other countries, such as the standards they do to ensure short customs transit times for its firms’ use when building infrastructure or the procedures exports when they reach their destination. that countries use to clear goods at the border. Cooperation on policy and trade facilitation can But common transport infrastructure also creates together go a long way toward eliminating delays at challenges. One is that it has significant implications borders. For example, for many years Guatemala and for public finances and may have asymmetric effects Honduras required identical paperwork and duplicate on the trade and GDP of individual countries. This processes on both sides of the border, but the red tape asymmetry raises the possibility that the countries was still expensive and time-consuming for busi- that build—and bear the cost of—large sections of the nesses. Some truck drivers even brought hammocks project may not be the ones that will gain the most to the border so they could wait out the lengthy pro- from it.85 Another challenge is the need to ensure cess in comfort. When both countries moved from a mutual compatibility in standards. An example of free trade area to a customs union, eliminating the how slight differences in infrastructure standards can need for complex rules of origin, transit times fell disrupt trade is the rail gauge—that is, the distance from 10 hours to just 15 minutes and trade increased between the two rails that form a railway track. Trains by 7 percent.86 Now paperwork is handled by a sin- cannot easily cross borders if the rail gauge standards gle online instrument. At the border, a digital reader differ across countries. Russia used broad-gauge device instantly scans a Quick Response (QR) code track (1,520 millimeters, or roughly 5 feet) in the 19th and quickly certifies—online—whether an importer century to protect it from the entry of trains from has already paid the VAT on the goods in the destina- the west, which ran on standard-gauge track (1,435 tion country.87 Another example is East Africa, where millimeters). For Russia, the 85-millimeter (or 3-inch) a combination of procedural simplification, introduc- difference served a strategic military purpose because tion of one-stop border posts, harmonization of vehi- troops and material could not easily enter the country cle standards, and enforcement of dwell time limits by rail. But in more tranquil times, the same 85 milli- helped to reduce the time to cross at the Malaba bor- meters have become a high trade barrier, preventing der post from two days to six hours for loaded trucks.88 goods from seamlessly crossing borders. Broad-gauge Trade facilitation has become an increasingly com- tracks are still used in successor states of the Soviet mon feature of trade agreements. It encourages coor- Union such as the Kyrgyz Republic, Tajikistan, and dination and cooperation among customs authorities, Uzbekistan, exacerbating the transport challenges expanding the benefits from improvements on both that these countries face because they are landlocked. sides of the border. For example, the Comprehensive In part because of the extensive delays when chang- and Progressive Agreement for Trans-Pacific Partner- ing cargo at borders, only about 5 percent of the goods ship commits members to adopting predictable and transported between Asia and Europe move by rail. transparent procedures and the advance electronic Synergies can arise across different types of infra- submission of import requirements. A problem, how- structure. For example, it is much cheaper to bundle ever, is that the reform requires a capacity, both tech- the laying of fiber-optic cable with the building of nical and monetary, that many developing countries electric or gas lines, roads, or railways than to create lack. communications, transport, and energy connectivity The WTO Trade Facilitation Agreement ratified separately. Such bundling has the further advantage in 2017 allows developing countries to reform at their of not prejudging the future importance of different own pace and with assistance provided by advanced types of international flows. It also does not presume countries. It serves as an example for other areas in the evolution of comparative advantage in any specific which cooperation and capacity are constraints on direction: a country is equipped to export goods by trade. As of August 2019, more than 63 percent of road or rail and digital services by cable. WTO members implemented the TFA, including 100 Seamless travel across borders requires cooper- percent of developed members, 62.5 percent of devel- ation not just on physical infrastructure, but also on oping members, and 26.8 percent of least developed soft infrastructure. When people cross borders, there countries. Cooperation beyond trade | 253 Notes 17. IMF (2019). 18. Pemberton and Loeprick (2019). 1. This section is on direct taxation. GVCs also pose chal- 19. Auerbach et al. (2017). lenges for indirect taxes such as the VAT, although these 20. Clavey et al. (2019). are more tractable (see Clavey et al. 2019). 21. For comparison, the U.S. Congress’s Joint Committee 2. UNCTAD (2015). on Taxation estimated the revenue impact of the Base 3. Crivelli, de Mooij, and Keen (2016). Erosion and Anti-abuse Tax (BEAT) at $149.6 billion over 4. The Inclusive Framework on BEPS, adopted by over 130 2018–27. See U.S. Congress, Joint Committee on Taxation countries, was launched in 2013. For more information, (2017) . see “International Collaboration to End Tax Avoidance,” 22. Pemberton and Loeprick (2019). Organisation for Economic Co-operation and Develop- 23. A third OECD/G20 Inclusive Framework proposal, sig- ment, Paris, http://www.oecd.org/tax/beps/. nificant economic presence, targets highly digitalized busi- 5. See “Global Forum on Transparency and Exchange of nesses and taxes revenues of such firms even if they lack Information for Tax Purposes,” Organisation for Eco- a physical presence. The allocation of profits would be by nomic Co-operation and Development, Paris, https:// formula. www.oecd.org/tax/transparency/. 24. Auerbach et al. (2017). 6. See “Multilateral Convention to Implement Tax 25. Auerbach et al. (2017). Treaty Related Measures to Prevent BEPS,” Organi- 26. Auerbach (2017b). sation for Economic Co-operation and Development, 27. IMF (2019). Paris, https://www.oecd.org/tax/treaties/multilateral 28. Hebous, Klemm, and Stausholm (2019). -convention-to-implement-tax-treaty-related-measures 29. Hebous, Klemm, and Stausholm (2019). -to-prevent-beps.htm. 30. Auerbach (2017a). 7. Beer and Loeprick (2015) suggest that the most relevant 31. This includes highly leveraged firms because debt- indicator of transfer pricing rules adoption are effective financed investments would no longer be subsidized. See documentation requirements. From 1994 to 2014, the Avi-Yonah and Clausing (2019). number of countries with “effective” transfer pricing 32. Grinberg (2017); Schön (2016). documentation rules increased from four to more than 33. Auerbach and Holtz-Eakin (2016); Auerbach et al. (2017). 80 (Cooper et al. 2016). Although that is a substantial 34. PCT (2017). increase, it did not result in comprehensive coverage of 35. Beer and Loeprick (2018). countries. Similarly, as of 2012 only 34 developing coun- 36. Although the focus here is on mandatory regulation, tries had formal transfer pricing rules (de Mooij and Liu similar factors arise in the context of private standard 2018). setting by lead firms in GVCs (including large retailers), 8. The split profit method is particularly useful in the GVC or collaborative efforts by firms and nongovernmental context where comparable market-based pricing is not organizations to set standards for products or production available for benchmarking transfer pricing. It is recog- processes used by firms that participate in the supply nized as one of the methods for transfer pricing in the chain, such as the Global Food Safety Initiative (GFSI) BEPS package, but the OECD guidance did not stipulate and GLOBALG.A.P. how and when practitioners should perform a “value 37. de Bruijn, Kox, and Lejour (2008). chain analysis” to determine whether the profit split 38. Trachtman (2014). method is the most appropriate method to price a related 39. Mattoo (2018). party transaction. Also, during consultations, transfer 40. MGI (2016). pricing experts disagreed over what constitutes a value 41. USITC (2014). chain analysis, and ultimately the 2018 guidance did not 42. Ferracane, Kren, and van der Marel (2018); Ferracane, cover the topic. Lee-Makiyama, and van der Marel (2018). 9. OECD (2015). 43. Ferracane, Kren, and van der Marel (2018); Ferracane, 10. IMF (2019). Lee-Makiyama, and van der Marel (2018). 11. Beer, Klemm, and Matheson (2018). 44. Schrems v. Data Prot. Comm’r, I.E.H.C. 310 (2014) at 12. Quak (2018). para. 73. 13. See Mansour and Rota-Graziosi (2013). 45. Members of the Fortune 500 would need to spend on 14. See PCT (2015). average $16 million each to avoid falling foul of the Euro- 15. In July 2019, France introduced a digital services tax (DST) pean Union’s GDPR, according to estimates reported in of 3 percent on revenue from digital services earned in the Financial Times (Khan 2017). Each company is expected France by large companies (some 30 mostly U.S.-based to hire on average five dedicated privacy employees (such multinational companies). The European Union has con- as data protection officers) and another five employees to sidered but not agreed on a DST. The United Kingdom deal partially with the new rules. Financial services and is planning to introduce a 2 percent DST in April 2020, technology companies face the biggest compliance costs. and other European countries such as Austria, the Czech 46. NASSCOM-DSCI (2013). Republic, Italy, Poland, and Spain are signaling that they 47. The recent United States–Mexico–Canada Agreement may introduce such a tax. (USMCA) follows the example of the CPTPP. 16. OECD (2019). 48. European Commission (2017). 254 | World Development Report 2020 4 9. U.S. Department of Commerce (2016). Campaign,” European Commission, http://ec.europa.eu 50. “List of Countries Which Have Signed, Ratified/Acceded /growth/sectors/electrical-engineering/red-directive to the African Union Convention on Cyber Security /common-charger_en. and Personal Data Protection,” https://au.int/sites 69. Chee (2018). /default/files/treaties/29560-sl-AFRICAN%20UNION%20 70. Petro (2017). CONVENTION%20ON%20CYBER%20SECURITY%20 71. Solomon (2019). AND%20PERSONAL%20DATA%20PROTECTION.pdf. 72. Anderson, Müller, and Sen (2018). 51. Connor (2014). 73. Connor (2016). 52. Issues that may be of concern to firms participating in 74. Morgan Lewis (2019). GVCs or to end consumers—such as the governance of 75. South Africa Competition Tribunal (2019). GVCs and the allocation of total profits associated with 76. Roberts, Vilakazi, and Simbanegavi (2017). the operation of a GVC along the value chain as a whole— 77. Daily Nation (2013). are not matters that can be typically addressed through 78. Licetti, Miralles, and Teh (2019). competition law. 79. See Connor (2012) and Linklaters (2018). 53. Most modern competition law regimes follow the 80. The merged entity was required to divest some opera- effects test. Under that test, a state has jurisdiction over tions in many of the countries concerned—see Holcim anticompetitive conduct when that conduct has pro- and Lafarge (2015). A noteworthy feature of this case was duced significant and foreseeable effects in the relevant the recognition of the value chain nature of activities. jurisdiction, regardless of whether the acts subject to a The U.S. Federal Trade Commission and the Canadian complaint took place within the territory of that state. Competition Bureau cooperated in order to ensure This test allows the reach of the domestic jurisdiction’s remedies would not disrupt cross-border supply chains. competition laws to extend outside the state’s borders. Thus Holcim was required to divest all its operations However, it also excludes jurisdiction over conduct in in Canada but also to sell several U.S. plants deemed foreign countries that does not have significant and fore- to be critical to the Canadian operation as part of an seeable domestic effects, and thus limits the likelihood integrated package to a single buyer. See  Competition for clashes in competition law enforcement between Bureau (2015).  different jurisdictions. Jurisdictions such as Canada, 81. Fox (2015). France, Japan, and the United Kingdom have adopted an 82. Behavior that has an effect only on foreign markets can- effects-based approach to jurisdiction (Zanettin 2002). not be addressed by national competition agencies. 54. Francois and Wooton (2001). 83. Fox (2015); Hoekman and Sabel (2019). 55. Govinda, Khumalo, and Mkhwanazi (2014). 84. Senegal faces an additional contradiction as the Eco- 56. OECD (2016). nomic Community of West African States (ECOWAS) 57. El Espectador (2015). does foresee the application of national competition law 58. “Colombia,” Observatory of Economic Complexity, Macro for conduct that does not affect trade among its member Connections Group, MIT Media Lab, Massachusetts states. Institute of Technology, Cambridge, MA, https://atlas 85. de Soyres, Mulabdic, and Ruta (2019). .media.mit.edu/en/profile/country/col/#Exports. 86. Gain and Alfaro de Morán (2019). 59. The companies involved were Daimler, DAF, Iveco, 87. Gain and Alfaro de Morán (2019). MAN, Scania, and Volvo/Renault. See “Antitrust/Cartel 88. Kunaka, Raballand, and Fitzmaurice (2019). Cases: 39824 Trucks,” European Commission, http:// ec.europa.eu/competition/elojade/isef/case_details.cfm? proc_code=1_39824. References 60. https://www.carteldigest.com/cartel-detail-page.cfm? Anderson, Robert D., Anna Caroline Müller, and Nivedita itemID=44. Sen. 2018. “Competition Policy in WTO Accessions: 61. Japan Fair Trade Commission (2009). Filling in the Blanks in the International Trading Sys- 62. European Commission (2012). tem.” In Trade Multilateralism in the Twenty-First Century: 63. Guyana Competition and Consumer Affairs Commission Building the Upper Floors of the Trading System through WTO (2019). Accessions, edited by Alexei Kireyev and Chiedu Osakwe, 64. Germany Bundeskartellamt (2017); Netherlands Author- 299–319. Cambridge, U.K.: Cambridge University Press. ity for Consumers and Markets (2017). Auerbach, Alan Jeffrey. 2017a. “Demystifying the Destination- 65. European Commission (2010). Based Cash Flow Tax.” NBER Working Paper 23881 66. Simpson Thacher (2019). (September), National Bureau of Economic Research, 67. FTC (2019). Cambridge, MA. 68. In 2009, 10 leading mobile phone manufacturers signed ——— —. 2017b. “Understanding the Destination-Based a memorandum of understanding (MoU) committing Approach to Business Taxation.” VOX CEPR Policy Portal, them to using micro-USB connectors for chargers. As October 26, Centre for Economic Policy Research, Lon- an exception, manufacturers could continue to use their don. https://voxeu.org/article/understanding-destination own connector if they offered an adapter, which allowed -based-approach-business-taxation. Apple to continue using its own connector. The MoU Auerbach, Alan Jeffrey, Michael P. Devereux, Michael Keen, expired in 2012. See “One Mobile Phone Charger for All and John Vella. 2017. “Destination-Based Cash Flow Cooperation beyond trade | 255 Taxation.” Working Paper WP 17/01, Centre for Business Development: Public Sector Governance Series. Wash- Taxation, Oxford University, Oxford, U.K. ington, DC: World Bank. Auerbach, Alan Jeffrey, and Douglas Holtz-Eakin. 2016. “The Crivelli, Ernesto, Ruud Aloysius de Mooij, and Michael Keen. Role of Border Adjustments in International Taxation.” 2016. “Base Erosion, Profit Shifting, and Developing Coun- Research, November 30, American Action Forum, Wash- tries.” FinanzArchiv: Public Finance Analysis 72 (3): 268–301. ington, DC. http://www.americanactionforum.org Daily Nation. 2013. “Omnia, Nyiombo Face Prosecution.” /research/14344/. June 6. https://www.zambiadailynation.com/2013/06/06 Avi-Yonah, Reuven S., and Kimberly Clausing. 2019. “Prob- /omnia-nyiombo-face-prosecution/. lems with Destination-Based Corporate Taxes and the de Bruijn, Roland, Henk Kox, and Arjan Lejour. 2008. “Eco- Ryan Blueprint.” Columbia Journal of Tax Law 8 (2): 229–56. nomic Benefits of an Integrated European Market for Baniya, Suprabha, Nadia Rocha, and Michele Ruta. 2019. Services.” Journal of Policy Modeling 30 (2): 301–19. “Trade Effects of the New Silk Road: A Gravity Analysis.” de Mooij, Ruud Aloysius, and Li Liu. 2018. “At a Cost: The Real Policy Research Working Paper 8694, World Bank, Wash- Effects of Transfer Pricing Regulations.” IMF Working ington, DC. Paper WP/18/69 (March 23), International Monetary Beer, Sebastian, Alexander Klemm, and Thornton Matheson. Fund, Washington, DC. 2018. “Tax Spillovers from U.S. Corporate Income Tax de Soyres, François, Alen Mulabdic, and Michele Ruta. 2019. Reform.” IMF Working Paper WP/18/166 (July), Interna- “Common Transport Infrastructure: A Quantitative tional Monetary Fund, Washington, DC. Model and Estimates from the Belt and Road Initiative.” Beer, Sebastian, and Jan Loeprick. 2015. “Taxing Income in Policy Research Working Paper 8801, World Bank, Wash- the Oil and Gas Sector: Challenges of International and ington, DC. Domestic Profit Shifting.” WU International Taxation Djankov, Simeon, Caroline L. Freund, and Cong S. Pham. Research Paper 2015–18, Vienna University of Economics 2010. “Trading on Time.” Review of Economics and Statistics and Business, Vienna. 92 (1): 166–73. ————. 2018. “The Cost and Benefits of Tax Treaties with El Espectador. 2015.  “Supuesto cartel del azúcar se remon- Investment Hubs: Findings from Sub-Saharan Africa.” taría a 1998.” June 2. https://www.elespectador.com Policy Research Working Paper 8623, World Bank, /noticias/economia/cartel-del-azucar-se-remontaria-1998 Washington, DC. -articulo-564010. Caliendo, Lorenzo, and Fernando Parro. 2015. “Estimates European Commission. 2010. “Antitrust: Commission Fines of the Trade and Welfare Effects of NAFTA.” Review of 11 Air Cargo Carriers €799 Million in Price Fixing Car- Economic Studies 82 (1): 1–44. tel.” Press release, November 9. https://europa.eu/rapid Chee, Foo Yun. 2018. “Amazon’s Use of Merchant Data /press-release_IP-10-1487_en.htm. under EU Microscope.” Reuters, September 19. https:// — —— —. 2012. “Antitrust: Commission Imposes €169 Million www.reuters.com/article/us-eu-amazon-com-antitrust Fine on Freight Forwarders for Operating Four Price /amazons-use-of-merchant-data-under-eu-microscope-id Fixing Cartels.” Press release, March 28. https://europa.eu USKCN1LZ1UV. /rapid/press-release_IP-12-314_en.htm. Clavey, Colin, Jonathan Leigh Pemberton, Jan Loeprick, and — —— —. 2017. “First Annual Review of the EU–U.S. Privacy Marijn Verhoeven. 2019. “International Tax Reform, Shield.” Fact Sheet (October 18), European Commission, Digitization, and Developing Economies.” Unpublished Brussels. working paper, World Bank, Washington, DC. — —— —. 2019. “Antitrust: Commission Fines Car Safety Equip- Competition Bureau. 2015. “Holcim/Lafarge Merger: Com- ment Suppliers €368 Million in Cartel Settlement.” Press petition Bureau Accepts the Sale of All of Holcim’s Oper- release (March 5), European Commission, Brussels. http:// ations in Canada.” News release, May 4, Competition europa.eu/rapid/press-release_IP-19-1512_en.htm. Bureau, Government of Canada, Ottawa. http://www Ferracane, Martina Francesca, Janez Kren, and Erik van der .competitionbureau.gc.ca/eic/site/cb-bc.nsf/eng/03919 Marel. 2018. “Do Data Policy Restrictions Impact the Pro- .html. ductivity Performance of Firms and Industries?” ECIPE Connor, John M. 2012. “Multiple Prosecutions Point to Huge Digital Trade Estimates Working Paper 1, European Cen- Damages from Auto-Parts Cartels.” AAI Working Paper ter for International Political Economy, Brussels. 12-06 (December 11), American Antitrust Institute, Wash- Ferracane, Martina Francesca, Hosuk Lee-Makiyama, and ington, DC. Erik van der Marel. 2018. “Digital Trade Restrictiveness ————. 2014. “Cartel Overcharges.” In The Law and Economics Index.” European Centre for International Political Econ- of Class Actions, edited by James Langenfeld, 249–387. omy, Brussels. Vol. 26 of Research in Law and Economics. Bingley, U.K.: Fox, Eleanor M. 2015. “Antitrust without Borders: From Emerald Publishing. Roots to Codes to Networks.” E15 Initiative Think Piece ————. 2016. “The Private International Cartels (PIC) Data (September), International Centre for Trade and Sus- Set: Guide and Summary Statistics, 1990–July 2016.” tainable Development and World Economic Forum, Revised 2d ed. (August 9), SSRN eLibrary. http://dx.doi Geneva. .org/10.2139/ssrn.2821254. Francois, Joseph Francis, and Ian Wooton. 2001. “Market Cooper, Joel, Randall Fox, Jan Loeprick, and Komal Mohin- Structure, Trade Liberalization, and the GATS.” CEPR Dis- dra. 2016. Transfer Pricing and Developing Economies: A cussion Paper 2669 (January), Centre for Economic Policy Handbook for Policy Makers and Practitioners. Directions in Research, London. 256 | World Development Report 2020 FTC (U.S. Federal Trade Commission). 2019. “FTC’s Bureau of /en/pressreleases/yearly-2009/mar/individual-000056 Competition Launches Task Force to Monitor Technology .html. Markets.” Press release, February 26, Washington, DC. Khan, Mehreen. 2017. “Companies Face High Cost to Meet https://www.ftc.gov/news-events/press-releases/2019/02 New EU Data Protection Rules.” Financial Times, Novem- /ftcs-bureau-competition-launches-task-force-monitor ber 19. https://www.ft.com/content/0d47ffe4-ccb6-11e7 -technology. -b781-794ce08b24dc. Gain, Bill, and Mayra Alfaro de Morán. 2019. “Leave Your Kunaka, C. G. Raballand, and M. Fitzmaurice. 2019. “How Hammocks at Home: How a Customs Union between Trucking Services Have Improved and May Contribute to Guatemala and Honduras Cut Trade Times from 10 Hours Economic Development.” In Industries without Smokestacks: to 15 Minutes.” Trade Post (blog), January 30, World Bank, Industrialization in Africa Reconsidered, edited by Richard S. Washington, DC. https://blogs.worldbank.org/trade/leave Newfarmer, John Page, and Finn Tarp. New York: Oxford -your-hammocks-home-how-customs-union-between University Press. http://fdslive.oup.com/www.oup.com -guatemala-and-honduras-cut-trade-times-10-hours. /academic/pdf/openaccess/9780198821885.pdf. Germany Bundeskartellamt. 2017. “Bundeskartellamt Licetti, M., G. Miralles, and R. Teh. 2019. “Competition- Imposes Fines on Harbour Towage Service Providers.” Related Provision in Preferential Trade Agreements.” In Press release, December 18. https://www.bundeskartel Handbook of Deep Integration Agreements, chap. 8. Washing- lamt.de/SharedDocs/Meldung/EN/Pressemitteilungen ton, DC: World Bank. /2017/18_12_2017_Hafenschlepper.html. Linklaters. 2018. “Pressure Points: Global Cartel Enforcement Govinda, Hariprasad, Junior Khumalo, and Siphamandla in 2018.” Insights (April), Linklaters, London. https://www Mkhwanazi. 2014. “On Measuring the Economic Impact: .linklaters.com/en-us/insights/publications/2018/april Savings to the Consumer Post Cement Cartel Bust.” /global-cartel-enforcement-in-2018. Paper presented at Competition Commission and Tri- Mansour, Mario, and Grégoire Rota-Graziosi. 2013. “Tax bunal’s Eighth Annual Conference on Competition Law, Coordination, Tax Competition, and Revenue Mobi- Economics, and Policy, Johannesburg, September 4–5. lization in the West African Economic and Monetary Grinberg, Itai. 2017. “A Destination-Based Cash Flow Tax Can Union.” IMF Working Paper WP/13/163 (July), Interna- Be Structured to Comply with World Trade Organization tional Monetary Fund, Washington, DC. Rules.” National Tax Journal 70 (4): 803–18. Mattoo, Aaditya. 2018. “Services Globalization in an Age Guyana Competition and Consumer Affairs Commission. of Insecurity: Rethinking Trade Cooperation.” Policy 2019. “CCAC Rules against Shipping Body in Price-Fixing Research Working Paper 8579, World Bank, Washington, Complaint.” Press release, April 12. https://ccac.gov.gy/ccac DC. -rules-against-shipping-body-in-price-fixing-complaint/. Mattoo, Aaditya, and Joshua P. Meltzer. 2018. “International Hebous, Shafik, Alexander D. Klemm, and Saila Stausholm. Data Flows and Privacy: The Conflict and Its Resolution.” 2019. “Revenue Implications of Destination-Based Cash- Journal of International Economic Law 21 (4): 769–89. Flow Taxation.” IMF Working Paper WP/19/7 (January 15), MGI (McKinsey Global Institute). 2016. Digital Globalization: International Monetary Fund, Washington, DC. The New Era of Global Flows. New York: McKinsey. Hoekman, Bernard M., and Charles Sabel. 2019. “Open Pluri­ Morgan Lewis. 2019. “Global Cartel Enforcement Report, lateral Agreements, International Regulatory Coopera- Year-End 2018.” Morgan, Lewis, and Bockius, Washing- tion, and the WTO.” EUI Working Paper RSCAS 2019/10, ton, DC. Robert Schuman Centre for Advanced Studies, European NASSCOM-DSCI (National Association of Software and University Institute, San Domenico di Fiesole, Italy. Services Companies–Data Security Council of India). Holcim and Lafarge. 2015. “Holcim and Lafarge Obtain 2013. “NASSCOM Update on EU Data Protection Regime.” Merger Clearances in the United States and Canada NASSCOM, New Delhi. Paving the Way to Closing Their Merger.” News release, Netherlands Authority for Consumers and Markets. 2017. May 4, Lafarge, Paris; Holcim, Zurich. https://www “Collaboration between Bundeskartellamt and ACM .lafargeholcim.com/sites/lafargeholcim.com/files/atoms Leads to Settlements in Towage Sector.” News, Decem- /files/05042015-press_finance-lafargeholcim_usa ber 18. https://www.acm.nl/en/publications/collaboration _canada_clearance_merger-uk.pdf. -between-bundeskartellamt-and-acm-leads-settlements Hummels, David L., and Georg Schaur. 2012. “Time as a Trade -towage-sector. Barrier.” NBER Working Paper 17758 (January), National OECD (Organisation for Economic Co-operation and Devel- Bureau of Economic Research, Cambridge, MA. opment). 2015. Measuring and Monitoring BEPS, Action 11: — —— —. 2013. “Time as a Trade Barrier.” American Economic 2015 Final Report. Paris: OECD. http://www.oecd.org/ctp Review 103 (7): 2935–59. /measuring-and-monitoring-beps-action-11-2015-final IMF (International Monetary Fund). 2019. “Corporate -report-9789264241343-en.htm. Taxation in the Global Economy.” Policy Paper 19/007 —— — —. 2016. “Annual Report on Competition Policy Develop- (March 10), IMF, Washington, DC. https://www.imf ments in Colombia, 2015.” Report DAF/COMP/AR(2016)11 .org/en/Publications/Policy-Papers/Issues/2019/03/08 (June 13), Competition Committee, Directorate for Finan- /Corporate-Taxation-in-the-Global-Economy-46650. cial and Enterprise Affairs, OECD, Paris. Japan Fair Trade Commission. 2009. “Cease and Desist Order —— — —. 2019. “Programme of Work to Develop a Consensus and Surcharge Payment Order against Air Freight For- Solution to the Tax Challenges Arising from the Digital- warders.” Press release, March 18. https://www.jftc.go.jp isation of the Economy: Inclusive Framework on BEPS.” Cooperation beyond trade | 257 May 31, OECD/G20 Base Erosion and Profit Shifting Simpson Thacher. 2019. “2018 Global Cartel Enforcement Project, OECD, Paris. Report.” January, Simpson Thacher, New York. https:// PCT (Platform for Collaboration on Tax). 2015. “Options for www.stblaw.com/docs/default-source/Publications Low-Income Countries’ Effective and Efficient Use of Tax /globalcartelenforcementreport_january2019.pdf. Incentives for Investment: A Report to the G-20 Develop- Solomon, Daina Beth. 2019. “Exclusive: Mexico Blamed ment Working Group by the IMF, OECD, UN, and World Walmart’s Size, Access to Rivals’ Data in Blocking App Bank.” October 15, International Monetary Fund; Center Deal.” Reuters, June 14. https://www.reuters.com/article for Tax Policy and Administration, Organisation for Eco- /us-walmart-mexico-exclusive/exclusive-mexico-blamed nomic Co-operation and Development; United Nations; -walmarts-size-access-to-rivals-data-in-blocking-app-deal and World Bank, Washington, DC. -idUSKCN1TF2IF. ————. 2017. “Update on Activities of the Platform for Collab- South Africa Competition Tribunal. 2019. “Order and Rea- oration on Tax.” June 27, International Monetary Fund; sons: Banks/FOREX Pre-Trial Matter.” June 12. https:// Center for Tax Policy and Administration, Organisation www.comptrib.co.za/info-library/case-press-releases for Economic Co-operation and Development; United /order-and-reasons-banks-forex-pre-trial-matter. Nations; and World Bank, Washington, DC. Trachtman, Joel P. 2014. “Mutual Recognition of Services Pemberton, Jonathan Leigh, and Jan Loeprick. 2019. “Low Tax Regulation at the WTO.” In WTO Domestic Regulation and Jurisdictions and Preferential Regimes: Policy Gaps in Services Trade: Putting Principles into Practice, edited by Aik Developing Economies.” Policy Research Working Paper Hoe Lim and Bart De Meester, 110–26. New York: Cam- 8778, World Bank, Washington, DC. bridge University Press. Petro, Greg. 2017. “Amazon’s Acquisition of Whole Foods Is UNCTAD (United Nations Conference on Trade and Devel- about Two Things: Data and Product.” Forbes, August 2. opment). 2015. World Investment Report 2015: Reforming https://www.forbes.com/sites/gregpetro/2017/08/02 International Investment Governance. Geneva: UNCTAD. /amazons-acquisition-of-whole-foods-is-about-two-things U.S. Congress, Joint Committee on Taxation. 2017. “Estimated -data-and-product/#12062f86a808. Budget Effects of the Conference Agreement for H.R. 1, the Quak, Evert-jan. 2018. “Tax Coordination and Tax Harmon- ‘Tax Cuts and Jobs Act’: Fiscal Years 2018–2027.” Document isation within the Regional Economic Communities in JCX-67-17 (December 18), Washington, DC. https://www Africa.” K4D Helpdesk Report (May 4), Institute of Devel- .jct.gov/publications.html?func=startdown&id=5053. opment Studies, Brighton, U.K. U.S. Department of Commerce. 2016. “Privacy Shield Roberts, Simon, Thando Selaelo Vilakazi, and Witness Framework.” Notice, Federal Register 81 (148), August 2, Simbanegavi. 2017. “Competition, Regional Integration, International Trade Administration, U.S. Department of and Inclusive Growth in Africa: A Research Agenda.” Commerce, Washington, DC. In Competition Law and Economic Regulation in Southern USITC (United States International Trade Commission). Africa: Addressing Market Power in Southern Africa, edited by 2014. Digital Trade in the U.S. and Global Economies, Part 2. Jonathan Klaaren, Simon Roberts, and Imraan Valodia, Publication 4485, Investigation 332–540 (August). Wash- 263–87. Johannesburg: WITS University Press. ington, DC: USITC. Schön, Wolfgang. 2016. “Destination-Based Income Taxation World Bank and OECD (Organisation for Economic Co- and WTO Law: A Note.” In Practical Problems in European and operation and Development). 2017. A Step Ahead: Competi- International Tax Law: Essays in Honour of Manfred Mössner, tion Policy for Shared Prosperity and Inclusive Growth. Wash- edited by Heike Jochum, Peter Essers, Michael Lang, Nor- ington, DC: World Bank. bert Winkeljohann, and Bertil Wiman, 429–51. Amster- Zanettin, Bruno. 2002. Cooperation between Antitrust Agencies dam: International Bureau of Fiscal Documentation. at the International Level. Oxford, U.K.: Hart Publishing. 258 | World Development Report 2020 Appendix A Databases used in this Report Country-specific and firm-level databases are used in some chapters. China Customs Statistics China Customs Statistics cover the merchandise passing through its customs, including goods from abroad entering customs warehouses, bonded areas, or special economic zones and goods leaving these areas for ship- ment abroad; goods for inward or outward processing or assembling and subsequent reexportation or reimpor- tation; goods on lease for one year or more; goods imported or exported by foreign-invested enterprises; and international aid or donations. Since 2014, duty-free goods have been included in China Customs Statistics. Since 2016, the countertrade of border residents has been included as “other” in Customs Statistics. http://english.customs.gov.cn/Statistics/Statistics?ColumnId=7 Deep Trade Agreements This World Bank database on the content of preferential trade agreements (PTAs) covers 189 countries and maps 52 provisions in 279 PTAs notified to the World Trade Organization and signed between 1958 and 2015. It also includes information on the legal enforceability of each provision. https://datacatalog.worldbank.org/dataset/content-deep-trade-agreements Developing Countries’ Trade and Market Access in the European Union and the United States Compiled by the World Bank, this database builds on existing national data sources for imports and tariffs covering the period 1997–2017: Eurostat’s COMEXT and the European Commission’s TARIC data sources for the European Union and the U.S. Census and U.S. International Trade Commission (USITC) data sources for the United States. https://datacatalog.worldbank.org/dataset/developing-countries%E2%80%99-trade-and-market-access-european -union-and-united-states-introducing Doing Business The World Bank’s Doing Business database provides objective measures of business regulations and their enforcement across 190 economies covering the period 2004–18. Economies are ranked on their ease of doing business from 1 to 190. https://datacatalog.worldbank.org/dataset/doing-business Economic Census (Mexico) The Economic Census includes all establishments producing goods, sellers of goods, and service providers to generate Mexico’s economic indicators at a high level of geographic, sectoral, and thematic detail. The Economic Census is conducted every five years by Mexico’s National Institute of Statistics and Geography (INEGI). Data are for 1993–2013. The data are proprietary, and access requires permission from the issuing agency. https://en.www.inegi.org.mx/programas/ce/2014 Appendix A Databases used in this Report | 259 ENAPROCE (Mexico) The National Survey on Productivity and Competitiveness of Micro, Small and Medium Enterprises (ENAPROCE) 2015 is conducted by Mexico’s National Institute of Statistics and Geography (INEGI). The data are proprietary, and access requires permission from the issuing agency. https://www.inegi.org.mx/programas/enaproce/2015/ Enterprise Surveys (Vietnam) Since 2001, Vietnam’s General Statistics Office (GSO) has conducted an annual census of enterprises operating in Vietnam. It covers all enterprises with more than 10 workers, and firms with fewer than 10 workers are surveyed. The data are proprietary, and access requires permission from the issuing agency. https://www.gso.gov.vn/Default_en.aspx?tabid=491 Enterprise Surveys (World Bank) The World Bank’s Enterprise Surveys offer an expansive array of economic data on 140,000 firms in 141 countries covering the period 2002–18. https://www.enterprisesurveys.org/ Eora The Eora global supply chain database consists of a multiregion input–output (MRIO) table model that pro- vides high-resolution input–output tables with matching environmental and social satellite accounts for 190 countries for the period 1990–2015. Eora26  is a complete global MRIO table, plus environmental satellite accounts, in a harmonized 26-sector classification. Eora is free for academic use at degree-granting academic institutions. All other users must license the data. https://worldmrio.com/ EU-KLEMS This database, a project of the European Union (EU), measures economic growth, productivity, employment creation, capital formation, and technological change at the industry level for all EU member states covering the period 1995–2015. http://www.euklems.net Exiobase Exiobase is a global detailed multiregional environmental extended supply-use table (MR-SUT) and input– output table (MR-IOT) covering 44 countries and five rest-of-world regions over the period 1995–2011 (version 3). The data are proprietary, and access requires a license from the issuing agency. https://www.exiobase.eu/ Exporter Dynamics Database This World Bank database includes indicators on exporter dynamics and concentration for 70 countries based on exporter-level customs data, most commonly covering the period 2005–12 and in some cases up to 2014. https://www.worldbank.org/en/research/brief/exporter-dynamics-database Global Trade Alert Global Trade Alert provides information on more than 10,000 state interventions since November 2008 that are likely to affect foreign commerce. It includes those affecting trade in goods and services, foreign investment, and labor force migration. https://www.globaltradealert.org GSMA Intelligence GSMA publishes mobile operator data, analysis, and forecasts covering the performance of all 1,400-plus operators and 1,200-plus mobile virtual network operators across 4,400 networks, 65 groups, and 237 countries and territories worldwide. Full unrestricted access to all data sets, tools, and research is available by subscription. https://www.gsmaintelligence.com 260 | World Development Report 2020 International Federation of Robotics The International Federation of Robotics provides worldwide market data on robotics surveys, studies, and statistics. Robotics data are based on annual surveys of robot suppliers and currently cover 75 countries (about 90 percent of the industrial robots market). The data are available for purchase. https://ifr.org/ IPUMS USA IPUMS USA collects, preserves, and harmonizes U.S. Census microdata and provides access to this data with enhanced documentation. Data include decennial censuses from 1790 to 2010 and American Community Surveys since 2000. https://usa.ipums.org/usa/ Large and Medium Manufacturing Industry Survey (Ethiopia) This census of large and medium manufacturing industries is conducted by the Ethiopian Central Statistical Agency and includes establishments with at least 10 employees. It covers the period 1996–2017. The data are proprietary, and access requires permission from the issuing agency. http://www.csa.gov.et/survey-report/category/17-large-and-medium-manufacturing-industry-survey Penn World Table The Penn World Table (PWT) contains information on the relative levels of income, output, input, and productivity worldwide. To date, 10 releases are available, differing in their country and period coverage. The most recent, the PWT 9.1 version, covers 182 countries over the period 1950–2017. https://www.rug.nl/ggdc/productivity/pwt/ Pew Research Center The Pew Research Center is a nonpartisan fact tank that informs the public about the issues, attitudes, and trends shaping the world. It conducts public opinion polling, demographic research, media content analysis, and other empirical social science research. The center does not take policy positions. It is a subsidiary of the Pew Charitable Trusts. https://www.pewresearch.org/ Producer and Consumer Support Estimates Agricultural policies address a wide range of issues, including providing sufficient food at reasonable prices for consumers, ensuring food safety, and improving environmental quality. The Organisation for Economic Co-operation and Development (OECD) has developed agricultural support indicators that, despite this diver- sity, express policy measures with numbers in a comparable way across OECD and other countries covering the period 1986–2017. https://www.oecd.org/countries/ukraine/producerandconsumersupportestimatesdatabase.htm Regional Trade Agreements The Regional Trade Agreements (RTAs) database was launched in 2009 as part of the Transparency Mechanism for RTAs of the World Trade Organization (WTO). It was developed and is maintained by the RTA Section of the WTO Trade Policies Review Division. The database is a repository of the legal texts and annexes of all RTAs notified to the WTO, preferential tariff and trade data provided by RTA parties, and other related documents. It currently covers 219 countries/territories over the period 1948–2019. https://rtais.wto.org/UI/PublicMaintainRTAHome.aspx Services Trade Restrictions Database This World Bank database provides comparable information on services trade policy measures for 103 countries, five sectors (telecommunications, finance, transportation, retail, and professional services), and key modes of delivery (modes 1, 3, and 4). The data are based on surveys that were mostly conducted in 2008. https://datacatalog.worldbank.org/dataset/services-trade-restrictions-database Appendix A Databases used in this Report | 261 TiVA The Trade in Value Added (TiVA) database, compiled by the Organisation for Economic Co-operation and Development (OECD), considers the value added by each country in the production of goods and services that are consumed worldwide. The latest (2018) release covers 64 economies over the period 2005–15 for 36 industries at the International Standard Industrial Classification Revision 4 (ISIC Rev. 4) level. https://www.oecd.org/sti/ind/measuring-trade-in-value-added.htm UN Comtrade UN Comtrade is a repository of official international trade statistics and relevant analytical tables. All data are available through the API portal. https://comtrade.un.org UN Trade Statistics This United Nations database covers international merchandise trade statistics, trade in services, and tourism statistics. https://unstats.un.org/unsd/tradekb/default.aspx/ UNCTAD-WB Nontariff Barriers This joint United Nations Conference on Trade and Development–World Bank database includes ad valorem equivalents (AVEs) of nontariff measures (NTMs), defined as the uniform tariffs that have the same trade impacts on the import of a product because of the presence of the NTMs. AVEs are available for 40 import- ing countries, as well as for the European Union and 200 exporting countries at the cross-sectional level. The AVE estimation is based on data in the World Bank’s World Integrated Trade Solution and World Development Indicators databases. https://datacatalog.worldbank.org/dataset/ad-valorem-equivalent-non-tariff-measures U.S. Census The mission of the U.S. Census Bureau is to serve as the United States’ leading provider of quality data about its people and economy. http://census.gov WDI World Development Indicators (WDI) is the primary World Bank collection of development indicators, compiled from officially recognized international sources. It presents the most current and most accurate global development data available and includes national, regional, and global estimates for 216 economies over the period 1960–2018. https://datacatalog.worldbank.org/dataset/world-development-indicators WIOD In the 2013 World Input–Output Database (WIOD), the World Input–Output Tables and underlying data cover 40 countries and a model for the rest of the world for the period 1995–2011. Data for 35 sectors are classified according to the International Standard Industrial Classification Revision 3 (ISIC Rev. 3). The tables adhere to the 1993 version of the System of National Accounts. In the 2016 WIOD database, the World Input–Output Tables and underlying data cover 43 countries and a model for the rest of the world for the period 2000–2014. Data for 56 sectors are classified according to the International Standard Industrial Classification Revision 4 (ISIC Rev. 4). The tables adhere to the 2008 version of the System of National Accounts. This is a collaborative project led by researchers at the University of Groningen. http://www.wiod.org/home WITS The  World Bank’s World Integrated Trade Solution (WITS)  database provides access to international merchandise trade, tariff, and nontariff measure data. https://wits.worldbank.org/about_wits.html 262 | World Development Report 2020 Women, Business, and the Law Women, Business, and the Law is a World Bank Group project collecting unique data on the laws and regulations that restrict women’s economic opportunities. The data offer objective and measurable benchmarks for global progress toward gender equality and cover 187 economies over the period 2009–18. https://datacatalog.worldbank.org/dataset/women-business-and-law World Bank Group–LinkedIn Digital Data for Development, Jobs, Skills, and Migration These data sets cover four categories of metrics: (1) industry employment shifts, (2) talent migration, (3) industry skills needs, and (4) skill penetration. LinkedIn and the World Bank Group plan to refresh the data annually at a minimum. The data sets cover 140 economies and the period 2015–18. https://datacatalog.worldbank.org/dataset/world-bank-group-linkedin-dashboard-dataset World Economic Outlook The World Economic Outlook (WEO) database of the International Monetary Fund (IMF) contains selected macroeconomic data series from the statistical appendix of the  World Economic Outlook report, which presents the IMF staff’s analysis and projections of economic developments at the global level, in major country groups, and in many individual countries. The WEO is released in April and September/October each year, is available for the period 1980 to the present for 194 economies, and also includes forecasts. https://www.imf.org/external/pubs/ft/weo/2019/01/weodata/index.aspx World Telecommunication/ICT Indicators Sponsored by the International Telecommunication Union (ITU), the United Nations specialized agency for information and communication technologies (ICTs), the World Telecommunication/ICT Indicators database contains time series data for the years 1960, 1965, 1970, and annually from 1975 to 2018 for more than 200 econo- mies and 180 telecommunication/ICT statistics. https://www.itu.int/en/ITU-D/Pages/About.aspx Worldscope Worldscope, under the auspices of the University of British Columbia, offers fundamental data on the world’s leading public and private companies, including annual and interim/quarterly data, detailed historical financial statement content, per share data, calculated ratios, and pricing and textual information. It covers over 80,000 companies across more than 120 countries and the period 1980 to the present. The data are proprietary, and access requires permission from the issuing agency. http://resources.library.ubc.ca/page.php?id=2165 WTO I-TIP The World Trade Organization’s Integrated Trade Intelligence Portal (I-TIP) provides a single entry point for information compiled by the WTO on over 25,000 trade policy measures. I-TIP Goods provides comprehensive information on nontariff measures applied by WTO members to the merchandise trade.  I-TIP Services, a joint initiative of the WTO and the World Bank, is a set of linked databases that provide information on members’ commitments under the WTO’s General Agreement on Trade in Services (GATS), services commitments in regional trade agreements, applied measures in services, and services statistics. https://www.wto.org/english/res_e/statis_e/itip_e.htm Appendix A Databases used in this Report | 263 Appendix B Glossary advanced manufacturing and services GVC. A country is part of an advanced manufacturing and services global value chain if it exports a high share of manufacturing and business services and has high backward GVC integration (see box 1.3). agribusiness or agrifood. The business of agricultural production and food processing. backward GVC participation. Importing inputs to produce goods or services that are exported. commodities GVCs. A country is part of a commodities global value chain if it predominantly exports commod- ities produced by agriculture and mining and has a small share of manufacturing exports and limited backward GVC integration (see box 1.3). co-movement. The common movement of two or more entities. deep integration agreement. Trade agreement that not only contains rules on tariffs and nontariff trade restrictions, but also regulates the business environment in other ways. Issues of deep integration include com- petition policy, investor rights, product standards, public procurement, and intellectual property rights. exchange rate elasticity of exports. The percentage increase in exports associated with a 1 percent change in the exchange rate. It is a measure of the responsiveness of exports to changes in currency value. forward GVC participation. Exporting domestically produced inputs to partners for the production of goods or services that they export. global production network. An organizational arrangement comprising interconnected actors coordinated by a global lead firm and producing goods or services across different countries and regions. global value chain (GVC). The series of stages required to produce a good or service that is sold to consumers, with each stage adding value and with at least two stages conducted in different countries. GVC activities (or stages). The activities required to produce a good or service in the context of a global value chain. Spread across several locations, these activities span the conception of the good or service to its end use and include research, design, production, marketing, and distribution. GVC intensification. An increase in the participation of a country, sector, or firm in a global value chain. Appendix B Glossary | 265 GVC participation (or integration). The engagement of a country, sector, or firm in at least one stage of a global value chain. Overall participation may take the form of two broad types: backward or forward participation. innovative activities GVC. A country is part of an innovative activities global value chain if it has high back- ward GVC integration, spends a large share of its GDP on research and development, and receives a large share of GDP from intellectual property (see box 1.3). lead firm. A firm that is the hierarchically dominant actor within a global value chain. limited manufacturing GVC. A country is part of a limited manufacturing global value chain if it exports a limited set of manufacturing products, often alongside commodities exports, and has medium backward GVC integration (see box 1.3). production fragmentation. The distribution of the production process across different countries and regions. relational GVC. A global value chain in which actors are engaged in long-term firm-to-firm relationships rather than anonymous spot market transactions.   sticky or rigid GVC relationship. A business relationship within a global value chain that is not easily changed. For example, it can correspond to a trade flow involving a supplier trading a product that is precisely customized for the buyer and for which the buyer cannot easily find another supplier. trade diversion. The process of diverting trade from a more efficient exporter to a less efficient one by means of a free trade agreement or a customs union. For example, when two countries sign a trade agreement, they could reduce their imports from the rest of the world and source their imports from each other. To the extent that this strategy of import reallocation has been triggered by the trade agreement, it can be considered a trade diversion. 266 | World Development Report 2020 ECO-AUDIT Environmental Benefits Statement The World Bank Group is committed to reducing its environmental foot- print. In support of this commitment, we leverage electronic publishing options and print-on-demand technology, which is located in regional hubs worldwide. Together, these initiatives enable print runs to be lowered and shipping distances decreased, resulting in reduced paper consumption, chemical use, greenhouse gas emissions, and waste. We follow the recommended standards for paper use set by the Green Press Initiative. The majority of our books are printed on Forest Stewardship Council (FSC)–certified paper, with nearly all containing 50–100 percent recycled content. The recycled fiber in our book paper is either unbleached or bleached using totally chlorine-free (TCF), processed chlorine–free (PCF), or enhanced elemental ­ c hlorine–free (EECF) processes. More information about the Bank’s environmental philosophy can be found at http://www.worldbank.org/corporateresponsibility.